.R-12/058F I September 2013 I www.er
United States
Environmental Protection
Agency
atershed modeling to assess the
sensitivity of streamflow, nutrient and
sediment loads to potential climate
change and urban development in
20 U.S. watersheds
-------
EPA/600/R-12/058F
September 2013
Watershed Modeling to Assess the Sensitivity of Streamflow, Nutrient, and
Sediment Loads to Potential Climate Change and Urban Development in
20 U.S. Watersheds
National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC 20460
-------
DISCLAIMER
This document has been reviewed in accordance with U.S. Environmental Protection Agency
policy and approved for publication. Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.
ABSTRACT
Watershed modeling was conducted in 20 large, U.S. watersheds to characterize the sensitivity of
streamflow, nutrient (nitrogen and phosphorus), and sediment loading to a range of plausible
mid-21st century climate change and urban development scenarios. The study also provides an
improved understanding of methodological challenges associated with integrating existing tools
(e.g., climate models, downscaling approaches, and watershed models) and data sets to address
these scientific questions. The study uses a scenario-analysis approach with a consistent set of
watershed models and scenarios applied to multiple locations throughout the nation. Study areas
were selected to represent a range of geographic, hydrologic, and climatic characteristics.
Watershed simulations were conducted using the Soil Water Assessment Tool (SWAT) and
Hydrologic Simulation Program—FORTRAN (HSPF) models. Scenarios of future climate
change were developed based on statistically and dynamically downscaled climate model
simulations representative of the period 2041-2070. Scenarios of urban and residential
development for this same period were developed from the EPA's Integrated Climate and Land
Use Scenarios (ICLUS) project. Future changes in agriculture and human use and management
of water were not evaluated.
Results provide an improved understanding of the complex and context-dependent relationships
between climate change, land-use change, and water resources in different regions of the nation.
As a first-order conclusion, results indicate that in many locations future conditions are likely to
be different from past experience. Results also provide a plausible envelope on the range of
streamflow and water quality responses to mid-21st century climate change and urban
development in different regions of the nation. In addition, in many study areas the simulations
suggest a likely direction of change of streamflow and water quality endpoints. Sensitivity
studies evaluating the implications of different methodological choices help to improve the
scientific foundation for conducting climate change impacts assessments, thus building the
capacity of the water management community to understand and respond to climate change.
This information is useful to inform and guide the development of response strategies for
managing risk.
Preferred Citation:
U.S. EPA (Environmental Protection Agency). (2013) Watershed modeling to assess the sensitivity of streamflow,
nutrient, and sediment loads to potential climate change and urban development in 20 U.S. watersheds. National
Center for Environmental Assessment, Washington, DC; EPA/600/R-12/058F. Available from the National
Technical Information Service, Alexandria, VA, and online at http://www.epa.gov/ncea.
11
-------
TABLE OF CONTENTS
LIST OF APPENDICES iv
LIST OF TABLES vi
LIST OF FIGURES ix
LIST OF ABBREVIATIONS xii
PREFACE xiv
AUTHORS, CONTRIBUTORS AND REVIEWERS xv
ACKNOWLEDGEMENTS xvi
1. EXECUTIVE SUMMARY 1-1
2. INTRODUCTION 2-1
2.1. About This Report 2-2
3. STUDY AREAS 3-1
4. MODELING APPROACH 4-1
4.1. MODEL BACKGROUND 4-2
4.1.1. HSPF 4-2
4.1.2. SWAT 4-4
4.2. MODEL SETUP 4-6
4.2.1. SWAT Setup Process 4-7
4.2.2. HSPF Setup Process 4-8
4.2.3. Watershed Data Sources 4-9
4.2.4. Baseline Meteorology Representation 4-14
4.3. SFMULATION OUTPUT AND ENDPOINTS 4-16
4.4. MODEL CALIBRATION AND VALIDATION 4-18
4.4.1. Hydrology 4-19
4.4.2. Water Quality 4-22
4.4.3. Accuracy of the Watershed Models 4-23
5. CLIMATE CHANGE AND URBAN DEVELOPMENT SCENARIOS 5-1
5.1. SCENARIO-BASED APPROACH 5-1
5.2. CLIMATE CHANGE SCENARIOS 5-2
5.2.1. Future Climate Models, Sources, and Downscaling 5-2
5.2.2. Translation of Climate Model Projections to Watershed Model Weather
Inputs 5-4
5.3. URBAN AND RESIDENTIAL DEVELOPMENT SCENARIOS 5-12
5.3.1. ICLUS Urban and Residential Development Scenarios 5-14
5.3.2. Mapping ICLUS Housing Density Projections to NLCD Land Use
Categories 5-14
6. STREAMFLOW AND WATER QUALITY SENSITIVITY TO DIFFERENT
METHODOLOGICAL CHOICES: ANALYSIS IN THE FIVE PILOT STUDY
AREAS 6-1
6.1. COMPARISON OF WATERSHED MODELS 6-1
in
-------
TABLE OF CONTENTS (continued)
6.1.1. Comparison of Model Calibration and Validation Performance 6-2
6.1.2. Comparison of Simulated Changes Using SWAT and HSPF 6-7
6.1.3. Sensitivity to Increased Atmospheric CO2 6-11
6.2. SENSITIVITY TO DIFFERENT METHODS OF DOWNSCALING
GCM OUTPUT 6-15
6.2.1. Climate Model Energy Inputs and PET Estimates 6-15
6.2.2. "Degraded" NARCCAP Climate Scenarios 6-16
6.2.3. Sensitivity of Flow and Water Quality to Approaches for Downscaling
GCM Projections 6-17
7. REGIONAL SENSITIVITY OF STREAMFLOW AND WATER QUALITY TO
CLIMATE CHANGE AND LAND DEVELOPMENT: RESULTS IN ALL
20 WATERSHEDS 7-1
7.1. SELECTION OF WATERSHED MODEL FOR USE IN ALL STUDY
AREAS 7-2
7.2. SENSITIVITY TO CLIMATE CHANGE SCENARIOS 7-3
7.3. SENSITIVITY TO URBAN AND RESIDENTIAL DEVELOPMENT
SCENARIOS 7-18
7.4. RELATIVE EFFECTS OF CLIMATE CHANGE AND URBAN
DEVELOPMENT SCENARIOS 7-20
7.5. SENSITIVITY TO COMBINED CLIMATE CHANGE AND URBAN
DEVELOPMENT SCENARIOS 7-24
7.6. WATER BALANCE INDICATORS 7-42
7.7. MODELING AS SUMPTIONS AND LIMITATIONS 7-60
7.7.1. Model Calibration 7-62
7.7.2. Watershed Model Selection 7-63
8. SUMMARY AND CONCLUSIONS 8-1
REFERENCES R-l
LIST OF APPENDICES
Appendix A. Model Setup Process A-l
Appendix B. Quality Assurance Project Plan (QAPP) Section 8: Model Calibration B-l
Appendix C. Climate Change and the Frequency and Intensity of Precipitation Events C-l
Appendix D. Model Configuration, Calibration and Validation for the ACF River Basin D-l
Appendix E. Model Configuration, Calibration and Validation for the Arizona
(Salt,Verde, and San Pedro) Basins E-l
Appendix F. Model Configuration, Calibration and Validation for the Susquehanna
River Basin F-l
IV
-------
LIST OF APPENDICES (continued)
Appendix G. Model Configuration, Calibration and Validation for the Minnesota River
Basin G-l
Appendix H. Model Configuration, Calibration and Validation for the Willamette River
Basin H-l
Appendix I. Model Configuration, Calibration and Validation for the Lake
Pontchartrain Drainages 1-1
Appendix J. Model Configuration, Calibration and Validation for the Tar and Neuse
River Basins J-l
Appendix K. Model Configuration, Calibration and Validation for the Nebraska (Loup
and Elkhorn River) Basins K-l
Appendix L. Model Configuration, Calibration and Validation for the Cook Inlet Basin L-l
Appendix M. Model Configuration, Calibration and Validation for the Georgia-Florida
Coastal Basins M-l
Appendix N. Model Configuration, Calibration and Validation for the Illinois River
Basin N-l
Appendix O. Model Configuration, Calibration and Validation for the Lake Erie
Drainages O-l
Appendix P. Model Configuration, Calibration and Validation for the New England
Coastal Basins P-l
Appendix Q. Model Configuration, Calibration and Validation for the Rio Grande
Valley Q-l
Appendix R. Model Configuration, Calibration and Validation for the Sacramento River
Basin R-l
Appendix S. Model Configuration, Calibration and Validation for the Southern
California Coastal Basins S-l
Appendix T. Model Configuration, Calibration and Validation for the South Platte
River Basin T-l
Appendix U. Model Configuration, Calibration and Validation for the Trinity River
Basin U-l
Appendix V. Model Configuration, Calibration and Validation for the Upper Colorado
River Basin V-l
Appendix W. Model Configuration, Calibration and Validation for the Powder and
Tongue River Basins W-l
Appendix X. Scenario Results for the Five Pilot Study Areas X-l
Appendix Y. Scenario Results for the 15 Nonpilot Study Areas Y-l
Appendix Z. Overview of Climate Scenario Monthly Temperature, Precipitation, and
Potential Evapotranspiration Z-l
-------
LIST OF TABLES
3-1. Summary of the 20 study areas 3-4
3-2. Current (2001) land use and land cover in the 20 study areas 3-7
4-1. Regrouping of the NLCD 2001 land-use classes for the HSPF and SWAT models 4-11
4-2. Calculated fraction impervious cover within each developed land class for each
study area based on NLCD 2001 4-12
4-3. Characteristics of NRCS soil hydrologic groups 4-12
4-4. Weather station statistics for the 20 study areas (1971-2000) 4-16
4-5. Summary of streamflow and water quality endpoints 4-18
4-6. Performance targets for hydrologic simulation (magnitude of annual and seasonal
relative mean error) from Donigian (2000) 4-20
4-7. Key hydrology calibration parameters for HSPF 4-22
4-8. Key hydrology calibration parameters for SWAT 4-22
4-9. Summary of SWAT model fit for initial calibration site (20 study areas) 4-27
4-10. Summary of HSPF model fit for initial calibration sites (five pilot study areas) 4-28
5-1. Climate models and source of model data used to develop climate change
scenarios 5-3
5-2. Climate change data available from each source used to develop climate scenarios 5-6
5-3. SWAT weather generator parameters and adjustments applied for scenarios 5-12
5-4. Comparison of PET estimation between different downscaling approaches 5-13
5-5. ICLUS projected changes in developed land area within different imperviousness
classes by 2050 5-16
6-1. Percent error in simulated total streamflow volume for 10-year calibration and
validation periods at initial and downstream calibration gages 6-3
6-2. Nash-Sutcliffe coefficient of model fit efficiency (E) for daily streamflow
predictions, 10-year calibration and validation periods at initial and downstream
calibration gages 6-3
6-3. Statistical comparison of HSPF and SWAT outputs at downstream station for the
five pilot sites across all climate scenarios 6-8
6-4. Effects of omitting simulated auxiliary meteorological time series on Penman-
Monteith reference crop PET estimates for "degraded" climate scenarios 6-17
6-5. Summary of SWAT-simulated total streamflow in the five pilot study areas for
scenarios representing different methods of downscaling 6-18
6-6. Summary of SWAT-simulated streamflow and water quality in the Minnesota
River study area for scenarios representing different methods of downscaling 6-19
6-7. Range of SWAT-projected changes in annual streamflow and pollutant loads for
combined mid-21st century NARCCAP climate change and ICLUS urban and
residential development scenarios 6-22
7-1. Downstream stations within each study area where simulation results are
presented 7-1
7-2. Average annual precipitation (in/yr and percent of baseline) for current conditions
and mid-21st century climate scenarios 7-4
7-3. Average annual temperature (°F and change from baseline) for current conditions
and mid-21st century climate scenarios 7-5
VI
-------
LIST OF TABLES (continued)
7-4. Average annual PET (in/yr and percent of baseline) for current conditions and
mid-21st century climate scenarios 7-6
7-5. Average annual SWAT-simulated actual ET (in/yr and percent of baseline) for
current conditions and mid-21st century climate scenarios 7-7
7-6. Changes in precipitation intensity for NARCCAP mid-21st century climate
scenarios 7-9
7-7. Simulated total streamflow volume (climate scenarios only; percent relative to
current conditions) for selected downstream stations 7-10
7-8. Simulated 7-day low flow (climate scenarios only; percent relative to current
conditions) for selected downstream stations 7-11
7-9. Simulated 100-year peak flow (log-Pearson III; climate scenarios only; percent
relative to current conditions) for selected downstream stations 7-12
7-10. Simulated changes in the number of days to streamflow centroid (climate
scenarios only; relative to current conditions) for selected downstream stations 7-13
7-11. Simulated Richards-Baker flashiness index (climate scenarios only; percent
relative to current conditions) for selected downstream stations 7-14
7-12. Simulated total suspended solids load (climate scenarios only; percent relative to
current conditions) for selected downstream stations 7-15
7-13. Simulated total phosphorus load (climate scenarios only; percent relative to
current conditions) for selected downstream stations 7-16
7-14. Simulated total nitrogen load (climate scenarios only; percent relative to current
conditions) for selected downstream stations 7-17
7-15. Projected mid-21st century impervious cover changes in study areas from ICLUS
for A2 emissions storyline 7-21
7-16. Simulated response to projected 2050 changes in urban and residential
development (percent or days relative to current conditions) for selected
downstream stations 7-22
7-17. Simulated range of responses of mean annual streamflow to mid-21st century
climate and land-use change at the HUC-8 and larger spatial scale 7-25
7-18. Simulated total streamflow volume (climate and land-use change scenarios;
percent relative to current conditions) for selected downstream stations 7-27
7-19. Simulated 7-day low flow (climate and land-use change scenarios; percent
relative to current conditions) for selected downstream stations 7-30
7-20. Simulated 100-year peak flow (log-Pearson III; climate and land-use change
scenarios; percent relative to current conditions) for selected downstream stations 7-33
7-21. Simulated change in the number of days to streamflow centroid (climate and land-
use change scenarios; relative to current conditions) for selected downstream
stations 7-36
7-22. Simulated Richards-Baker flashiness index (climate and land-use change
scenarios; percent relative to current conditions) for selected downstream stations 7-39
7-23. Coefficient of variation of SWAT-simulated changes in streamflow by study area
in response to the six NARCCAP climate change scenarios for selected
downstream stations 7-43
vn
-------
LIST OF TABLES (continued)
7-24. Coefficient of variation of SWAT-simulated changes in streamflow by
NARCCAP climate scenario for selected downstream stations 7-44
7-25. Simulated total suspended solids load (climate and land-use change scenarios;
percent relative to current conditions) for selected downstream stations 7-45
7-26. Simulated total phosphorus load (climate and land-use change scenarios; percent
relative to current conditions) for selected downstream stations 7-48
7-27. Simulated total nitrogen load (climate and land-use change scenarios; percent
relative to current conditions) for selected downstream stations 7-51
7-28. Simulated percent changes in water balance statistics for study areas (NARCCAP
climate with land-use change scenarios; median percent change relative to current
conditions) 7-54
Vlll
-------
LIST OF FIGURES
3-1. Locations of the 20 study areas with HUC 8-digit watershed boundaries 3-2
3-2. Distribution of precipitation and temperature among the study areas 3-3
3-3. Apalachicola-Chattahoochee-Flint basins study area 3-9
3-4. Arizona: Salt and Verde River section of study area 3-10
3-5. Arizona: San Pedro River section of study area 3-11
3-6. Cook Inlet basin study area 3-12
3-7. Georgia-Florida Coastal Plain study area 3-13
3-8. Illinois River basin study area 3-14
3-9. Lake Erie drainages study area 3-15
3-10. Lake Pontchartrain drainage study area 3-16
3-11. Minnesota River basin study area 3-17
3-12. Nebraska: Loup and Elkhorn River basins study area 3-18
3-13. New England Coastal basins study area 3-19
3-14. Powder and Tongue River basins study area 3-20
3-15. Rio Grande Valley study area 3-21
3-16. Sacramento River basin study area 3-22
3-17. Southern California Coastal basins study area 3-23
3-18. South Platte River basin study area 3-24
3-19. Susquehanna River basin study area 3-25
3-20. Tar andNeuse River basins study area 3-26
3-21. Trinity River basin study area 3-27
3-22. Upper Colorado River basin study area 3-28
3-23. Willamette River basin study area 3-29
4-1. Example of weak correlation of rainfall and flow in the Dismal River at Thedford,
NE (USGS 06775900) in the Loup River basin 4-24
6-1. Comparison of model calibration fit to streamflow for the calibration initial site 6-4
6-2. Sensitivity of model fit for total streamflow volume to temporal change 6-5
6-3. Sensitivity of model fit for streamflow to spatial change 6-5
6-4. Comparison of baseline adjusted model fit efficiency for total suspended solids
monthly loads for calibration site (left) and downstream site (right) 6-6
6-5. Comparison of baseline adjusted model fit efficiency for total phosphorus
monthly loads for calibration site (left) and downstream site (right) 6-6
6-6. Comparison of baseline adjusted model fit efficiency for total nitrogen monthly
loads for calibration site (left) and downstream site (right) 6-6
6-7. SWAT and HSPF simulated changes in total streamflow in pilot watersheds
(expressed relative to current conditions) 6-7
6-8. SWAT and HSPF simulated changes in TSS at downstream station in pilot
watersheds (expressed relative to current conditions) 6-9
6-9. SWAT and HSPF simulated changes in total phosphorus load in pilot watersheds
(expressed relative to current conditions) 6-11
6-10. SWAT and HSPF simulated changes in total nitrogen load in pilot watersheds
(expressed relative to current conditions) 6-12
IX
-------
LIST OF FIGURES (continued)
6-11. Differences between SWAT projections of mid-21st century streamflow and water
quality (median across six NARCCAP scenarios) with and without representation
of increased atmospheric CC>2 6-14
6-12. Consistency in SWAT model projections of mean annual streamflow at
downstream stations with downscaled (NARCCAP, BCSD) and GCM projections
oftheGFDLGCM 6-20
6-13. Consistency in SWAT model projections of mean annual streamflow at
downstream stations with downscaled (NARCCAP, BCSD) and GCM projections
oftheCGCMS GCM 6-21
7-1. Ratio of winter (January-March) to summer (July-September) runoff volume
under current and mid-21st century NARCCAP climate scenarios 7-19
7-2. Box plots of the distribution of the ratio of winter (January-March) to summer
(July-September) runoff volume normalized to the ratio under current conditions 7-20
7-3. Comparison of simulated responses of mean annual streamflow to urban
development and climate change scenarios—HSPF model 7-24
7-4. Simulated total future streamflow volume relative to current conditions
(NARCCAP climate scenarios with urban development) for selected stations 7-28
7-5. Median simulated percent changes in total future streamflow volume for six
NARCCAP scenarios relative to current conditions by JTUC-8 (median of
NARCCAP climate scenarios with urban development) 7-29
7-6. Simulated 7-day low flow relative to current conditions (NARCCAP climate
scenarios with urban development) for selected downstream stations 7-31
7-7. Median simulated percent changes in 7-day average low flow volume for six
NARCCAP scenarios relative to current conditions by JTUC-8 (median of
NARCCAP climate scenarios with urban development) 7-32
7-8. Simulated 100-year peak flow relative to current conditions (NARCCAP climate
scenarios with urban development) for selected downstream stations 7-34
7-9. Median simulated percent changes in 100-year peak flow for six NARCCAP
scenarios relative to current conditions by JTUC-8 (median of NARCCAP climate
scenarios with urban development) 7-35
7-10. Simulated change in days to streamflow centroid relative to current conditions
(NARCCAP climate scenarios with urban development) for selected downstream
stations 7-37
7-11. Median simulated change in the number of days to streamflow centroid for six
NARCCAP scenarios relative to current conditions by JTUC-8 (median of
NARCCAP climate scenarios with urban development) 7-38
7-12. Simulated Richards-Baker flashiness index relative to current conditions
(NARCCAP climate scenarios with urban development) for selected downstream
stations 7-40
7-13. Simulated absolute changes in the Richards-Baker flashiness index for six
NARCCAP scenarios relative to current conditions by JTUC-8 (median of
NARCCAP climate scenarios with urban development) 7-41
-------
LIST OF FIGURES (continued)
7-14. Simulated total suspended solids load relative to current conditions (NARCCAP
climate scenarios with urban development) for selected downstream stations 7-46
7-15. Median simulated percent changes in total suspended solids loads for six
NARCCAP scenarios relative to current conditions by HUC-8 (median of
NARCCAP climate scenarios with urban development) for selected downstream
stations 7-47
7-16. Simulated total phosphorus load relative to current conditions (NARCCAP
climate scenarios with urban development) for selected downstream stations 7-49
7-17. Median simulated percent changes in total phosphorus loads for six NARCCAP
scenarios relative to current conditions by HUC-8 (median of NARCCAP climate
scenarios with urban development) 7-50
7-18. Simulated total nitrogen load relative to current conditions (NARCCAP climate
scenarios with urban development) for selected downstream stations 7-52
7-19. Median simulated percent changes in total nitrogen loads for six NARCCAP
scenarios relative to current conditions by HUC-8 (median of NARCCAP climate
scenarios with urban development) 7-53
7-20. Median simulated percent changes in watershed Dryness Ratio for six NARCCAP
scenarios relative to current conditions (median of NARCCAP climate scenarios
with urban development) 7-55
7-21. Median simulated percent changes in watershed Low Flow Sensitivity for six
NARCCAP scenarios relative to current conditions (median of NARCCAP
climate scenarios with urban development) 7-56
7-22. Median simulated percent changes in watershed Surface Runoff Fraction for six
NARCCAP scenarios relative to current conditions (median of NARCCAP
climate scenarios with urban development) 7-57
7-23. Median simulated percent changes in watershed Snowmelt Fraction for six
NARCCAP scenarios relative to current conditions (median of NARCCAP
climate scenarios with urban development) 7-58
7-24. Median simulated percent changes in watershed Deep Recharge for six
NARCCAP scenarios relative to current conditions (median of NARCCAP
climate scenarios with urban development) 7-59
XI
-------
LIST OF ABBREVIATIONS
ACE air, climate, and energy
AET actual evapotranspiration
ANOVA analysis of variance
BASINS Better Assessment Science Integrating Point and Nonpoint Sources
BCSD bias-corrected and statistically downscaled
CAT Climate Assessment Tool
CCSM Community Climate System Model
cfs cubic feet per second
CGCM3 Third Generation Coupled Global Climate Model
CMIP3 Coupled Model Intercomparison Project Phase 3
CN curve number
CRCM Canadian Regional Climate Model
CV coefficient of variation
DEM digital elevation model
E Nash-Sutcliffe coefficient of model fit efficiency
EI' Garrick's baseline adjusted coefficient of model fit efficiency
ET evapotranspiration
FTable hydraulic functional table (in HSPF)
GCM global climate model
GFDL Geophysical Fluid Dynamics Laboratory global climate model
GFDL hi res Geophysical Fluid Dynamics Lab. 50-km global atmospheric time slice model
GIS geographic information system
HadCM3 Hadley Centre Coupled Model, version 3
HRM3 Hadley Region Model 3
HRU hydrologic response unit
HRU Hydrologic response unit
HSG hydrologic soil group
HSPF Hydrologic Simulation Program—FORTRAN
HUC hydrologic unit code
HUC-2 HUC 2-digit watershed
HUC-4 HUC 4-digit watershed
HUC-8 HUC 8-digit watershed
HUC-10 HUC 10-digit watershed
ICLUS Integrated Climate and Land Use Scenarios
EVIPLND impervious land segment (in HSPF)
INFILT nominal infiltration rate parameter (in HSPF)
IPCC Intergovernmental Panel on Climate Change
LZETP lower zone evapotranspiration parameter
LZSN lower soil zone nominal soil moisture storage
MSL mean sea level
MUSLE Modified Universal Soil Loss Equation
NARCCAP North American Regional Climate Change Assessment Program
NARR North American Regional Reanalysis
NCAR National Center for Atmospheric Research
xn
-------
LIST OF ABBREVIATIONS (continued)
ND no data
NHD National Hydrography Dataset
NLCD National Land Cover Dataset
NRCS Natural Resource Conservation Service
PCS Permit Compliance System
PERLND pervious land segment (in HSPF)
PET potential evapotranspiration
PRMS Precipitation Runoff Modeling System
QAPP Quality Assurance Project Plan
RCHRES stream reach segment (in HSPF)
RCM regional climate model
RCM3 Regional Climate Model, version 3
SERGoM Spatially Explicit Regional Growth Model
SPARROW Spatially-Referenced Regression On Watershed attributes
STATSGO State Soil Geographic Database
SWAT Soil Water Assessment Tool
TMDL total maximum daily load
TN total nitrogen
TP total phosphorus
TSS total suspended solids
UCI user control input file (in HSPF)
U.S. EPA U.S. Environmental Protection Agency
USGS U.S. Geological Survey
USLE Universal Soil Loss Equation
WDM watershed data management binary file (for HSPF)
WinHSPF Windows interface to Hydrologic Simulation Program—FORTRAN
WRFP Weather Research and Forecasting Model
WXGEN weather generator (in SWAT)
Xlll
-------
PREFACE
This report was prepared by U.S. Environmental Protection Agency (EPA)'s Air, Climate, and
Energy (ACE) research program, located within the Office of Research and Development. The
ACE research program is designed to address the increasingly complex environmental issues we
face in the 21st century. The overarching vision of ACE is to provide the cutting-edge scientific
information and tools to support EPA's strategic goals of protecting and improving air quality
and taking action on climate change in a sustainable manner.
Climate change presents a risk to the availability and quality of water resources necessary to
support people and the environment. EPA, with Contractor support from Tetra Tech, Inc.,
recently completed a large-scale modeling effort to assess the sensitivity of streamflow and water
quality in different regions of the nation to a range of mid-21st century climate change and urban
development scenarios. This report describes the methods, models, scenarios, and results of this
project.
Responding to climate change is a complex issue. The information in this report is intended to
inform and help build the capacity of EPA and EPA clients to understand and respond to the
challenge of climate change. This final report reflects consideration of peer review and public
comments received on an External Review Draft report released in March, 2013 (EPA/600/R-
12/058A).
xiv
-------
AUTHORS, CONTRIBUTORS AND REVIEWERS
The National Center for Environmental Assessment, Office of Research and Development, was
responsible for preparing this final report. An earlier draft report was prepared by Tetra Tech,
Inc., under EPA Contracts EP-C-05-061 and EP-C-08-004.
AUTHORS
Tetra Tech, Inc.
Jonathan Butcher
Andrew Parker
Saumya Sarkar
Scott Job
Mustafa Faizullabhoy
Peter Cada
Jeremy Wyss
Texas A&M University
Raghavan Srinivasan
Pushpa Tuppad
Deb Debjani
AQUA TERRA Consultants
Anthony Donigian
John Imhoff
Jack Kittle
Brian Bicknell
Paul Hummel
Paul Duda
U.S. Environmental Protection Agency, Office of Research and Development
Thomas Johnson
Chris Weaver
Meredith Warren (ORISE Fellow)
Daniel Nover (AAAS Fellow)
REVIEWERS
This report was much improved by many excellent and thoughtful comments provided by
reviewers Dao Nguyen Khoi, Timothy Randhir, Susanna Tak Yung Tong, and Chong-Yu Xu.
We are also grateful for comments on an earlier draft of this report provided by EPA staff David
Bylsma, Chris Clark, and Steve Klein.
xv
-------
ACKNOWLEDGEMENTS
We acknowledge and thank the entire project team at Tetra Tech, Inc., Texas A&M University,
AQUA TERRA, Stratus Consulting, and FTN Associates for their support contributing to the
development of this report. We also thank Seth McGinnis of the National Center for
Atmospheric Research (NCAR) for processing the North American Regional Climate Change
Assessment Program (NARCCAP) output into change statistics for use in the watershed
modeling. NCAR is supported by the National Science Foundation. We acknowledge the
modeling groups, the Program for Climate Model Diagnosis and Intercomparison and the
WCRP's Working Group on Coupled Modeling for their roles in making available the WCRP
Coupled Model Intercomparison Project Phase 3 (CMIP3) multimodel data set. Support of this
data set is provided by the Office of Science, U.S. Department of Energy.
xvi
-------
1. EXECUTIVE SUMMARY
There is growing concern about the potential effects of climate change on water resources. The
2007 Fourth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC)
states that warming of the climate system is now unequivocal (IPCC, 2007). Regionally variable
changes in the amount and intensity of precipitation have also been observed in much of the
United States (Groisman et al., 2012). Climate modeling experiments suggest these trends will
continue throughout the 21st century, with continued warming accompanied by a general
intensification of the global hydrologic cycle (IPCC, 2007; Karl et al., 2009; Kharin et al., 2013).
Over the same time horizon, human population is expected to continue to increase, with
accompanying changes in land use and increased demand on water resources. In many areas,
climate change is expected to exacerbate current stresses on water resources from population
growth and economic and land-use change, including urbanization (IPCC, 2007). Responding to
this challenge requires an improved understanding of how we are vulnerable and development of
strategies for managing future risk.
This report describes watershed modeling in 20 large, U.S. drainage basins (6,000-27,000 mi2)
to characterize the sensitivity of streamflow, nutrient (nitrogen and phosphorus), and sediment
loading to a range of potential mid-21st century climate futures; to assess the potential interaction
of climate change and urbanization in these basins; and to improve our understanding of
methodological challenges associated with integrating existing tools (e.g., climate models,
downscaling approaches, and watershed models) and data sets to address these scientific
questions.
Study areas were selected to represent a range of geographic, hydroclimatic, physiographic, and
land-use conditions, while also meeting practical criteria such as the availability of data to
calibrate and validate watershed models. Climate change scenarios are based on mid-21st
century climate model projections downscaled with regional climate models (RCMs) from the
North American Regional Climate Change Assessment Program (NARCCAP; Mearns, 2009)
and the bias-corrected and statistically downscaled (BCSD) data set described by Maurer et al.
(2007). Urban and residential development scenarios are based on the U.S. Environmental
Protection Agency (EPA)'s national-scale Integrated Climate and Land Use Scenarios (ICLUS)
project (U.S. EPA, 2009c). Watershed modeling was conducted using the Hydrologic
Simulation Program—FORTRAN (HSPF) and Soil and Water Assessment Tool (SWAT)
watershed models.
Climate change scenarios based on global climate model (GCM) simulations in the NARCCAP
and BCSD data sets project a continued general warming trend throughout the nation over the
next century, although the magnitude of the warming varies from place to place. Wetter winters
and earlier snowmelt are likely in many of the northern and higher elevation watersheds.
Changes in other aspects of local climate, such as the timing and intensity of precipitation, show
greater variability and uncertainty. ICLUS urban and residential development scenarios project
continued growth in urban and developed land over the next century throughout the nation with
most growth occurring in and around existing urban areas. Model simulations of watershed
response to these changes provide a national-scale perspective on the range of potential changes
in streamflow and water quality in different regions of the nation. Simulations evaluating the
1-1
-------
variability in watershed response using different approaches for downscaling climate data and
different watershed models provide guidance on the use of existing models and data sets for
assessing climate change impacts. Key findings are summarized below.
There is a high degree of regional variability in the model simulated responses of different
streamflow and water quality endpoints to a range of potential mid-21st century climatic
conditions throughout the nation. Comparison of watershed simulations in all 20 study areas
for the 2041-2070 time horizon suggests the following hydrologic changes may occur:
• Potential streamflow volume decreases in the Rockies and interior southwest, and
increases in the east and southeast coasts.
• Higher peak streamflow will increase erosion and sediment transport; loads of nitrogen
and phosphorus are also likely to increase in many watersheds.
• Many watersheds are likely to experience significant changes in the timing of streamflow
and pollutant delivery. In particular, there will be a tendency to shift from
snowmelt-dominated spring runoff systems to rain-dominated systems with greater
winter runoff.
• Changes in nutrient and sediment loads are generally correlated with changes in
hydrology.
Changes in watershed water balance and hydrologic processes are likely in many regions of
the nation. Changes in streamflow are determined by the interaction of changes in precipitation
and evapotranspiration (ET). Model simulations in this study suggest that in many regions of the
nation, the fraction of streamflow derived from surface stormflow will increase, while
groundwater-supported baseflow and recharge to deep groundwater aquifers may decrease.
The simulated responses of streamflow and water quality endpoints to climate change
scenarios based on different climate models and downscaling methodologies span a wide
range in many cases and sometimes do not agree in the direction of change. The ultimate
significance of any given simulation of future change will depend on local context, including the
historical range of variability, thresholds and management targets, management options, and
interaction with other stressors. The simulation results in this study do, however, clearly
illustrate that the potential streamflow and water quality response in many areas could be large.
Given these uncertainties, successful climate change adaptation strategies will need to
encompass practices and decisions to reduce vulnerabilities and risk across a range of potential
future climatic conditions.
Simulated responses to increased urban development scenarios are small relative to those
resulting from climate change at the scale of modeling in this study. This is likely due to the
relatively small changes in developed lands as a percent of total watershed area at the large
spatial scale of watersheds in this study. The finest spatial scale reported in this study is that of
an 8-digit hydrologic unit code (HUC), and most urbanized areas are located on larger rivers
1-2
-------
downstream of multiple 8-digit HUCs. Over the whole of individual study areas, urban and
residential growth scenarios represented changes in the amount of developed land on the order of
<1 to about 12% of total watershed area, and increases in impervious surfaces on the order of
0 to 5% of total watershed area. As would be expected, such small changes in development did
not have a large effect on streamflow or water quality at larger spatial scales. It is well
documented, however, that urban and residential development at higher levels can have
significant impacts on streamflow and water quality. At smaller spatial scales where changes in
developed lands represent a larger percentage of watershed area, the effects of urbanization are
likely to be greater. The scale at which urbanization effects may become comparable to the
effects of a changing climate is uncertain.
Simulation results are sensitive to methodological choices such as different approaches for
downscaling global climate change simulations and use of different watershed models.
Watershed simulations in this study suggest that the variability in watershed response resulting
from a single GCM downscaled using different RCMs can be of the same order of magnitude as
the ensemble variability between the different GCMs evaluated. Watershed simulations using
different models with different structures and methods for representing watershed processes
(HSPF and SWAT in this study) also resulted in increased variability of outcomes. SWAT
simulations accounting for the influence of increased atmospheric carbon dioxide (€62) on
evapotranspiration significantly affected results. One notable insight from these results is that, in
many watersheds, increases in precipitation amount and/or intensity, urban development, and
atmospheric CO2 can have similar or additive effects on streamflow and pollutant loading (e.g., a
flashier runoff response with higher high and lower low flows).
Significance and next steps. The model simulations in this study contribute to a growing
understanding of the complex and context-dependent relationships between climate change,
land-use change, and water resources in different regions of the nation. As a first order
conclusion, results indicate that in many locations future conditions are likely to be different
from past experience. In the context of decision making, being aware and planning for this
uncertainty is preferable to accepting a position that later turns out to be incorrect. Results also
provide a plausible envelope on the range of streamflow and water quality responses to mid-21st
century climate change and urban development in different regions of the nation. In addition, in
many study areas the simulations suggest a likely direction of change of streamflow and water
quality endpoints. This information can be useful in planning for anticipated but uncertain future
conditions. Sensitivity studies evaluating the implications of different methodological choices
help to improve the scientific foundation for conducting climate change impacts assessments,
thus building the capacity of the water management community to understand and respond to
climate change.
Understanding and responding to climate change is complex, and this study is only an
incremental step towards fully addressing these questions. It must be stressed that results are
conditional upon the methods, models, and scenarios used in this study. Scenarios represent a
plausible range but are not comprehensive of all possible futures. Several of the study areas are
also complex, highly managed systems; all infrastructure and operational aspects of water
management are not represented in full detail. Successful climate change adaptation strategies
will need to encompass practices and decisions to reduce vulnerabilities across a wide range of
plausible future climatic conditions. It is the ultimate goal of this study to build awareness of the
1-3
-------
potential range of future watershed response so that where simulations suggest large and
potentially disruptive changes, the management community will respond to build climate
resiliency.
1-4
-------
2. INTRODUCTION
It is now generally accepted that human activities including the combustion of fossil fuels and
land-use change have resulted, and will continue to result, in long-term changes in climate
(IPCC, 2007; Karl et al., 2009). The 2007 Fourth Assessment Report of the IPCC states that
"warming of the climate system is unequivocal, as is now evident from observations of increases
in global average air and ocean temperatures, widespread melting of snow and ice and rising
global average sea level" (IPCC, 2007). Regionally variable changes in the amount and intensity
of precipitation have also been observed in much of the United States (Allan and Soden, 2008;
Groisman et al., 2012). Climate modeling experiments suggest these trends will continue
throughout the 21st century, with continued warming accompanied by a general intensification of
the global hydrologic cycle (IPCC, 2007; Karl et al., 2009; Kharin et al., 2013; Emori and
Brown, 2005). While uncertainty remains, particularly for precipitation changes at regional
spatial scales, the presence of long-term trends in the record suggests many parts of the United
States could experience future climatic conditions unprecedented in recent history.
Water managers are faced with important questions concerning the implications of climate
change for water resources. Changes in climate will vary over space and time. The hydrologic
response to climate change will be further influenced by the attributes of specific watersheds,
including physiographic setting, land use, pollutant sources, and human use and management of
water. Runoff is generally expected to increase at higher latitudes and in some wet tropical
areas, and decrease over dry and semiarid regions at mid-latitudes due to decreases in rainfall
and higher rates of evapotranspiration (IPCC, 2007; Karl et al., 2009). Northern and
mountainous areas that receive snow in the winter are likely to see increased precipitation
occurring as rain versus snow. In addition, most regions of the United States are anticipated to
experience increasing intensity of precipitation events; that is, warming-induced intensification
of the global hydrologic cycle will increase the fraction of total precipitation occurring in large
magnitude events. Precipitation changes can result in hydrologic effects that include changes in
the amount and seasonal timing of streamflow, changes in soil moisture and groundwater
recharge, changes in land cover and watershed biogeochemical cycling, changes in nonpoint
pollutant loading to water bodies, and increased demands on water infrastructure, including
urban stormwater and other engineered systems. Regions exposed to increased storm intensity
could experience increased coastal and inland flooding. Such changes challenge the assumption
of stationarity that has been the foundation for water management for decades (e.g., Milly et al.,
2008).
Changes in climate and hydrology will also affect water quality. Although less studied, potential
effects include changes in stream temperature and hydrologic controls on nutrient, sediment, and
dissolved constituent loads to water bodies. Hydrologic changes associated with climate change
could also influence pollutant loading from urban and agricultural lands. Previous studies
illustrate the sensitivity of stream nutrient loads, sediment loads, and ecologically relevant
streamflow characteristics to changes in climate (e.g., see Poff et al., 1996; Williams et al., 1996;
Murdoch et al., 2000; Monteith et al., 2000; Chang et al., 2001; Bouraoui et al., 2002; SWCS,
2003; Marshall and Randhir, 2008; Wilson and Weng, 2011; Tong et al., 2011). A review
(Whitehead et al., 2009) details progress on these questions but emphasizes that still relatively
little is known about the link between climate change and water quality.
2-1
-------
Many watersheds are currently impacted by existing stressors, including land-use change, water
withdrawals, pollutant discharges, and other factors. It is important to recognize that climate
change will not act independently, but will interact in complex and poorly understood ways with
existing and future changes in nonclimatic stressors. One area of concern is the interaction of
climate change and urban development in different watershed settings. Throughout this century,
urban and residential development is expected to increase throughout much of the nation
(U.S. EPA, 2009c). Stormwater runoff from roads, rooftops, and other impervious surfaces in
urban and residential environments is a well-known cause of stream impairment (Walsh et al.,
2005; Paul and Meyer, 2001). Changes in rainfall associated with climate change will have a
direct effect on stormwater runoff (Pyke et al., 2011). More generally, changes in climate could
exacerbate or ameliorate the impacts of other nonclimatic stressors. This understanding is
particularly important because in many situations, the only viable management strategies for
adapting to future climatic conditions involve improved methods for managing and addressing
nonclimatic stressors.
Understanding and adapting to climate change is complicated by the scale, complexity, and
inherent uncertainty of the problem. We currently have a limited ability to predict long-term
(multidecadal) future climate at the local and regional scales needed by decision makers
(Sarewitz et al., 2000). It is therefore not possible to know with certainty the future climatic
conditions to which a particular watershed will be exposed. Scenario analysis using simulation
models is a useful and common approach for assessing vulnerability to plausible but uncertain
future conditions (Lempert et al., 2006; Sarewitz et al., 2000; Volkery and Ribeiro, 2009).
Evaluation of multiple scenarios can provide understanding of the complex interactions
associated with watershed response to climate change and other watershed stressors, and identify
uncertainties associated with changes in different drivers (such as climate and land-use change)
and uncertainties associated with different analytical approaches and methods. This information
is useful for developing an improved understanding of system behavior and sensitivity to a wide
range of plausible future climatic conditions and events, identifying how we are vulnerable to
these changes, and ultimately to guide the development of robust strategies for reducing risk in
the face of changing climatic conditions (Sarewitz et al., 2000; Lempert et al., 2006; Johnson and
Weaver, 2009).
2.1. ABOUT THIS REPORT
This report describes the structure—including methods, models, scenarios, and results—of a
large-scale watershed modeling study designed to address gaps in our knowledge of the
sensitivity of U.S. streamflow, nutrient (nitrogen and phosphorus), and sediment loading to
potential mid-21st century climate change. Modeling also considers the potential interaction of
climate change with future urban and residential development in these watersheds and provides
insights concerning the effects of different methodological choices (e.g., method of downscaling
climate change data, choice of watershed model, etc.) on simulation results.
Watershed modeling was conducted in 20 large U.S. watersheds using a scenario analysis
approach. Study sites were selected to represent a range of geographic, hydrologic, and climatic
characteristics throughout the nation.
Model projections consider the effects of climate change alone, urban and residential
development alone, and the combined effects of climate change and urban development on
_
-------
streamflow, total nitrogen (TN), total phosphorus (TP), and total suspended solids (TSS) loads.
Climate change scenarios were based on downscaled climate model projections from two
sources; the NARCCAP and the BCSD archive from the Bureau of Reclamation/Santa Clara
University/Lawrence Livermore. Scenarios of urban and residential development were based on
projections from EPA's ICLUS project.
All 20 watersheds were modeled with the SWAT model using a consistent set of climate and
land-use change scenarios. In a subset of five study watersheds, referred to as pilot sites,
additional simulations were conducted to address methodological questions related to the
conduct of climate change impacts assessments. In these watersheds, a second watershed model,
the HSPF, was run using the same climate and land-use scenarios used with SWAT to assess the
influence of different watershed models on watershed simulations. Pilot watersheds were also
evaluated for additional climate change scenarios to assess hydroclimatic sensitivity to different
methods of downscaling climate data. All watershed models are constructed at a scale
approximating HUC-lOs, but the finest spatial resolution of model calibration and output was on
the order of HUC-8 watersheds.
As with any study of this type, simulation results are conditional on the specific methods,
models, and scenarios used. Given the difficulty and level of effort involved with modeling at
this scale, it was necessary to standardize model development for efficiency. Several of the
study areas are complex, highly managed systems. We do not attempt to represent all these
operational aspects in full detail. Future changes in agriculture and human use and management
of water were also not evaluated.
This report consists of a main volume and 26 appendices. The main volume describes the study
methods, models, scenarios, and results. The appendices contain additional information on
model setup, calibration, and additional modeling results (at HUC 8-digit spatial scale) not
included in the main report. Supplementary data sets summarizing SWAT simulation results at
all 20 study areas are also available at EPA's ICLUS web page
http://map3.epa.gov/ICLUSonline/.
2-3
-------
3. STUDY AREAS
This project evaluates watershed response to climate change and urban development scenarios in
20 large drainage basins, ranging in size from approximately 6,000 to 27,000 mi2, located
throughout the contiguous United States and Alaska (see Figure 3-1 below). Study areas were
selected based on both geographic and practical considerations. Sites were selected to represent
a broad range of geographic, physiographic, land use, and hydroclimatic settings (see Table 3-1).
Site selection also considered the availability of necessary data for calibration and validation of
watershed models, including a selection of U.S. Geological Survey (USGS) streamflow
monitoring gages (at varying spatial scales) and an adequate set of water quality monitoring data
(e.g., USGS National Water Quality Assessment study areas). Finally, study areas were selected
to leverage, where possible, preexisting calibrated watershed models.
The 20 study areas selected cover a wide range of geology and climate (see Table 3-1), with
elevations ranging from sea level to over 14,000 feet, average annual temperatures from 34 to
68°F, and average annual precipitation ranging from 15 to 66 inches. Figure 3-2 shows the
distribution of average annual precipitation and temperature among the study sites, indicating a
wide range of climatic conditions, from dry to wet and cold to warm. The ratio of winter
(January-March) to summer (July-September) precipitation varies from about 0.1 to 11 while
the fraction of runoff derived from snowmelt ranges from 0 to 54%. The study areas also sample
all of the Level I ecoregions in the contiguous United States (CECWG, 1997), with the exception
of the Tropical Wet Forests ecoregion (present within the contiguous United States only in
southern Florida). Many of the study areas are in the Eastern Temperate Forests ecoregion, but
this region occupies most of the eastern half of the contiguous United States.
The selected study areas also cover a range of land-use conditions, with agricultural land
occupying from 0 to 78% of the land area and urbanized areas (impervious plus developed
pervious land) occupying up to 38%. Overall imperviousness of the study areas (at
approximately the FIUC-4 scale) ranges from near zero to about 14%; however, individual
subwatersheds within a study area have substantially greater imperviousness. For instance,
within the Apalachicola-Chattahoochee-Flint River watersheds (ACF) study area the individual
modeling subbasins (at approximately the FIUC-10 scale) range from 0.15 to 27.44%
impervious.
A detailed summary of current land use and land cover in the 20 study areas is shown in Table
3-2, based on 2001 data from the National Land Cover Dataset (NLCD).
5-1
-------
to
Powder and Tongue
Nebraska: Loup and
Elkhorn River Basin
Lake Erie
Drainages
Minnesota
River Basin
Willamette
River
Basin
New England
Coastal Basins
Upper Colorado
River Basin
Susquehanna
River Basin
Tar and Neuse
River Basins
Southern
California
Coastal
Basins
Atlantic
Ocean
Rio Grande
Valley
Apalachicola-
Chattahoochee-
Flint Basins
Arizona: Trinity
Salt, Verde, Rjver Basjn
and San Pedro
Georgia-Florida
Coastal Plain
Lake
Pontchartrain
Drainage
Gulf of
Mexico
1,200
Kilometers
Study Areas
NAD 1983 Albers meters
Gulf of
Alaska
Figure 3-1. Locations of the 20 study areas with HUC 8-digit watershed boundaries.
-------
70
60
"c"
750 -
0
s
t40 -
CU
£
"ro
| 30 -
cu
I20
10
0 -
3
» LPont
» Willa
* ACF* GaFIa
.. r * TarNeu
* NewEng
* Susq • Trin
* lllin * Sac
LErie
* Cook » Minn
* Neb
SOP';V°wTon *Ariz*S°Cal
RioGra
0 40 50 60 70 80
Average Annual Temperature (F)
Figure 3-2. Distribution of precipitation and temperature among the study
areas.
Note: Precipitation and temperature are averages over the weather stations used in simulation for the modeling period
(approximately 1970-2000, depending on model area).
The USGS (Seaber et al., 1987) has classified watershed drainage areas in a hierarchical system
in which each hydrologic unit is assigned a Hydrologic Unit Code (HUC). The first four levels
of the hierarchy (occupying eight digits) identify the region (HUC-2), subregion (HUC-4), basin
(HUC-6), and subbasin (HUC-8). The United States contains 222 HUC-4s with an average size
of 16,800 mi2. The 20 study areas selected for this study are of a similar scale to HUC-4 basins,
ranging in size from approximately 6,000 to 27,000 mi , but do not correspond exactly with the
boundaries of established HUC-4 basins. Each study area comprises from 7 to 19 HUC 8-digit
watersheds. The individual HUC 8-digit watersheds in the study areas have a median size of
1,164 mi2, and an interquartile range from 805 to 1,808 mi2. In some cases study areas are
composed of a single, contiguous watershed. In other cases, study areas include several adjacent
but noncontiguous watersheds (e.g., separate rivers draining to the coast). Where possible,
watersheds strongly influenced by upstream dams, diversions, or other human interventions were
avoided to simplify modeling.
Maps of the individual study areas are provided in Figures 3-3 through 3-23. Detailed
descriptions of each study area are presented in Appendices D through W, which describe model
development and calibration for the individual study areas.
5-3
-------
Table 3-1. Summary of the 20 study areas
Study area
Apalachicola-
Chattahoochee-Flint Basins
(Pilot Site)
Arizona: Salt, Verde, and
San Pedro (Pilot Site)
Cook Inlet Basin
Georgia-Florida Coastal
Plain
Illinois River Basin
Lake Erie Drainages
Lake Pontchartrain
Drainage
Minnesota River Basin
(Pilot Site)
Site ID
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Location
(states)
GA, AL,
FL
AZ
AK
GA,FL
IL, IN, WI
OH, IN,
MI
LA, MS
MN,IA,
SD
Total
area
(mi2)
19,283
14,910
22,243
17,541
17,004
11,682
5,852
16,989
Elevation
range
(ft MSL)
0-4,347
1,918-11,407
0-18,882
0-485
365-1,183
339-1,383
0-502
683-2,134
Average
precip
(in/yr)
54.26
19.67
28.50
53.21
38.25
38.15
66.33
28.26
Averag
etemp
(°F)
63.43
56.81
34.16
68.24
49.00
49.10
66.64
43.90
Ratio winter
to summer
runoff
2.01
2.06
0.11
1.29
1.24
2.60
1.70
0.50
Fraction of
runoff as
snowmelt
(%)
0.7
9.3
53.8
0.1
13.3
13.4
0.5
14.8
Level I
ecoregions
Eastern Temperate
Forests
Temperate Sierras,
Southern Semi-arid
Highlands, North
America Deserts
Marine West Coast
Forests, Northwest
Forested Mountains
Eastern Temperate
Forests
Eastern Temperate
Forests
Eastern Temperate
Forests
Eastern Temperate
Forests
Great Plains,
Eastern Temperate
Forests
Major cities
Atlanta, GA
Flagstaff, AZ;
Sierra Vista, AZ
Anchorage, AK
Tallahassee, FL;
Tampa, FL
Chicago, IL;
Milwaukee, WI;
Peoria, IL
Fort Wayne, IN;
Cleveland, OH;
Akron, OH
New Orleans,
LA;
Baton Rouge, LA
Mankato, MN,
Minneapolis, MN
-------
Table 3-1. Summary of the 20 study areas (continued)
Study area
Nebraska: Loup and
Elkhom River Basins
New England Coastal
Basins
Powder and Tongue River
Basins
Rio Grande Valley
Sacramento River Basin
Southern California Coastal
Basins
South Platte River Basin
Susquehanna River Basin
(Pilot Site)
Site ID
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
Location
(states)
NE
MA,NH,
ME
MT,WY
NM,CO
CA
CA
CO,WY
PA, NY,
MD
Total
area
(mi2)
22,095
10,359
18,800
18,959
8,316
8,322
14,668
27,504
Elevation
range
(ft MSL)
1,069-4,292
0-5,422
2,201-13,138
4,726-14,173
17-10,424
0-11,488
4,291-14,261
0-3,141
Average
precip
(in/yr)
26.10
48.45
17.70
15.18
37.47
20.21
16.82
41.30
Average
temp
48.35
46.23
44.15
44.71
57.45
61.20
43.46
48.26
Ratio winter
to summer
runoff
0.91
1.41
1.18
0.52
1.61
5.94
0.49
2.06
Fraction of
runoff as
snowmelt (%)
12.6
21.1
30.2
23.8
17.6
4.9
28.3
16.6
Level I ecoregions
Great Plains
Northern Forests,
Eastern Temperate
Forests
Great Plains, North
American Deserts,
Northwestern
Forested Mountains
Northwest Forested
Mountains, North
American Deserts,
Temperate Sierras
Mediterranean
California,
Northwest Forested
Mountains
Mediterranean
California
Great Plains,
Northwest Forested
Mountains
Eastern Temperate
Forests, Northern
Forests
Major cities
No major cities
Portland, ME,
Greater Boston,
MA
No major cities
Santa Fe, NM;
Albuquerque, NM
Chico, CA;
Reading, CA
Greater Los
Angeles, CA
Fort Collins, CO;
Denver, CO
Scranton, PA;
Harrisburg, PA
-------
Table 3-1. Summary of the 20 study areas (continued)
Study area
Tar and Neuse River
Basins
Trinity River Basin
Upper Colorado River
Basin
Willamette River Basin
(Pilot Site)
Site ID
TarNeu
Trin
UppCol
Willa
Location
(states)
NC
TX
CO,UT
OR
Total
area
(mi2)
9,972
17,949
17,865
11,209
Elevation
range
(ft MSL)
0-854
0-2,150
4,323-14,303
8-10,451
Average
precip
(in/yr)
49.91
40.65
16.36
58.38
Average
temp
(°F)
59.91
64.78
41.73
51.19
Ratio winter
to summer
runoff
1.59
1.45
0.31
10.99
Fraction of
runoff as
snowmelt (%)
3.3
1.6
42.4
4.5
Level I ecoregions
Eastern Temperate
Forests
Great Plains,
Eastern Temperate
Forests
Great Plains,
Eastern Temperate
Forests
Marine West Coast
Forests, Northwest
Forested Mountains
Major cities
Raleigh, NC;
Durham, NC;
Greenville, NC
Dallas, TX
Grand Junction,
CO;
Edwards, CO
Portland, OR;
Salem, OR;
Eugene, OR
MSL = mean sea level
Notes: Precipitation and temperature are averages over the weather stations used in simulation for the modeling period (approximately 1970-2000, depending on model area).
The ratio of winter (January-March) to summer (July-September) runoff and the fraction of runoff as snowmelt are derived from the calibrated SWAT model applications
described in this report.
-------
Table 3-2. Current (2001) land use and land cover in the 20 study areas
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Total area
(mi2)
19,283
14,910
22,243
17,541
17,004
11,682
5,852
16,989
22,095
10,359
18,800
18,959
Water
(%)
1.8
0.2
2.55
0.9
1.9
1.1
3.3
3.0
0.8
4.2
0.1
0.3
Barren
(%)
0.4
0.3
18.97
0.4
0.1
0.1
0.4
0.1
0.1
0.5
0.7
1.0
Wetland
(%)
9.3
0.3
7.59
25.7
1.4
2.7
32.3
4.9
3.2
7.6
1.7
2.1
Forest
(%)
47.9
41.9
24.10
33.5
10.3
13.0
23.1
2.9
1.1
63.6
10.0
35.3
Shrub
(%)
9.6
56.0
38.11
10.1
2.1
1.5
14.3
4.6
64.5
2.2
85.5
54.2
Pasture/hay
(%)
9.1
0.1
0.05
7.2
3.6
5.8
10.3
5.9
1.1
4.5
0.6
4.1
Cultivated
(%)
12.4
0.1
0.11
10.9
62.6
61.2
4.5
72.1
26.5
1.1
1.0
0.7
Developed
pervious*
(%)
7.3
1.0
0.58
8.8
11.9
11.2
8.5
5.5
2.4
10.8
0.4
1.7
Impervious
(%)
2.0
0.2
0.24
2.5
6.2
3.5
3.2
1.1
0.4
5.6
0.1
0.5
Snow/ice
(%)
0.0
0.0
7.70
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
-------
Table 3-2. Current (2001) land use and land cover in the 20 study areas (continued)
Study area
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
Total area
(mi2)
8,316
8,322
14,668
27,504
9,972
17,949
17,865
11,209
Water
(%)
0.5
0.6
0.9
1.1
4.5
3.7
0.5
0.9
Barren
(%)
0.5
0.6
1.0
0.4
0.2
0.3
3.8
0.9
Wetland
(%)
2.0
0.4
2.3
1.2
14.1
7.8
1.6
1.8
Forest
(%)
22.4
10.6
23.7
61.1
33.5
16.4
53.9
56.2
Shrub
(%)
48.3
50.9
46.4
1.8
10.0
30.6
33.9
12.3
Pasture/hay
(%)
2.3
1.0
1.5
17.1
7.3
20.6
3.2
12.5
Cultivated
(%)
19.7
2.8
16.5
9.8
21.1
7.0
1.1
8.2
Developed
pervious*
(%)
3.6
19.4
5.0
5.9
7.7
9.4
1.0
4.7
Impervious
(%)
0.7
13.8
2.1
1.5
1.7
4.2
0.4
2.5
Snow/ice
(%)
0.0
0.0
0.7
0.0
0.0
0.0
0.7
0.0
oo
*Developed pervious land includes the pervious portion of open space and low, medium, and high density land uses.
-------
Legend
Hydrography
Water (Nal. Atlas Dataset)
US Census Populated Places
^•1 Municipalities (pop 2 50.000)
County Boundaries
Watershed with HUCSs
AthensCarke
County
ChaUahoochee-
__ Lake Harding
03130005 t ^
-------
Hydrography
Water (Nat. Alias Dataset)
US Census Populated Places
^•1 Municipalities (pop > 50 COO)
| County Boundaries
Watershed withHUCSs
Salt and Verde
River Basins
San Pedro
River Basin
Big
C h i no -W i 11 i a m s o n
Valley
(15060201
Lower Verd
(15060203)
Tonto
(15060105)
Carrizo
(15060104) ^
White
(15060102)
Upper Salt
(15060103)
seven
ervoir
Black
(15060101)
GCRP Model Areas - Salt and Verde River Basins
Base Map
Figure 3-4. Arizona: Salt and Verde River section of study area.
3-10
-------
Salt and Verde
River Basins
San Pedro
River Basin
Lower San Pedro
(15050203)
San Pedro
River
Upper San Pedro
(15050202)
Hydrography
Interstate
Water (Nat. Atlas Dataset)
US Census Populated Places
Municipalities (pop a 50,000)
County Boundaries
Watershed with HUC8s
GCRP Model Areas - San Pedro River Basin
Base Map
Figure 3-5. Arizona: San Pedro River section of study area.
3-11
-------
Hydrography
^H Water (Nat. Atlas Dataset)
US Census Populated Places
^H Municipalities (pop 2 50,000)
County Boundaries
Chulitna
River
(19020502)
] Watershed with HUCSs
Talkeetna
River
(19020503)
Upper Susitna
River
(19020501)
Lower Susitna
River
(19020505)
Upper Kenai
Peninsula
(19020302)
Anchor Point
Gw/fof
Alaska
GCRP Model Areas - Cook Inlet Basin
Base Map
Figure 3-6. Cook Inlet basin study area.
-------
Alapaha
0311*02021
Hydrography
Water (Nat Atlas Dataset)
US Census Populated Places
^H Municipalities (pop > 50,000)
I I County Boundaries
~~] Watershed with HUCSs
Withlacoochee
Ochlockonee
(03120002)
Upper Suwannee
i (03110201)
Jacksonville
Aucilla
(03110103)
Lower
Suwannee
(03-110205)
S (03110206)
r ii— •
PithlacKaotee
(03100207)
Spring H
_,^HH
Tampa
illsboTouqn
0*3100205)
GULF OF
MEXICO
St. Petersburg
GCRP Model Areas - Georgia-Florida Coastal Plain
Base Map
Figure 3-7. Georgia-Florida Coastal Plain study area.
3-13
-------
Hydrography
Water (Nat. Atlas Dalasel)
US Census Populated Places
•I Municipalities (pop > 50,000)
I I County Boundaries
Watershed with HUCSs
Grand Rapids
Michigan
Lake
Michigan
Lower Fox
[071200.07,)
Kankakee
(07J-20'0'01)
Lower Mlmois-
Senachwine
Vermilion
(07130002)
Mackinaw
(07130004)
Bloomington
Lowerlllinpis-
Lake'Chautauaua
(07130003)
GCRP Model Areas - Illnois River Basins
Base Map
Figure 3-8. Illinois River basin study area.
3-14
-------
Legend
Hydrography
Water (Nat. Atlas Dataset)
US Census Populated Places
H Municipalities (pop £ 50.000)
I I County Boundaries
Watershed withHUCSs
Ti,ffin
(04100006)
Black-Roky
,041100,
Lower Maumee
Huron-
Vermilion
(04100012)
St. Joseph
(04100003)
Sandusky
(04100011)
Blanchard
oin oooos)
Auglai
(04100007)
Columbus
GCRP Model Areas - Lake Erie Drainages
Base Map
Figure 3-9. Lake Erie drainages study area.
-------
Hydrography
Water (Nal Atlas Dataset)
US Census Populated Places
•I Municipalities (pop 2 50.000)
I I County Boundaries
Watershed with HUCSs
Mississippi
»——
Louisiana
Bayou Sara-
Thompson
(08070201)
Tangipahoa
08070205)
Amite
(08070202)
Tickpaw
-(08070203)
Liberty Bayou-
Teh efun eta
(08090201)
Baton Rouge.
Lake
Pontchartrain
Lake Maurepas
08070204)
Eastern Louisiana
Coastal
(08090203)
New Orleans
GCRP Model Areas - Lake Pontchartrain Drainage
Base Map
Figure 3-10. Lake Pontchartrain drainage study area.
3-16
-------
Hydrography
Water (Nal. Atlas Dataset)
US Census Populated Places
Municipalities (pop 2 50.000)
County Boundaries
Watershed with HUCSs
„ North Dakota
pper Minnesota
(07020001)
Minneapolis
Hawk^Yellow Medicine
:ower Minnesota
(070200012) {
edwood
(07020006)
Middle Minnesota
(07020007)
Cottonwood
(07020008)
-- Mankato
South Dakota
Le Sueur
(07020011)
_L
Watonwan
(07020010)
Blue Earth
(07020009)
GCRP Model Areas - Minnesota River Basin
Base Map
Figure 3-11. Minnesota River basin study area.
3-17
-------
oo
Hydrography
Water (Nat Alias Dataset)
US Census Populated Places
Municipalities (pop 2 50,000)
County Boundaries
"2 Watershed with HUCSs
Upper North Loup
(10210006)
North ^i
Fork ]l Logan
Elkhorn \\ (1022'0004
0220002
Upper Elkhorn
(10220001)
Upper Middle Loup
' (10210001)
Calamus
(10210008)
Dismal
(10210002)
Lower Elkhorn
(10220003) /
Loup
(10210009)
North Loup
(102100071)
North South L'pup
Platte (10210604)
Lower Middle Loup
(10210003)
—X _M(Jd_
(10210005)
GCRP Model Areas - Nebraska: Loup and Elkhorn River Basins
Base Map
Figure 3-12. Nebraska: Loup and Elkhorn River basins study area.
-------
- Hydrography
•I Water (Nat. Atlas Dataset)
US Census Populated Places
^H Municipalities (pop > 50,000)
| | County Boundaries
Watershed with HUCSs
New Hampshire
W-
Presumpscot
(01060001)
Stfco
(01060002)
Pemigewasset
(01070001)
Winnipesaukee i i
(01070002)-
Pisc'ataqua-Salmon Falls
Merrimack
(01070006)
1
Contoocook
(01070003)
Charles
(01090001)
Lowell
Lynn
Boston
Nashua
(01070004)
Lr
oncord
(01070005)
Taunton
Rhode
Island
GCRP Model Areas - New England Coastal Basins
Base Map
Figure 3-13. New England Coastal basins study area.
3-19
-------
Hydrography
•I Water (Nat Atlas Dataset)
US Census Populated Places
•I Municipalities (pop ? 50.000)
I I County Boundaries
Watershed withHUCSs
Powder
(10090207)
Upper Tongue
(10090101)
Little
Powder
(10090208)
Clear
(10090206)
Crazy
Woman
(10090205)1 Upper Powder
(10090202)
iddle
Fork
Powder
(10090201)
South Fork
Powder
(10090203
Salt
(10090204)
Antelope Hills
GCRP Model Areas - Powder and Tongue River Basins
Base Map
Figure 3-14. Powder and Tongue River basins study area.
3-20
-------
Hydrography
•I Water (Nat. Atlas Dataset)
US Census Populated Places
•I Municipalities (pop > SO COO)
I I County Boundaries
Watershed with HUCSs
Saguache
(13010004)
Rio Grande
Headwaters
13010001)
San Luis
13010003)
lamosa-Trinchera
(13010002)-/
New Mexico
Upper Rio Grande
(13020101)
Rio Chama
(13020102)
Jemez
(13020202)
Albuquerque
Rio Grande
Santa Fe
(13020201)
Cedar Grove
Rio Grande-
Albuquerque
(13020203)
GCRP Model Areas - Rio Grande Valley
Base Map
Figure 3-15. Rio Grande Valley study area.
3-21
-------
Hydrography
Water (Nat Alias Dataset)
US Census Populated Places
•I Municipalities (pop > 50.000)
I County Boundaries
Model Subbasins
Sacramento-LowerCow-
Lower Clear
(18020101)
otto n wood
Headwaters
(18020113)
Upper Cow-Battle
(18020118)
Upper Elder-
Upper Thomes
(18020114)
Mill-Big Chico
(18020119)
Lower Thomes
(18020103)
Upper Stony
(18020115)
^'er Butte
(18020120)s
Lower Butte
^('18020105)
Sacramento-
Stone Corral
Citrus Heights
Folsom
Sacramento
GCRP Model Areas - Sacramento River Basin
Base Map
Figure 3-16. Sacramento River basin study area.
3-22
-------
to
Legend
- Hydrography
Water (Nat. Atlas Dasascti
US Census Populated Places
^H Municipalities (pop 2 90,000)
I I County Boundaries
f~~| Watershed with HUC8s
Santa Clarita
Lancaster
Palmdale
Santa Clara
(18070102)
Ventura
18070101)
Los Angeles
,8070105)
I
Thousand Oaks
Los Angeles
San Gabriel
(18070106)
San Bernardino
Santa" rVlonica.Bav
(18070104)
Santa Monica
Santa Ana
(18070203)
Glendale
Pasadena
Redlands
-7---.
Riverside
Seal Beat
(180702011)
San J a cm to
(18070202)
Newport'Bay
(18070204)
iso-San Onofre
Mission Viejo
Santa Margarita
(18070302)
GCRP Model Areas - Southern California Coastal Basins
Base Map
Figure 3-17. Southern California Coastal basins study area.
-------
Lone Tree, _
(10190008)
Crow
(10190009)
Cache
La Poudre
(10190007)
Hydrography
Water (Nat. Atlas Dataset)
US Census Populated Places
Municipalities (pop > 50,000)
County Boundaries
Watershed with HLJCSs
Big Thompson
(10190006)
Middle South Platte-
Cherry Creek
(10190003)
St. Vrain
(10190005)-»p
Boulder
Denver
Clear
(10190004)
BIJOU
(10190011)
Upper South
' Platte/
South Platte
Headwater
(10190001)
Kiowa
(10190010)
Colorado
Springs
0 15 30
GCRP Model Areas - South Platte River Basin
Base Map
Figure 3-18. South Platte River basin study area.
3-24
-------
Lake Ontario
Upper Susquehanna
(2050101 j '
Chenango
(2050102)
ego-Wappasenin
(2050103)
Chemung
(2050105)
Tioga
••I (2050104)
Upper Susquehanna-Tunkhanno
PJne Pennsylvania
(2050205)
Middle West Branch
Susquehanna
(2050203)
UppeJ'
Susquehanna-lScka wanna
(ZOSftttfr)
Sinnemahpning
(205p*20'2)
Lower West Brancru
usquehanna
2050206)
0}
r f Bald Eagle
K-
-------
to
Legend
- Hydrography
Water (Nat. Atlas Oataset)
US Census Populated Places
^H Municipalities (pop £ 50,000)
[ | County Boundaries
H Watershed with HUCSs
Rocky Mount
North Carolina '
Fishing
(03020102)
Upper Tar
(03020101)
Lower Tar
(03020103)
Pamlico
(03020104)
ontentnea
(03020203)
Upper Neuse
(03020201)
Middle Neuse
(03020202)
r Neuse
3020204)
Jacksonville
GCRP Model Areas - Tar and Neuse River Basins
Base Map
Figure 3-20. Tar and Neuse River basins study area.
-------
Upper West
Fork Trinity
{120301'01)
Longview
Lower-Trinity
Tehuacana
(12030201:).
Hydrography
^H Water (Nat. Atlas Dataset)
US Census Populated Places
H Municipalities (pop 5 50.000)
I | County Boundaries
~1 Watershed with HUCSs
GCRP Model Areas - Trinity River Basin
Base Map
Figure 3-21. Trinity River basin study area.
3-27
-------
to
oo
- Hydrography
Water (Nat. Atlas Datasot)
US Census Populated Places
HI Municipalities (pop £ 50,000)
I I County Boundaries
r"~| Watershed with HUCBs
Colorado Headwaters
(14010001)
Parachute-
Roan
(14010006)
Eagle
(14010003)
Blue
(14010002
Colorado Headwaters
1 Plateau
14010005)
Highlands
Ranch
Roaring Fork
(14010004)
Grand
Junction
Lower Gunnison
(14020005)
East-
Taylor
(14020001)
Upper Gunnison
(14020002)
Uncompahgre
1114020006)
Tomichl—
(14020003)
GCRP Model Areas - Upper Colorado River Basin
Base Map
Figure 3-22. Upper Colorado River basin study area.
-------
Hydrography
Water (Nal. Atlas Dataset)
US Census Populated Places
^H Municipalities (pop > 50.000)
County Boundaries
Watershed with HUCSs
Tualatin
(17090010)
S
Yamhill
(17090008)
Molalla-Puddmg
(17090009)
Clackatnas
(17090011)
-Upper Willamette
(17090003)
South Santiam
(17090006)
North Santiam
17090005)
Mckenzie
(17090004)
Washington
Middle Fork
Willamette
(17090001)
Coastal Fo
Willamette
(17090002)
GCRP Model Areas - Willamette River Basin
Base Map
Figure 3-23. Willamette River basin study area.
3-29
-------
4. MODELING APPROACH
This study uses dynamic watershed models to simulate the watershed response to potential
mid-21st century climate change scenarios, urban and residential development scenarios, and
combined climate change and urban development scenarios. Watershed models were developed
for 20 large-scale study areas (approximately HUC-4 scale) located throughout the contiguous
United States and Alaska. The study also evaluates the sensitivity of modeling results to
different methodological choices for assessing climate change impacts, such as the use of climate
change scenarios based on different methods of downscaling GCM projections and the use of
different watershed models.
A watershed model is a useful tool for providing a quantitative linkage between external forcing
and in-stream response. It is essentially a series of algorithms applied to watershed
characteristics and meteorological data to simulate naturally occurring, land-based processes
over an extended period, including hydrology and pollutant transport. Many watershed models
are also capable of simulating in-stream processes. After a model has been set up and calibrated
for a watershed, it can be used to quantify the existing loading of pollutants from subbasins or
from different land-use categories and can also be used to assess the effects of a variety of
management scenarios.
Five of the 20 sites were selected as "pilot" sites: the Minnesota River watershed (Minn), the
Apalachicola-Chattahoochee-Flint River watersheds (ACF), the Willamette River watershed
(Willa), the Salt/Verde/San Pedro River watershed (Ariz), and the Susquehanna River watershed
(Susq). Pilot sites were selected in part due to previous experiences of the study team in
applying watershed models in these areas, and in part because they provide a representative cross
section of the full set of 20 study areas from a regional, meteorological, geographic, and land-use
perspective. Pilot sites were used for testing and comparing model development and application
methods, as well as for evaluating the sensitivity of modeling results to different types of climate
change scenarios and use of different watershed models. Analysis of the pilot site results led to
the selection of a reduced, more streamlined approach for the remaining 15 sites using one
watershed model and a reduced set of climate change scenarios.
Two watershed models were selected for initial application to the five pilot study sites: HSPF
(Bicknell et al., 2001, 2005) and SWAT (Neitsch et al., 2005). Each of these models has been
widely used for hydrologic and water quality applications for regulatory purposes, such as the
development of pollutant load allocations under the Total Maximum Daily Load (TMDL)
provisions of the Clean Water Act. Both models are also in the public domain with open-source
code, enabling ready replication of results. They both provide dynamic simulation with a
subdaily or daily time step and can be built from readily available spatial coverages, but are
sufficiently efficient to allow implementation of multiple runs for model calibration or scenario
application purposes. Both models have also been used in previous studies of watershed
responses to climate change (e.g., Taner et al., 2011; and Tong et al., 2011 for HSPF; Luo et al.,
2013; Wilson and Weng, 2011; Marshall and Randhir, 2008; and Ficklin et al., 2009 for SWAT).
Application of both HSPF and SWAT to the five pilot watersheds allowed assessment of the
variability associated with use of different watershed models in simulating watershed response to
4-1
-------
climate change. The two model frameworks exhibited similar skill in reproducing observations
at the large spatial scales addressed in this project (see Section 4.4.3); however, SWAT is based
on a plant growth model that can explicitly represent the impacts of altered temperature,
moisture, and CC>2 regimes on plants and the resulting impacts on the water balance and pollutant
transport. The analysis of the pilot site results (see Section 6) emphasized the potential
importance of these processes. Therefore, the SWAT model was applied in all 20 study areas.
HSPF and SWAT are each described in more detail below
4.1. MODEL BACKGROUND
4.1.1. HSPF
The HSPF (Bicknell et al., 2001, 2005) is a comprehensive, dynamic watershed and receiving
water quality modeling framework that was originally developed in the mid-1970s. During the
past several decades, it has been used to develop hundreds of EPA-approved TMDLs, and it is
generally considered among the most advanced hydrologic and watershed loading models
available. The hydrologic portion of HSPF is based on the Stanford Watershed Model
(Crawford and Linsley, 1966), which was one of the pioneering watershed models developed in
the 1960s. The HSPF framework is developed modularly with many different components that
can be assembled in different ways, depending on the objectives of a project. The model
includes three major modules:
• PERLND for simulating watershed processes on pervious land areas
• IMPLND for simulating processes on impervious land areas
• RCHRES for simulating processes in streams and vertically mixed lakes
All three of these modules include many subroutines that calculate the various hydrologic and
water quality processes in the watershed. Many options are available for both simplified and
complex process formulations.
HSPF models hydrology as a water balance in multiple surface and subsurface layers and is
typically implemented in large watersheds at an hourly time step. The water balance is simulated
based on Philip's infiltration (Bicknell et al., 2001, 2005) coupled with multiple surface and
subsurface stores (interception storage, surface storage, upper zone soil storage, lower zone soil
storage, active groundwater, and inactive [deep] groundwater). Potential evapotranspiration
(PET) is externally specified to the model.
As implemented in HSPF, the infiltration algorithms represent both the continuous variation of
infiltration rate with time as a function of soil moisture and the areal variation of infiltration over
the land segment. The infiltration capacity, the maximum rate at which soil will accept
infiltration, is a function of both the fixed and variable characteristics of the watershed. Fixed
characteristics include soil permeability and land slopes, while variables are soil surface
conditions and soil moisture content. A linear probability function is used to account for spatial
variation (Bicknell et al., 2005). The primary parameters controlling infiltration are INFILT, an
index to mean soil infiltration rate (in/hr) and LZSN, the lower soil zone nominal soil moisture
4-2
-------
storage. Specifically, the mean infiltration capacity over a land segment at any point in time,
IBAR, is calculated as
IBAR =
INFILT
(LZS/
VNFEXP
/LZSN'
• INFFAC
4-1
where LZS is the current lower soil zone storage, INFEXP is an exponent typically set to a value
of 2, and INFFAC is an adjustment factor to account for frozen ground effects.
Neither INFILT nor LZSN is directly observable or provided in soils databases and both must be
refined in calibration. As INFILT is not a maximum rate nor an infiltration capacity term, its
values are normally much less than published infiltration rates, soil percolation test results, or
permeability rates from the literature (U.S. EPA, 2000).
Sediment erosion in HSPF uses a method that is formally similar to, but distinct from, the
universal soil loss equation (USLE) sediment-detachment approach coupled with transport
capacity based on overland flow. Nutrients may be simulated at varying levels of complexity,
but are most typically represented by either buildup/washoff or sediment potency approaches on
the land surface coupled with user-specified monthly concentrations in interflow and
groundwater.
Spatially, the watershed is divided into a series of subbasins representing the drainage areas that
contribute to each of the stream reaches. The stream network (RCHRES) links the surface runoff
and groundwater flow contributions from each of the land segments and subbasins and routes
them through water bodies. The stream model includes precipitation and evaporation from the
water surfaces as well as streamflow contributions from the watershed, tributaries, and upstream
stream reaches. It also simulates a full range of stream sediment and nutrient processes,
including detailed representations of scour, deposition, and algal growth.
The version of HSPF used in this study is the Windows interface to Hydrologic Simulation
Program—FORTRAN (WinHSPF) as distributed with BASINS version 4.0. WinHSPF is a
Windows interface to HSPF and is a component of the EPA's Better Assessment Science
Integrating point and Nonpoint Sources (BASINS) Version 4.0 (U.S. EPA, 2001, 2009a, 2009c).
WinHSPF itself is a user interface to HSPF that assists the user in building User Control Input
(UCI) files (containing model input parameters) from geographic information system (GIS) data
(Duda et al., 2001). After the UCI file is built, WinHSPF is used to view, understand, and
modify the model representation of a watershed. HSPF can be run from within WinHSPF. The
actual model executable engine distributed with BASINS is called WinHSPFLt, which can be
run in batch mode independent of the BASINS/WinHSPF interface. The model code for HSPF
is stable and well documented. Detailed descriptions of the model theory and user control input
are provided in Bicknell et al. (2001, 2005).
WinHSPF also provides access to the Climate Assessment Tool (CAT), which is a component of
BASINS 4.0. BASINS CAT facilitates watershed-based assessments of the potential effects of
_
-------
climate variability and change on water and watershed systems (namely streamflow and pollutant
loads) using the HSPF model (U.S. EPA, 2009a, 2009b). BASINS CAT is capable of creating
climate change scenarios that allow users to assess a wide range of what //questions related to
climate change.
4.1.2. SWAT
The SWAT model was developed by the U.S. Department of Agriculture to simulate the effect of
land management practices on water, sediment, and agricultural chemical yields in large,
complex watersheds with varying soils, land use, and management conditions over long periods
of time (Neitsch et al., 2005). SWAT requires data inputs for weather, soils, topography,
vegetation, and land use to model water and sediment movement, nutrient cycling, and numerous
other watershed processes. SWAT is a continuous model appropriate for long-term simulations.
SWAT, as implemented in this study, employs a curve number approach (SCS, 1972) to estimate
surface runoff and then completes the water balance through simulation of subsurface flows,
evapotranspiration, soil storages, and deep seepage losses. The curve number approach requires
a daily time step. PET is typically calculated internally by SWAT based on other weather inputs.
SWAT provides an option for subdaily Green-Ampt infiltration, but this is infrequently used.
The curve number approach is popular because parameters are simple and readily available. The
curve number approach estimates the depth of daily runoff (Q) from rainfall depth (P), initial
abstractions (la, depth), a storage parameter (S, depth), and a curve number (CN), as (SCS,
1972):
(P-fo)2
P-Ia + S 4_2
la is typically assumed to be 20% of S (indeed, this is a hard-coded default in SWAT). In units
of millimeters for S, this yields:
.
(P + 0.8S) CN
4_3
The curve number is estimated as a function of land use, cover, condition, hydrologic soil group
(HSG), and antecedent soil moisture. SWAT provides capabilities to automatically adjust the
CN based on soil moisture, plant evapotranspiration, slope, and the presence of frozen ground.
The conceptual simplicity of the curve number approach also introduces some potential
problems. Specifically, the curve number was developed as a design methodology to estimate
average runoff volume of a specific return period, given average total event rainfall of the same
return period. It was not designed to predict runoff from specific individual events or runoff
from more frequent smaller events, and applicability to continuous simulation is inexact,
4-4
-------
especially at small spatial scales. For a summary of these issues and their potential implications
in continuous simulation modeling, see Garen and Moore (2005).
Sediment yield and erosion are calculated by SWAT using the Modified Universal Soil Loss
Equation (MUSLE; Williams, 1975). The MUSLE is based on several factors, including surface
runoff volume, peak runoff rate, area of hydrologic response unit (HRU), soil erodibility, land
cover and management, support practice, topography, and a coarse fragment factor. MUSLE
implicitly combines the processes of sediment detachment and delivery. Nutrient load
generation and movement are simulated using overland runoff and subsurface flow.
A key feature of SWAT is the incorporation of an explicit plant growth model, including plant
interactions with water and nutrient stores. The transformation of various nitrogen and
phosphorus species is simulated in detail in the soil; however, concentrations of nutrients in
groundwater discharges are user specified, as in HSPF.
In-stream simulation of sediment in SWAT 2005 includes a highly simplified representation of
scour and deposition processes. Nutrient kinetics in receiving waters are based on the numeric
representation used in the QUAL2E model but implemented only at a daily time step.
SWAT is generally considered to be an effective tool for watershed simulation that is especially
appropriate for estimating streamflow and cumulative pollutant loads in agricultural and rural
watersheds (see review by Gassman et al., 2007). Bosch et al. (2011) found that SWAT was an
effective tool for estimating hydrology, sediment, and nutrient loads in Lake Erie watersheds, but
performed less well in urbanized settings. SWAT has some potential weaknesses relative to
HSPF for the simulation of urban lands because it is typically run using a curve number approach
at a daily time step while HSPF is typically run at an hourly time step using Philip infiltration.
The daily time step is insufficient to resolve details of urban runoff hydrographs that have
important implications for stability of small stream channels, while the curve number approach
can result in poor resolution of surface versus subsurface flow pathways (Garen and Moore,
2005). The impacts of these differences are, however, believed to be minor at the larger spatial
scales addressed in this study.
An important component of the SWAT model is the weather generator (WXGEN). SWAT
requires daily values of precipitation, maximum and minimum temperature, solar radiation,
relative humidity, and wind speed. The user may read these inputs from a file or generate the
values using SWAT's weather generator model based on monthly average data summarized over
a number of years (Neitsch et al., 2005). The weather generator model (Sharpley and Williams,
1990) can be used to generate climatic data or to fill in gaps in weather data. The weather
generator first independently generates precipitation for the day. Maximum temperature,
minimum temperature, solar radiation, and relative humidity are then generated based on the
presence or absence of rain for the day. Finally, wind speed is generated independently.
The version of SWAT used in this study is SWAT 2005 as distributed with ArcSWAT 2.1,
which was the most recent stable version of SWAT available at the start of this study.
ArcSWAT 2.1 is an ArcGIS-ArcView extension and a graphical user input interface for the
SWAT watershed model (TAMU, 2010). As with HSPF, the underlying executable code can be
run in batch mode independent of the user interface. Unlike HSPF, the SWAT code is
4-5
-------
continuously evolving, with frequent enhancements and bug fixes. For a detailed description of
the version of SWAT used here, see Neitsch et al. (2005).
4.2. MODEL SETUP
Watershed models were configured to simulate each study area as a series of hydrologically
connected subbasins. Each study area was subdivided into subbasin-scale modeling units.
Continuous simulations of streamflow, total nitrogen, total phosphorus, and total suspended
solids were then made for each unit using meteorological, land use, soil, and stream data.
Many study areas are highly managed systems influenced by humans, including dams, water
transfers and withdrawals, point source discharges and other factors. Given the difficulty of
modeling at the large spatial scale in this study, detailed representation of all management was
not possible. The following assumptions were made to simplify modeling among all 20 study
areas:
External boundary conditions (where needed), such as upstream inflows and pollutant
loads, are assumed constant.
Interactions with deep groundwater systems are assumed constant.
Large-scale shifts in natural cover type in response to climate change are not simulated.
Point source discharges and water withdrawals are assumed constant at current levels.
Only large dams that have a significant impact on hydrology at the HUC-8 (subbasin)
scale are included in the models. Where these dams are simulated, an approximation of
current operating rules (using a target storage approach) is assumed to apply in all future
scenarios.
Human adaptation response to climate change, such as shifts in water use or cropping
practices, are not simulated.
The project team consisted of multiple modelers working in different locations. To ensure
consistency of results, a common set of procedures and assumptions was established (e.g., see
Appendix A). Both HSPF and SWAT were implemented using a HRU approach to upland
simulation. An HRU consists of a unique combination of land use/land cover, soil, and land
management practice characteristics, and thus represents areas of similar hydrologic response.
Individual land parcels included within an HRU are expected to possess similar hydrologic and
load generating characteristics and can thus be simulated as a unit. The HRU approach is the
default for SWAT but is also good practice with HSPF. Consistent with the broad spatial scale
of the models, the land cover component is aggregated into a relatively small number of
categories (e.g., forest, wetland, range, grass/pastureland, crop, developed pervious, low-density
impervious, and high-density impervious).
4-6
-------
Initial preparation of spatial data was done primarily in ArcGIS for the entire study area.
Processed GIS inputs were then used in ArcSWAT (which runs as an extension in ArcGIS), and
imported into B ASINS4 (which uses Map Window GIS) to complete the setup of SWAT and
HSPF, respectively. Spatial data sources are discussed in more detail in Section 4.2.3.
Additional initial setup tasks included identification of weather stations, streamflow gaging and
water quality monitoring locations, and major watershed features that significantly affect the
water balance, such as presence of major lakes, reservoirs, and diversions.
4.2.1. SWAT Setup Process
SWAT model setup used the ArcSWAT extension in ArcGIS. The general procedure for SWAT
setup is described below; a more detailed modeling protocol used for this project is included in
Appendix A.
Subbasin boundaries and reach hydrography for each study area were generally defined from
NHDPlus catchments (U.S. EPA, 2010) aggregated to approximately the HUC-10 spatial scale.
The subbasin and reach shapefiles were imported into the SWAT interface and subbasin
parameters were calculated automatically.
Study area boundaries were configured to minimize the presence of large reservoirs due to the
difficulty of representing operational rules. Models included only major reservoirs that have a
significant effect on streamflow at the scale of HUC-8s or greater. Inclusion of reservoirs was
left to the discretion of individual modelers; however, the reservoirs included are generally those
that drain an area greater than a single HUC-8 and provide a retention time of half a year or
greater. If a reservoir was located at the terminus of the model area, it was generally ignored so
that the model represented input to, rather than output from, the terminal reservoir. Models
include point source discharges from major permitted facilities (greater than 1 million gallons
per day [MGD] discharge). It was also necessary to define an upstream boundary condition
"point source" for study areas where the model did not extend to the headwaters (e.g.,
Sacramento River basin).
HRUs were developed from an intersection of land use, slope, and major soils, using the
geospatial data sources described in Section 4.2.3. In the HRU analysis, SWAT was used to
classify the slopes into two categories: above and below 10%. A single breakpoint was chosen
to represent major differences in runoff and erosive energy without creating an unmanageable
number of individual HRUs. The State Soil Geographic Database (STATSGO) soils coverage
was assigned using the dominant component method in which each soil polygon is represented
by the properties of the dominant constituent soil. The NLCD 2001 land use coverage was
loaded directly into ArcSWAT without modification. The default NLCD class to SWAT class
mapping was appropriate for most areas. Impervious percentage was assigned to developed
land-use classes in the SWAT urban database using values calculated from the NLCD
impervious coverage. The same assumptions were applied for the future developed land-use
classes (i.e., the future classes have the same total and connected impervious fractions as the
corresponding existing urban land uses). HRUs were created by overlaying land use, soil, and
slope at appropriate cutoff tolerance levels to prevent the creation of large numbers of
insignificant HRUs. Land-use classes were retained if they occupied at least 5% of the area of a
subbasin (with the exception of developed land uses, which were retained regardless of area).
Soils were retained if they occupied at least 10% of the area within a given land use in a
_
-------
subbasin. Slope classes were retained if they occupied at least 5% of the area within a given soil
polygon. Land uses, soils, and slope classes that fall below the cutoff value are reapportioned to
the dominant classes so that 100% of the watershed area is modeled (Winchell et al., 2008).
The SWAT models were linked to meteorological stations contained in EPA's BASINS 4
meteorological data set (U.S. EPA, 2008). The models used observed time series for
precipitation and temperature; other weather data were simulated with the SWAT weather
generator, as discussed in Section 4.2.4. Elevation bands were turned on if necessary to account
for orographic effects in areas with a sparse precipitation network and significant elevation
changes. This was generally appropriate where elevations within subbasins spanned a range of
250 m or more. Daily curve number hydrology with observed precipitation and air temperature
was used.
Land management operations were assigned, primarily to account for agricultural practices. For
urban lands, the USGS regression method for pollutant load estimation was specified. In-stream
water quality options started with program defaults.
The target time period for simulation was 31 water years, with the first year dropped from
analysis to account for model spinup (initialization). Some weather stations may have been
absent for the spinup year, but SWAT fills in the missing records using the weather generator.
The remaining 30 years span a period for which the supplied weather data were complete and
included the year 2000 (with the exception of the Loup/Elkhorn basins in Nebraska, for which
the simulation period ended in 1999 due to the termination of a number of precipitation gauges
before the end of 2000).
4.2.2. HSPF Setup Process
HSPF models were developed on a common basis with the SWAT models using the same
geospatial data, but only for the five pilot watersheds. Subbasin boundaries and reach
hydrography were defined using the same NHDPlus catchments as the SWAT models. The
HRUs for HSPF were calculated from the SWAT HRUs, but differ in that soils were aggregated
into hydrologic soil group. Pervious (PERLND) and impervious (IMPLND) land areas are
specified and simulated separately in HSPF, whereas SWAT specifies an impervious fraction for
different land-use categories.
The WinHSPF interface distributed with BASINS (U.S. EPA, 2001) was used to create the user
control input (UCI) and watershed data management (WDM) files. A starter UCI file was
prepared that assigned default values for HRUs. Initial parameter values were based on previous
modeling where available. For areas without previous modeling, hydrologic parameters were
based on recommended ranges in BASINS Technical Note 6 (U.S. EPA, 2000) and related to soil
and meteorological characteristics where appropriate. Snowmelt simulation used the simplified
degree-day method.
The stage-storage-discharge hydraulic functional tables (FTables) for stream reaches were
generated automatically during model creation. The WinHSPF FTable tool calculates the tables
using relationships to drainage area. FTables were adjusted in WinHSPF if specific information
was available to the modeler. Hydraulic characteristics for major reservoirs and flow/load
characteristics for major point sources were defined manually based on available information.
4-8
-------
Nutrients on the land surface were modeled as inorganic nitrogen, inorganic phosphorus, and
total organic matter. The latter was transformed to appropriate fractions of organic nitrogen and
organic phosphorus in the linkage to the stream. The in-stream simulation represented total
nitrogen and total phosphorus as general quality constituents subject to removal approximated as
an exponential decay process. Initial values for decay rates were taken from studies supporting
the USGS SPARROW model (e.g., Alexander et al., 2008).
4.2.3. Watershed Data Sources
The HSPF and SWAT models each use identical geospatial and other input data sources as
described below.
4.2.3.1. Watershed Boundaries and Reach Hydrography
Subbasin boundaries and reach hydrography (with connectivity) for both SWAT and HSPF were
defined using NFIDPlus data (U.S. EPA, 2010), which is a comprehensive set of digital spatial
data representing the surface water of the United States including lakes, ponds, streams, rivers,
canals, and oceans. NHDPlus provided catchment/reach flow connectivity, allowing for creation
of large model subbasins with automation. NHDPlus incorporates the National Hydrography
Dataset (NHD), the National Elevation Dataset, the NLCD, and the Watershed Boundary
Dataset. A MapWindow script was developed to automate (with supervision) the aggregation of
NHDPlus catchments/reaches into model subbasins and reaches. The general approach was to
first run the aggregation script with a smaller target subbasin size (i.e., create several hundred to
a thousand subbasins), then run the script again to create watersheds of the target model size
(comparable to the HUC-10 spatial scale). The two-tiered approach has several benefits; it was
found to be more time efficient, it allowed for greater control over the final basin size, and it
provided a midpoint that could be used to redefine subbasin boundaries to match specified
locations, such as gaging stations and dams/diversions.
Each delineated subbasin was conceptually represented with a single stream assumed to be a
completely mixed, one-dimensional segment with a constant cross section. For the HSPF model,
reach slopes were calculated based on Digital Elevation Model (DEM) data, and stream lengths
were measured from the original NHD stream coverage. Assuming representative trapezoidal
geometry for all streams, mean stream depth and channel width were estimated using regression
curves that relate upstream drainage area to stream dimensions developed for three regions in the
Eastern United States. Existing and more detailed models provided additional site-specific
information on channel characteristics for some watersheds (e.g., Minnesota River; Tetra Tech,
2008b).
The SWAT model also automatically calculates the initial stream geometric values based on
subbasin drainage areas, standard channel forms, and elevation, using relationships developed for
numerous areas of the United States. Channel slope is automatically calculated from the DEM.
4.2.3.2. Elevation
Topography was represented by digital elevation models (OEMs) with a resolution of 30 meters
obtained from USGS' National Elevation Dataset (Gesch et al., 2002). Multiple DEM coverages
were grouped and clipped to the extent of the model watershed area (with a 10-mile buffer to
allow for unforeseen changes to watershed boundaries).
4-9
-------
4.2.3.3. Land Use and Land Cover
The SWAT and HSPF models use a common land use platform representing current (calibration)
conditions and derived from the 2001 NLCD (Homer et al., 2004, 2007). The 2001 NLCD land
cover was used to ensure consistency between all models for the project. The 2001 land use was
chosen rather than the 2006 coverage because it is closer in time to the calibration period of the
models, which typically runs through 2002/3. The 2001 land use is assumed to apply throughout
the baseline model application period.
Some additional processing of the NLCD data was necessary. Several of the land use classes
were aggregated into more general categories to provide a more manageable set of HRUs. The
developed land classes were kept separate for SWAT but aggregated for HSPF. This is because
SWAT assigns percent imperviousness to total developed area, whereas HSPF explicitly
separates developed pervious and impervious areas. The regrouping of the NLCD classes for
SWAT and HSPF is shown in Table 4-1.
The percent impervious area was specified for each developed land class from the NLCD Urban
Impervious data coverage. The NLCD 2001 Urban Imperviousness coverage was clipped to the
extent of the model watershed area (with 10-mile buffer) to calculate the impervious area. The
percent impervious area was then specified by combining data from the 2001 NLCD Land Cover
and Urban Impervious data products. Specifically, average percent impervious area was
calculated over the whole basin for each of the four developed land use classes. These
percentages were then used to separate out impervious land. The analysis was performed
separately for each of the 20 study areas, since regional differences occur. Table 4-2 presents the
calculated 2001 impervious areas for each study area.
4.2.3.4. Soils
Soils data were implemented using SWAT's built-in STATSGO (USDA, 1991) national soils
database. The SWAT model uses the full set of characteristics of dominant soil groups directly,
including information on infiltration, water holding capacity, erodibility, and soil chemistry. A
key input is infiltration capacity, which is used, among other things, to estimate the runoff curve
number. Curve numbers are a function of hydrologic soil group, vegetation, land use, cultivation
practice, and antecedent moisture conditions. The Natural Resource Conservation Service
(NRCS; SCS, 1972) has classified more than 4,000 soils into four hydrologic soil groups (HSGs)
according to their minimum infiltration rate for bare soil after prolonged wetting. The
characteristics associated with each HSG are provided in Table 4-3.
In the HSPF setup the HRUs are not based directly on dominant soils; instead, these were
aggregated to represent HSGs. The HSGs include special agricultural classes (A/D, B/D, and
C/D) in which the first letter represents conditions with artificial drainage and the second letter
represents conditions without drainage. The first designator was assumed to apply to all crop
land, while the second designator was assumed for all other land uses.
4-10
-------
Table 4-1. Regrouping of the NLCD 2001 land-use classes for the HSPF and
SWAT models
NLCD class
1 1 Water"
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
3 1 Barren Land
41 Forest — Deciduous
42 Forest — Evergreen
43 Forest — Mixed
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91-97 Wetland (emergent)
98-99 Wetland (nonemergent)
SWAT class
WATR (water)
WATR (water)
URLD (Urban Residential— Low Density)
URMD (Urban Residential — Medium Density)
URHD (Urban Residential— High Density)
UIDU (Urban Industrial and High Intensity)
SWRN (Range-Southwestern U.S.)
FRSD (Forest— Deciduous)
FRSE (Forest — Evergreen)
FRST (Forest— Mixed)
RNGB (Range— Brush)
RNGE (Range — grasses)
HAY
AGRR (Agricultural Land-Row Crops)
WETF (Wetlands— Forested), WETL (Wetlands),
WETN (Wetlands— Nonforested)
WATR (water)
HSPF class
WATER
BARREN
DEVPERV (Developed Pervious)
IMPERV (Impervious)
BARREN
FOREST
SHRUB
GRASS
BARREN
GRASS
AGRI (Agriculture)
WETL (Wetlands)
WATER
"Water surface area is usually accounted for as reach area.
4-11
-------
Table 4-2. Calculated fraction impervious cover within each developed land
class for each study area based on NLCD 2001
Site ID
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
Open space (%)
8.04
7.37
10.11
7.20
8.83
7.30
7.53
6.59
8.34
8.22
7.42
8.76
5.95
7.75
6.41
6.90
7.17
7.74
9.78
9.56
Low intensity (%)
30.16
29.66
29.79
31.87
32.36
32.53
32.91
29.20
29.68
32.81
31.64
32.36
30.02
35.39
33.46
31.26
30.90
31.65
31.89
32.31
Medium intensity (%)
60.71
53.71
61.48
60.14
61.24
60.72
60.11
55.01
60.14
60.90
59.16
60.49
55.41
61.31
60.79
60.90
61.05
60.78
60.48
61.49
High intensity (%)
89.90
73.85
87.17
87.47
88.70
86.75
88.08
83.31
86.59
87.25
85.99
84.32
81.20
88.83
86.76
85.41
87.31
89.15
87.41
88.94
Table 4-3. Characteristics of NRCS soil hydrologic groups
Soil group
A
B
C
D
Characteristics
Sandy, deep, well drained soils; deep loess; aggregated silty soils
Sandy loams, shallow loess, moderately deep and moderately well drained soils
Clay loam soils, shallow sandy loams with a low permeability horizon impeding
drainage (soils with a high clay content), soils low in organic content
Heavy clay soils with swelling potential (heavy plastic clays), water-logged soils,
certain saline soils, or shallow soils over an impermeable layer
Minimum infiltration
capacity (in/hr)
0.30-0.45
0.15-0.30
0.05-0.15
0.00-0.05
4.2.3.5. Point Source Discharges
The primary objective of this study is to examine relative changes that are potentially associated
with changes in climate and land use. From that perspective, point source discharges can be
4-12
-------
characterized as a nuisance parameter. However, point sources that are large enough relative to
receiving waters to affect the observed streamflow and nutrient loads in river systems need to be
included to calibrate the models. This is done in a simplified way, and the point sources were
then held constant for future conditions, allowing analysis of relative change. Only the major
dischargers, typically those with a discharge rate greater than 1 MGD were included in the
models. The major dischargers account for the majority of the total flow from all permitted
discharges in most watersheds, so the effect on the calibration of omitting smaller sources is
relatively small, except perhaps during extreme low streamflow conditions. Data were sought
from the EPA's Permit Compliance System database for the major dischargers in the watersheds.
Facilities that were missing TN, TP, or total suspended solids (TSS) concentrations were filled
with a typical pollutant concentration value from the literature based on Standard Industrial
Classification (SIC) code. The major dischargers were represented at long-term average flows,
without accounting for changes over time or seasonal variations.
4.2.3.6. Atmospheric Deposition
Atmospheric deposition can be a significant source of inorganic nitrogen to watersheds and
water bodies. SWAT2005 allows the user to specify wet atmospheric deposition of nitrate
nitrogen. This is specified as a constant concentration across the entire watershed. Wet
deposition of ammonia and dry deposition of nitrogen is not addressed in the SWAT2005 model.
HSPF allows the specification of both wet and dry deposition of both nitrate nitrogen (NOs) and
ammonium nitrogen (NH4), and both were included in the model. Dry deposition is specified as
a loading series, rather than concentration series. Because wet deposition is specified as a
concentration, it will vary in accordance with precipitation changes in future climate scenarios,
whereas the dry deposition series (HSPF only) is assumed constant for future scenarios.
Total oxidized nitrogen (NOX) emissions in the United States remained relatively constant to a
first approximation across the model period considered in this study from the early 1970s up
through 2002 (U.S. EPA, 2002). There is strong geographic variability in atmospheric
deposition, but much smaller year-to-year variability at the national scale over this period
(Suddick and Davidson, 2012). The National Acid Deposition Program (NADP;
http://nadp.sws.uiuc.edu/) monitors wet deposition across the country and produces yearly
gridded maps of NOs and NH4 wet deposition concentrations. Dry deposition rates are
monitored (and interpreted with models) by the EPA Clean Air Status and Trends Network
(http://epa.gov/castnet/javaweb/index.html). Results for year 2000 were selected as generally
representative and each study watershed was characterized by a spatial average wet deposition
concentration (and dry deposition loading rate for HSPF). Atmospheric deposition of
phosphorus and sediment was not considered a significant potential source and is not addressed
in the models.
4.2.3.7. Impoundments, Diversions, and Withdrawals
The hydrology of many large watersheds in the United States is strongly impacted by
anthropogenic modifications, including large impoundments and withdrawals for consumptive
use. It is necessary to take these factors into account to develop a calibrated model. At the same
time, these anthropogenic factors constitute a problem for evaluating responses to future
changes, as there is no clear basis for evaluating future changes in reservoir operations or water
4-13
-------
withdrawals. In addition, information on impoundments, withdrawals, and trans-basin water
imports is often difficult to obtain. The approach taken in this project is to minimize the
importance of impoundments and withdrawals by focusing on relative changes between present
and future conditions with these factors held constant. In this way, the results that are presented
are estimates of the change that may be anticipated based on changes to meteorological and land
use forcing within the subject study area, with other factors held constant. Simulation results do
not account for potential future changes in water management.
The general approach adopted for this project was to select study areas by avoiding major human
interventions (e.g., reservoirs) in the flow system where possible, to ignore relatively minor
interventions, and where necessary to represent significant interventions in a simplified manner.
In the first instance, study watersheds were delineated to avoid major reservoirs where possible.
For example, the model of the Verde River watershed (Arizona) is terminated at the inflow to
Horseshoe Reservoir. In some cases, as in the Sacramento River watershed, an upstream
reservoir is treated as a constant boundary condition because information on future reservoir
management responses to climate change was not available.
Impoundments, withdrawals, and water imports that do not have a major impact on downstream
streamflow were generally omitted from the large scale models. Inclusion or omission of such
features was a subjective choice of individual modelers; however, it was generally necessary to
include such features if they resulted in a modification of flow at downstream gages on the order
of 10% or more. Where these features were included they were represented in a simplified
manner: (1) impoundments were represented by simplified (two-season) stage-discharge
operating rules, developed either from documented operational procedures or from analysis of
monitored discharge; (2) large withdrawals were represented as either annual or monthly
constant average rates; and (3) major trans-basin water imports were also represented as either
annual or monthly constant average rates depending on availability of data. Use of surface water
for irrigation was simulated only in those basins where it was determined during calibration that
it was a significant factor in the overall water balance. These simplifying assumptions decrease
the quality of model fit during calibration and validation, but provide a stable basis for the
analysis of relative response to climate and land-use change within the basin.
More detailed information about the representation of impoundments and other anthropogenic
influences on hydrology in each study area are presented in Appendices D through W.
4.2.4. Baseline Meteorology Representation
Time series of observed meteorological data (for both SWAT and HSPF) were obtained from the
2006 BASINS 4 Meteorological Database (U.S. EPA, 2008). The database contains records for
16,000 stations from 1970-2006, set up on an hourly basis, and has the advantage of providing a
consistent set of parameters with missing records filled and daily records disaggregated to an
hourly time step. The disaggregation was performed using automated scripts that distribute the
daily data using one of several nearby hourly stations, using the one whose daily total is closest
to the daily value being disaggregated. If the daily total for the hourly stations being used were
not within a specified tolerance of the daily value, the daily value was distributed using a
triangular distribution centered at the middle of the day. The process involved extensive quality
control review; however, the true temporal distribution of precipitation at daily stations is
unknown and the automated approach can occasionally result in anomalously high hourly
__
-------
estimates. Both factors introduce some irreducible uncertainty into hydrologic simulations using
disaggregated daily precipitation stations.
A typical site-specific watershed project would assemble additional weather data sources to
address under-represented areas, but this requires significant amounts of additional quality
control and data processing to fill gaps and address accumulated records. It was assumed that
the use of the BASINS 2006 data was sufficient to produce reasonable results at the broad spatial
scale that is the focus of this project, particularly for evaluating the relative magnitude of change.
Significant orographic variability was accounted for through the use of lapse rates because the
available stations typically under-represent high mountain areas.
The required meteorological time series for both SWAT and HSPF (as implemented for this
project) included precipitation, air temperature, and either calculated PET or time series required
to generate PET. SWAT uses daily meteorological data, while HSPF requires hourly data.
Stations were selected to provide a common 30-year or more period of record (or one that could
be filled from an approximately co-located station).
Table 4-4 presents a summary of annual precipitation and temperature observations for each of
the study areas from 1971-2000. For more specific details on the meteorological data used in
each of the study areas, refer to the model calibration reports provided in Appendices D through
W.
PET is the third major weather time series input to the watershed models. As evapotranspiration
is typically the largest outgoing term in the water balance, watershed models are highly sensitive
to the specification of PET, particularly for simulating low streamflow conditions and events.
Many watershed modeling efforts perform well with simplified approaches to estimating PET,
such as the Hamon method (included as an option in the BASINS data set), which depend
primarily on air temperature. However, the robustness of watershed model calibrations
conducted with simplified PET is suspect under conditions of climate change, since a variety of
other factors that influence PET, such as wind speed and cloud cover, are also likely to change.
Therefore, we implemented Penman-Monteith PET, which employs a full energy balance
(Monteith, 1965; Jensen et al., 1990). The implementation varies slightly between SWAT and
HSPF. In SWAT, the full Penman-Monteith method (Allen et al., 2005) is implemented as an
internal option in the model and includes feedback from crop height simulated by the plant
growth model. For HSPF, Penman-Monteith reference evapotranspiration at each weather
station was calculated externally using the SWAT2005 model subroutines with observed
precipitation and temperature. In both cases, the additional inputs to the energy balance (solar
radiation, wind movement, cloud cover, and relative humidity) were provided by the SWAT
weather generator, which relies on monthly conditional probability statistics for each of these
inputs. An evaluation of the BASINS meteorological data set indicated substantial amounts of
missing data for these inputs (especially for solar radiation and cloud cover); hence, the SWAT
weather generator was preferred to enable consistent 30-year simulations. HSPF does not
simulate crop growth, so monthly coefficients are incorporated in the model to convert reference
crop PET to values appropriate to different crop stages using the Food and Agriculture
Organization (FAO) method (Allen et al., 1998).
4-15
-------
Table 4-4. Weather station statistics for the 20 study areas (1971-2000)
Model area
Apalachicola-Chattahoochee-
Flint Basins
Arizona: Salt, Verde, and San
Pedro
Cook Inlet Basin
Georgia-Florida Coastal
Plain
Illinois River Basin
Lake Erie Drainages
Lake Pontchartrain Drainage
Minnesota River Basin
Nebraska: Loup and Elkhom
River Basins
New England Coastal Basins
Powder and Tongue River
Basins
Rio Grande Valley
Sacramento River Basin
Southern California Coastal
Basins
South Platte River Basin
Susquehanna River Basin
Tar and Neuse River Basins
Trinity River Basin
Upper Colorado River Basin
Willamette River Basin
Number of
precipitation
stations
37
29
14
51
72
57
26
39
81
52
37
53
28
85
50
60
40
64
47
37
Average annual
precipitation total
(inches)
54.26
19.67
28.50
53.21
38.25
38.15
66.33
28.26
26.10
48.45
17.70
15.18
37.47
20.21
16.82
41.30
49.91
40.65
16.36
58.38
Number of
temperature stations
22
25
14
37
47
41
15
32
31
36
30
41
18
33
23
27
28
32
39
29
Average annual
temperature (°F)
63.43
56.81
34.16
68.24
49.00
49.10
66.64
43.90
48.35
46.23
44.15
44.71
57.45
61.20
43.46
48.26
59.91
64.78
41.73
51.19
4.3. SIMULATION OUTPUT AND ENDPOINTS
Simulations focused on streamflow, total nitrogen, total phosphorus, and total suspended solids
loads. Output from both models was analyzed as daily time series over the 30-year analysis
period. Several summary metrics or endpoints were also calculated based on the daily time
series. Because of calibration uncertainty inherent in modeling at this scale, estimates of relative
change between historical and future simulations are most relevant. In addition to basic
streamflow statistics, comparisons are made for 100-year flood peak (fit with Log Pearson type
III distribution; USGS, 1982), average annual 7-day low flow, Richards-Baker flashiness index
(a measure of the frequency and rapidity of short-term changes in streamflow; Baker et al.,
2004), and days to the centroid of mass for the annual streamflow on a water-year basis (i.e.,
4-16
-------
days from previous October 1 at which half of the streamflow for the water year is achieved, an
important indicator of changes in the snow accumulation and melt cycle). For the Log Pearson
III estimator, use of a regionalized skew coefficient is not appropriate to climate change scenario
applications as the regional map represents existing climate. Therefore, the K factor is estimated
using the skew coefficient from the model output only, without any weighting with the regional
estimate.
Each of the streamflow endpoints discussed in the preceding section has been calculated for each
scenario at the output of each HUC-8 contained within a study area. Several other summary
measures of the water balance, largely drawn from the work of Kurd et al. (1999), are
summarized as averages at the whole-watershed scale. These are the Dryness Ratio (fraction of
precipitation that is lost to ET as reported by the SWAT model), Low Flow Sensitivity
(expressed as the rate of baseflow generation by shallow groundwater, tile drainage, and lateral
subsurface flow pathways in units of cfs/mi2), Surface Runoff Fraction (the fraction of total
streamflow from the uplands that occurs through overland flow pathways), Snowmelt Fraction
(the fraction of total streamflow from the uplands that is generated by melting snow), and Deep
Recharge Rate (the annual average depth of water simulated as recharging deep aquifers that do
not interact with local streams). Table 4-5 provides a summary of streamflow and water quality
endpoints evaluated in this study.
The mobilization and transport of pollutants will also be affected by climate and land-use
change, both as a direct result of hydrologic changes and through changes in land cover and plant
growth. Monthly and annual loads of sediment, phosphorus, and nitrogen are likely the most
useful and reliable measures of water quality produced by the analysis. Accordingly, the focus
of comparison among scenarios is on monthly and average annual loads for total suspended
solids, total nitrogen, and total phosphorus. As with the streamflow simulation, it is more robust
to examine relative rather than absolute changes in simulated pollutant loads when comparing
scenarios to current conditions. Thus, we also calculate and express results as percent changes.
All models are calibrated and validated, but in many cases current loads are imprecisely known
due to limited monitoring data.
Because the sediment load in rivers/streams is often dominated by channel adjustment processes,
which are highly site specific and occur at a fine spatial scale, it is anticipated that precision in
the simulation of sediment and sediment-associated pollutant loads will be relatively low.
Nutrient balances can also be strongly affected by biological processes in the channels, which
can only be roughly approximated at the scale of modeling undertaken. It should also be noted
that the modeling makes the following assumptions that limit uses for absolute (as opposed to
relative) simulations of future pollutant loads: (1) external boundary conditions (if needed), such
as upstream inflows and pollutant loads, are constant; (2) point source discharges and water
withdrawals are assumed constant at current rates; (3) no provision is made for human adaptation
in rural land management, such as shifts in crop type in response to climate change; and (4) plant
growth responses to climate change are simulated to the extent they are represented in the SWAT
plant growth model; however, large-scale shifts in natural cover type in response to climate
change are not simulated.
4-17
-------
Table 4-5. Summary of streamflow and water quality endpoints
Endpoint
Future Flow Volume
Average Seven Day
Low Flow
100 Year Peak Flow
Days to Flow Centroid
Richards-Baker
Flashiness Index
Dryness Ratio
Low Flow Sensitivity
Surface Runoff
Fraction
Snowmelt Fraction
Deep Recharge
AET
PET
Total Suspended
Solids (TSS)
Total Phosphorus (TP)
Total Nitrogen (TN)
Dimension
\}lt
L3/t
L3/t
f(days)
dimensionless
dimensionless
L/t
dimensionless
dimensionless
L/t
L/t
Lit
mass/t
mass/f
mass/t
Description
Average of simulated streamflow
volume per unit time
Average annual 7-day low
streamflow event volume
Estimated peak streamflow rate
based on annual flow maxima
series, Log Pearson III method
Number of days from the previous
October 1 (start of water year) at
which half of the streamflow
volume for that water year is
achieved
Indicator of the frequency and
rapidity of short term changes in
daily streamflow rates
Fraction of input precipitation lost
toET
Rate of baseflow contributions
from shallow groundwater, tile
drainage, and lateral subsurface
flow pathways, depth per unit time
Fraction of streamflow contributed
by overland flow pathways
Fraction of streamflow contributed
by snowmelt
Depth of water recharging deep
aquifers per unit time
Actual depth of evapotranspiration
lost to the atmosphere per unit time
Theoretical potential
evapotranspiration as depth per unit
time, assuming moisture not
limiting
Mass load of suspended sediment
exiting stream reach per unit time
Mass load of total phosphorus
exiting reach per unit time
Mass load of total nitrogen exiting
stream reach per unit time
Calculation
Sum of annual streamflow volume
simulated by the watershed model
Lowest 7-day-average streamflow
simulated for each year
Log Pearson III extreme value estimate
following USGS (1982), based on
simulated annual maxima series
Count of days to 50% of simulated total
annual streamflow volume for each
water year.
Analyzed by method given in Baker et
al. (2004), applied to daily streamflow
series for each year
Calculated as (precipitation -
outflow)/precipitation for consistency
with Kurd etal. (1999)
Sum of simulated streamflow from
shallow groundwater, tile drainage, and
lateral subsurface flow pathways
divided by area.
Surface runoff divided by total outflow.
Estimated as water equivalent of
simulated snowfall divided by total
precipitation
Total water volume simulated as lost to
deep recharge divided by area
Evapotranspiration simulated by the
watershed model
Potential evapotranspiration simulated
by the Penman-Monteith method
(Jensen et al., 1990)
Sum of simulated mass exiting a stream
reach
Sum of simulated mass exiting a stream
reach
Sum of simulated mass exiting a stream
reach
4.4. MODEL CALIBRATION AND VALIDATION
The watershed models were calibrated and validated in each of the study areas in accordance
with the project Quality Assurance Project Plan (QAPP; see Appendix B). The following section
4-18
-------
provides a brief summary of calibration and validation methods and results. Detailed description
of calibration and validation methods and results for the individual study areas are presented in
Appendices D through W.
Calibration refers to the adjustment of model parameters to reproduce or fit simulation results to
observed data. Calibration is required for parameters that cannot be deterministically and
uniquely evaluated from topographic, climatic, physical, and chemical characteristics of the
watershed and compounds of interest. Validation is performed by application of the calibrated
model to a different period of observed data to test the robustness of the calibrated parameter set.
If the model exhibited a significant degradation in performance in the validation period, the
calibration process is repeated until results are considered acceptable.
The calibration and validation approach for the study areas was to first focus on a single HUC-8
within the larger study area (preferably one for which some modeling was already available
along with a good record of flow gaging and water quality monitoring data), and then extend the
calibration to adjacent areas with modifications as needed to achieve a reasonable fit at multiple
spatial scales. Each HUC-8 watershed was generally subdivided into approximately 8 subbasins,
approximating the HUC-10 watershed scale.
The base period of observed data used for calibration and validation was approximately 1970 to
2000, with some variation depending on availability of meteorological data, while the base land
use was from 2001 NLCD. In watersheds with significant land-use change, moving back too far
from 2001 may not provide a firm basis for calibration. Therefore, calibration generally focused
on approximately the 1991-2001 time period, although the full 1971-2000 period was used for
comparison to future changes. Validation was typically performed on the period before 1991
and/or data from post-1991 at different locations.
4.4.1. Hydrology
The goal of hydrologic calibration for both HSPF and SWAT was to achieve error statistics for
total streamflow volume, seasonal streamflow volume, and high and low streamflow within the
range recommended by Lumb et al. (1994) and Donigian (2000) while also maximizing the
Nash-Sutcliffe coefficient of model fit efficiency (£). Standardized spreadsheet tools were
developed to help ensure consistency in the calibration and validation process across watersheds,
and to provide a standardized set of statistics and graphical comparisons to data. These statistics
were used to adjust appropriate model parameters until a good statistical match was shown
between the model output and observed data.
Lumb et al. (1994) and Donigian (2000) recommend performance targets for HSPF based on
relative mean errors calculated from simulated and observed daily average streamflow.
Donigian classified these into qualitative ranges, which were modified slightly in this project for
application to both HSPF and SWAT (see Table 4-6). In general, hydrologic calibration
endeavored to achieve a "good" level of model fit where possible. It is important to note that the
tolerance ranges are intended to be applied to mean values and that individual events or
observations may show larger differences and still be acceptable (Donigian, 2000).
4-19
-------
Table 4-6. Performance targets for hydrologic simulation (magnitude of
annual and seasonal relative mean error) from Donigian (2000)
Model component
1 . Error in total volume
2. Error in 50% lowest streamflow volumes
3. Error in 10% highest streamflow volumes
4. Error in storm volume
5. Winter volume error
6. Spring volume error
7. Summer volume error
8. Fall volume error
9. Error in summer storm volumes
Very good (%)
<5
<5
<10
<10
<15
<15
<15
<15
<25
Good (%)
5-10
5-10
10-15
10-20
15-30
15-30
15-30
15-30
25-50
Fair (%)
10-15
10-25
15-25
20-30
30-50
30-50
30-50
30-50
50-75
Poor (%)
>15
>25
>25
>30
>50
>50
>50
>50
>75
The Nash-Sutcliffe coefficient of model fit efficiency (E) is also widely used to evaluate the
performance of models that predict time series. Nash and Sutcliffe (1970) define E as:
Zfa-^)
t(o,-o)2
4-4
where Ot andPj represent members of a set of n paired time series observations and predictions,
respectively, and O is the mean of the observed values. E ranges from minus infinity to 1.0, with
higher values indicating better agreement. The coefficient represents the ratio of the mean
square error to the variance in the observed data, subtracted from unity (Wilcox et al., 1990). A
value of zero for E indicates that the observed mean is as good a predictor of time series values
as the model, while negative values indicate that the observed mean is a better predictor than the
model. A value of E greater than 0.7 is often taken as an indicator of a good model fit
(Donigian, 2000). Note, however, that the value depends on the time basis on which the
coefficient is evaluated. That is, values of E for monthly average streamflow are typically
noticeably greater than values of E for daily streamflow, as watershed models, in the face of
uncertainty in the representativeness of precipitation records, are often better predictors of
interseasonal trends than of intraseasonal variability. Moriasi et al., (2007) recommend a Nash-
Sutcliffe E of 0.50 or better (applied to monthly sums) as an indicator of adequate hydrologic
calibration when accompanied by a relative error of 25% or less.
A potential problem with the use of E is that it depends on squared differences, making it overly
sensitive to extreme values (Legates and McCabe, 1999). This is particularly problematic for
sparse time series, such as water quality observations, in which poor estimation of one or a few
high outliers may strongly influence the resulting statistic. It is an even greater problem for the
4-20
-------
comparison of model output to load estimates based on sparse concentration data, as these
estimates are themselves highly uncertain (using point-in-time grab samples to represent daily
averages and interpolating to unobserved days), further increasing the leverage associated with
high outliers.
To address these issues and lessen the effect of outliers, Garrick et al. (1978) proposed use of a
baseline adjusted coefficient of model fit efficiency, E\, which depends on absolute differences
rather than squared differences:
4-5
Garrick's proposed statistic is actually more general, allowing 0 ' to be a baseline value that may
be a function of time or of other variables, rather than simply the mean. E\ 'may be similar to or
greater or less than E for a given set of predictions and measurements depending on the type of
outliers that are present.
For most watershed models, E is an appropriate measure for the fit of streamflow time series in
which complete series of observations are known with reasonable precision. E\ 'is a more
appropriate and stable measure for the comparison of simulated pollutant loads to estimates
based on sparse observed data.
4.4.1.1. Flow Calibration Adjustments
HSPF and SWAT hydrology calibration adjustments were made for a range of sensitive model
parameters selected to represent key watershed processes affecting runoff (U.S. EPA, 2000;
Neitsch et al., 2005; see Tables 4-7 and 4-8, respectively, for selected key parameters most
frequently adjusted). The adjustment of other parameters and the degree of adjustment to each
parameter vary by watershed. Details are provided in the individual calibration reports for each
of the watersheds in Appendices D through W.
4-21
-------
Table 4-7. Key hydrology calibration parameters for HSPF
Parameter name
INFILT
AGWRC
LZSN
BASETP
KMELT
PET factor
DEEPFR
LZETP
Definition
Nominal infiltration rate parameter
Ground-water recession rate
Lower zone nominal soil moisture storage
ET by riparian vegetation
Degree-day melt factor
Potential evapotranspiration
Fraction of groundwater inflow that will enter deep groundwater
Lower zone E-T parameter
Table 4-8. Key hydrology calibration parameters for SWAT
Parameter name
CN
ESCO
SLTRLAG
ALPHA_BF
GW_DELAY
CANMAX
OV_N, CH_N2, CH_N1
Sol_AWC
Bank storage and recession rates
Snow parameters SFTMP, SMTMP,
SMFMX and SMFMN
TIMP
CHJC1
Definition
Curve numbers — varied systematically by land use
Soil evaporation compensation factor
Surface runoff lag coefficient
Baseflow alpha factor
Groundwater delay time
Maximum canopy storage
Manning's "w" values for overland flow, main channels, and tributary channels
Available water capacity of the soil layer, mm water/mm of soil
Bank storage and recession rates
Snowfall temperature, snowmelt base temperature, maximum melt rate for snow
during year, and minimum melt rate for snow during year
Snow pack temperature lag factor
Effective hydraulic conductivity in tributary channel alluvium
4.4.2. Water Quality
The models in this study are designed to simulate total nitrogen, total phosphorus, and total
suspended solids. The first objective of calibration was to reduce the relative absolute deviation
between simulated and estimated loads to below 25% if possible. The water quality calibration
focuses on the replication of monthly loads, as specified in the project QAPP (see Appendix B).
While a close match to individual, instantaneous concentration observations cannot be expected
given the approach taken in the model simulations of water quality, the calibration also examined
the general relationship of observed and predicted concentrations with the intent of minimizing
bias relative to streamflow regime or time of year. Comparison to monthly loads presents
challenges, as monthly loads are not observed. Instead, monthly loads must be estimated from
4-22
-------
scattered concentration grab samples and continuous streamflow records. As a result, the
monthly load calibration is inevitably based on the comparison of two uncertain numbers.
Nonetheless, calibration is able to achieve a reasonable agreement. The direct comparison of
estimated and simulated monthly loads was supplemented by detailed examinations of the
relationships of streamflow to loads and concentrations, and the distribution of concentration
prediction errors versus streamflow, time, and season to help minimize bias in the calibration.
For application on a nationwide basis, it was assumed that total suspended solids and total
phosphorus loads will likely exhibit a strong positive correlation to streamflow (and associated
erosive processes), while total nitrogen loads, which often have a dominant subsurface loading
component, will not (Allan, 1986; Burwell et al., 1975; Follett, 1995). Accordingly, total
suspended solids and total phosphorus loads were estimated from observations using a
flow-stratified log-log regression approach, while total nitrogen loads were estimated using a
flow-stratified averaging estimator, consistent with the findings of Preston et al. (1989).
4.4.2.1. Water Quality Calibration Adjustments
Water quality calibration began with sediment transport processes. Observed suspended solids
concentrations are the result of multiple processes, including sediment detachment, sediment
transport in overland flow, and channel scour and deposition processes. The sediment
detachment routines for both SWAT and HSPF were related to USLE parameters available in the
soils database. For most basins, calibration focuses on sediment transport in overland flow,
using the peak rate or transport rate factors available in both models. Channel scour and
deposition processes were modified where needed to achieve a fit to observations or where
detailed work with prior models provided a basis for modifying the default parameters.
In HSPF, nitrogen loading from the land surface was simulated as a buildup/washoff process,
while phosphorus was simulated as sediment-associated. Both nitrogen and phosphorus also
were simulated with dissolved-phase loads from interflow and groundwater discharge.
Calibration for nutrients in HSPF primarily addressed adjustments of the buildup/washoff
coefficients or sediment potency (concentration relative to sediment load) factors and monthly
subsurface discharge concentrations. In SWAT, the nutrient simulation is intimately linked to
the plant growth model, but is sensitive to initial nutrient concentrations and the ability of plants
to withdraw nutrients from various soil layers. In watersheds where significant channel scour
was simulated, the nutrient content of scoured sediment was also an important calibration
parameter.
4.4.3. Accuracy of the Watershed Models
The quality of model fit varies with the study area and parameter considered. In general, the full
suite of SWAT models for the 20 watersheds—after calibration—provide a good to excellent
representation of the water balance at the monthly scale and a fair to good representation of
hydrology at the daily scale (see Table 4-9 for the initial calibration site results). The quality of
model fit to hydrology as measured at multiple stations (HUC-8 spatial scale and larger)
throughout the watershed was, not surprisingly, better when a spatial calibration approach was
used. At all calibration and validation sites, the median monthly Nash-Sutcliffe E coefficient
from the SWAT models was 0.74 for both the pilot and nonpilot study areas. More detailed
calibration and validation results for each study area are provided in Appendices D through W.
4-23
-------
Less precise model fit to observations resulted in several study areas for various reasons. In
addition to differences in individual modeler preferences and skill, Low E coefficients in the Rio
Grande Valley likely reflect insufficient knowledge of operations of the many reservoirs in the
basin. Calibrating watershed hydrology was problematic in systems dominated by large-scale
interactions with regional groundwater systems—notably, Verde River in Arizona and the
Loup/Elkhorn River system in the Nebraska sandhills. Both HSPF and SWAT use simplistic
storage reservoir representations of groundwater in which water can percolate from the soil
profile into local shallow groundwater storage, from which it is gradually released following an
exponential decay pattern characterized by a recession coefficient. Perennial streamflow in the
Verde River is sustained by groundwater discharges of nonlocal origin that derive from the
upstream Chino basin. The Loup and Elkhorn Rivers drain highly porous sands where surface
runoff is minimal and streamflow in some tributary rivers is nearly constant and only weakly
correlated to rainfall patterns (e.g., see Figure 4-1), a situation that is difficult to address in a
rainfall-runoff model without linking to a true groundwater simulation model.
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1989 to 9/30/1999)
300
200
o
100
O-89 A-91
O-92
A-94
O-95
A-97
O-98
Month
Figure 4-1. Example of weak correlation of rainfall and flow in the Dismal
River at Thedford, NE (USGS 06775900) in the Loup River basin.
Different modelers handled the situation in these two regions in different ways. For the Verde
River (where both HSPF and SWAT were applied) the regional groundwater inflow was
specified as an external forcing time series. This has the advantage of allowing the model
calibration to focus on rainfall-runoff events that are responsible for most year-to-year variability
in streamflow and most pollutant transport. The major disadvantage is that there is not a clear
means to specify how this groundwater forcing might respond to changes in climate. Instead,
results for the Verde River show relative changes that would be expected under the assumption
that the regional groundwater discharge does not change.
For the Loup and Elkhorn River basins, a reasonable fit to both calibration and validation periods
was obtained by specifying extremely slow groundwater recession rates in conjunction with use
4-24
-------
of the soil crack flow option (which allows a fraction of rainfall to flow directly to groundwater)
in the sandhill region. This approach can replicate the major observed features of the water
balance, although it does not achieve a high degree of precision in explaining day-to-day
variability in observed streamflow. Further, the simulated groundwater discharges are
responsive to changes in climate forcing. However, use of this approach comes at a cost due to
the way that groundwater is simulated in the SWAT model. Specifically, SWAT simulates
baseflow discharge on a given day as a function of discharge on the previous day, modified by
the recession coefficient, plus the effects of new recharge to groundwater. Groundwater
discharge at the start of the simulation is constrained to be zero. Use of a very slow recession
rate gives a reasonable fit to the calibration and validation periods in this study area; however, it
also results in very slow convergence of estimated groundwater discharge from the initial zero.
This resulted in a situation in which it took approximately 10 years for streamflow to reach levels
in line with observations. Thus, simulated streamflow for the early years are often zero. Adding
a longer spinup period does not resolve the problem as the low recession rate results in a
nonstationary solution in which baseflow continues to gradually increase over time and the
simulated streamflow eventually overshoots observations during the calibration period if the
spinup period is extended. Due to this issue, change scenario results are presented only for the
20-year calibration and validation periods in the Loup and Elkhorn River study area.
Calibration and validation for water quality is subject to higher uncertainty than streamflow
calibration due to limited amounts of monitoring data and a simplified representation of the
multiple complex processes that determine in-stream pollutant concentrations. The primary
objective of water quality simulation in this project is to assess relative changes in pollutant
loads, but loads are not directly observed. Inferring loads from point-in-time concentration data
and streamflow introduces another layer of uncertainty into the calibration process. Calibration
also examined observed versus predicted concentrations; however, SWAT, as a daily curve
number model, does not have a high level of skill in simulating instantaneous concentrations,
particularly during high flow events, and is better suited to the simulation of loads at the weekly
to monthly scale.
As with the hydrology calibration, the reliability of the models for simulating changes in water
quality appears to increase with calibration at multiple locations. In general, it is more difficult
to obtain a high level of precision for simulated water quality than for hydrology in a watershed
model, as the processes are complex, the data typically sparse, and any errors in hydrology tend
to be amplified in the water quality simulation. The water quality calibration is based on loads,
but loads are not directly observed. Instead, loads are inferred from sparse concentration
monitoring data and streamflow gaging. Thus, both the simulated and "observed" loads are
subject to considerable uncertainty. Comparison based on concentrations can also be
problematic, as most water quality samples are grab samples that represent points in time and
space, whereas model output is integrated over a stream segment and may produce large
apparent errors due to small shifts in timing. Finally, most stations at the HUC-8 scale include
upstream point sources, which often have a strong influence on low-flow concentrations and load
estimates. Limited knowledge about point source loads thus also creates a challenge for the
water quality calibration. In most cases, the pollutant load simulations from the SWAT model
appear to be in the fair to good range (see Table 4-9)—except in a few cases where parameters
were extended from one station to another watershed without adjustment, giving poor results.
This suggests limits to the reliability of simulation results in the portions of watersheds for which
4-25
-------
calibration was not pursued. Nonetheless, simulations of the relative response to climate change
and land development scenarios are more reliable than for the actual observed future values—as
long as the significant processes that determine pollutant load and transport within a watershed
are represented.
HSPF model calibration for the five pilot sites provided a somewhat stronger fit to daily
streamflow in four of the five watersheds (see Table 4-10), presumably at least in part due to
HSPF's use of subdaily precipitation. In two models, the fit to total suspended solids load was
notably worse for HSPF, apparently due to the difficulties in adjusting the more complex channel
scour and deposition routines of this model with limited data and on a compressed schedule.
4-26
-------
Table 4-9. Summary of SWAT model fit for initial calibration site (20 study areas)
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Initial calibration/
validation watershed
Upper Flint River
Verde River
Kenai River
Ochlockonee River
Iroquois River
Cuyahoga River
Amite River
Cottonwood River
Elkhorn River
Saco River
Tongue River
Saguache Creek
Sacramento River
Santa Ana River
South Platte River
Initial
calibration/
validation
USGS gage
02349605
09504000
15266300
02329000
05526000
04208000
07378500
05317000
06800500
01066000
06306300
08227000
11377100
11066460
06714000
Hydrology
caL/val. yr
1993-2002/
1983-1992
1992-2002/
1982-1992
1992-2001/
1982-1991
1992-2002/
1982-1992
1992-2001/
1982-1992
1990-2000/
1980-1990
1995-2004/
1985-1994
1992-2002/
1982-1992
1989-1999/
1978-1988
1993-2003/
1983-1993
1993-2003/
1983-1993
1993-2003/
1983-1993
1992-2001/
1983-1992
1991-2001/
1981-1991
1991-2000/
1981-1990
Total volume
cal./val.
(daily and
monthly E)
0.62/0.56
0.88/0.83
0.03/-1.0
0.88/0.32
0.68/0.55
0.80/0.75
0.71/0.80
0.79/0.90
0.70/0.67
0.77/0.71
0.61/0.62
0.70/0.73
0.79/0.69
0.95/0.90
0.79/0.74
0.91/0.83
0.42/0.52
0.70/0.66
0.61/0.76
0.71/0.84
0.72/0.7
0.83/0.82
0.47/0.07
0.53/0.31
0.75/0.57
0.94/0.92
0.63/0.59
0.75/0.68
0.74/0.52
0.86/0.63
Total
volume
cal./val.
(% error)
7.28/3.33
-2.46/5.68
-18.96/19.49
4.25/-5.S4
-16.99/-2.98
-3.32/-13.38
-1.61/-0.93
-5.41/-0.84
-2.59/-8.81
1.08/0.67
9.26/-9.9S
-4.92/32.99
10.23/10.06
3.71/1.61
9.82/-16.28
Water quality
cal./val yr
1999-2002/
1991-1998
1993-2002/
1986-1992
1985-2001/
1972-1984
1992-2002/
1982-1992
1985-2001/
1978-1984
1990-2000/
1980-1990
1984-1994/ND
1993-2000/
1986-1992
1990-1995/
1979-1989
1993-2003/
1983-1993
1993-2003/
1982-2002
1985-2003/
1973-1984
1997-200 I/
1973-1996
1998-2000/ND
1993-2000/ND
TSS monthly
load cal./val.
(% error)
-9/17
16.9/-42.6
66.4/64.1
9.S/-6.6
38/39
67.9/69.8
9.2/NA
9.2/9
59.6/66.8
-9/3.2
-21.8/-3.4
57.3/41
-2/-5S
19/NA
86.6/NA
TP monthly
load cal./val.
(% error)
-50/-30
83.5/31.4
83.2/82.18
-7.4/-S.8
5/-1
23.9/-12.5
2.4/NA
9.3/-21.6
24.2/34.9
9.6/-11.5
8.8/35.1
-46.9/-6S3.98
-8/-33
-14.7/NA
-14/NA
TN monthly
load cal./val.
(% error)
-18/9
-14.4/-15.9
57.3/50.4
-8/-5
56/60
35.8/13.7
-8.9/NA
-8.9/-1.3
28.1/18.1
27.5/26.3
3.9/31.5
-28.3/-909.1
-135/-156
-5.5/NA
6.1/NA
to
-------
Table 4-9. Summary of SWAT model fit for initial calibration site (20 study areas) (continued)
Study area
Susq
TarNeu
Trin
UppCol
Willa
Initial calibration/
validation watershed
Raystown Branch of the
Juniata River
Contentnea Creek
Trinity River
Colorado River
Tualatin River
Initial
calibration/
validation
USGS gage
02050303
02091500
08066500
09070500
14207500
Hydrology
caL/val. yr
1995-2005/
1985-1995
1993-2003/
1983-1993
1992-2001/
1982-1991
1992-2002/
1982-1992
1995-2005/
1985-1995
Total volume
cal./val.
(daily and
monthly E)
0.29/0.42
0.67/0.66
0.68/0.64
0.86/0.74
0.62/0.47
0.74/0.76
0.83/0.78
0.86/0.82
0.49/0.39
0.88/0.81
Total
volume
cal./val.
(% error)
-5.41/16.3
-3.98/-1.18
-6.88/0.70
8.18/0.93
-4.76/-12.1
Water quality
cal./val yr
1991-2000/ 1990
1993-2003/
1983-1993
1985-2001/
1972-1984
1992-2002/ —
ND
1991-1995/
1986-1990
TSS monthly
load cal./val.
(% error)
-10.1/-33.6
-19.9/9.9
9.2/-17.4
0.4/NA
-12/-7
TP monthly
load cal./val.
(% error)
-0.5/-9.2
15.9/5.3
3/-21.58
47.4/NA
-114/-105
TN monthly
load cal./val.
(% error)
28.6/43.9
-5.6/5.3
-3.8/-31.9
15.1/NA
-727-66
to
oo
Table 4-10. Summary of HSPF model fit for initial calibration sites (five pilot study areas)
Study area
ACF
Ariz
Minn
Susq
Willa
Initial calibration/
validation watershed
Upper Flint River
Verde River
Cottonwood River
Raystown Branch of the
Juniata River
Tualatin River
Initial
calibration/
validation
USGS gage
02349605
09504000
05317000
02050303
14207500
Hydrology
caL/val. yr
1993-2002/
1983-1992
1992-2002/
1982-1992
1992-2002/
1982-1992
1995-2005/
1985-1995
1995-2005/
1985-1995
Total volume
cal./val. (daily
and monthly E)
0.71/0.65
0.93/0.90
0.48/0.45
0.85/0.66
0.75/0.78
0.69/0.86
0.70/0.55
0.90/0.87
0.73/0.81
0.96/0.92
Total
volume
cal./val.
(% error)
5.50/5.79
2.43/6.31
1.61/14.78
-0.16/-8.0
-3.92/-9.80
Water quality
cal./val yr
1999-2002/
1991-1998
1993-2002/
1986-1992
1993-2002/
1986-1992
1991-2000/ 1990
1991-1995/
1986-1990
TSS monthly
load cal./val.
(% error)
-117/-78
31/-41
7.5/13.1
-78.2/-89J
3.0/4.8
TP monthly
load cal./val.
(% error)
-S9/-23
87/66
23/15.8
26.0/21.5
-1.2/-9.3
TN monthly
load cal./val.
(% error)
-30/-22
1.6/-2.7
15.4/16.2
7.0/17.2
2.21-63
-------
5. CLIMATE CHANGE AND URBAN DEVELOPMENT SCENARIOS
Watershed simulations were conducted using SWAT and HSPF in each study area to assess the
sensitivity of streamflow, total nitrogen, total phosphorus, and total suspended solids loads to a
range of plausible mid-21st century climate change and urban development scenarios. Climate
change scenarios are based on downscaled climate model projections for mid-21st century from
the NARCCAP and BCSD (Maurer et al., 2007) data sets. Fourteen climate scenarios were
applied to the five pilot sites, and a subset of 6 climate scenarios from the NARCCAP archive
were applied to the nonpilot sites. Scenarios of urban and residential development were based on
projections from EPA's ICLUS project (U.S. EPA, 2009c).
Simulations were conducted to assess the response to climate change scenarios alone, urban and
residential development scenarios alone, and combined climate change and urban development
scenarios. The following sections discuss the use and implementation of climate change and
urban development scenarios in this study.
5.1. SCENARIO-BASED APPROACH
The scientific uncertainties related to our understanding of the physical climate system are large,
and they will continue to be large for the foreseeable future. It is beyond our current capabilities
to predict with accuracy decadal (and longer) climate changes at the regional spatial scales of
relevance for watershed processes (e.g., see Cox and Stephenson, 2007; Stainforth et al., 2007;
Raisanen, 2007; Hawkins and Sutton, 2009; among many others). The uncertainties associated
with socioeconomic trajectories, technological advances, and regulatory changes that will drive
greenhouse gas emissions changes (and land-use changes) are even larger and less potentially
tractable.
Faced with this uncertainty, an appropriate strategy is to take a scenario-based approach to the
problem of understanding climate change impacts on water quality. A scenario is a plausible
description of how the future may develop, based on a coherent and internally consistent set of
assumptions about driving forces and key relationships (IPCC, 2007). Scenarios are used in
assessments to provide alternative views of future conditions considered likely to influence a
given system or activity. By systematically exploring the implications of a wide range of
plausible alternative futures, or scenarios, we can reveal where the greatest vulnerabilities lie.
This information can be used by decision makers to help understand and guide the development
of response strategies for managing climate risk. A critical step in this approach is to create a
number of plausible future states that span the key uncertainties in the problem. The goal is not
to estimate a single, "most likely" future trajectory for each study watershed, but instead to
understand, to the extent feasible, how big an envelope of potential future impacts we are unable
to discount and must therefore incorporate into future planning.
Note that for climate change studies, the word "scenario" is often used in the context of the IPCC
greenhouse gas storylines. The IPCC emissions scenarios describe alternative development
pathways, covering a range of demographic, economic, and technological driving forces that
affect greenhouse gas emissions. This can produce some confusion when phrases like "climate
change scenarios" are used to refer to the future climates simulated using these greenhouse gas
5-1
-------
storylines. For the purposes of this study, "scenario" is a generic term that can be applied to any
defined future, including a climate future or a land-use future, among others.
5.2. CLIMATE CHANGE SCENARIOS
It is standard practice when assessing climate change impacts to consider an ensemble of climate
change scenarios based on different climate models and emissions pathways. Use of a single
model run is not considered scientifically rigorous because different GCMs often produce very
different results, and there is no consensus in the climate modeling community that any model is
comprehensively better or more accurate than the others (e.g., see Gleckler et al., 2008).
Different methods of "downscaling" GCM model output to finer spatial scales can also influence
the variability among models.
5.2.1. Future Climate Models, Sources, and Downscaling
To sample across this model-based uncertainty, this project focused on six climate change
scenarios derived from four GCMs covered by the regional downscaling efforts of the
NARCCAP (http://www.narccap.ucar.edu). NARCCAP uses higher-resolution RCMs to
dynamically downscale output from four of the GCMs used in the IPCC 4th Assessment Report
(IPCC, 2007) to a 50 x 50 km grid over North America. This downscaled output is archived for
the two 30-year periods (1971-2000 and 2041-2070) at a temporal resolution of 3 hours.
NARCCAP uses the IPCC's A2 greenhouse gas storyline (which at the time of development was
a relatively "pessimistic" future greenhouse gas trajectory, but is now more middle-of-the-road
compared to current trends and the most recently developed scenarios). We note that, by mid-
21st Century, the different IPCC greenhouse gas storylines have not yet diverged much, so
impact of the choice of any one particular storyline is diminished compared to later in the
century.
At the time we initiated the watershed modeling, six downscaled scenarios were available from
NARCCAP, and we are using these six as our common set of climate scenarios across all the
20 watersheds, as listed in Table 5-1.
One of the objectives of this work was to investigate the influence of downscaling approaches on
watershed model simulations. To evaluate the sensitivity of our results to downscaling
methodology, we ran the watershed models in the five pilot sites with eight additional scenarios
(also listed in Table 5-1) derived from the same four GCMs used in NARCCAP: four scenarios
interpolated to station locations directly from the GCM output (without downscaling), and four
scenarios based on the BCSD statistically downscaled climate projections described by Maurer et
al. (2007), and served at: http://gdo-dcp.ucllnl.org/downscaled_cmip3_projections/. The BCSD
data provides monthly mean surface air temperature and precipitation rates for the contiguous
United States (along with portions of Canada and Northern Mexico) at a horizontal grid spacing
of 1/8 degree (roughly 12 x 12 km2) for the period 1950-2099.
The BCSD climate projections use statistical downscaling to interpret GCMs to a finer resolution
based on current observations. The principal potential weakness of this approach is an
assumption of stationarity. That is, the assumption is made that the relationship between large-
scale precipitation and temperature and local precipitation and temperature in the future will be
the same as in the past. Thus, the method can successfully account for orographic effects that are
5-2
-------
observed in current data, but not for impacts that might result from the interaction of changed
wind direction and orographic effects. A second assumption included in the bias-correction step
of the BCSD method is that any biases exhibited by a GCM for the historical period will also be
exhibited in simulations of future periods.
Table 5-1. Climate models and source of model data used to develop climate
change scenarios
Scenario #
Climate model(s) (GCM/RCM)
NARCCAP (dynamically downscaled)
1
2
3
4
5
6
CGCM3/CRCM
HadCM3/HRM3
GFDL/RCM3
GFDL/GFDL hi res
CGCM3/RCM3
CCSM/WRFP
GCM (without downscaling)
7
8
9
10
CGCM3
HadCM3
GFDL
CCSM
BCSD (statistically downscaled)
11
12
13
14
CGCM3
HadCM3
GFDL
CCSM
Model Abbreviations:
CGCM3: Third Generation Coupled Global Climate Model
http://www.ee.gc.ca/ccmac-cccma/default.asp?lang=En&n=4A642EDE-l
HadCM3: Hadley Centre Coupled Model, version 3
http://www-pcmdi.llnl.gov/ipcc/model_documentation/HadCM3.htm
GFDL: Geophysical Fluid Dynamics Laboratory GCM
http://www-pcmdi.llnl.gov/ipcc/model_documentation/GFDL-cm2.htm
CCSM: Community Climate System Model
http://www-pcmdi.llnl.gov/ipcc/model_documentation/CCSM3.htm
CRCM: Canadian Regional Climate Model
http://www.ee.gc.ca/ccmac-cccma/default.asp?lang=En&n=4A642EDE-l
RCM3: Regional Climate Model, version 3
http://users.ictp.it/~pubregcm/RegCM3/
HRM3: Hadley Region Model 3
http://precis.metoffice.com/
WRFP: Weather Research and Forecasting Model
http://www.wrf-model.org/index.php
GFDL hi res: Geophysical Fluid Dynamics Laboratory 50-km global atmospheric time slice
http://www-pcmdi.llnl.gov/ipcc/model_documentation/GFDL-cm2.htm
5-3
-------
The BCSD scenarios, while all derived from the A2 climate storyline, do not in all cases use the
output of the exact same GCM run that was used to construct the NARCCAP archive.
Specifically, the BCSD results for the Geophysical Fluid Dynamics Laboratory global climate
model (GFDL) and Third Generation Coupled Global Climate Model (CGCM3) GCMs use
exactly the same GCM output as NARCCAP, but BCSD results for Hadley Centre Coupled
Model, version 3 (HadCMS) and Community Climate System Model (CCSM) use different runs
of the A2 scenario than used by NARCCAP. The HadCMS run used in NARCCAP was a
custom run generated specifically for NARCCAP and has not been downscaled for the BCSD
archive. The CCSM run used in NARCCAP is run number 5, which is not available in the
CMIP3 archive used by BCSD. Instead, BCSD uses the HadCMS run 1 and CCSM run 4 from
the CMIP3 archive for the A2 scenario. As a result, the most direct comparisons between the
NARCCAP and BCSD data sets are for the GFDL and CGCM3 GCM models. However, we
still expect comparisons between NARCCAP and BCSD for the HadCMS and CCSM to provide
useful insights when considered along with the GFDL and CGCM3 comparisons. These
scenarios were evaluated only at the five pilot study areas.
Scenarios for the five pilot sites also examined use of the direct output from the GCM runs used
to drive the NARCCAP downscaling (i.e., no downscaling). Comparison of results from these
scenarios to full dynamical downscaling is expected to inform the accuracy with which simpler
methods can be used to address watershed response. These scenarios were evaluated only at the
five pilot study areas.
Table 5-1 summarizes the climate change scenarios used in this study and also contains a
numbering key for shorthand reference to climate scenarios. For example, climate scenario 2
refers to the HadCMS GCM, downscaled with the Hadley Region Model 3 (HRM3) RCM. All
14 scenarios are applied in the five pilot sites. Only scenarios 1 through 6 are applied for the
nonpilot sites.
5.2.2. Translation of Climate Model Projections to Watershed Model Weather Inputs
Even the 50-km NARCCAP scale is relatively coarse for watershed modeling. In this study,
meteorological time series for input to the watershed models were created using a "change
factor" or "delta change" method (Anandhi et al., 2011). Using this approach, a period of
baseline observed weather data was selected for each study area (to which the watershed models
have been calibrated), and the data series adjusted or perturbed to represent a specific type of
climate change projected by a climate model (i.e., a climate change scenario). The benefits of
the change factor approach include its simplicity, elimination of the need for bias correction, and
ability to create spatially variable climate change scenarios that maintain the observed historical
spatial correlation structure among different watershed locations. Specifically, there is a
tendency for GCMs to generate too many low-intensity events and to under-simulate the
intensity of heavy events (Sun et al., 2006; Dai, 2006). The frequency and duration of large
events can have significant effects on hydrology, pollutant loading, and other watershed
processes. Applying the model-derived change factors to the observed precipitation time series
mitigates this problem. Limitations of this approach include the inability to adjust the number
and timing of precipitation events (e.g., to add precipitation events on dry days), and potential
bias introduced through the selection of an arbitrary historical base period as the template for
5-4
-------
future climate time series. In addition, climate models do not necessarily archive all the
meteorological forcing variables required to run watershed models.
Monthly change factors derived from climate models for each climate change scenario were
calculated by comparing simulated monthly average values for baseline (1971-2000) and future
(2041-2070) climate conditions. It should be noted that the intention is not to simulate the
impacts of change in land use and climate that occurred over the decades from 1971 to 2000.
Rather, the 1971-2000 meteorological data is assumed to provide a static estimate of natural
climate variability under "current" land-use conditions, which are defined by the selection of the
2001 NLCD baseline land cover.
Change statistics from the climate models were interpolated to locations corresponding to each
of the BASINS meteorological stations and SWAT weather generator stations used in the
watershed models. Change factors were used to perturb existing records of hourly observed
precipitation and temperature using the CAT (U.S. EPA, 2009b). CAT permits the sequential
modification of weather records to introduce a number of alterations, each reflecting various
assumptions concerning the regional manifestations of climate change. Precipitation records can
be modified by (1) multiplying all records by an empirical constant reflecting projected climate
change to simulate a shift in total precipitation, applied uniformly to all periods and intensity
classes, (2) selective application of such a multiplier to specific seasons or months, (3) selective
application of the multiplier to a range of months or years within the record, and (4) selective
application of the multiplier to storm events of a specific size or intensity class. Modification
can be iteratively applied to more than one event size class, allowing changes in frequency and
intensity as well as changes in overall volume of precipitation to be represented. Temperature
records can be modified by adding or subtracting a constant to all values in the record, or
selective application to certain months or years within the record.
The third meteorological time series required by the watershed models is PET, which is
calculated based on other meteorological time series as described in Section 5.2.2.3.
The full suite of statistics available to calculate PET using the Penman-Monteith energy balance
method is not available for the statistically downscaled model runs or the nondownscaled GCM
archives. Data availability is summarized in Table 5-2 and assumptions for creating PET time
series in the absence of specific data sets is discussed in Section 5.2.2.3.
It is important to note that using this approach, multiyear climate change scenarios created by
perturbing multiple years of historical weather data are representative of a single, future time
period and do not represent continuous climatic change during this period (i.e., they are not
transient simulations). Instead, the variability in multiyear scenarios created in this way provides
a snapshot of the natural variability in climate based on historical conditions.
5.2.2.1. Temperature Changes
Monthly variations (deltas) to the temperature time series throughout the entire time period were
applied using the BASINS CAT. Monthly adjustments based on each scenario were used and a
modified HSPF binary data (WDM) file was created. The temperature time series were adjusted
based on an additive change using the monthly deltas (temperature difference in Kelvin [K])
calculated from the 2041-2070 to 1971-2000 climate simulation comparison. Beginning with
5-5
-------
the HSPF WDM, an automated script then creates the SWAT observed temperature files (daily
maximum and daily minimum).
Table 5-2. Climate change data available from each source used to develop
climate scenarios
Scenario
#
RCM
GCM
Temp.
Prec.
Dew point
temp
Solar
radiation
Wind
speed
Min
temp.
Max
temp.
Prec. bin
data
NARCCAP RCM-downscaled scenarios
1
2
3
4
5
6
CRCM
HRM3
RCM3
GFDL hi res
RCM3
WRFP
CGCM3
HadCM3
GFDL
GFDL
CGCM3
CCSM
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
n/a
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Driving GCMs of the NARCCAP andBCSD scenarios (Le., no downscaling)
1
8
9
10
CGCM3
HadCM3
GFDL
CCSM
X
X
X
X
X
X
X
X
X
n/a
n/a
X
X
n/a
X
X
X
n/a
X
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
BCSD statistically downscaled scenarios
11
12
13
14
CGCM3
HadCM3
GFDL
CCSM
X
X
X
X
X
X
X
X
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
Note: X indicates data are available; n/a indicates not available.
5.2.2.2. Precipitation Changes
Relative changes in the frequency and intensity of precipitation events associated with climate
change may prove to be more influential in determining future patterns of discharge than changes
in overall (annual, seasonal) precipitation. Appendix C provides a summary review of recent
literature on potential changes in the precipitation regime, including volume and intensity, and
the ability of climate models to simulate these changes.
As a general pattern, warming of the lower atmosphere is projected to lead to a more vigorous
hydrologic cycle, characterized by increases in global precipitation, and proportionally larger
increases in high-intensity precipitation events (Trenberth et al., 2007). Much of the United
States is anticipated to experience an increasing proportion of annual precipitation as larger,
more intense events (Kundzewicz et al., 2007; Groisman et al., 2012). Increasing intensity of
precipitation could increase direct runoff during events and increase nonpoint source loading of
sediment, nutrients, and other pollutants to streams (Gutowski et al., 2008). To ensure that
model simulations embody the most important dimensions of climate change affecting watershed
5-6
-------
response, it is important that climate change scenarios represent potential changes in
precipitation intensity-frequency-duration relationships.
The most rigorous approach to applying the downscaled climate scenario results to modification
of the existing precipitation series would be to undertake a detailed analysis (by month) of the
distribution of precipitation event volumes and intensities. Working on an event basis is
important because many of the existing precipitation time series in the BASINS meteorological
data set are disaggregated from daily totals. However, analyzing volume-event data for each of
the climate scenarios for all the precipitation stations was not feasible and the ability of the
climate models to correctly simulate event durations is suspect.
Using the change factor method, future climate time series are constructed by applying changes
to observed precipitation time series that represent the ratio between historical simulations and
future climate simulations in a given climate model. No modifications were made to the number
of rainfall events in the observed record. The following approach was developed to apply
changes in intensity in the baseline precipitation time series.
Total accumulated precipitation data for different percentile bins (for each station location by
month) were provided by NARCCAP for the dynamically downscaled climate change scenarios.
The data consisted of total simulated precipitation volume (over 30 years) and the 0-25, 25-50,
50-70, and 70-90, and >90 percentile bins of the 3-hour intensity distribution (relative to the
existing intensity distribution). These intensity percentiles yield information on where
precipitation intensification occurs, but represent fixed 3-hour windows, not discrete event
volumes, as required for the CAT program. Most of the climate scenarios showed increases in
precipitation volume in the larger events, while volume in the smaller events remained constant
or decreased. The net effect of this was an increase in the proportion of annual precipitation
occurring in larger events. Analysis of the comprehensive (percentile, total volume) climate
scenario data showed that, for most weather stations, the change in the lower percentiles of the
intensity distribution appeared to be relatively small compared to the changes above the 70th
percentile. However, in some cases (e.g., in Arizona), there is greater change in the 25-50*
percentile bin.
Analyses of observed changes in precipitation during the 20* and early 21st century indicate that
more than half of the precipitation increase has occurred in the top 10 or 5% of events (Karl and
Knight, 1998; Alexander et al., 2006). However, GCMs have been shown to systematically
underestimate the frequency of heavy events in the top few percentages (Trenberth et al., 2003;
Sun et al., 2006; Dai, 2006). Therefore, the top 30% range is selected as a compromise that
accounts for intensification but remains within the general skill of the climate models.
To account for changes in intensity, climate change scenarios were thus created using the delta
method by applying climate change adjustments separately to precipitation events >70*
percentile and events <70* percentile, while maintaining the appropriate mass balance as
described below.
Percentile bin-intensity data were available only for climate scenarios 1 through 6
(RCM-downscaled scenarios). Bin data were not available for climate scenarios 7 through 14
(GCM and statistically downscaled scenarios). Two approaches were developed to account for
5-7
-------
intensification of precipitation, depending on whether precipitation bin data were available.
Each approach is discussed in detail below.
Approach 1: Precipitation Bin Data Are Available
For scenarios where bin data was available (the six NARCCAP scenarios) the following
approach was used. For these data, the change in the volume above the 70* percentile intensity
can be taken as an index of the change in the top 30% of events. At the same time, it is necessary
to maintain mass balance by honoring the predicted relative change in total volume. This can be
accomplished mathematically as follows:
Let the ratio of total volume in a climate scenario (¥2) relative to the baseline scenario volume
(Vi) be given by r = (V^Vi). Further assume that the total event volume (V) can be decomposed
into the top 30% (Vn) and bottom 70% (Vi). These may be related by a ratio s = VH/VL. To
conserve the total volume we must have:
V,=rV, 5-1
Equation 5-1 can be rewritten to account for intensification of the top 30% of events (Vn) by
introducing an intensification parameter, q:
5-2
Substituting for the first instance of VH,I=S VL,\ in eq 5-2 yields:
5-3
In eq 5-3 the first term represents the change in the volume of the lower 70% of events and the
second term the change in the top 30%. This provides multiplicative factors that can be applied
to event ranges using the BASINS CAT program on a month-by-month basis.
The intensification parameter, q, can be calculated by defining it relative to the lower 70% of
values (i.e., from 0 to 70* percentile). Specifically (r - rqs), which represents the events below
the 70th percentile, can be written as the ratio of the sum of the volumes below the 70th percentile
in a climate scenario relative to the sum of the volumes below the 70th percentile for the current
condition:
(C?7o)
7ol 5.4
-------
where (Qio)i and (1370)2 are the sum of the volumes reported up to the 70* percentile for a month
for the current condition and future condition respectively.
Solving eq 5-4 for q yields:
q = (\-Alr)ls 5_5
where A is defined as A =
(670)1
In sum, for each month at each station the following were calculated:
V,
r =
from the summary of the climate scenario output,
s - H/ftT _T/r ^ from the existing observed precipitation data for the station, sorted
/\y ~*H)
into events and postprocessed to evaluate the top 30% (F#) and
bottom 70% (VL) event volumes. The numerator is calculated as
the difference between total volume and the top 30% volume,
rather than directly from VL to correct for analyses in which some
scattered precipitation is not included within defined "events."
The s value was calculated by month and percentile (for every
station, every month) using the observed precipitation time-series
data that forms the template for the delta method representation of
future climate time series.
q = (1 - AIT) IS where A is obtained from the percentile bin climate scenario output
summary
The multiplicative adjustment factors for use in the CAT tool can then be assembled as:
r (1 - qs), for the events below the 70* percentile, and
r (1 + q\ for the events above the 70th percentile.
In addition to the typical pattern of increasing rainfall occurring in large events, this approach is
applicable for the cases in which there is a relative increase in the low-percentile intensities. In
those cases, the change in the 70th percentile intensity is relatively small and tends to be less than
current conditions under the future scenario, resulting in q being a small negative number. In
such cases, application of the method results in a decrease in the fraction of the total volume
5-9
-------
belonging to the larger events, with a shift to the smaller events—thus approximating observed
increases in intensity for smaller events.
In general, it is necessary to have -\ 1 (increasing precipitation), the future volume representing the
climate scenario (Vi can be defined as:
V2=VlL+r • VIH 5_6
where r*is the change applied to the upper range (>30%), VH is the volume in the top 30%, and
VL is the volume in the bottom 70% of events.
Rearranging eq 5-6 and expressing r = r + -(r -1) • IL/VIH , the overall change is satisfied, as:
V =V + r* -V = V + r-V -V + r-V = r (V + V ) = r- V
V1 'IL ^ ' V\H 'IL ^ ' V\H 'IL ^ ' V\L ' \V\H ^ '\L/ ' V\^ 5-7
Further, as r > 7, r is always positive.
For the case of r < 1 (decreasing precipitation), an across-the-board decrease in precipitation was
applied as follows:
V2 = r • V\L + r • V\H S-8
5-10
-------
The adjustment factors can then be assembled as follows:
For the events above the 70th percentile, if
r > 1, then user*
r < 1, then use r.
For the events below the 70th percentile, if
r > 1, then use 1 (no change)
r < 1, then use r.
5.2.2.3. Potential Evapotranspiration Changes
Potential evapotranspiration is an important parameter that is sensitive to climate change and
urban development. In this study, PET is simulated with the Penman-Monteith energy balance
method. In addition to temperature and precipitation, the Penman-Monteith method requires dew
point (or relative humidity), solar radiation, and wind as inputs. Because only a few stations
have time series for all four additional variables that are complete over the entire 1971-2000
period, these variables are derived from the SWAT 2005 statistical weather generator (Neitsch et
al., 2005). This is done internally by SWAT. For HSPF implementation a stand-alone version of
the weather generator code was created and used to create time series for each of the needed
variables at each BASINS meteorological station based on the nearest SWAT weather generator
station after applying an elevation correction.
The SWAT weather generator database (.wgn) contains the statistical data needed to generate
representative daily climate data for the different stations. Adjustments to the wgn file
parameters were made using monthly change statistics for the NARCCAP dynamically
downscaled scenarios. Specifically solar radiation, dew point temperature, and wind speed were
adjusted for each scenario (see Table 5-3).
The probability of a wet day following a dry day in the month and the probability of a wet day
following a wet day in the month were kept the same as in the original SWAT climate generator
file for the station. Climate models showed a systematic bias, likely introduced by the scale
mismatch (between a 50-km grid and a station observation) for weather generator parameters
like wet day/dry day timing, resulting in too many trace precipitation events relative to observed.
Thus it was not possible to use climate models to determine changes in these parameters. Also,
an analysis of the dynamically downscaled 3-hourly time series for the Canadian Regional
Climate Model (CRCM) downscaling of the CGCM3 GCM at five randomly selected locations
in the southeast, southwest, mid-Atlantic, upper Midwest, and Pacific Northwest demonstrated
that the probability that a rainy day is followed by a rainy day (transition probability) in the
model output did not change significantly at any of the sample locations.
For the BCSD climate scenarios, information on these additional meteorological variables is not
available. Many of these outputs are also unavailable from the archived nondownscaled GCM
output. For these scenarios it was assumed that the statistical parameters remained unchanged at
current conditions. While the lack of change is not physically realistic (e.g., changes in rainfall
will be associated with changes in cloud cover and thus with changes in direct solar radiation
5-11
-------
reaching the land surface), this reflects the way in which output from these models is typically
used.
Table 5-3. SWAT weather generator parameters and adjustments applied
for scenarios
SWAT wgn file
parameter
SOLARAV1
DEWPT1
WNDAV1
Description
Average daily solar radiation for month (MJ/m2/day)
Average daily dew point temperature in month (°C)
Average daily wind speed in month (m/s)
Adjustment applied
Adjusted based on Surface Downwelling Shortwave
Radiation change (%)
Additive Delta value provided for climate scenario for
each month
Adjusted based on 10-meter Wind Speed change (%)
Inconsistencies in the available data among different scenarios required special treatment. One
of theNARCCAP scenario archives (Scenario 5: CGCM3 downscaled with regional climate
model, version 3 [RCM3]) does not include solar radiation, which may be affected by changes in
cloud cover. Current condition statistics for solar radiation contained in the weather generator
were used for this scenario. This does not appear to introduce a significant bias as the resulting
changes in PET fall within the range of those derived from the other NARCCAP scenarios.
Table 5-4 compares the reference crop estimates of Penman-Monteith PET for the five pilot
watersheds. This is the PET used directly by the HSPF model, while the SWAT model performs
an identical calculation internally, and then adjusts actual evapotranspiration (AET) for crop
height and leaf area development. Because PET is most strongly a function of temperature, a
fairly consistent increase in PET is simulated for most basins. It can be seen from the figures in
Appendix Z, however, that the statistically downscaled and nondownscaled GCM scenarios
(scenarios 7-14) that do not include solar radiation, dew point, and wind time series consistent
with the simulated precipitation and temperature, generally provide higher estimates of PET than
do the dynamically downscaled models. This issue is explored in more detail in Section 6.2.
5.3. URBAN AND RESIDENTIAL DEVELOPMENT SCENARIOS
Watershed simulations were also conducted to assess the sensitivity of study areas to potential
mid-21st century changes in urban and residential development.
5-12
-------
Table 5-4. Comparison of PET estimation between different downscaling approaches
Scenario type
Climate scenario
ACF
(GA, AL, FL)
Minnesota River
(MN, SD)
Salt/Verde/San
Pedro (AZ)
Susquehanna
(PA, NY, MD)
Willamette (OR)
annual average PET
(in)
difference from
NARCCAP mean
annual average PET
(in)
difference from
NARCCAP mean
annual average PET
(in)
difference from
NARCCAP mean
annual average PET
(in)
difference from
NARCCAP mean
annual average PET
(in)
difference from
NARCCAP mean
NARCCAP dynamically
downscaled
1. CRCM-CGCM3
60.32
1.46%
58.57
2.92%
83.67
0.47%
43.78
1.79%
44.18
-0.37%
5.RCM3-
CGCM3
58.59
-1.46%
55.24
-2.92%
82.89
-0.47%
42.24
-1.79%
44.51
0.37%
Nondownscaled
GCM
7. CGCM3
59.85
0.67%
56.22
-1.21%
84.19
1.09%
42.91
-0.23%
45.24
2.01%
BCSD
statistically
downscaled
11. CGCM3
64.75
8.90%
63.90
12.29%
85.01
2.07%
51.15
18.94%
50.73
14.41%
NARCCAP dynamically
downscaled
3. RCM3-GFDL
60.46
2.81%
54.92
-4.44%
81.32
-0.98%
43.06
0.43%
45.44
1.70%
4. GFDL
(high res)
57.16
-2.81%
60.02
4.44%
82.93
0.98%
42.69
-0.43%
43.91
-1.70%
Nondownscaled
GCM
9. GFDL
67.88
15.42%
64.99
13.08%
86.73
5.60%
50.18
17.05%
49.16
10.04%
BCSD
statistically
downscaled
13. GFDL
65.97
12.17%
63.65
10.75%
84.74
3.18%
50.17
17.02%
49.17
10.06%
-------
5.3.1. ICLUS Urban and Residential Development Scenarios
Projected changes in urban and residential development were acquired from EPA's ICLUS
project (U.S. EPA, 2009c). ICLUS has produced seamless, national-scale change scenarios for
developed land that are compatible with the assumptions about population growth, migration,
and economic development that underlie the IPCC greenhouse gas emissions storylines. ICLUS
projections were developed using a demographic model coupled with a spatial allocation model
that distributes the population as housing units across the landscape. Specifically, population is
allocated to 1-hectare (ha) pixels, by county, using the Spatially Explicit Regional Growth Model
(SERGoM). The model is run for the conterminous United States and output is available for
each emissions storyline by decade to 2100. The final spatial data sets provide decadal
projections of housing density and impervious surface cover as a function of population for the
period 2000 through 2100 (U.S. EPA, 2009c).
Data from the ICLUS project are composed of grid-based housing density estimates with 100-m
cells, whose values are set equal to units/ha x 1,000. Existing housing densities were estimated
using a variety of sources and models, and future housing densities developed under various
scenarios for each decade through 2100. For the existing housing density grid, two types of
"undevelopable" area where residential development was precluded were masked out during the
production—a comprehensive spatial data set of protected lands (including land placed in
conservation easements), and land assumed to be commercial/industrial under current conditions.
Undevelopable commercial/industrial land use was masked out according to the SERGoM
method (U.S. EPA, 2009c) that eliminated commercial, industrial, and transportation areas that
preclude residential development, identified as "locations (1-ha cells) that had >25%
urban/built-up land cover with lower than suburban levels of housing density."
The ICLUS projections used in this study thus do not account for potential growth in
commercial/industrial land use. It is also important to note that the ICLUS projections do not
explicitly account for changes in rural or agricultural land uses. These categories change in the
analysis based on ICLUS only when they convert to developed land.
5.3.2. Mapping ICLUS Housing Density Projections to NLCD Land Use Categories
The ICLUS projections used in this study are for changes in housing density and impervious
cover. This data cannot be used directly with the SWAT and HSPF watershed models, which
require land use data consistent with the NLCD. It was therefore necessary to translate between
ICLUS projections and NLCD land-use classes.
In addition, ICLUS housing density class estimates and the NLCD developed classes do not have
a one-to-one spatial relationship because they are constructed on different underlying scales.
ICLUS represents housing density based largely on the scale of census block groups. As a result,
it represents the overall density within a relatively large geographic area when compared to the
30x30 meter resolution of NLCD 2001 land cover and can represent a mix of different NLCD
classes. Therefore, land-use changes must be evaluated on a spatially aggregated basis at the
scale of model subbasins.
Baseline land use, derived from the 2001 NLCD, contains four developed land classifications
(NLCD classes 21 through 24), nominally representing "developed, open space" (less than 20%
5-14
-------
impervious), developed, low intensity (20-49% impervious), developed, medium intensity
(50-79% impervious), and developed, high intensity (greater than 80% impervious). Impervious
fractions within each developed NLCD land-use class were estimated separately for each study
area, using the 2001 NLCD Land Cover and Urban Impervious data products. ICLUS land-use
change scenarios were implemented by modifying the existing land-use distribution in the
watershed models.
ICLUS estimates housing density on a continuous scale. To process the data more efficiently,
the data were reclassified into 10 housing density ranges. In each study area, the ICLUS housing
density ranges were cross-tabulated with NLCD 2001 classes based on percent imperviousness.
It was assumed that the number of housing units changes, but that the characteristic percent
impervious values for each NLCD developed class remains constant. The change in land area
needed to account for the change in impervious area was then back calculated.
To represent the net change in future land cover, the change in developed land use was added (or
subtracted) from the existing totals in each subbasin. Land area was then removed from each
undeveloped NLCD class (excluding water and wetlands) according to their relative ratios in
each subbasin to account for increases in developed area. If the undeveloped land area was not
sufficient to accommodate the projected growth, development on wetlands was allowed. The
reductions in undeveloped land were distributed proportionately among modeled soils (in
SWAT) or hydrologic soil groups (in HSPF). The new developed lands were then assumed to
have the parameters of the most dominant soil and lowest HRU slope in the subbasin. For
HSPF, the changed area was implemented directly in the area table of the user control input
(.uci) file. For SWAT, the land-use change was implemented by custom code that directly
modified the SWAT geodatabase that creates the model input files.
The gains (and losses) in NLCD class interpreted from ICLUS were tabulated separately for each
subbasin. In almost every case, the gains far exceeded the losses and a net increase was
projected in all four NLCD developed classes. However, in a few cases there was an overall loss
of the lowest density NLCD class. This tended to occur when a subbasin was already built out,
and ICLUS projected redevelopment at a higher density.
The projected overall changes in developed land for 2050 as interpreted to the NLCD land-cover
classes and used for modeling are presented in Table 5-5. Note that even in areas of expected
high growth (e.g., the area around Atlanta in the ACF basin), new development by 2050 is
expected to constitute only a small fraction of the total watershed area at the scale of the study
areas in this project. The highest rate of land-use change in the studied watersheds is Coastal
Southern California, at 11.7%. (Note that the ICLUS project does not cover the Cook Inlet
watershed in Alaska. Urban and residential development scenarios were thus not evaluated at
this study area.)
5-15
-------
Table 5-5. ICLUS projected changes in developed land area within different
imperviousness classes by 2050
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
Change,
<20%
impervious
class (km2)
+665.2
+92.1
ND
+873.9
+353.5
+152.1
+307.2
+71.3
+8.9
+238.6
+1.3
+139.0
+103.6
+162.0
+329.4
+211.1
+492.4
+978.9
+56.9
+75.8
Change,
20-49%
impervious
class (km2)
+809.7
+87.0
ND
+776.1
+1,506.6
+204.8
+308.3
+142.9
+18.7
+327.2
+0.5
+228.8
+58.1
+1,001.0
+1,364.6
196.2
+306.6
+1,896.7
+168.1
+193.4
Change,
50-79%
impervious
class (km2)
+212.3
+16.0
ND
+361.5
+447.5
+51.0
+91.4
+60.9
+4.1
+215.5
+0.1
+57.1
+29.5
+1,089.1
+473.5
+69.6
+107.4
+891.1
+66.3
+95.0
Change,
>80%
impervious
class (km2)
+90.8
+1.3
ND
+102.2
+116.2
+15.6
+23.4
+18.5
+1.6
+59.2
0.0
+7.4
+8.2
+114.1
+83.6
+25.6
+29.2
+304.3
+8.3
+33.3
Total change in
developed land
(km2)
+1,778.0
+196.4
ND
+2,113.8
+2,424.0
+423.4
+730.1
+293.5
+33.2
+840.4
+1.9
+432.4
+199.3
+2,466.2
+2,251.1
+502.5
+935.6
+4,071.0
+299.6
+397.6
Increase as
percent of
study area (%)
+3.56
+0.51
ND
+4.65
+5.50
+1.40
+4.82
+0.67
+0.06
+3.13
+0.00
+0.88
+0.93
+11.72
+5.93
+0.71
+3.66
+8.76
+0.65
+1.37
Note: The ICLUS project does not cover the Cook Inlet watershed. Results shown are total new developed area,
including pervious and impervious fractions.
5-16
-------
6. STREAMFLOW AND WATER QUALITY SENSITIVITY TO DIFFERENT
METHODOLOGICAL CHOICES: ANALYSIS IN THE FIVE PILOT STUDY
AREAS
One goal of this study was to assess the implications of different methodological choices for
conducting climate change impacts assessments on the variability of simulation results.
Sensitivity studies in the five pilot study areas allow assessment of the variability resulting from
the use of different watershed models, and variability resulting from use of climate change
scenarios developed using different methods of downscaling GCM output. The five pilot study
areas are the Minnesota River, ACF, Susquehanna, Willamette, and Salt/Verde/San Pedro
Rivers. In each of these sites, independent simulations were conducted using the SWAT and
HSPF watershed models, and in addition to the six dynamically downscaled NARCCAP
scenarios, an additional set of climate change scenarios was evaluated, four based on the BCSD
statistically downscaled data set, and four based directly on GCMs with no downscaling. This
section presents a summary of these results.
6.1. COMPARISON OF WATERSHED MODELS
The magnitude of the additional variability introduced by choice of a hydrologic model is of
interest when simulating hydrologic responses to climate change and urban development. Two
different watershed models, SWAT and HSPF, were calibrated and applied to the five pilot study
areas. Evaluation of different watershed models can be considered an extension of the
scenario-based, ensemble approach commonly used in climate change studies. Detailed
examination of the calibration of each model in the five pilot study areas and the results of
change scenarios conducted with each model are presented in separate sections and the
appendices to this report.
HSPF and SWAT take different approaches to watershed simulation and have different structures
and algorithms, resulting in different strengths and weaknesses. Most notably, the two models
differ in the way that they represent infiltration and plant-climate interactions. SWAT (in
standard application mode) simulates rainfall-runoff processes using a curve number approach,
operating at a daily time step. The curve number approach first partitions incoming moisture
into direct runoff and a remainder that is available for infiltration. In contrast, HSPF simulates
rainfall-runoff processes using Green-Ampt infiltration, in which infiltration into the soil is
simulated first, with the remainder available for direct runoff or surface storage.
HSPF is typically run at a subdaily time step, usually hourly for large watersheds, and has a more
sophisticated representation of runoff, infiltration, and channel transport processes than does
SWAT. SWAT's advantage is that it incorporates a plant growth model (including
representation of changes in atmospheric CC>2 concentration) and can therefore simulate some of
the important feedbacks between plant growth and hydrologic response. Both models simulate
evapotranspiration of soil water stores, but HSPF does this using empirical monthly coefficients
relative to potential evapotranspiration, while SWAT incorporates a plant growth model that can,
in theory, dynamically represent plant transpiration of soil moisture.
6-1
-------
6.1.1. Comparison of Model Calibration and Validation Performance
Models were calibrated and validated using multiple measures as summarized previously in this
report and described in detail in Appendices D-W. Calibration of both models was conducted in
accordance with the modeling QAPP (see Appendix B; Tetra Tech, 2008a) for each of the five
pilot study areas. Development and setup of the two watershed models proceeded from a
common basis, with both models using the same subbasin delineations, land use coverage (2001
NLCD), soils coverage (STATSGO), hydrography, digital elevation model, impervious area
fractions for developed land classes, and point source and dam representations. Other aspects of
model setup were designed to be similar, although it was not possible to be identical because of
differences in the way the two models conceptualize discretization of the land surface. For
instance, hydrologic response units (the fundamental building blocks of the upland simulation)
were created as an overlay of land use and HSG for HSPF, while SWAT uses an overlay of land
use and STATSGO dominant soil, associating various other properties from the soil database in
addition to HSG with the model hydrologic response units. In addition, HSPF simulates
impervious surfaces as a separate land use, while SWAT assigns an impervious fraction to an
underlying land use.
Calibration/validation locations and observed data series were the same for both models.
Further, the calibration of both models was guided by prespecified statistical analyses that were
performed using identical spreadsheet setups obtained from a common template. Despite these
commonalities, the scope of the modeling effort in this study required that models be developed
by different modeling teams, with inevitable differences in results. To reduce the likelihood of
bias, model calibration assignments were structured so that the same team did not apply both
HSPF and SWAT to a single study area, and each watershed model was implemented by at least
three different modeling teams for the pilot studies.
6.1.1.1. Streamflow Results
This section examines hydrologic simulations as compared to observed streamflow records based
on total volume error and the daily Nash-Sutcliffe coefficient of model fit efficiency. Model
performance is first examined in terms of the quality of fit for the initial calibration watershed,
followed by similar analyses for the largest-scale downstream watershed. Intercomparisons then
provide some insight into model performance relative to temporal change (calibration vs.
validation period) and relative to spatial change within each study area (calibration watershed vs.
downstream watershed).
Summary results for percent error in total volume and the Nash-Sutcliffe E coefficient for daily
streamflow are shown in Tables 6-1 and 6-2, respectively, for the initial calibration site along
with the calibration fit for the most downstream gage in the watershed. In general, the quality of
model fit is good for both models. In most, but not all cases, the quality of model fit is slightly
better (smaller magnitude of percent error, larger E coefficient) for the HSPF simulations (e.g.,
see Figure 6-1 for the calibration period). This is likely due in large part to the use of daily
precipitation in SWAT versus hourly precipitation in HSPF, although the advantage accruing to
HSPF is muted by the fact that many of the "hourly" precipitation input series used are actually
disaggregated from daily totals. Monthly values of Nash-Sutcliffe E are higher for both models,
but attention is called to the daily scale because it better reflects the models' ability to separate
surface and subsurface flow pathways. Note that E is low for the Arizona initial site on the
6-2
-------
Verde River because streamflow is dominated by relatively constant deep groundwater
discharges.
Table 6-1. Percent error in simulated total streamflow volume for 10-year
calibration and validation periods at initial and downstream calibration
gages
Study area
Apalachicola-
Chattahoochee-Flint (ACF)
Salt/Verde/San Pedro (Ariz)
Minnesota River (Minn)
Susquehanna (Susq)
Willamette (Willa)
Model
HSPF
SWAT
HSPF
SWAT
HSPF
SWAT
HSPF
SWAT
HSPF
SWAT
Initial site calibration
5.50
7.28
2.43
-2.46
1.61
-5.41
-0.16
-5.41
-3.92
-4.76
Initial site validation
5.79
3.33
6.31
5.68
14.78
-0.84
-8.00
-16.30
-9.80
12.10
Downstream calibration
16.79
16.53
4.48
9.43
-4.25
7.89
1.79
-9.74
2.58
-4.96
Table 6-2. Nash-Sutcliffe coefficient of model fit efficiency (E) for daily
streamflow predictions, 10-year calibration and validation periods at initial
and downstream calibration gages
Study area
Apalachicola-Chattahoochee-
Flint (ACF)
Salt/Verde/San Pedro (Ariz)
Minnesota River (Minn)
Susquehanna (Susq)
Willamette (Willa)
Model
HSPF
SWAT
HSPF
SWAT
HSPF
SWAT
HSPF
SWAT
HSPF
SWAT
Initial site calibration
0.71
0.62
0.48
0.03
0.75
0.79
0.70
0.29
0.80
0.49
Initial site validation
0.65
0.56
0.45
-1.00
0.78
0.74
0.55
0.42
0.81
0.39
Downstream calibration
0.72
0.64
0.53
0.22
0.92
0.63
0.77
0.45
0.88
0.67
6-3
-------
• HSPF
• SWAT
Ariz
Susq
Willa
• HSPF
• SWAT
ACF CeAZ MNRiver Susq Willm
Figure 6-1. Comparison of model calibration fit to streamflow for the
calibration initial site.
Note: Figures compare calibration results for HSPF and SWAT. Total volume error is converted to its absolute value.
The ability of the model to assess relative changes in response to altered climate forcing is of
paramount importance in this project. Some insight on this topic can be gained by looking at the
sensitivity of model fit to temporal and spatial changes in application. Figure 6-2 summarizes
the sensitivity to temporal changes by examining the percent error in the calibration period and
the validation test. It is interesting to observe that for both the ACF and the Minnesota River, the
SWAT model achieved an improvement in total volume error during the validation period.
These are the two study areas with the greatest amount of row crop agriculture, and the results
may reflect SWAT's ability to reflect changing responses of crops to changes in climate over the
last 20 years.
Figure 6-3 examines model sensitivity to spatial scale, comparing performance during the
calibration period for the initial calibration target gage (HUC-8 spatial scale) and the most
downstream gage in the model (approximately HUC-4 spatial scale). The left panel shows the
change in the absolute magnitude of percent error, while the right panel shows the change in E.
A smaller magnitude of change in total volume error or a larger increase in E represents better
performance. The changes in total volume errors are generally small, regardless of whether
detailed spatial calibration was pursued. In most cases, the models achieved an improvement in
E in going from the smaller to the larger scale.
6.1.1.2. Water Quality Results
The water quality calibration compared simulated monthly loads to monthly load estimates
obtained from a stratified regression on (typically sparse) observed data. To compare these
results between models, the baseline adjusted E\' coefficient of model fit efficiency is most
appropriate. Results are summarized graphically for the calibration period at the calibration
initial site and downstream site in Figures 6-4 through 6-6. For suspended solids and total
phosphorus, the performances of the two models are similar, while HSPF appears to provide a
somewhat better fit for total nitrogen.
6-4
-------
HSPF-Cal
HSPF-Val
SWAT-Cal
SWAT-Val
-20
ACF Ariz Minn Susq
Study Area
Willa
Figure 6-2. Sensitivity of model fit for total streamflow volume to temporal
change.
u
I1"
I8
r
= 3
I.
J.J
ACf Aril Minn
• HSPT
• SWAT
O.K
0.2
O.lb
0.1
0.0-j
0
•o.os
-0.1
-0.15
0.2
JUJ
Acr
Susq
Willj
• HSPf
• SWAT
Figure 6-3. Sensitivity of model fit for streamflow to spatial change.
Note: Change in percent total volume error represents the difference in the absolute value of percent error in going from the
initial calibration site to a larger scale, typically the furthest downstream site. Change in E represents the difference in the Nash-
Sutcliffe E coefficient in going from the calibration site to the larger-scale site.
6-5
-------
• HSPF
• SWAT
ACF Ariz Minn Susq Willa
• HSPF
• SWAT
ACF Ariz Minn Susq Willa
Figure 6-4. Comparison of baseline adjusted model fit efficiency for total
suspended solids monthly loads for calibration site (left) and downstream site
(right).
• HSPF
• SWAT
Minn Susq Willa
Ariz Minn Susq Willa
• HSPF
• SWAT
Figure 6-5. Comparison of baseline adjusted model fit efficiency for total
phosphorus monthly loads for calibration site (left) and downstream site
(right).
• HSPF
• SWAT
Minn Susq Willa
Minn Susq Willa
• HSPF
• SWAT
Figure 6-6. Comparison of baseline adjusted model fit efficiency for total
nitrogen monthly loads for calibration site (left) and downstream site (right).
6-6
-------
6.1.1.3. Summary of Relative Model Performance
In general, the HSPF model provides a somewhat better fit to observed streamflow and water
quality data for the calibration periods. The effect is most noticeable in the coefficient of model
fit efficiency (£) for daily streamflow, where the HSPF approach of applying Philip infiltration
using hourly precipitation appears to yield an advantage over the SWAT daily curve number
method. However, relative performance of the two models is more similar as the analysis moves
to the validation period or to other sites for which detailed calibration has not been undertaken.
Most importantly, both models appear to be capable of performing adequately.
6.1.2. Comparison of Simulated Changes Using SWAT and HSPF
Figure 6-7 compares HSPF and SWAT simulated changes in mean annual streamflow at the
downstream station of each of the five pilot watersheds for all 28 combinations of climate and
land-use change scenarios (expressed as a percent of the baseline conditions, representing
approximately 1970-2000). In general, the mean annual streamflow results provided by the two
models are similar, as is shown quantitatively below. One notable difference is for the
Minnesota River where SWAT projects higher flows relative to HSPF under future climate
conditions—an issue that is explored further in Section 6.1.3. Note that points plotting close to
or on top of each other for a given study site in Figure 6-7 are scenarios representing the same
climate change scenario with and without changes in urban development.
200%
+ ACF
• Ariz
AMinn
XSusq
XWilla
50% 100% 150%
HSPF Total Flow Volume - Percent of Baseline
200%
Figure 6-7. SWAT and HSPF simulated changes in total streamflow in pilot
watersheds (expressed relative to current conditions).
6-7
-------
Table 6-3 provides a statistical comparison of the HSPF and SWAT results at the downstream
station. Three types of tests are summarized. The first is a t-test on the series of paired means
(HSPF and SWAT for each climate and land use scenario), which has a null hypothesis that the
mean of the differences between the series is not significantly different from zero. The second
test is a two-way analysis of variance (ANOVA) that looks at choice of watershed model (HSPF
or SWAT) as blocks and climate scenario as treatment. The null hypotheses for this test are that
the difference between series for a given source of variance is zero. The third test is a linear
regression on SWAT results as a function of HSPF results. Where the models are in full
agreement, the intercept of such a regression should not be significantly different from zero and
the slope should not be significantly different from unity.
For mean annual streamflow, both models produce similar results with a high Pearson correlation
coefficient. The null hypothesis from the Mest that the mean difference is zero cannot be
rejected. However, the two-way ANOVA shows that both the choice of watershed model and
the climate scenario are significant sources of variability in streamflow, with probability values
(p-value) well less than 0.1. Together these results suggest that the SWAT and HSPF results are
similar in the aggregate, but may contain an underlying systematic shift. A regression analysis
shows that the slope coefficient for SWAT and HSPF is 0.93, with a 95% confidence interval
that does not overlap 1.0, and an intercept of 1,262 that also does not overlap zero. Thus, SWAT
projects a somewhat smaller response to increased rainfall, but results in higher baseflow
estimates (likely due to the effects of increased CC>2 on evapotranspiration, as explained further
below).
Table 6-3. Statistical comparison of HSPF and SWAT outputs at
downstream station for the five pilot sites across all climate scenarios
Measure
Mean annual flow (cfs)
TSS load (t/yr)
TP load (t/yr)
TN load (t/yr)
Paired f-test on sample means
HSPF Mean
SWAT Mean
Pearson Correlation
t- statistic
p (two-tail)
20,546
20.435
0.989
0.616
0.539
2,398,714
2,865,178
0.733
-3.123
0.002
2,748
3,344
0.644
-4.783
O.001
35,346
43,275
0.948
-7.385
O.001
Two-way ANOVA on watershed model and climate scenario
p value — Model
p value — Climate
0.001
O.001
0.071
0.960
0.006
0.999
0.044
1.000
Linear regression; SWAT result as a function of HSPF result
Intercept
Intercept, 95%
confidence
Coefficient (slope)
Coefficient (slope)
95% confidence
1,261.7
695-1,828
0.933
0.911-0.956
141,717
-363,064-646,498
1.136
0.964-1.307
954.0
431-1,477
0.870
0.702-1.038
-1,173.1
-4,194-1,848
1.257
1.189-1.326
6-8
-------
The comparison for total suspended solids is obscured by the extremely large projected increases
under certain scenarios for the Arizona basins (Verde River, in this case). Those increases are
mostly due to simulated channel erosion, for which both models are likely to be highly uncertain
because future simulated peak flows are outside the range of calibration data. Figure 6-8 shows
the simulated total suspended solids results but with the x-axis truncated to exclude these
extreme results for the Verde River. Results for the other four pilot sites appear generally
consistent between models, although simulated increases from SWAT are generally less than
those from HSPF for the ACF, Susquehanna, and Willamette. In part this is due to differences in
the baseline simulation. For example, HSPF simulations show less channel transport and much
smaller total suspended solids loads at the mouth of the Susquehanna than does SWAT for the
baseline scenario, resulting in a larger relative change with increased future streamflow. The
difference between results for SWAT and HSPF may also reflect the effects of increased
atmospheric CO2 concentration and longer growing periods simulated by SWAT, leading to
more litter cover and reduced soil erosion.
400%
• Ariz
AMinn
XSusq
XWilla
0%
0% 100% 200% 300%
HSPF TSS Load - Percent of Baseline
400%
Figure 6-8. SWAT and HSPF simulated changes in TSS at downstream
station in pilot watersheds (expressed relative to current conditions).
Note: HSPF simulation for climate scenarios 9 (GFDL, nondownscaled GCM), 10 (CCSM, nondownscaled GCM), 12 (HadCMS,
BCSD), and 13 (CCSM, BCSD) yield increases in simulated total suspended solids load of greater than 400% and are omitted
from this plot.
For total suspended solids, the baseline load is higher in SWAT than in HSPF for three of the
five watersheds; thus the statistical comparison (see Table 6-3) shows a higher mean load from
6-9
-------
SWAT, even though the percentage increases are often smaller. The Mest on means shows that
this difference is highly significant. However, the ANOVA show that neither the model choice
nor the climate scenario is a significant explanatory variable for the variance at the 95%
confidence level. The regression analysis shows that the intercept is large, but not significantly
different from zero, while the slope is not significantly different from 1. Together these statistics
indicate that the total suspended solids simulation is subject to considerable uncertainty and that
differences between sites are more important than other factors.
Results for total phosphorus are generally similar to those seen for total suspended solids, with
much more extreme increases projected by both models for the Verde River (Ariz; see Figure
6-9). HSPF simulations are especially high due to an assumption of phosphorus concentrations
in scoured channel sediment. SWAT tends to simulate higher rates of increases for total nitrogen
(see Figure 6-10) than does HSPF (likely due to more rapid cycling of organic matter), with the
notable exception of the ACF study area. However, it appears that projections of total nitrogen
at the downstream end of the ACF may be significantly underestimated in the calibrated SWAT
model. Total nitrogen varies little in the Susquehanna model due to small changes in streamflow
and significant point source contributions.
For both total nitrogen and total phosphorus the choice of model is a significant factor in the
ANOVA and higher mean loads are produced by SWAT. The slope of a regression of SWAT on
HSPF is not significantly different from 1 for total phosphorus, consistent with the solids
simulation, but the intercept is significantly different from zero, indicating differences in the
baseflow simulation of total nitrogen. For total nitrogen, the intercept is not significantly
different from zero, but the slope is significantly greater than 1, suggesting that SWAT projects a
greater increase in total nitrogen loads under future climate conditions.
In sum, the comparison of relative response to change scenarios indicates that the two models
provide generally consistent results for hydrology, with differences that may be in part due to the
inclusion of explicit representation of several processes in SWAT (increased atmospheric CC>2,
changes in planting time, changes in crop growth and litter production, and changes in nutrient
recycling rates) that are not automatically included in HSPF. Water quality results exhibit
greater variability between the models, due in large part to the uncertainty inherent in model
calibration.
An additional contributing cause to differences in results from the two models is the extent to
which spatial calibration of the model was pursued, which was left to modeler judgment. In all
study areas, initial calibration and validation was pursued at an "initial calibration" gage and
monitoring station at an HUC-8 spatial scale. The calibration results were then carried to the
larger study area. At this point, individual modeler preferences introduced some variability into
results. Some modelers undertook detailed spatial adjustments to parameters; others extended
the initial parameter set with only minor modifications. With more spatial adjustments a higher
degree of fit is generally to be expected for model calibration—although this does not necessarily
result in better performance in model validation. In general, only limited spatial calibration
adjustments beyond the initial parameter set was carried out for the Minnesota River,
Susquehanna, and Willamette SWAT models and also for the Susquehanna HSPF model.
6-10
-------
200%
*ACF
• Ariz
AMinn
XSusq
XWilla
0%
0% 50% 100% 150% 200%
HSPF Total Phosphorus Load - Percent of Baseline
Figure 6-9. SWAT and HSPF simulated changes in total phosphorus load in
pilot watersheds (expressed relative to current conditions).
Note: 22 HSPF simulations for Ariz ranging from 200 to 875% are omitted.
Due to the potential influence of modeler choice and skill, it is cautioned that the results should
not be interpreted as a true head-to-head comparison of the two models, as the results for any
given watershed may be skewed by exogenous factors such as modeler calibration strategy.
Instead, it is most relevant to examine relative performance and potential inconsistencies
between simulations using the two models.
6.1.3. Sensitivity to Increased Atmospheric
A key difference between HSPF and SWAT is that SWAT has a dynamic plant growth module
with ability to represent changes in atmospheric CC>2 on plant growth and water loss to ET. We
performed paired sets of SWAT simulations with and without increased CC>2 for all five pilot
sites to assess the sensitivity of streamflow and water quality endpoints to the effects of
increased atmospheric CO2 concentrations.
6-11
-------
200%
cu
c
ro
CO
M—
2 150%
c
cu
£
cu
a.
g 100%
c
cu
- 50%
to
0%
• Ariz
AMinn
XSusq
XWilla
0% 50% 100% 150% 200%
HSPF Total Nitrogen Load - Percent of Baseline
Figure 6-10. SWAT and HSPF simulated changes in total nitrogen load in
pilot watersheds (expressed relative to current conditions).
IPCC estimates of future atmospheric CO2 concentrations under the assumptions of the A2
emissions scenario (the basis of climate and land-use change scenarios in this study) call for an
increase from 369 ppmv CC>2 in 2000 to about 532 ppmv (using the ISAM model reference run)
or 522 ppmv (using the Bern-CC model reference run) in 2050 (Appendix II in IPCC, 2001).
Plants require CC>2 from the atmosphere for photosynthesis. An important effect of increased
atmospheric CC>2 is a reduction in the time plant leaf stomata must be open to obtain the CC>2
needed for growth, resulting in reduced water loss as transpiration (Leakey et al., 2009; Cao et
al., 2010; Ainsworth and Rogers, 2007). This effect can potentially counterbalance projected
increases in transpiration associated with increased air temperatures. It may also reduce water
stress on plants, resulting in greater biomass and litter production, which in turn will influence
pollutant loads.
In the past it has been argued that these effects, long documented at the leaf and organism level,
might not translate to true ecosystem effects. However, recent research, particularly results from
the Free-Air CC>2 Enrichment (FACE) experiments (Leakey et al., 2009) suggests that significant
reductions in evapotranspiration do occur at the ecosystem level with increased atmospheric CC>2
concentrations. Although there are differences in responses among plant species, with lesser
effects with C^ photosynthesis, the magnitude of the response to CC>2 levels projected by the
mid-21st century appears to be on the order of a 10% reduction in evapotranspiration response
(e.g., Bernacchi et al., 2007). Further, a recent study by Cao et al. (2010) suggests that up to
6-12
-------
25% of the temperature increase projected for North America could result directly from
decreased plant evapotranspiration under increased CC>2 concentrations.
SWAT includes a plant growth module that accounts for the effects of changes in atmospheric
CC>2 concentration on stomatal conductance using the equation developed by Easterling et al.
(1992). Using this approach, increased CC>2 leads to decreased leaf conductance, which in turn
results in an increase in the canopy resistance term in the PET calculation. The model also
simulates the change in radiation use efficiency of plants as a function of CO2 concentration
using the method developed by Stockle et al. (1992). Figure 6-11 shows the differences between
projected mid-21st century streamflow and water quality endpoints in the five pilot sites
simulated using SWAT with and without representation of the effects of increased atmospheric
CC>2 concentrations (SWAT projections for the six NARCCAP climate scenarios incorporating
the ICLUS future land use for each watershed). These simulations suggest increases in mean
annual streamflow from 3 to 38% due to increased CO2, with a median of 11%, in the same
range as the results summarized by Leakey et al. (2009). Simulations also suggest increased
atmospheric CC>2 results in increased pollutant loads. Total suspended solids loads show
increases from 3 to 57%, with a median of 15%. Total phosphorus loads increase from 0 to 29%,
with a median of 6%. Total nitrogen loads increase from zero to 34%, with a median of 6%.
The large increases in total suspended solids loads indicate that the effects of higher runoff under
increased atmospheric CO2 (largely due to greater soil moisture prior to rainfall events) may
outweigh benefits associated with greater ground cover—a finding that could have important
land management implications in the midwestern watersheds, including many of the Great Lakes
drainages. For the nutrients, the simulated load increases are less than for streamflow and total
suspended solids increases. This presumably is due to the fact that increased atmospheric CO2
concentrations allow greater plant growth per unit of water, resulting in greater uptake and
sequestration of nutrients, and thus smaller increases in nutrient loads relative to streamflow and
total suspended solids.
The response to increased atmospheric CO2 concentration varies greatly by study area, with the
greatest effect simulated by SWAT for the Minnesota River basin and the smallest effect for the
Willamette basin. The large effect in the Minnesota River basin apparently occurs because the
land in this basin is predominantly in high-biomass corn-soybean rotation agricultural cropland
with precipitation and evapotranspiration in approximate balance. In contrast, the Willamette
basin is dominated by evergreen forest and has a moisture surplus for much of the year.
Ficklin et al. (2009), working with the SWAT model in the San Joaquin watershed in California,
also showed that increased atmospheric CC>2 could cause a significant relative decrease in
simulated evapotranspiration and a corresponding increase in water yield relative to simulations
that did not account for increased CC>2. However, Luo et al. (2013) recently suggested that the
approach used in SWAT to estimate the effects of CO2 on evapotranspiration is appropriate only
for arable land and may overestimate C (^-associated reductions from forest, pasture, and range
land. This remains an important topic for further investigation.
6-13
-------
ACF Central Minnesota Susquehanna Willamette
Arizona River
Figure 6-11. Differences between SWAT projections of mid-21st century
streamflow and water quality (median across six NARCCAP scenarios) with
and without representation of increased atmospheric
Note: Figure shows model simulation with increased CO2 minus projection with CO2 assumed constant at current levels.
Several important feedback loops other than the CC>2 effect on stomatal conductance are also
included in the SWAT plant growth model. First, planting, tillage, fertilization, and harvest
timing for crops (and start and end of growth for native plants) is represented by heat unit
scheduling relative to existing climate normals, allowing automatic adjustment in timing under a
changed temperature regime. Evapotranspiration is also simulated with the full
Penman-Monteith method, allowing dynamic simulation of leaf area development and crop
height, both of which impact ET. Finally, organic matter residue accumulation and degradation
on the land surface are dynamically simulated as a function of plant growth, and the effects of
altered cover on land surface erosion are represented.
All these factors are of potential importance in examining response to climate change. In
contrast to SWAT, HSPF does not automatically compute these adjustments. Instead, the user
would need to estimate changes in monthly parameters such as the lower zone evapotranspiration
coefficient (LZETP) and erosion cover externally and bring them into the model. While not well
understood, use of calibrated parameters in HSPF without these modifications could introduce
error to simulations under climatic conditions different from those during the calibration period.
6-14
-------
6.2. SENSITIVITY TO DIFFERENT METHODS OF DOWNSCALING GCM OUTPUT
A variety of methods for downscaling large-scale GCM output to local scale projections are
available. Both the selection of an underlying GCM and the choice of downscaling method have
a significant influence on the streamflow and water quality simulations. Indeed, in some basins
(e.g., Minnesota River, ACF) the difference among watershed model simulations as driven by
the six NARCCAP dynamically downscaled scenarios appears to be noticeably greater than the
range of model simulations driven by BCSD statistically downscaled or nondownscaled GCM
scenarios. The results of the larger ensemble leads to the observation that incorporating
additional information, either from dynamic RCMs or via statistical methods, can increase the
range of variability of simulated changes.
6.2.1. Climate Model Energy Inputs and PET Estimates
PET is calculated using the Penman-Monteith PET energy balance approach. The BCSD and
nondownscaled GCM scenarios do not provide all the required meteorological time series (see
Table 5-2 in Section 5.2.1.). As a result, PET for these scenarios was estimated using current
climate statistics for solar radiation, dew point, and wind time series. Comparisons presented in
Appendix Z suggest that PET estimates for the and GCM scenarios (scenarios 7-14) that do not
include solar radiation, dew point, and wind time series that are consistent with the simulated
precipitation and temperature are noticeably higher than estimates of PET derived from the
dynamically downscaled models that do provide these time series.
A comparison of the effects of data availability on PET calculations can be done through
comparison of scenarios that are based on the identical underlying GCM runs for CGCM3 and
GFDL that were each dynamically downscaled with two different RCMs (as discussed in
Section 5.2.2.). Annual average PET estimates from these pairs are generally close to one
another, but may differ by up to 4.5% from their mean (see Table 5-4). For the CGCM3 model,
PET generated from the nondownscaled GCM is similar to that from the dynamically
downscaled scenarios, but PET calculated from the statistically downscaled scenario is from 2 to
19% higher. This appears to be due to the fact that dew point temperature, which has an
important impact on PET, is provided with the CGCM3 GCM but is not available from the
BCSD scenarios (see Table 5-2 above). The difference is smallest for the Salt/Verde/San Pedro
River basins in Arizona, where dew point temperature is very low and not expected to change
much under future climates. In contrast, the GFDL model does not provide dew point
temperature from the nondownscaled GCM. For that model, both the nondownscaled and
statistically downscaled climate change scenarios produce higher PET estimates than the
NARCCAP dynamically downscaled scenarios. As with CGCM3, the smallest effect is seen in
the Salt/Verde/San Pedro River basins in Arizona, and the largest effect in the Susquehanna
basin, where a greater change in dew point temperature and relative humidity is projected. The
observed sensitivity of PET estimates to climate variables other than air temperature and
precipitation suggests that simulation of future climates that does not account for changes in the
full suite of variables that influence PET could thus introduce significant biases into the
simulated water balance. Further investigation of this phenomenon was pursued through use of
"degraded" NARCCAP climate scenarios, as described below.
6-15
-------
6.2.2. "Degraded" NARCCAP Climate Scenarios
To provide a consistent basis for comparison, all scenarios were created with a common
minimum set of variables. Specifically, NARCCAP provided data on changes in precipitation
intensity (bin data), solar radiation, wind, and humidity that were not available in the GCM and
BCSD based scenarios. The following steps were taken to develop a consistent set of climate
scenario input series that differ only in the underlying climate model and downscaling technique:
• Representation of intensification in each of the NARCCAP dynamically downscaled
scenarios was based on Approach 2 in Section 5.2.2., which assumes that all increases in
precipitation occur in the top 30% of events, rather than using the direct analysis of
intensity changes provided by NARCCAP.
• Complete information on changes in weather generator statistics for dew point
temperature, solar radiation, and wind speed was removed for the NARCCAP
dynamically downscaled scenarios, consistent with the information available for the
BCSD scenarios. Incomplete information on these variables provided by the
nondownscaled GCMs was also removed. (For the nondownscaled GCMs this affects
weather scenarios 7, 9, and 10—see Table 5-2 above).
• Penman-Monteith PET was recalculated with the revised set of climate variables.
• Simulations use current land use to remove land-use change effects.
Note that these simplified or "degraded" NARCCAP scenarios are used only for the comparisons
presented in this section. Results presented in subsequent sections of this report use the
scenarios that contain all available meteorological information.
Comparison of the PET series generated with full climatological data to the degraded series in
which only precipitation and temperature are updated illustrates the effect of including these
additional variables (see Table 6-4). Further, the effect of individual meteorological time series
is discernible because the original set lacked solar radiation for Scenario 5, dewpoint temperature
for Scenario 9, and wind speed for Scenario 10 (see Table 5-2). Dewpoint temperature (which
tends to increase in future, warmer climates) has the biggest impact. Including a climate
model-simulated dewpoint that is consistent with the scenario temperature and precipitation
regime results in a reduction in estimated annual PET of about 11% across all the meteorological
stations used for the five pilot watersheds. The effect appears to be greater at higher latitudes.
The reduction in PET from including simulated dewpoint is around 10-20% for the Minnesota,
New York, Oregon, and Pennsylvania stations, but only 3-10% for the Alabama, Arizona,
Florida, and Georgia stations. In contrast, for Scenario 9 (for which dewpoint temperature was
not available), the original PET series were on average 1.9% higher than the degraded series.
Omission of solar radiation or wind speed results from the climate scenario appears to have at
most a minor impact on the estimated PET.
In retrospect, these results suggest that a better approach to simulation of PET in cases where the
climate models do not provide dewpoint would be to assume that relative humidity remains
6-16
-------
constant and recalculate a new dewpoint based on the relative humidity and climate-modified air
temperature, thus providing a more physically realistic estimate of vapor pressure deficit.
Table 6-4. Effects of omitting simulated auxiliary meteorological time series
on Penman-Monteith reference crop PET estimates for "degraded" climate
scenarios
State
AL
AZ
FL
GA
MN
NY
OR
PA
All (%)
All
(in/yr)
Climate Scenario (GCM/RCM)
1
CGCM3/
CRCM
-4.87%
-2.38%
-7.14%
-9.30%
-14.68%
-23.27%
-15.82%
-17.62%
-12.53%
-6.36
2
HadCM3/
HRM3
-4.44%
-3.01%
-8.48%
-7.21%
-10.30%
-16.99%
-14.28%
-12.54%
-9.93%
-5.27
3
GFDL/
RCM3
-5.21%
-4.12%
-7.45%
-7.79%
-13.73%
-17.68%
-7.75%
-14.77%
-9.97%
-5.16
4
GFDL/
GFDL hi
res
-10.90%
-3.59%
-16.69%
-18.01%
-10.30%
-20.62%
-12.90%
-18.93%
-12.62%
-6.48
5
CGCM3/
RCM3
-5.76%
-2.97%
-9.04%
-10.15%
-16.46%
-22.95%
-13.67%
-18.59%
-12.86%
-6.42
6
CCSM/
WRFP
-4.47%
-3.08%
-9.02%
-7.27%
-21.16%
-18.30%
-13.29%
-13.40%
-12.48%
-6.31
7
CGCM3
(not down-
scaled)
-4.89%
-0.99%
-7.35%
-8.71%
-13.83%
-23.01%
-12.73%
-17.96%
-11.37%
-5.63
9
GFDL
(not down-
scaled)
2.66%
2.69%
2.92%
1.79%
1.68%
-1.29%
0.11%
0.28%
1.19%
0.90
10
CCSM (not
down-
scaled)
-7.11%
-3.02%
-10.91%
-14.04%
-16.46%
-20.48%
-10.17%
-17.28%
-12.39%
-6.55
Note: Auxiliary time series are solar radiation, dewpoint temperature, and wind. Scenario 5 did not have a solar radiation time
series; Scenario 9 did not have a dewpoint temperature time series; Scenario 10 did not have a wind time series. Results are
averages across entire study area. See Table 5-1 for details of the climate scenarios.
These results suggest that downscaling approaches that omit dewpoint temperature can introduce
significant biases. Specifically, simulation without adjusting for future changes in dewpoint
temperature is likely to overestimate PET, leading to an underestimation of soil moisture and
streamflow.
6.2.3. Sensitivity of Flow and Water Quality to Approaches for Downscaling GCM
Projections
The effect of downscaling approach on the variability of watershed model simulations can be
investigated quantitatively by comparing the results from simulations based on degraded
NARCCAP, GCM, and BCSD scenarios. Table 6-5 presents results obtained with current land
use and the SWAT watershed model (with increased atmospheric CO2) at the most downstream
gage in each study area. Table 6-6 presents detailed results for multiple streamflow and water
quality parameters in the Minnesota River study area. Differences among results with different
downscaling methods are qualitatively similar for HSPF output (not shown).
6-17
-------
Table 6-5. Summary of SWAT-simulated total streamflow in the five pilot
study areas for scenarios representing different methods of downscaling
Study area
Apalachicola-Chattahoochee-
Flint (ACF)
Salt/Verde/San Pedro (Ariz)
Minnesota River (Minn)
Susquehanna (Susq)
Willamette (Willa)
Downscaling
method
NARCCAP
BCSD
GCM
NARCCAP
BCSD
GCM
NARCCAP
BCSD
GCM
NARCCAP
BCSD
GCM
NARCCAP
BCSD
GCM
Number of
scenarios
6
4
4
6
4
4
6
4
4
6
4
4
6
4
4
Median
(cms)
710.4
675.5
655.0
19.4
24.0
26.0
229.5
236.8
238.3
834.8
935.7
868.7
878.8
833.0
843.3
Maximum
(cms)
818.8
722.0
750.7
24.5
28.4
27.0
274.3
286.3
277.0
855.5
948.4
1,017.1
951.8
1,003.7
970.7
Minimum
(cms)
478.6
655.3
581.3
12.9
21.3
19.9
149.4
209.7
124.4
705.6
879.2
807.0
763.6
800.3
810.6
CV
0.208
0.042
0.105
0.233
0.122
0.131
0.230
0.153
0.301
0.068
0.035
0.106
0.086
0.108
0.082
Notes: Results shown are for most downstream station in each study area; coefficient of variation (CV) = standard deviation
divided by the mean. Climate scenarios are degraded to a common basis of scenario precipitation and air temperature
information only.
Results show considerable variability among climate models and downscaling techniques in
different basins and for different streamflow and water quality endpoints. No consistent pattern
attributable to downscaling method is evident for the case in which all climate model outputs are
evaluated using a common basis of precipitation and air temperature only. As was discussed in
Section 6.1.3., the additional information on other meteorological variables can have a profound
effect on PET and watershed responses.
It is noteworthy that the dynamically downscaled results may differ significantly from the
statistically downscaled results from the same GCM, and that the results may also be quite
different when the same GCM is downscaled with a different RCM (e.g., refer to Table 5-1 and
compare climate scenarios 1 and 5 for CGCM3, also 3 and 4 for the GFDL). As noted in Section
5.2., direct comparison between NARCCAP and BCSD downscaling of a single GCM can only
be reliably undertaken for the GFDL and CGCM3 models, because slightly different GCM runs
were used to produce NARCCAP and BCSD results for other GCMs.
6-18
-------
Table 6-6. Summary of SWAT-simulated streamflow and water quality in
the Minnesota River study area for scenarios representing different methods
of downscaling
Endpoint
Total Streamflow (cms)
100-Yr High Flow (cms)
7 Day Average Low
Flow (cms)
Total Suspended
Solids (MT/yr)
Total Phosphorus
(MT/yr)
Total Nitrogen (MT/yr)
Downscaling
method
NARCCAP
BCSD
GCM
NARCCAP
BCSD
GCM
NARCCAP
BCSD
GCM
NARCCAP
BCSD
GCM
NARCCAP
BCSD
GCM
NARCCAP
BCSD
GCM
Number of
scenarios
6
4
4
6
4
4
6
4
4
6
4
4
6
4
4
6
4
4
Median
229.5
236.8
238.3
3,415.4
3,960.2
3,565.7
27.7
25.8
28.2
1,926,166
2,002,421
1,914,800
36,304
40,579
38,747
2,700
3,073
2,889
Maximum
274.3
286.3
277.0
3,700.2
5,055.0
4,432.3
38.5
37.9
37.0
2,520,444
2,428,565
2,557,634
42,119
44,936
42,087
3,283
3,453
3,162
Minimum
149.4
209.7
124.4
3,155.7
3,617.6
2,508.7
14.3
22.3
12.9
896,806
1,376,608
633,793
25,843
32,451
21,538
2,007
2,356
1,489
CV
0.230
0.153
0.301
0.058
0.153
0.227
0.353
0.247
0.395
0.385
0.265
0.460
0.191
0.150
0.264
0.194
0.183
0.292
MT = metric ton
Notes: Results shown are for most downstream station in each study area; coefficient of variation (CV) = standard deviation
divided by the mean. Climate scenarios are degraded to a common basis of scenario precipitation and air temperature
information only.
Both the GFDL and CGCM3 A2 scenario runs for 2041-2070 were downscaled with two
different NARCCAP RCMs—with one RCM (RCM3) in common between the two. A
comparison in terms of the ratio of simulated future mean annual streamflow to simulated current
mean annual streamflow, using SWAT, is made in Figure 6-12 for the GFDL and in Figure 6-13
for the CGCM3 model. For both GCMs, the NARCCAP downscaling, BCSD downscaling, and
nondownscaled GCM output produce relatively consistent results for the Willamette and
Susquehanna basins, but diverge for the Minnesota River. For the Arizona basin, the two
different downscaling approaches diverge for the GFDL but not the CGCM3 GCM. Elevated
coefficients of variation (CVs) on mean annual streamflow in both the Minnesota River and
Arizona basins appear to be largely due to the difference in downscaling results obtained with
6-19
-------
the GFDL high-resolution regional model, which suggests lower flow than other dynamically
downscaled interpretations of the GFDL GCM.
9nn°/
cu
c
"3
in
ro
CO
M-
o
cu
u
*- 1 nn°/
Q.
E
_3
o
5
_o
u_
1
no/
I
V
X
*
* 1
• • •
ACF Ariz Minn SL
KLMo
• High Res
A BCSD
X Raw GFDL
i 2
jsq Willa
Figure 6-12. Consistency in SWAT model projections of mean annual
streamflow at downstream stations with downscaled (NARCCAP, BCSD)
and nondownscaled GCM projections of the GFDL GCM.
Note: The climate change scenarios used in this analysis are simplified to include changes only in air temperature and
precipitation (variables common to the NARCCAP, BCSD, and GCM data sets) to provide a common basis for comparison.
Figures 6-12 and 6-13 demonstrate that a single GCM may yield rather different results
depending on the RCM used for dynamical downscaling. In the current state of the science it
does not appear that the use of dynamical downscaling reduces uncertainty; however, use of
multiple downscaling approaches helps to inform the potential range of climate futures.
To date, relatively few comparisons of RCM model performance in the NARCCAP data sets
have been undertaken. An exception is the study of Wang et al. (2009) for the Intermountain
Region of the Western United States. Significant orographic effects in this area lead to a
complex combination of precipitation annual and semiannual cycles that form four major climate
regimes in this area. Wang et al. compared results from six RCMs over this region to the North
American Regional Reanalysis (NARR) precipitation study (Mesinger et al., 2006) and found
that each model produces its own systematic bias in the central Intermountain Region where the
four different climate regimes meet. All six of the RCMs appeared to produce simulated annual
cycles that are too strong and winter precipitation that is too high under current conditions. The
BCSD statistical approach can correct this for current conditions; however, the statistical
approach would not account for any future large-scale changes in the interaction of the major
climate regimes.
6-20
-------
cu
c
ro
CO
c
cu
H
cu
E
_3
o
I
0%
•
1
•
+ RCM3
• CRCM
A BCSD
XRawCGCMB
A
* *
...» A
ACF
Ariz
Minn
Susq
Willa
Figure 6-13. Consistency in SWAT model projections of mean annual
streamflow at downstream stations with downscaled (NARCCAP, BCSD)
and nondownscaled GCM projections of the CGCM3 GCM.
Note: The climate change scenarios used in this analysis are simplified to include changes only in air temperature and
precipitation (variables common to the NARCCAP, BCSD, and GCM data sets) to provide a common basis for comparison.
Wang et al. (2009) also demonstrate that the different RCMs are largely consistent in the
Cascade Range (OR, WA), where the dominant upper level flow first encounters land, which fits
with the reduced level of variability between downscaling methods noted for the Willamette
study area. The differences among RCMs reported by Wang et al., and the difference from
NARR, are greatest on the windward side of the Rocky Mountains in Colorado and remain large
into Arizona. Interestingly, the apparent wet bias of the CRCM and dry bias of most other
RCMs relative to NARR in Arizona reported by Wang et al. does not appear to carry through
into the future scenarios reported here, suggesting that the RCMs may be providing different
simulated solutions to the future interaction of large-scale climate regimes in this area.
In addition to uncertainties in representing climate forcing at the watershed level, as discussed in
this section, previous sections have shown that the results are sensitive to the selection of a
watershed model, and to modeler skill in calibrating the model. Furthermore, the results are
undoubtedly also sensitive to feedback loops that are not incorporated into the models. Results
produced in this study thus likely do not span the full range of potential future impacts (even
conditional on the A2 storyline) for the reasons given above, among others. Nonetheless, the
range of uncertainty is considerable, and generally covers the zero point, as is summarized at
selected downstream analysis points shown in Table 6-7.
6-21
-------
Table 6-7. Range of SWAT-projected changes in annual streamflow and
pollutant loads for combined mid-21st century NARCCAP climate change
and ICLUS urban and residential development scenarios
Downstream location
ACF: Apalachicola River Outlet
Ariz: Verde River ab Tangle Creek
Susq: Susquehanna River Outlet
Minn: Minnesota River Outlet
Willa: Willamette River Outlet
Change in flow
(%)
-26. 9 to +23. 6
-29.4 to +26.7
-10.0 to +11.0
-14.3 to +62.1
-8.4 to +15. 9
Change in total
solids load
(%)
-47.2 to +6.1
-52. 6 to +118.4
-15.6 to +17.8
-22.9 to +122.9
-10. 3 to +24. 5
Change in total
nitrogen load
(%)
-4.6 to +25. 6
-7.2 to +46. 6
+32.1 to +61. 9
+4. 9 to +71.0
-10. 9 to +3. 3
Change in total
phosphorus load
(%)
-6. 6 to +73.1
-32.8 to +63.4
+6. 3 to +28.1
-6.3 to +59.5
-13. 3 to +4.2
The ranges shown in Table 6-7 suggest that for 2041-2070 conditions it is not possible in most
cases to even state the sign of change in watershed response with a high degree of assurance
unless one is willing to assert that one of the RCMs is more reliable than another. Rather, the
results tell us that the range of potential responses is large.
Based on the analysis presented here, however, the differences in simulation results in our study
are largely a result of combined differences in the underlying GCM and the downscaling
approach used, and more specifically, largely a result of heterogeneity in simulated precipitation
amounts and patterns. For the 2041-2070 timeframe, these warming-induced increases in
simulated PET are generally insufficient to overcome this range of variability in projected
precipitation. This may not be the case, however, for more distant future simulation
periods—given continually increasing temperature and PET, evapotranspiration increases are
likely to ultimately exceed the range of variability in projected precipitation in many basins,
resulting in more uniform decreases in runoff.
6-22
-------
7. REGIONAL SENSITIVITY OF STREAMFLOW AND WATER QUALITY TO
CLIMATE CHANGE AND LAND DEVELOPMENT: RESULTS IN ALL
20 WATERSHEDS
This section presents simulation results in all 20 study areas using SWAT. Model simulations
evaluate the effects of mid-21st century climate change alone (see Section 7.1.), urban and
residential development alone (see Section 7.2.), and the combined effects of climate change and
urban development (see Section 7.4) on streamflow, TN, TP, and TSS. Scenarios also assume
future increases in atmospheric CC>2. Results are presented for a single representative analysis
point in each study area (see Table 7-1). For study areas composed of a single watershed, this is
the outlet (pour point) of the entire study area. For study areas composed of multiple, adjacent
watersheds draining to the coast, the analysis point reported here is at or near the outlet of the
largest river within the study area. Results for additional locations within each study area are
presented in Appendix X for the five pilot study areas and in Appendix Y for the other 15 study
areas.
Table 7-1. Downstream stations within each study area where simulation
results are presented
Study area
Apalachicola-Chattahoochee-Flint Basins (ACF)
Southern California Coastal (SoCal)
Cook Inlet Basin (Cook)
Georgia-Florida Coastal Plain (GaFla)
Illinois River Basin (Illin)
Lake Erie Drainages (LErie)
Lake Pontchartrain Drainage (LPont)
Nebraska: Loup and Elkhorn River Basin (Neb)
Minnesota River Basin (Minn)
Tar and Neuse River Basins (TarNeu)
New England Coastal Basins (NewEng)
Powder and Tongue River Basin (PowTon)
Rio Grande Valley (RioGra)
Sacramento River Basin (Sac)
Arizona: Salt, Verde, and San Pedro (Ariz)
South Platte River Basin (SoPlat)
Susquehanna River Basin (Susq)
Trinity River Basin (Trin)
Upper Colorado River Basin (UppCol)
Willamette River Basin (Willa)
Location presenting results
Apalachicola R at outlet
Los Angeles R at outlet
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Elkhom R at outlet
Minnesota R at outlet
Neuse R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Salt River near Roosevelt
S. Platte R at outlet
Susquehanna R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
7-1
-------
7.1. SELECTION OF WATERSHED MODEL FOR USE IN ALL STUDY AREAS
Resource limitations for this study precluded the application of SWAT and HSPF in all 20 study
areas. Analyses at Pilot sites were used to select a single model for application in all 20 study
areas. Analyses in the Pilot sites show HSPF and SWAT are each capable of providing a good
fit to streamflow and pollutant loads for existing conditions. The quality of fit depends in part on
the strategy and skill of the individual modeler. In this study, the quality of fit was also
influenced by the availability in certain areas of preexisting, calibrated models which were
adapted for use as compared to locations where new models were developed and calibration
subject to resource limitations.
For the purposes of this study, the SWAT model was considered to have a technical advantage
because it can account for the influence of changes in atmospheric CO2 concentration and other
feedback responses of plant growth to climate change. HSPF does not automatically account for
these effects. While it uncertain how well SWAT is able to represent the complex processes
affecting plant growth, nutrient dynamics, and water budgets under changing climate (see Luo et
al., 2013), it was considered important to include some representation of these processes to better
understand potential watershed sensitivity to a wide range of conditions. In addition, there are
also practical advantages to the choice of SWAT, as the model is somewhat easier to set up and
calibrate than is HSPF.
Conversely, the HSPF model proved generally better able to replicate observations during
calibration, as shown in Section 6.1.1., although the difference between HSPF and SWAT model
performance was small for the selected response variables. HSPF is often able to provide a
better fit to streamflow after calibration due to the use of hourly precipitation and a more
sophisticated algorithm compared to SWAT's daily curve number approach—although this
advantage is diminished by the need to use disaggregated daily total rainfall to drive the models
in many areas. Increased accuracy in hydrology—especially the accurate partitioning between
surface and subsurface runoff—should also provide increased accuracy in the simulation of
sediment yield and the transport of sediment-associated nutrients. However, at the larger
watershed scales studied here (HUC-8 and greater), such advantages will tend to diminish as
observations reflect the integration of flows and loads from multiple subwatersheds driven by
multiple weather stations. Further, SWAT is generally considered to perform better under
limited calibration and thus may have an advantage for extension to changed conditions of land
use and climate (Gassman et al., 2007).
The file structure of the HSPF model is also considerably more efficient for implementing and
running multiple scenarios. SWAT's use of the curve number approach to hydrology and a daily
time step can also cause difficulties in representing the full hydrograph and introduces
uncertainties into the simulation of erosion and pollutant loading as a function of surface flow
(Garen and Moore, 2005). This is a concern in particular for the simulation of urban hydrology
at small spatial scales; however, these concerns are of lesser importance at the larger spatial
scales that are the focus of this study.
Given that both models were capable of performing adequately, the SWAT model was selected
for use in the 15 nonpilot watersheds due to its integrated plant growth model and practical
advantages of ease of calibration.
7-2
-------
It should be recognized that there are other feedback cycles that are not incorporated in either
model, such as the potential for any increased rate of catastrophic forest fires (Westerling et al.,
2006), changes to vegetative communities as a result of pests and disease (Berg et al., 2006), and
human adaptations such as shifts to different crops and agricultural management strategies
(Polsky and Easterling, 2001).
7.2. SENSITIVITY TO CLIMATE CHANGE SCENARIOS
This section presents the results of SWAT simulations in all 20 study areas for climate change
scenarios alone (that is, with land use held constant at existing conditions). In general, the
different climate scenarios provide a consistent picture of temperature increases by mid-century
(on the order of 2 to 3°C or 3 to 6°F), although there do appear to be systematic differences
between the scenarios (for example, the NARCCAP scenario using the GFDL model downscaled
with RCM3 typically is the coolest scenario for the watersheds studied here). In contrast,
changes in precipitation between the historical and future periods differ widely across climate
change scenarios, with some producing increases and some decreases in total precipitation.
Projected mid-21st century precipitation, air temperature, PET, and simulated AET (from SWAT)
for each of the six NARCCAP climate change scenarios in each study area are shown in Tables
7-2 through 7-5. For Cook Inlet (Alaska) results are shown only for the three NARCCAP
scenarios that provide climate projections for this portion of Alaska. The projected future
climate annual average as a percent of baseline resulting from each of the six NARCCAP
scenarios is shown for precipitation, PET, and AET; absolute change is shown for the annual
average temperature. It should be noted that while the projected future average annual
temperature increases in all cases, PET does not always increase. This is particularly noticeable
in some of the southwestern study areas (e.g., Rio Grande Valley) where at least some future
climate scenarios project increases in humidity and cloudiness that offset the temperature impact
on PET. While shown here for comparison to PET, AET is a model input, not a model output.
AET is driven by PET, but can also be limited by lack of soil moisture and is affected by changes
in the seasonal timing of both precipitation and plant growth.
7-3
-------
Table 7-2. Average annual precipitation (in/yr and percent of baseline) for current conditions and mid-21s
century climate scenarios
Study area
ACF — Apalachicola-Chattahoochee-Flint Basins
Ariz — Arizona: Salt, Verde, and San Pedro
Cook — Cook Inlet Basin
GaFla — Georgia-Florida Coastal Plain
Illin — Illinois River Basin
LErie — Lake Erie Drainages
LPont — Lake Pontchartrain Drainage
Minn — Minnesota River Basin
Neb — Nebraska: Loup and Elkhorn River Basins
NewEng — New England Coastal Basins
PowTon — Powder and Tongue River Basins
RioGra — Rio Grande Valley
Sac — Sacramento River Basin
SoCal — Southern California Coastal Basins
SoPlat— South Platte River Basin
Susq — Susquehanna River Basin
TarNeu — Tar and Neuse River Basins
Trin — Trinity River Basin
UppCol — Upper Colorado River Basin
Willa — Willamette River Basin
Current
conditions
52.14
19.38
24.22
52.98
37.63
36.88
64.76
27.61
24.43
46.42
13.85
12.20
35.81
19.62
15.93
39.73
48.90
42.83
15.88
55.43
CRCM_
cgcm3 (%)
105.1
87.4
NO
101.3
101.5
102.4
96.0
102.3
99.5
106.1
99.1
89.1
102.3
96.2
95.4
106.6
99.5
94.8
90.3
106.5
HRM3
hadcm3 (%)
114.3
94.3
118.3
117.3
114.2
114.2
109.2
106.7
103.4
113.2
100.2
91.1
88.6
117.1
92.2
109.2
122.3
110.4
97.3
101.1
RCM3
gfdl (%)
106.2
110.4
NO
106.5
103.9
104.9
106.4
110.3
103.4
107.7
104.8
106.5
95.8
97.0
97.5
103.6
112.6
98.6
108.3
97.6
GFDL
slice (%)
97.2
85.9
113.9
95.3
104.1
109.0
92.5
97.8
86.2
107.4
86.5
90.6
99.6
95.5
87.1
105.4
103.2
83.4
95.7
88.4
RCM3 cgcm3
(%)
111.2
98.5
NO
112.0
105.3
104.0
100.9
110.7
106.3
104.7
105.6
88.3
99.1
99.4
98.9
105.7
108.0
101.8
94.8
105.1
WRFP ccsm
(%)
90.4
87.9
122.6
85.1
93.3
91.7
87.8
112.1
104.8
98.1
120.0
99.4
96.3
87.6
101.2
97.9
92.4
105.9
95.2
94.5
Median ratio
(%)
105.6
91.1
118.3
103.9
104.0
104.5
98.5
108.5
103.4
106.7
102.5
90.8
97.7
96.6
96.5
105.6
105.6
100.2
95.4
99.4
-------
Table 7-3. Average annual temperature (°F and change from baseline) for current conditions and mid-21st
century climate scenarios
Study area
ACF — Apalachicola-Chattahoochee-Flint Basins
Ariz — Arizona: Salt, Verde, and San Pedro
Cook — Cook Inlet Basin
GaFla — Georgia-Florida Coastal Plain
Illin — Illinois River Basin
LErie — Lake Erie Drainages
LPont — Lake Pontchartrain Drainage
Minn — Minnesota River Basin
Neb — Nebraska: Loup and Elkhorn River Basins
NewEng — New England Coastal Basins
PowTon — Powder and Tongue River Basins
RioGra — Rio Grande Valley
Sac — Sacramento River Basin
SoCal — Southern California Coastal Basins
SoPlat— South Platte River Basin
Susq — Susquehanna River Basin
TarNeu — Tar and Neuse River Basins
Trin — Trinity River Basin
UppCol — Upper Colorado River Basin
Willa — Willamette River Basin
Current
conditions
64.33
56.41
33.13
68.29
49.57
49.13
66.48
44.18
47.94
46.32
44.84
44.72
58.23
61.38
45.06
48.18
59.93
64.91
40.80
51.48
CRCM_
cgcm3
+3.81
+4.93
NO
+3.56
+5.36
+5.19
+3.77
+5.61
+5.20
+4.97
+4.77
+5.13
+4.16
+3.58
+4.98
+4.98
+4.28
+4.35
+5.20
+3.79
HRM3_
hadcm3
+4.16
+5.19
+5.20
+3.99
+4.66
+4.65
+4.53
+5.29
+5.10
+4.81
+4.97
+5.37
+4.76
+3.97
+5.20
+4.98
+4.51
+4.66
+5.14
+4.37
RCM3_ gfdl
+3.62
+4.35
NO
+3.45
+4.38
+4.29
+3.61
+4.01
+3.88
+4.07
+3.81
+4.20
+3.75
+3.72
+4.14
+4.16
+3.83
+3.97
+4.13
+2.80
GFDL_ slice
+4.49
+4.96
+3.99
+4.36
+4.84
+4.75
+4.13
+5.02
+5.09
+4.12
+4.71
+5.84
+3.47
+3.27
+5.51
+4.72
+4.18
+4.45
+5.53
+3.03
RCM3_
cgcm3
+3.45
+4.75
NO
+3.32
+4.75
+4.67
+3.41
+4.60
+4.53
+4.67
+4.50
+5.02
+3.94
+3.98
+4.93
+4.59
+3.70
+3.79
+4.90
+3.59
WRFP_
ccsm
+4.35
+4.62
+5.30
+3.68
+5.36
+5.11
+3.79
+4.90
+4.65
+4.50
+4.27
+4.74
+4.06
+3.57
+4.77
+4.60
+4.14
+4.38
+5.04
+3.57
Median
change
+3.98
+4.84
+5.20
+3.62
+4.80
+4.71
+3.78
+4.96
+4.87
+4.58
+4.61
+5.08
+4.00
+3.65
+4.96
+4.66
+4.16
+4.36
+5.09
+3.58
-------
Table 7-4. Average annual PET (in/yr and percent of baseline) for current conditions and mid-21st century
climate scenarios
Study area
ACF — Apalachicola-Chattahoochee-Flint Basins
Ariz — Arizona: Salt, Verde, and San Pedro
Cook — Cook Inlet Basin
GaFla — Georgia-Florida Coastal Plain
Illin — Illinois River Basin
LErie — Lake Erie Drainages
LPont — Lake Pontchartrain Drainage
Minn — Minnesota River Basin
Neb — Nebraska: Loup and Elkhorn River Basins
NewEng — New England Coastal Basins
PowTon — Powder and Tongue River Basins
RioGra — Rio Grande Valley
Sac — Sacramento River Basin
SoCal — Southern California Coastal Basins
SoPlat— South Platte River Basin
Susq — Susquehanna River Basin
TarNeu — Tar and Neuse River Basins
Trin — Trinity River Basin
UppCol — Upper Colorado River Basin
Willa— Willamette River Basin
Current
conditions
62.04
81.27
16.56
65.82
42.91
45.27
59.19
49.36
61.94
43.22
55.39
54.48
66.77
64.41
53.25
43.81
56.38
77.27
38.14
43.64
CRCM
cgcm3 (%)
101.2
103.6
NO
99.9
112.3
102.0
101.4
106.3
100.4
103.3
101.3
94.4
99.2
99.2
102.0
102.9
100.5
99.5
106.8
97.4
HRM3
hadcm3 (%)
103.8
103.8
106.2
101.1
110.5
100.7
106.9
110.4
100.5
105.9
102.7
100.1
103.0
100.2
103.4
107.6
100.9
100.1
107.7
102.0
RCM3 gfdl
(%)
101.6
100.3
NO
99.6
109.3
99.6
103.0
98.6
97.0
100.3
99.0
90.4
102.5
99.9
100.3
101.2
99.2
99.0
103.9
100.5
GFDL
slice (%)
97.5
103.0
99.7
100.3
111.2
101.1
103.9
110.5
101.4
100.6
102.3
99.9
98.8
99.0
104.1
101.4
100.1
99.8
108.8
98.9
RCM3
cgcm3 (%)
98.3
102.6
NO
98.6
110.0
100.2
99.3
99.7
98.1
100.6
100.3
95.9
98.8
100.0
101.9
99.5
99.0
98.1
105.8
98.6
WRFP
ccsm (%)
105.2
106.4
104.1
100.6
111.2
101.3
101.2
94.2
97.9
101.3
99.0
92.6
101.6
99.2
101.4
104.9
100.1
98.7
106.2
98.7
Median
ratio (%)
101.4
103.3
104.1
100.1
110.9
100.9
102.2
103.0
99.2
100.9
100.8
95.1
100.4
99.6
102.0
102.1
100.1
99.3
106.5
98.8
-------
Table 7-5. Average annual SWAT-simulated actual ET (in/yr and percent of baseline) for current conditions
and mid-21st century climate scenarios
Study area
ACF — Apalachicola-Chattahoochee-Flint Basins
Ariz — Arizona: Salt, Verde, and San Pedro
Cook — Cook Inlet Basin
GaFla — Georgia-Florida Coastal Plain
Illin — Illinois River Basin
LErie — Lake Erie Drainages
LPont — Lake Pontchartrain Drainage
Minn — Minnesota River Basin
Neb — Nebraska: Loup and Elkhorn River Basins
NewEng — New England Coastal Basins
PowTon — Powder and Tongue River Basins
RioGra — Rio Grande Valley
Sac — Sacramento River Basin
SoCal — Southern California Coastal Basins
SoPlat— South Platte River Basin
Susq — Susquehanna River Basin
TarNeu — Tar and Neuse River Basins
Trin — Trinity River Basin
UppCol — Upper Colorado River Basin
Willa— Willamette River Basin
Current
conditions
32.22
14.47
7.95
30.86
22.90
22.75
29.83
21.64
18.00
23.31
16.83
10.32
15.26
8.75
13.06
23.73
29.48
27.58
13.13
19.84
CRCM
cgcm3 (%)
106.1
86.8
NO
98.5
101.5
94.4
100.1
96.1
97.8
103.3
93.3
84.2
99.2
97.2
96.1
104.8
97.2
95.0
91.7
87.9
HRM3
hadcm3 (%)
110.6
94.8
109.1
101.1
103.4
95.3
107.0
99.9
101.8
110.4
94.9
87.9
97.6
102.7
94.0
108.4
99.9
99.9
98.0
92.9
RCM3 gfdl
(%)
106.6
102.8
NO
99.7
101.3
96.0
101.3
94.9
100.6
103.5
96.7
98.0
94.7
92.9
96.0
102.6
98.0
97.0
101.2
85.4
GFDL
slice (%)
106.2
86.3
103.6
98.9
101.0
97.2
103.4
97.2
94.3
102.5
83.7
88.7
97.0
93.7
90.4
103.7
97.8
90.1
98.6
82.9
RCM3
cgcm3 (%)
104.9
97.3
NO
99.1
101.7
93.3
99.0
95.7
98.9
104.3
97.2
85.2
95.9
96.8
97.1
102.0
97.4
96.4
94.0
88.7
WRFP ccsm
(%)
102.8
89.8
108.6
95.0
98.9
90.8
98.2
92.7
97.9
104.8
105.4
94.7
94.9
94.1
98.5
104.8
95.6
97.8
95.5
88.3
Median
ratio (%)
106.2
92.3
108.6
99.0
101.4
94.9
100.7
95.9
98.4
103.9
95.8
88.3
96.5
95.5
96.0
104.2
97.6
96.7
96.7
88.1
-------
In addition to changes in precipitation amount, this study considers the impacts of changes in
precipitation intensity, which may have significant effects on the partitioning between surface
and subsurface flows and associated generation of pollutant loads. As described in Section 5.2.,
a change factor approach was used to modify historical meteorological time series to represent
mid-21st century climate futures projected by a variety of downscaled (and nondownscaled)
GCM projections. Potential intensification of precipitation is represented by reapportioning the
net change in precipitation volume according to GCM forecasts of the distribution of event
intensities above and below the 70* percentile of the distribution of current (1971-2000) rainfall
events. Under current conditions, the fraction of rainfall volume occurring in events above the
70th percentile ranges from a low of 61% (Cook Inlet and Willamette) to a high of 93%
(Southern California Coastal). Projected mid-21st century changes in precipitation intensity from
the six NARCCAP scenarios, shown in Table 7-6, are mixed. Across all study areas there is an
average increase in the fraction of total volume above the 70* percentile of the current
distribution of 1.19 percentage points. However, for most study areas the six NARCCAP
scenarios are not in full agreement as to whether intensification of precipitation (as defined
relative to the 70* percentile event) will increase. An increase in the volume in high-intensity
events is consistently projected across all six of the NARCCAP mid-21st century projections in
only six of the 20 study areas (Susq, Minn, Cook, LErie, Illin, and NewEng). Two RCM/GCM
combinations (HRM3_hadcm3 = Scenario 2) and (RCM3_cgcm3 = Scenario 5) project increases
in intensity in all study areas. No study area is expected to have a decrease in precipitation
volume in high intensity events across all NARCCAP scenarios, while six study areas (Cook,
Illin, LErie, Minn, NewEng, and Susq) are projected to have an increase in high-intensity events
across all six NARCCAP scenarios. By far the largest increases in high-intensity events are
projected for the Cook Inlet watershed in Alaska, followed by the Upper Colorado basin.
The simulated watershed responses to mid-21st century climate change scenarios are shown in
Tables 7-7 through 7-14. For endpoints other than days to streamflow centroid, the results are
displayed as a percentage relative to the current baseline (generally, 1972-2003), allowing
comparison across multiple basins with different magnitudes of streamflow and pollutant loads.
For Cook Inlet (Alaska), the results are shown only for the three NARCCAP scenarios that
provide climate projections for this portion of Alaska.
Table 7-7 summarizes results for total average annual streamflow volume, with results ranging
from 62% to 240% of current average flows. Results for 7-day low streamflow and 100-year
peak flows (estimated with log-Pearson III fit) are shown in Tables 7-8 and 7-9, respectively.
The Kenai River has by far the greatest increase in 7-day low flows because warmer
temperatures alter the snow/ice melt regime, while the largest increases in 100-year peak flows
are for the Neuse River on the east coast.
Table 7-10 summarizes the estimated change in days to streamflow centroid relative to the start
of the water year. Many stations show negative shifts, indicating earlier snowmelt resulting in an
earlier center of streamflow mass. In contrast, several stations show positive shifts due to
increased summer precipitation.
Results for the Richards-Baker flashiness index (see Table 7-11) show generally small
percentage changes, with a few exceptions. Baker et al. (2004) suggest that changes on the order
-------
of 10% or more may be statistically significant. It is likely, however, that the focus on larger
watersheds reduces the observed flashiness response.
Table 7-6. Changes in precipitation intensity for NARCCAP mid-21st
century climate scenarios
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
CRCM cgcm3
(%)
3.28
-0.19
ND
2.74
1.95
2.25
2.49
1.92
1.08
2.55
1.67
-0.04
1.87
0.24
-0.20
3.28
2.59
1.32
-0.10
2.50
HRM3 hadcm3
(%)
1.35
0.53
7.51
1.60
1.56
2.01
0.71
1.22
1.56
1.51
1.54
0.96
0.47
1.83
0.85
1.58
1.38
1.24
1.95
2.59
RCM3 gfdl
(%)
1.87
0.03
ND
2.33
1.09
1.64
2.50
1.94
1.61
1.74
1.66
1.64
0.00
-0.76
1.13
2.08
1.55
0.61
2.36
0.46
GFDL slice
(%)
-0.68
0.36
3.71
-1.23
0.87
1.81
-0.48
0.09
-0.12
0.36
-1.09
0.37
-1.06
-0.28
-0.15
0.41
-0.42
-0.66
0.79
-2.63
RCM3 cgcm3
(%)
2.43
0.79
ND
2.99
1.25
1.12
1.61
1.38
0.99
1.21
1.59
0.91
2.28
0.29
1.27
1.77
1.21
0.14
2.06
2.71
WRFG ccsm
(%)
-0.12
-0.43
4.62
-0.89
0.20
0.27
-0.87
0.43
0.14
0.15
0.97
0.52
1.42
-0.15
0.51
0.41
-0.02
0.46
0.71
0.97
Note: Potential change in precipitation intensity is shown as the change total volume of precipitation event above the 70th
percentile of the current (1971-2000) distribution of rainfall event volumes.
Simulated changes in pollutant loads (TN, TP, TSS) are summarized in Tables 7-12 through
7-14. The patterns are generally similar to changes in streamflow. Increases in pollutant loads
are suggested for many watersheds, but there are also basins where loads decline, mostly due to
reduced flows.
7-9
-------
Table 7-7. Simulated total streamflow volume (climate scenarios only; percent relative to current conditions) for
selected downstream stations
Station
Apalachicola R at outlet
Salt River near Roosevelt
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Minnesota R at outlet
Elkhom R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Los Angeles R at outlet
S. Platte R at outlet
Susquehanna R at outlet
Neuse R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
CRCM cgcm3
(%)
107
80
ND
114
94
116
96
109
117
108
101
72
104
92
90
109
103
98
86
116
HRM3 hadcm3
(%)
122
80
154
153
125
150
110
113
125
115
85
69
89
138
74
106
158
146
95
106
RCM3 gfdl
(%)
108
149
ND
128
101
120
115
147
137
111
140
112
98
102
90
106
137
106
116
105
GFDL slice
(%)
88
75
132
92
102
136
84
86
68
111
70
66
98
103
65
108
110
62
89
92
RCM3 cgcm3
(%)
124
94
ND
156
105
122
106
146
138
106
130
69
100
106
107
111
125
118
92
114
WRFP ccsm
(%)
73
73
167
75
78
88
77
162
143
94
240
84
99
84
119
90
86
134
91
98
Median
(%)
107
80
154
121
101
121
101
130
131
109
115
71
99
103
90
107
118
112
91
105
-------
Table 7-8. Simulated 7-day low flow (climate scenarios only; percent relative to current conditions) for selected
downstream stations
Station
Apalachicola R at outlet
Salt River near Roosevelt
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Minnesota R at outlet
Elkhom R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Los Angeles R at outlet
S. Platte R at outlet
Susquehanna R at outlet
Neuse R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
CRCM cgcm3
(%)
97
58
ND
104
85
104
73
115
119
110
102
81
101
96
93
91
94
26
85
131
HRM3 hadcm3
(%)
120
77
267
141
123
184
106
136
133
140
92
64
91
114
87
120
170
167
94
113
RCM3 gfdl
(%)
105
130
ND
121
97
126
88
201
151
130
145
120
95
98
97
104
135
64
121
108
GFDL slice
(%)
85
87
280
95
91
132
74
81
48
118
67
62
96
98
74
89
113
23
85
83
RCM3 cgcm3
(%)
113
79
ND
136
100
128
89
182
148
124
127
74
99
100
102
107
125
70
91
127
WRFP ccsm
(%)
64
90
401
78
70
58
62
228
154
120
235
86
93
92
113
86
70
85
90
102
Median
(%)
101
83
280
113
94
127
81
159
140
122
115
77
95
98
95
98
119
67
91
111
-------
Table 7-9. Simulated 100-year peak flow (log-Pearson III; climate scenarios only; percent relative to current
conditions) for selected downstream stations
Station
Apalachicola R at outlet
Salt River near Roosevelt
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Minnesota R at outlet
Elkhom R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Los Angeles R at outlet
S. Platte R at outlet
Susquehanna R at outlet
Neuse R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
CRCM cgcm3
(%)
119
119
ND
130
120
96
105
84
126
114
118
90
105
83
132
107
71
97
78
116
HRM3 hadcm3
(%)
144
101
132
145
153
106
150
83
117
130
113
77
98
89
127
130
292
106
84
130
RCM3 gfdl
(%)
110
104
ND
129
107
87
108
96
109
111
133
108
125
161
98
106
161
107
97
114
GFDL slice
(%)
90
68
125
94
99
93
99
88
92
138
82
66
117
95
126
128
111
60
91
79
RCM3 cgcm3
(%)
128
120
ND
157
128
93
105
90
139
89
121
72
102
127
151
172
224
86
94
116
WRFP ccsm
(%)
94
66
132
107
97
92
65
96
103
80
146
92
131
77
150
100
63
106
84
95
Median
(%)
114
102
132
130
114
93
105
89
113
112
119
83
111
92
129
118
136
102
87
115
to
-------
Table 7-10. Simulated changes in the number of days to streamflow centroid (climate scenarios only; relative to
current conditions) for selected downstream stations
Station
Apalachicola R at outlet
Salt River near Roosevelt
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Minnesota R at outlet
Elkhom R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Los Angeles R at outlet
S. Platte R at outlet
Susquehanna R at outlet
Neuse R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
CRCM_ cgcm3
-2
-18
ND
-3
-12
-2
-14
-13
-12
-17
-6
25
-4
5
-12
-18
-14
16
-11
3
HRM3_ hadcm3
-2
41
-3
17
6
-4
13
-19
6
-14
-3
6
-7
48
-20
16
23
21
-14
-8
RCM3_ gfdl
1
28
ND
25
-3
1
-24
-6
1
-19
1
3
-4
-3
-14
-6
30
30
-7
-1
GFDL_ slice
8
17
-5
-8
-12
0
-7
-15
-15
-13
-16
11
-1
10
-19
-12
-12
3
-10
3
RCM3_ cgcm3
-6
-6
ND
-5
-2
10
-6
-3
-6
-9
-4
14
-3
-3
-3
-6
10
6
-8
1
WRFP_ ccsm
1
53
-1
11
-15
-8
-11
2
2
-18
7
17
-8
1
-12
0
-5
37
-10
8
Median
-1
22
-3
4
-7
-1
-9
-10
-2
-16
-3
13
-4
3
-13
-6
2
18
-10
2
-------
Table 7-11. Simulated Richards-Baker flashiness index (climate scenarios only; percent relative to current
conditions) for selected downstream stations
Station
Apalachicola R at outlet
Salt River near Roosevelt
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Minnesota R at outlet
Elkhom R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Los Angeles R at outlet
S. Platte R at outlet
Susquehanna R at outlet
Neuse R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
CRCM cgcm3
(%)
106
81
ND
93
106
99
105
104
95
101
102
109
124
103
99
107
96
71
101
101
HRM3 hadcm3
(%)
125
102
94
62
104
101
105
112
98
103
108
117
103
119
91
111
113
68
107
105
RCM3 gfdl
(%)
109
121
ND
76
103
99
106
107
94
99
104
95
112
100
101
107
115
72
111
100
GFDL slice
(%)
94
98
102
117
106
100
104
100
95
101
100
119
109
105
87
110
98
73
105
97
RCM3 cgcm3
(%)
125
103
ND
59
105
100
104
109
96
98
103
103
116
105
108
112
103
69
104
101
WRFP ccsm
(%)
90
119
96
187
104
96
102
108
94
93
109
106
123
99
106
103
91
68
101
102
Median
(%)
108
102
96
84
105
100
104
108
95
100
104
108
114
104
100
109
101
70
105
101
-------
Table 7-12. Simulated total suspended solids load (climate scenarios only; percent relative to current conditions)
for selected downstream stations
Station
Apalachicola R at outlet
Salt River near Roosevelt
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Minnesota R at outlet
Elkhorn R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Los Angeles R at outlet
S. Platte R at outlet
Susquehanna R at outlet
Neuse R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
CRCM_
cgcm3 (%)
125
89
ND
121
116
123
100
107
122
118
108
60
139
71
91
117
106
63
80
124
HRM3
hadcm3 (%)
146
79
234
176
142
169
115
119
131
128
84
53
94
111
87
108
199
124
90
111
RCM3
gfdl (%)
129
184
ND
138
115
126
128
187
147
117
169
114
122
81
94
108
162
62
124
109
GFDL_
slice (%)
93
66
196
90
128
153
83
77
60
122
66
49
118
81
80
115
115
27
82
90
RCM3_
cgcm3 (%)
144
106
ND
181
120
129
111
197
162
111
153
59
99
84
100
118
143
83
89
121
WRFP ccsm
(%)
53
74
244
74
90
86
71
225
162
85
351
71
108
65
104
84
82
113
85
97
Median
(%)
127
84
234
130
118
128
106
153
139
118
131
59
113
81
93
112
129
73
87
110
-------
Table 7-13. Simulated total phosphorus load (climate scenarios only; percent relative to current conditions) for
selected downstream stations
Station
Apalachicola R at outlet
Salt River near Roosevelt
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Minnesota R at outlet
Elkhorn R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Los Angeles R at outlet
S. Platte R at outlet
Susquehanna R at outlet
Neuse R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
CRCM cgcm3
(%)
138
82
ND
115
107
118
113
97
118
111
107
54
100
53
90
128
112
124
79
100
HRM3 hadcm3
(%)
152
83
89
171
112
150
131
115
124
118
86
43
86
88
78
106
230
163
88
98
RCM3 gfdl
(%)
134
155
ND
135
107
132
135
151
138
111
163
127
104
71
99
111
169
130
119
96
GFDL slice
(%)
118
70
90
89
113
148
94
97
65
115
67
51
115
60
72
127
120
83
81
94
RCM3_
cgcm3 (%)
148
106
ND
173
108
117
115
138
145
106
148
41
95
62
108
115
166
135
84
100
WRFP_
ccsm (%)
106
88
113
76
99
88
83
160
147
94
324
67
108
54
111
109
94
160
83
96
Median
(%)
136
86
90
125
108
125
114
126
131
111
127
53
102
61
95
113
143
132
84
97
-------
Table 7-14. Simulated total nitrogen load (climate scenarios only; percent relative to current conditions) for
selected downstream stations
Station
Apalachicola R at outlet
Salt River near Roosevelt
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Minnesota R at outlet
Elkhorn R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Los Angeles R at outlet
S. Platte R at outlet
Susquehanna R at outlet
Neuse R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
CRCM cgcm3
(%)
116
90
ND
127
103
128
123
126
93
119
109
49
99
93
86
162
111
121
73
104
HRM3 hadcm3
(%)
125
91
200
160
118
158
141
130
97
128
91
38
89
140
70
147
189
165
82
97
RCM3 gfdl
(%)
115
142
ND
135
106
162
143
163
145
117
165
125
100
131
91
147
154
125
110
95
GFDL slice
(%)
106
86
175
112
110
191
106
105
88
121
71
47
110
98
63
156
118
80
76
89
RCM3_
cgcm3 (%)
122
105
ND
166
108
125
120
158
104
114
148
37
98
90
109
150
144
136
80
103
WRFP_
ccsm (%)
95
84
223
85
93
94
91
171
107
101
320
64
107
101
116
132
99
164
79
93
Median
(%)
116
90
200
131
107
143
121
144
101
118
128
48
100
100
89
149
131
130
80
96
-------
For most measures in most watersheds, there is a substantial amount of variability between
scenario projections based on different methods of downscaling GCM outputs. This reflects our
uncertainty in predicting future climate, especially the future joint distribution of precipitation
and potential evapotranspiration that is fundamental to watershed response, and reinforces the
need for an ensemble approach for evaluating the range of potential responses.
Climate change could also alter the seasonal dynamics of streamflow and nutrient loading.
Seasonal effects are investigated here in summary form through calculation of the ratio of winter
(January-March) to summer (July-September) runoff volume averaged over all HUC-8s in a
study area. More detailed results showing simulated changes in streamflow by month are
presented in Appendices X and Y. The different study areas have very different seasonal runoff
volume ratios under current conditions, ranging from a winter:summer low of 0.11 in the Cook
Inlet basin to a high of 11 in the Willamette River basin. The average ratios under the mid-21st
century NARCCAP climate change scenarios are shown relative to the current ratio in Figure
7-1. In most cases, the future climate scenarios span the current ratio; however, in the case of the
South Platte and Upper Colorado study areas, currently dominated by snowmelt runoff from the
Rocky Mountains, all future climate scenarios project an increase in the ratio. In some basins the
range of future projected seasonal runoff ratios is quite large. For the Salt, Verde, and San Pedro
River basins (Ariz) the average future ratios by climate scenario range from 0.8 to 5.4, depending
on whether the climate scenario projects greater increases in the summer monsoon or winter
rainy period, while in the Lake Erie drainages (LErie) the range is from 1.5 to 7.8. The
distribution for each of the six NARCCAP climate scenarios is summarized in Figure 7-2. There
are clear differences between the different scenarios, with some projecting a much greater
increase in the winter: summer runoff ratio than others.
7.3. SENSITIVITY TO URBAN AND RESIDENTIAL DEVELOPMENT SCENARIOS
This section presents the results of SWAT simulations in all 20 study areas for mid-21st century
urban and residential development alone (that is, with climate held constant at existing
conditions). Results in the pilot study areas (see Section 6.) suggested that effects of urban and
residential development by 2050 on streamflow and pollutant loads is likely to be comparatively
small relative to the potential range of impacts associated with climate change. This is largely a
reflection of the scale of the analysis: at the scale of large (HUC-4 to HUC-8) watersheds,
developed land is rarely a large portion of the total land area. Significant effects may occur in
smaller subbasins where extensive new land development occurs.
Over the full extent of individual study areas, current impervious surface area ranges from near
zero to 13.8% of the total area, while projected changes (increases) in impervious cover area
range from 0 to 5.3% of the total area (see Table 7-15). While several fast-growing metropolitan
areas are included within the study areas, the impact of these areas is diminished at larger spatial
scales. At the HUC-8 and larger scale, it is not surprising that projected changes in urban and
residential development have only a relatively small effect compared to climate change, which
affects all portions of a watershed. The largest response of total streamflow volume to land-use
change at the full-basin scale is simulated for the Trinity River in Texas, where total flow
increased by 6%, while the estimated 100-year peak flow decreased and days to streamflow
centroid increased (i.e., later runoff). This reflects increases in development upstream in the
Dallas-Fort Worth metropolitan area. A stronger response to land development is seen at smaller
7-18
-------
spatial scales where development can account for a larger fraction of watershed area.
Development effects are also more likely be reflected in high or low streamflow statistics. For
example, in the Los Angeles River projected changes in urban and residential development result
in little change in model-simulated total streamflow volume, but the 100-year peak flow
increases by nearly 25%.
1 nn nn
J.UU.UU
J^
'S
c
cc
>- 1 n nn .
QJ 1U.UU
E
of Winter to Sum
i-*
b
o
g
+j
re
QC
01 n
. J.U
£
LL.
U
^
*
A
*
M
<
__.
' '
* '
O
o
U
w
03
ID
I
I
1
1
1
1
1
1
1
1
1
1
1
1
1 o
1
1
':•
-li—
i
i
i
i
i
i
i
i
i
i
i
i
i
C _OJ
= LU
1
•4-J
C
o
X.
c
c
X
- >
OJ
i
.__
QD
NewEn
X
->*'-
I
c
o
o
*
O3
(5
O
ft
__.
U
TO
l/l
s
,__
U
o
LO
•4-J
o
LO
i
O"
tn
LO
I
-^
H
(U
L_
O3
__,
C
H
.__
n
o
U
Q.
Q.
^^
• •
"^J"
^
.__
O3
i
/- j.
^Current
• 2
A3
X4
X5
• 6
Figure 7-1. Ratio of winter (January-March) to summer (July-September)
runoff volume under current and mid-21st century NARCCAP climate
scenarios.
Notes: Results are averages over all HUC-8s simulated within a study area. Climate scenarios are (RCM and GCM): (1)
CRCM_cgcm3, (2) HRM3_hadcm3, (3) RCM3_gfdl, (4) GFDL High Res_gfdl, (5) RCM3_cgcm3, and (6) WRFP_ccsm.
The simulated watershed responses to projected mid-21st century urban and residential
development are shown in (see Table 7-16). Results across all 20 watersheds are small, as would
be expected given the small changes in developed lands, when expressed as a fraction of total
watershed area, at the scale of modeling in this study. Larger effects are likely in smaller
subbasins within the study areas where urban and residential development is concentrated. Note
that results are not available for the Kenai River (Cook Inlet, AK study area) because ICLUS
projections do not include Alaska.
7-19
-------
|3.0
CL
O
CD ^-°
"ro
CD
"ro
or
1_
CD
1 ^ "
W
i_
CD
"£ 1 n
C I .U
§
£ 0.5
n n
X
X
X
I
T |
I ^ J
X
1 2 3
J
L
I
I
X
4
5
6
Climate Scenario
Figure 7-2. Box plots of the distribution of the ratio of winter
(January-March) to summer (July-September) runoff volume normalized to
the ratio under current conditions.
Notes: The box shows the interquartile range, with median indicated by a horizontal line, and the whiskers extend 1.5 times the
interquartile range. Outliers beyond the whiskers are shown by individual points. The data are averages over all HUC-8s
simulated within a study area. Climate scenarios are (RCM and GCM): (1) CRCM_cgcm3, (2) HRM3_hadcm3, (3) RCM3_gfdl,
(4) GFDL High Res_gfdl, (5) RCM3_cgcm3, and (6) WRFP_ccsm.
7.4. RELATIVE EFFECTS OF CLIMATE CHANGE AND URBAN DEVELOPMENT
SCENARIOS
The changes in urban and residential development projected by ICLUS for 2050 suggest changes
may be large locally but are small relative to the area of basins modeled in this study (see Table
5-5). Urban and residential development has long been recognized as a source of hydrologic
changes and water quality degradation at local scales in developing areas (e.g., U.S. EPA, 1984).
The cumulative impacts of development, however, tend to be relatively small at the larger basin
scale evaluated in this study simply because only a small fraction of most HUC-4 scale
watersheds is developed or projected to be developed by 2050.
7-20
-------
Table 7-15. Projected mid-21s century impervious cover changes in study
areas from ICLUS for A2 emissions storyline
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
Current (2001)
impervious cover (%)
2.04
0.19
0.24
2.50
6.19
3.48
3.24
1.06
0.38
5.59
0.08
0.55
0.73
13.80
2.06
1.50
1.70
4.17
0.37
2.51
Projected mid-21s* century impervious
cover (%)
3.06
0.30
ND
3.86
8.22
3.88
4.56
1.28
0.39
6.74
0.08
0.81
0.95
19.11
4.27
1.69
2.55
7.37
0.61
3.06
Change in
impervious cover (%)
1.02
0.11
ND
1.36
2.03
0.40
1.32
0.22
0.01
1.15
0.00
0.26
0.22
5.31
2.21
0.19
0.85
3.20
0.24
0.55
The relative magnitude of effects from urban development versus climate change in our
simulations can be examined by looking at changes in mean annual streamflow. Figure 7-3
compares the HSPF simulated change in mean annual streamflow in the pilot study areas for
mid-21st century urban and residential development compared to the six NARCCAP climate
change scenarios. The results summarize the range of responses across selected HUC-8
subbasins and calibration locations contained within each study area. Table 7-17 compares the
range of SWAT simulated changes in mean annual streamflow in all study locations for mid-21st
century urban and residential development and the six NARCCAP climate change scenarios.
Results summarize the ranges at the HUC-8 and larger scale within the study areas.
7-21
-------
Table 7-16. Simulated response to projected 2050 changes in urban and residential development (percent or
days relative to current conditions) for selected downstream stations
Station
Apalachicola R at outlet
Salt River near Roosevelt
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Minnesota R at outlet
Elkhorn R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Los Angeles R at outlet
S. Platte R at outlet
Susquehanna R at outlet
Neuse R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
Study
area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
Total flow
(%)
100.3
100.1
ND
100.3
102.4
100.5
100.8
100.2
100.3
100.4
100.0
100.1
100.1
101.4
102.8
100.2
101.7
106.4
100.1
99.9
7-day low
flow (%)
100.4
100.0
ND
99.9
104.0
100.8
102.6
100.3
100.3
100.5
100.0
100.1
100.1
101.3
100.7
100.7
105.2
188.1
100.6
100.0
100-yr peak
flow (%)
100.3
100.2
ND
100.6
102.1
101.4
101.6
99.9
101.5
101.4
100.0
100.4
99.9
114.4
101.1
99.7
102.1
74.2
100.3
100.1
Days to flow
centroid
-0.1
0.1
ND
0.3
1.0
0.2
0.2
0.3
0.0
0.0
0.0
0.0
-0.1
0.0
0.9
0.1
0.7
3.7
-0.1
0.0
Richards-Baker
flashiness (%)
100.0
100.3
ND
99.5
98.4
100.9
100.4
100.1
102.8
101.3
100.0
100.2
100.4
103.9
103.9
100.1
99.1
68.8
99.8
100.7
TSS load
(%)
100.6
100.2
ND
100.4
100.5
100.6
98.7
98.0
100.1
101.2
100.0
101.1
99.7
106.6
103.9
100.2
102.3
61.9
100.0
99.7
TP load
(%)
101.1
100.4
ND
108.9
100.2
101.3
106.8
99.3
100.1
103.8
100.0
95.4
102.1
138.2
104.0
99.7
106.7
110.0
100.8
99.9
TN load
(%)
100.5
100.2
ND
102.5
99.2
99.6
103.9
99.5
99.8
102.0
100.0
99.6
104.7
111.1
103.4
99.2
103.3
106.2
100.2
102.5
to
to
-------
Simulations using both HSPF and SWAT show a smaller range of response to projected future
changes in urban development than to projected climate change. As discussed previously, at the
spatial scale of these simulations projected future changes in developed land were a relatively
small fraction of total watershed area. At smaller spatial scales, however, the effects of urban
and residential development could be greater. Results for pollutant loads are similar to those for
streamflow.
The simulated response to land-use change is also sensitive to model choice—or, more precisely,
an interaction between the model and the way in which the ICLUS is interpreted. In the SWAT
setup, there are representations of both directly connected (effective) and disconnected
impervious area. New developed land use implied by ICLUS is identified to the model as a total
area in a given development density class, then subdivided by the model into pervious and
impervious fractions using basin-specific estimates of total and effective impervious area. The
effective impervious fraction for a given development category is calculated from the 2000
NLCD and assumed invariant. The model then assumes that the effective impervious area has a
curve number of 98, while the remaining disconnected impervious area provides a small
modification to the curve number assigned to the pervious fraction of the HRU.
In contrast, HSPF has pervious (PERLND) and impervious (IMPLND) land uses, but does not
distinguish a separate disconnected impervious class. For HSPF, the new developed area in
ICLUS is assigned to the relevant pervious and impervious land-use fractions based on the
basin-specific percent imperviousness for the land-use class. In essence, this means that
somewhat greater future connected imperviousness is being specified to the HSPF model than is
specified to the SWAT model. While the two approaches are rather different, they are consistent
with typical modeling practice for the two models.
Several other details of the SWAT modeling process adopted in this study affect results. The
approach to implementing changes in urban development in SWAT was to remove land from
existing undeveloped and nonexempt land uses and reassign it to new developed classes that
have the parameters of the most dominant soil and lowest HRU slope in the subbasin. In some
cases (particularly when a subbasin is already largely developed) the dominant soil in the
watershed may have characteristics different from the soils and slopes of the remaining
undeveloped land. For HSPF, the urban land uses are not associated with a specific soil or HSG.
In addition, a special circumstance occurs in the Willamette SWAT model. In that model, new
developed land primarily comes from dense forest cover. The model tends to simulate greater
evapotranspiration for urban grass than for intact evergreen forest, which appears to offset
increases in total streamflow volume due to increased impervious area.
The effects of land-use change on simulated streamflow extremes can be more dramatic in basins
where strong growth is expected, but also tend to be smaller than the range of simulated climate
responses. For example, in the ACF basin, land-use change alone can increase the simulated
100-year flood peak by up to 27%, but the range of responses to the six NARCCAP climate
scenarios is from 17 to 66%.
7-23
-------
100%
80o/0 |-~r— -c-h_ange_due_tp_
urbanization
60% + -/
o
_ 40% +-
CO
3
i 20% +-
<
S o%
o>
1 -20%
|> -40%
CD
u -60%
-80%
-100% --
Change due to
climate
Figure 7-3. Comparison of simulated responses of mean annual streamflow
to urban development and climate change scenarios—HSPF model.
Note: The blue area represents the range of responses to the six NARCCAP RCM-downscaled 2050 climate scenarios across the
different HUC-8 scale reporting sites (with no change in land use). The red bars represent the maximum response to land-use
change among the reporting sites (with no change in climate). Results are shown for Apalachicola River at outlet (ACF), Sat
River near Roosevelt (Ariz), Minnesota River at outlet (Minn), Susquehanna River at outlet (Susq), and Willamette River at
outlet (Willa).
7.5. SENSITIVITY TO COMBINED CLIMATE CHANGE AND URBAN
DEVELOPMENT SCENARIOS
This section presents the results of SWAT simulations in all 20 study areas for the combined
effects of mid-21st century climate change and urban and residential development scenarios.
Simulation results are generally consistent with results for climate scenarios alone (presented in
Section 7.2.) given the relatively small response to projected urban and residential development
at the spatial scale of modeling in this study. Results are presented for selected locations in each
study area in Tables 7-18 through 7-27. For study sites comprised of a single watershed, results
are shown for a downstream outlet. For study sites comprised of multiple adjacent basins results
are shown for a single representative basin, typically the largest. These same results for each
study area are also shown as scatterplots in Figures 7-4 through 7-24, followed by maps showing
the simulated median values for the six NARCCAP scenarios at the HUC-8 scale within study
areas. It should be noted that use of the median values alone without taking into account the full
range of simulated responses to all scenarios is potentially misleading. Median values are
presented here only as an indicator of variability between study areas and should not alone be
considered indicative of broad regional trends. It should also be noted that simulation results for
7-24
-------
Kenai River in the Cook Inlet basin do not include urban and residential development scenarios.
ICLUS projections are not available for the Alaska study area, but are anticipated to be small.
Table 7-17. Simulated range of responses of mean annual streamflow to mid-
21st century climate and land-use change at the HUC-8 and larger spatial
scale
ACF
Ariz
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
Climate Change Response
Minimum (%)
-45.73
-35.29
-39.73
-22.20
-22.89
-24.75
-23.39
-79.14
-12.55
-42.49
-45.38
-20.79
-26.91
-53.04
-23.80
-13.65
-60.57
-20.21
-17.51
Maximum (%)
24.84
152.52
69.85
34.00
72.13
21.82
85.38
72.64
19.80
206.01
19.86
10.29
62.19
59.23
25.79
61.60
125.65
22.93
23.21
Land-Use Change Response
Minimum (%)
0.00
0.00
0.01
0.00
0.00
0.00
0.00
0.00
0.02
0.00
-0.07
-0.03
-3.60
-1.00
0.00
0.28
7.09
-0.38
-1.18
Maximum (%)
0.68
1.48
7.36
11.90
1.84
1.24
0.19
0.27
0.76
0.00
0.13
0.47
6.36
2.82
0.23
4.31
34.91
0.47
0.00
Note: Cook Inlet basin is not shown because ICLUS land-use change information is not available. Results based on SWAT
simulations for the six NARCCAP climate change scenarios and ICLUS 2050 projected changes in developed land.
The simulated ranges of total streamflow volume changes shown in Figure 7-4 suggest several
observations. The first is that increases in streamflow volume for the Kenai River (Cook Inlet
basin) are on average larger than for other basins. Perhaps more importantly, for a majority of
the basins the different downscaled models do not provide a consistent sign for changes in
streamflow for the 2041-2070 period, with some simulating increases and some decreases. The
models are in complete agreement as to the sign of change only for Kenai River (increase). It is
7-25
-------
also worth noting that the Weather Research and Forecasting Model (WRFP) downscaling of the
CCSM GCM often seems to be an outlier relative to the other models.
Figure 7-5 shows the median simulated annual streamflow volume (as the median over the six
NARCCAP scenarios; expressed as percent of baseline conditions) at the HUC-8 spatial scale for
each study area. On this map, a neutral gray tone represents no change from current conditions
(100% of current conditions). Browns indicate streamflow volumes less than current, with
greater color intensity reflecting lower streamflow; blues represent flow volumes greater than
current, with greater intensity reflecting higher flows. Simulated median values suggest a
general trend of decreasing streamflow volume in the central Rockies, accompanied by increases
in streamflow in the northern plains. Only moderate changes are seen for the west coast and
Mississippi Valley, while streamflow volume generally increases on the east coast.
In addition to streamflow volume, changes in the timing and rate of streamflow can also affected
by climate change. At a national scale, the number of days to the streamflow centroid—the point
at which half the streamflow volume of an average year is achieved (calculated from the
October 1 start of the water year)—is a useful measure of changes in the seasonal distribution of
streamflow. Figure 7-11 shows that the centroid of streamflow comes earlier in the year in
model-simulated response to warmer temperatures for many of the snow-melt dominated basins,
particularly Cook Inlet in Alaska and higher elevations in the Rockies, but also for many basins
in the southeast. The latter result reflects changes in precipitation timing, with increased winter
precipitation and decreased summer precipitation. Several of the western basins have later dates
for the streamflow centroid due to a substantial increase in model-simulated spring or summer
precipitation relative to winter snowpack that counteracts the effects of earlier snowmelt.
Appendices X and Y provide more detailed information about seasonal shifts in streamflow
timing in the study areas.
The geographic distribution of 100-year peak flows (Log-Pearson III) fit is displayed in Figure
7-9 and shows considerably more heterogeneity. Simulated peak flows increase in many basins,
but show less of a clear pattern (see Figure 7-8). Peak flows tend to decline in the area of the
Southwest where total streamflow volumes decline, while the greatest increases are seen in
Alaska and the populated areas of the east and west coast. The increase in 100-year peak flows
is generally greater (or, in some instances, the reduction less) than the change in total streamflow
volume, consistent with the findings of Taner et al. (2011) for Lake Onondaga.
Results also suggest a large (factor of 5) increase in low flows for the Kenai River (see Figure
7-6). This reflects greater dry season melt rates of ice under a warmer climate in Alaska. The
models also consistently show large declines in low flows for the Rio Grande Valley.
7-26
-------
Table 7-18. Simulated total streamflow volume (climate and land-use change scenarios; percent relative to
current conditions) for selected downstream stations
Station
Apalachicola R at outlet
Salt River near Roosevelt
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Minnesota R at outlet
Elkhorn R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Los Angeles R at outlet
S. Platte R at outlet
Susquehanna R at outlet
Neuse R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
CRCM_
cgcm3 (%)
107
80
ND
115
96
117
96
110
117
108
101
73
104
92
92
109
104
102
86
116
HRM3
hadcm3 (%)
122
80
154
154
126
151
111
113
126
116
85
69
89
140
76
107
160
150
95
106
RCM3
gfdl (%)
108
149
ND
128
103
120
116
147
137
111
140
112
98
104
92
106
138
110
116
104
GFDL slice
(%)
89
75
132
93
104
136
85
86
68
112
70
66
98
103
67
108
111
66
89
92
RCM3_
cgcm3 (%)
124
94
ND
157
106
123
107
146
138
106
130
69
100
107
110
111
127
122
92
114
WRFP_
ccsm (%)
73
73
167
75
79
89
78
162
143
94
240
84
99
85
121
90
88
138
91
98
Median
(%)
108
80
154
122
103
122
102
130
131
110
115
71
99
103
92.27
108
119
116
91
105
to
-------
to
-------
VO
0 200 400 800
Kilometers
Figure 7-5. Median simulated percent changes in total future streamflow volume for six NARCCAP scenarios
relative to current conditions by HUC-8 (median of NARCCAP climate scenarios with urban development).
Note: Cook Inlet results do not include land-use change.
-------
Table 7-19. Simulated 7-day low flow (climate and land-use change scenarios; percent relative to current
conditions) for selected downstream stations
Station
Apalachicola R at outlet
Salt River near Roosevelt
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Minnesota R at outlet
Elkhorn R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Los Angeles R at outlet
S. Platte R at outlet
Susquehanna R at outlet
Neuse R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
CRCM_
cgcm3 (%)
98
58
ND
105
88
105
76
115
119
112
102
81
101
98
94
92
100
33
85
131
HRM3
hadcm3 (%)
120
77
267
141
126
184
108
137
133
141
92
64
91
115
88
121
175
199
94
113
RCM3
gfdl (%)
105
131
ND
121
100
127
91
202
152
131
145
120
95
99
98
105
139
87
122
108
GFDL
slice (%)
86
87
280
95
94
133
77
82
48
119
67
62
96
100
75
90
118
36
86
82
RCM3_
cgcm3 (%)
113
79
ND
136
103
129
92
182
148
125
127
74
99
101
103
108
129
93
92
127
WRFP_
ccsm (%)
64
90
401
78
73
59
64
228
154
121
235
86
93
93
114
87
74
102
91
102
Median
(%)
101
83
280
113
97
128
84
159
141
123
115
77
95
99
96
98
123
90
91
111
-------
ylfVW
4L/U/0
_C
m
(0 3 r\r\ O/
fQ J(JL)/o
•5
•N
§ 250%
p
o_
> T rmo/
5 ZUU/o
^2
^L
J£
O -] Rf)Q/
s
i nn^
lUU/o
rpiO/
JU/O "
nQ£
u/o
I
_
X
j;
zl
O
15
a:
_ro
7;
LL.
1
A
<^k
•K
01
ie: Maume
LLJ
^J
^H
•
JZ
0
E
ro
a:
OJ
.t;
0
Q.
^J
•
A
m
X
^
•
X
j-
0
E
re
a;
ro
i: Minnesot
c
I
A
JZ
zt
0
E
15
TV
C.
0
UJ
01
1
JZ
ZI
E
ro
OL
^
: Merrimac
00
c
LLJ
3
(U
*
A
•
X
JZ
ro
OL
O)
00
O
Lz
o
J
o
0.
A
I
HI
Z
2^
UJ
ZI
cr
n
5
Grande at ,
0
'cc
o
be
T
j;
^
o
ra
u.
i
ramento Ri
u
re
1/1
U
re
LO
•
i.
JZ
ZI
o
In
cc
"3
00
c
10
3
ro
u
0
—
•
^•^
f
A
c
o
1_
HI
c
OJ
X
•r^
(D
C£
HI
IS
0.
ro
0.
o
1/1
i
*
V
JZ
^
0
E
ro
Q^
ro
usque hann
i/>
1
t/)
A
A
V
az
D
0
E
15
0£
HI
ra
ZI
HI
zi
HI
ro
i —
_
^p
^
*
¥
JZ
Z)
o
15
oe
±i
1—
t:
1—
A
T
0)
c
—
a
ra
ro
K
c
Colorado R
o
u
a.
Q.
"
^.
If
••
™"
~~
X
JZ
D
E
ro
ae
OJ
: Willamett
_ro
i
Climate Scenario
* l-CRCM_cgcm3
• 2-HRM3_hadcm3
4 3-RCM3_gfdl
,1 /trpil cliro
H-OrUL bMCc
t 5-RCM3_cgcm3
*6-WRFP_ccsm
— Median
Figure 7-6. Simulated 7-day low flow relative to current conditions (NARCCAP climate scenarios with urban
development) for selected downstream stations.
-------
to
0 200 400 800
Kilometers
7-day Low Flow
Median Change (%)
Figure 7-7. Median simulated percent changes in 7-day average low flow volume for six NARCCAP scenarios
relative to current conditions by HUC-8 (median of NARCCAP climate scenarios with urban development).
Note: Cook Inlet results do not include land-use change.
-------
Table 7-20. Simulated 100-year peak flow (log-Pearson III; climate and land-use change scenarios; percent
relative to current conditions) for selected downstream stations
Station
Apalachicola R at outlet
Salt River near Roosevelt
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Minnesota R at outlet
Elkhorn R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Los Angeles R at outlet
S. Platte R at outlet
Susquehanna R at outlet
Neuse R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
CRCM_
cgcm3 (%)
117
119
ND
131
121
96
107
84
128
116
118
90
105
100
132
108
71
97
78
116
HRM3
hadcm3 (%)
145
101
132
145
155
107
152
83
117
134
113
77
98
112
126
130
294
107
83
131
RCM3 gfdl
(%)
110
104
ND
130
109
88
110
96
110
113
133
108
122
194
101
107
163
108
97
114
GFDL
slice (%)
90
68
125
95
103
94
100
87
93
141
82
66
117
124
129
129
113
60
91
79
RCM3_
cgcm3 (%)
128
121
ND
158
129
94
107
89
139
90
121
72
102
158
163
173
227
87
93
116
WRFP_
ccsm (%)
94
66
132
107
98
93
66
96
102
82
146
92
131
93
152
101
64
107
84
95
Median
(%)
114
102
132
130
115
94
107
88
114
115
119
83
111
118
131
118
138
102
87
115
-------
250%
a
200% -
41
O.
0
o
150%
100%
0%
Climate Scenario
•» l-CRCM_cgcm3
• 2-HRM3_hadcm3
- 3-RCM3jgfdl
4-GFDL_slice
i 5-RCM3_cgcm3
6-WRFP_ccsm
— Median
Figure 7-8. Simulated 100-year peak flow relative to current conditions (NARCCAP climate scenarios with
urban development) for selected downstream stations.
-------
0 200 400 800
Kilometers
Figure 7-9. Median simulated percent changes in 100-year peak flow for six NARCCAP scenarios relative to
current conditions by HUC-8 (median of NARCCAP climate scenarios with urban development).
Note: Cook Inlet results do not include land-use change.
-------
Table 7-21. Simulated change in the number of days to streamflow centroid (climate and land-use change
scenarios; relative to current conditions) for selected downstream stations
Station
Apalachicola R at outlet
Salt River near Roosevelt
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Minnesota R at outlet
Elkhorn R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Los Angeles R at outlet
S. Platte R at outlet
Susquehanna R at outlet
Neuse R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
CRCM_
cgcm3
-2
-18
-3
-11
-2
-14
-13
-12
-17
-6
25
-4
6
-11
-18
-13
17
-11
o
6
HRM3_
hadcm3
-2
41
-3
17
6
-4
14
-19
6
-14
-3
6
-7
48
-19
16
23
23
-14
-8
RCM3_ gfdl
1
28
25
-2
1
-23
-6
1
-19
1
3
-4
-3
-13
-6
31
31
-7
-1
GFDL_ slice
8
17
-5
-8
-12
0
-7
-15
-15
-13
-16
11
-1
10
-18
-12
-11
4
-10
o
6
RCM3_
cgcm3
-6
-5
-5
-1
10
-5
o
— J
-6
-9
-4
14
o
— J
-3
-2
-5
11
7
-8
1
WRFP_
ccsm
1
54
-1
11
-14
-8
-11
2
2
-18
7
17
-8
0
-11
0
-5
25
-11
8
Median
-1
22
-3
4
-6
-1
-9
-9
-2
-16
-3
13
-4
3
-12
-6
3
20
-10
2
-------
fin
•o
0
£ ™
c 30 -
a
u. 2U
a
« m
Q
c
O
C
n
5
~f-\J
X
ff
X
—
^
0
E
*-*
ro
DC
IT!
Apalachicol
'•^
m
.
A
X
.
^
Ol
Si
o
0
DC
C
i
DC
1
f
_c
D
0
E
to
DC
C
QJ
O
0
u
A
^rv
\/
_c
^
0
E
fi
cc:
(U
la: Suwane
re
H
i
t
c
1
o
Ol
CO
ra
oc
'o
c
'^
^
I
j=
^
0
E
TO
QC
Ol
E
ro
.
±1
c
I—
iz
£
f
OJ
— '
Ol
m
—
j_
C
Colorado R
o
u
a.
Q.
—*
^
.n
^
Q
E
ro
DC
01
: Willamett
JS
•—
*
Climate Scenario
* l-CRCM_cgcm3
• 2-HRM3 hadcm3
* 3-RCM3_gfdl
4-GFDL slice
i 5-RCM3 cgcm3
£. WRPP rrtm
— Median
Figure 7-10. Simulated change in days to streamflow centroid relative to current conditions (NARCCAP climate
scenarios with urban development) for selected downstream stations.
-------
oo
0 200 400 800
Kilometers
Change in Days
To Flow Centroid
Figure 7-11. Median simulated change in the number of days to streamflow centroid for six NARCCAP
scenarios relative to current conditions by HUC-8 (median of NARCCAP climate scenarios with urban
development).
Note: Cook Inlet results do not include land-use change.
-------
Table 7-22. Simulated Richards-Baker flashiness index (climate and land-use change scenarios; percent relative
to current conditions) for selected downstream stations
Station
Apalachicola R at outlet
Salt River near Roosevelt
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Minnesota R at outlet
Elkhorn R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Los Angeles R at outlet
S. Platte R at outlet
Susquehanna R at outlet
Neuse R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
CRCM_
cgcm3 (%)
106
81
ND
93
105
100
105
105
97
102
102
109
124
104
103
107
95
71
101
102
HRM3
hadcm3 (%)
125
103
94
62
103
102
105
112
101
104
108
117
103
125
95
111
112
69
107
105
RCM3 gfdl
(%)
109
121
ND
76
102
100
106
108
97
100
104
95
113
103
105
107
114
72
111
100
GFDL
slice (%)
94
98
102
116
106
101
104
101
96
102
100
120
109
105
91
110
97
73
105
98
RCM3_
cgcm3 (%)
126
103
ND
59
105
100
104
109
98
99
103
103
117
108
113
112
102
70
103
101
WRFP_
ccsm (%)
90
119
96
185
103
97
102
108
97
94
109
106
124
104
110
103
90
68
101
102
Median
(%)
107
103
96
84
104
100
104
108
97
101
104
108
115
105
104
109
100
70
104
102
VO
-------
g
1
u
•5
c
5
a.
X
c
m
EC
snai
60%
/i no/
H
.c
^
0
E
ts
en
re
': Apalachicol
.
*
X
^^
a
a)
;/>
O
0
1=
§
be:
•H*
M
X
*
.c
^
0
E
ts
L-.
C
(U
o
0
u
X
^M
A
•
I
^
D
0
E
ts
o:
01
aFIa: Suwane
a
i
c
1
o
ro
fl
CO
ts
EC
l/l
'o
c
=
A
_z
0
E
—
OS
CY"
U
Erie: Maume
t
,
j
_z
0
E
ts
L-
acramento Ri
u
m
•
i
.c
^
0
E
ts
rf
v>
•il: LosAngele
u
0
i
s
c
1_
^
c
111
X
—
cc
1)
t;
ro
0.
t/i
ro
0.
L/1
1
.c
^
0
E
ns
rf*
ro
Susquehann
^
1/1
_n
D
0
E
ts
(Y
OJ
VI
a
01
id
m
•z.
i™
£
jr
^
0
E
ts
C£
.t!
C
H
c
H
i
0)
— '
01
m
1_
C
1: Colorado R
o
u
a.
a.
j;
D
E
ts
LT
0)
la: Willamett
g
Climate Scenario
* 1-CRCM cgcm3
• 2-HRM3_hadcm3
3-RCM3_gfdl
4-GFDL_slice
1 5-RCM3_cgcm3
• 6-WRFP_ccsm
— Median
Figure 7-12. Simulated Richards-Baker flashiness index relative to current conditions (NARCCAP climate
scenarios with urban development) for selected downstream stations.
-------
0 200 400 800
Kilometers
Flasniness Index
Absolute Change
Figure 7-13. Simulated absolute changes in the Richards-Baker flashiness index for six NARCCAP scenarios
relative to current conditions by HUC-8 (median of NARCCAP climate scenarios with urban development).
Note: Cook Inlet results do not include land-use change.
-------
Regional differences also occur in the degree of agreement among simulated watershed
responses to climate change scenarios. Table 7-23 shows the CV (standard deviation divided by
the mean) for SWAT-simulated percentage changes in different streamflow endpoints at the
downstream location of each study site for the six NARCCAP scenarios (calculated without
land-use change to isolate the impacts of climate). The CV for total streamflow is large at some
stations, such as Salt River and Tongue River, indicating poor model agreement on the
magnitude of change. Note that CVs on total streamflow are artificially reduced at some stations
(e.g., Colorado River, Sacramento River) due to the presence of constant upstream boundary
conditions (representing interbasin transfers for the Colorado and releases from an upstream dam
on the Sacramento River). The largest divergences among simulated high flows are seen at
different stations than the largest divergences among total streamflow volume estimates.
CVs were also calculated reflecting the variability in response across the selected downstream
stations for all study areas for each NARCCAP climate change scenario. Table 7-24 shows these
values along with the average absolute difference from the median of all scenarios for each
NARCCAP scenario. For total streamflow volume, the CCSM downscaled with WRFP has both
the greatest station-to-station variability (highest CV) and largest average absolute difference
from the median of all six simulations.
Simulated changes in pollutant loads are shown in Tables 7-25 through 7-27, and Figures 7-14
through 7-19. Changes in projected pollutant loads are qualitatively similar to those seen for
response to climate change only, but further increased in areas with significant new urban
development. In general, projected changes in pollutant loads follow a pattern similar to the
changes in total streamflow volume. Total suspended solids loads (see Figure 7-15) increase in
most basins, except for declines in the Rocky Mountain and Southwest study areas where overall
streamflow decreases. The large increases in solids loads for some basins (especially sand bed
rivers in the west) are mostly driven by channel scour. These results should be considered highly
uncertain given the simplified approach to channel scour included in SWAT version 2005 and
the differences among individual models in calibration to channel scour. The regional pattern for
total phosphorus loads is similar, as much of the total phosphorus load is driven by erosion (see
Figure 7-17), with the notable exception of the Cook Inlet basin in Alaska. The regional pattern
for total nitrogen loads is also generally similar, with some additional variability associated with
the interactions of plant growth and erosion (see Figure 7-19).
Changes in the timing of nutrient load delivery may be even more significant for ecological
impacts (c.f, Tu, 2009; Wilson and Weng, 2011; Tong et al., 2011; Marshall and Randhir, 2008).
Potential ecological impacts of changes in timing of pollutant delivery simulated in the rich data
set generated by this study remain to be evaluated.
7.6. WATER BALANCE INDICATORS
Several additional endpoints—identified here as water balance indicators—were calculated for
each study area. Water balance indicators are defined in Section 4.3. This section presents
results describing the SWAT-simulated changes in these indicators in response to the six mid-
21st century NARCCAP climate change and urban development scenarios.
Table 7-28 provides a summary of water balance indicators for each study area. Figures 7-20
through 7-24 show the median values for changes in water balance metrics for simulations using
7-42
-------
the six NARCCAP climate change scenarios at each study location. As stated previously,
median values are presented here only as an indicator of variability within and among study
areas and should not alone be considered indicative of broad regional trends. Appendices X and
Y provide more detailed results for changes in water balance indicators including analysis at
additional locations in each study area.
Table 7-23. Coefficient of variation of SWAT-simulated changes in
streamflow by study area in response to the six NARCCAP climate change
scenarios for selected downstream stations
Station
Apalachicola R at outlet
Salt River near Roosevelt
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Minnesota R at outlet
Elkhorn R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Los Angeles R at outlet
S. Platte R at outlet
Susquehanna R at outlet
Neuse R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
Total flow
0.038
0.091
0.021
0.089
0.023
0.035
0.023
0.066
0.064
0.005
0.293
0.039
0.003
0.032
0.044
0.005
0.055
0.079
0.013
0.008
100-yr peak
0.037
0.060
0.001
0.043
0.039
0.004
0.070
0.004
0.024
0.046
0.039
0.028
0.016
0.100
0.029
0.057
0.534
0.036
0.006
0.030
7-day low flow
0.043
0.067
0.172
0.053
0.033
0.137
0.029
0.198
0.128
0.009
0.273
0.056
0.001
0.005
0.019
0.017
0.101
0.378
0.020
0.028
7-43
-------
Table 7-24. Coefficient of variation of SWAT-simulated changes in
streamflow by NARCCAP climate scenario for selected downstream stations
RCM/GCM
CRCM_cgcm3
HRM3_hadcm3
RCM3_gfdl
GFDL_slice
RCM3_cgcm3
WRFP_ccsm
Total flow
CV
0.016
0.068
0.026
0.049
0.036
0.167
Average absolute
difference from
median (%)
14.66
15.38
19.54
18.37
16.06
25.25
100-yr peak flow
CV
0.032
0.166
0.035
0.048
0.108
0.063
Average absolute
difference from
median (%)
14.97
19.76
18.52
17.56
25.00
19.83
7-day low flow
CV
0.058
0.163
0.073
0.264
0.068
0.571
Average absolute
difference from
median (%)
27.95
23.45
27.32
20.10
21.52
31.89
7-44
-------
Table 7-25. Simulated total suspended solids load (climate and land-use change scenarios; percent relative to
current conditions) for selected downstream stations
Station
Apalachicola R at outlet
Salt River near Roosevelt
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Minnesota R at outlet
Elkhorn R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Los Angeles R at outlet
S. Platte R at outlet
Susquehanna R at outlet
Neuse R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
CRCM_
cgcm3 (%)
126
89
ND
121
117
123
99
104
122
119
108
61
138
75
95
118
108
64
80
124
HRM3
hadcm3 (%)
147
79
234
177
142
170
113
117
131
129
84
54
94
121
91
108
201
126
90
111
RCM3
gfdl (%)
128
184
ND
139
115
127
125
183
147
119
169
115
121
86
98
109
164
64
124
108
GFDL
slice (%)
93
66
196
90
128
154
82
76
60
123
66
50
118
85
84
116
117
28
82
89
RCM3_
cgcm3 (%)
145
106
ND
182
121
130
110
192
162
112
153
60
99
90
104
118
145
85
89
121
WRFP_
ccsm (%)
53
74
244
74
91
87
70
219
162
86
351
72
108
69
108
85
84
115
85
97
Median
(%)
127
84
234
130
119
128
104
150
139
119
131
60
113
86
97
112
131
74
87
110
-------
^ nn°/
ttj
"HJ
m T C-O0/,
CO
|
c
H
£
•o
O
i
crw
X
.E
^
0
E
ro
LL.
fD
^palachicol
|
<
A
X
t
^
(D
y,
o
0
CC
c
§
en
'>_
^f
t
X
j:
^
0
E
ro
ry
—
c
O
0
u
i
T
X
.c
^
0
E
-
C£.
su
la: Suwane
to
In
I
*
C
1
o
t>
ra
u
DC
ce
w
'o
c
_
—
u
X
•
X
^
0
E
ro
CC
a>
ie: Maume
LLJ
—^
i
. -_.J
j:
D
0
E
re
fY*
-•
.ti
0
D.
I
5
X
.n
D
0
E
-
CC
re
: Minnesot
c
Z-
t
x
j:
^
0
E
15
OL
C
0
.i.-
LJJ
01
i
•JB
.n
^
TO
c;
_i/
: Merrimac
c
LLJ
=J
*
^
D
0
E
TO
ff-
HJ
v\
a
±
b
aj
2
I™
1
-.
X
j:
^
0
E
T5
ry
-^
,t!
H
c
™
A
0)
(U
m
j_
C
Colorado R
o
u
Q.
i
*
x
^
Q
E
%
ry
OJ
: Willamett
JS
•^
5
• l-CRCM_cgcm3
• 2-HRM3 hadcm3
. 3-RCM3_gfdl
4-GFDL slice
1, 5-RCM3_cgcm3
*6-WRFP ccsm
— Median
Figure 7-14. Simulated total suspended solids load relative to current conditions (NARCCAP climate scenarios
with urban development) for selected downstream stations.
-------
0 200 400 800
Kilometers
Figure 7-15. Median simulated percent changes in total suspended solids loads for six NARCCAP scenarios
relative to current conditions by HUC-8 (median of NARCCAP climate scenarios with urban development) for
selected downstream stations.
Note: Cook Inlet results do not include land-use change.
-------
Table 7-26. Simulated total phosphorus load (climate and land-use change scenarios; percent relative to current
conditions) for selected downstream stations
Station
Apalachicola R at outlet
Salt River near Roosevelt
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Minnesota R at outlet
Elkhorn R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Los Angeles R at outlet
S. Platte R at outlet
Susquehanna R at outlet
Neuse R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
CRCM_
cgcm3 (%)
139
82
ND
125
107
121
123
97
118
116
107
51
102
78
93
128
123
148
80
100
HRM3
hadcm3 (%)
153
84
89
190
112
155
144
115
124
125
86
40
88
128
81
106
259
188
88
98
RCM3
gfdl (%)
136
156
ND
149
107
136
147
151
138
116
163
125
106
102
104
110
184
153
120
97
GFDL_
slice (%)
119
70
90
96
113
151
103
97
65
120
67
49
117
83
75
127
134
98
82
94
RCM3_
cgcm3 (%)
150
107
ND
189
108
120
125
138
145
111
148
37
97
89
113
114
183
155
84
100
WRFP_
ccsm (%)
107
88
113
82
99
89
89
160
148
97
324
64
110
71
115
108
103
187
84
96
Median
(%)
138
86
90
137
107
128
124
126
131
116
127
50
104
86
99
112
158
154
84
97
oo
-------
VO
,nno.
jsUU/o
a /!jU/o
M
(U
00
•!•• T Ar\0/
^ 200%
•g
u
Cu 1 cn0/
i IjU/o
•a
F 1 nn°/
1 UU /o
n%
u/o
X
.E
ZJ
0
S
a:
IB
0
.y
jz
u
ro
a
0.
~
§
A
>
*i
be
o
a
E
ro
u
ro
LO
(J
ro
•
A
IK
f
j:
^
0
ts
ce
in
W>
C
<£
w
"ro
U
0
A
T
A
^•^^
c
o
£
u
c
u
I
<-
IB
££
OD
a.
^
00
a.
o
in
-.. .
f
•
JZ
^
0
It!
rv
ro
c
c
ro
_r
S
cr
Ul
ZJ
1
[/I
A
—
^fc
^
j=
^
0
s
ae
on
r
ai
zi
HI
IB
|
B
*
*. x
JR^
JZ
"^
0
ro
ce
±i
i—
iz
K
•
T
01
.^
09
m
ro
t/i
c
ce
o
•o
"o
i,
o
u
a.
^
^^•P
j;
"D
g
ro
ce
u
•*-*
E
i
JB
i
Climate Scenario
* l-CRCM_cgcm3
• 2-HRM3_hadcm3
4 3-RCM3_gfdl
4-QFDL_slice
••1 5-RCM3_cgcm3
6-WRFP_ccsm
Figure 7-16. Simulated total phosphorus load relative to current conditions (NARCCAP climate scenarios with
urban development) for selected downstream stations.
-------
0 200 400 800
Kilometers
Total Phosphorus
Median Change (%)
Figure 7-17. Median simulated percent changes in total phosphorus loads for six NARCCAP scenarios relative
to current conditions by HUC-8 (median of NARCCAP climate scenarios with urban development).
Note: Cook Inlet results do not include land-use change.
-------
Table 7-27. Simulated total nitrogen load (climate and land-use change scenarios; percent relative to current
conditions) for selected downstream stations
Station
Apalachicola R at outlet
Salt River near Roosevelt
Kenai R at Soldotna
Suwanee R at outlet
Illinois R at Marseilles, IL
Maumee R at outlet
Amite R at outlet
Minnesota R at outlet
Elkhorn R at outlet
Merrimack R at outlet
Tongue R at outlet
Rio Grande R below Albuquerque
Sacramento R at outlet
Los Angeles R at outlet
S. Platte R at outlet
Susquehanna R at outlet
Neuse R at outlet
Trinity R at outlet
Colorado R near State Line
Willamette R at outlet
Study area
ACF
Ariz
Cook
GaFla
Illin
LErie
LPont
Minn
Neb
NewEng
PowTon
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
CRCM_
cgcm3 (%)
117
90
ND
129
103
127
130
126
93
123
109
50
104
125
89
161
120
140
73
106
HRM3
hadcm3 (%)
126
91
200
167
117
158
152
130
97
131
91
38
94
159
72
146
207
187
82
98
RCM3 gfdl
(%)
116
142
ND
139
105
161
153
163
145
121
165
127
105
154
95
146
166
142
111
97
GFDL slice
(%)
107
87
175
113
109
190
113
104
88
124
71
48
113
102
65
155
125
93
76
91
RCM3_
cgcm3 (%)
123
105
ND
171
107
125
127
158
104
116
148
37
103
96
112
149
155
153
80
105
WRFP_
ccsm (%)
96
85
223
86
93
94
95
170
107
103
320
65
111
101
120
131
105
186
79
95
Median
(%)
117
91
200
134
106
142
128
144
100
122
128
49
104
113
92
147
140
148
80
97
-------
to
Q> ZjU/o
"3
U)
re
CO
NM
v
U
IH
^ 150% -
•o
1
fc i nn°/s
soy
f
•
.E
13
0
15
Of
ra
^palachicol
A
X
t
*i
s
o
0
cc
c
1
'u-
"*
X
.E
13
0
— •
ro
[Y1
—
C
.ii
O
0
U
i
„.
X
•
.E
0
ro
C£
111
la: Suwane
ro
<2
^B
£
AV
•F
c
o
ra
CJ
CD
CSL
i/i
'o
C
=
X
.A.
*
.c
0
ro
at
03
ie: Maume
—
M
X
A
V
^
0
15
a:
.*;
0
a.
I
9
x
J5
0
ro
a:
ro
: Minnesot
5
_ . .
"A"
-
f
j:
13
0
15
p^
n
0
UJ
01
*
9
^
1
15
nr-
: Merrimac
DC
C
LLJ
3
z
A
X
X
.E
13
ro
cr
0)
ttO
o
F-
o
5
a.
A
U
E"
01
13
f-i
ej|
HI
C
5
.0
ro
(D
Q
C£
^^^
1
j:
0
15
^
>
rarnento Ri
U
ro
to
U
ro
ui
*
1
.c
13
0
ts
cc
Los Angele
ro
U
0
^
*
c
ELI
C
(U
X
—
UJ
i
s
uo
ro
0.
1/1
4
j:
0
ro
fY
ro
usquehann
m
1
uo
B
..
¥
•
_n
0
15
oc
on
3
(U
z
13
m
m
H
\v-
X
.E
0
ro
cc:
.ti
l—
if
K
f
-------
0 200 400 800
Kilometers
Figure 7-19. Median simulated percent changes in total nitrogen loads for six NARCCAP scenarios relative to
current conditions by HUC-8 (median of NARCCAP climate scenarios with urban development).
Note: Cook Inlet results do not include land-use change.
-------
Table 7-28. Simulated percent changes in water balance statistics for study
areas (NARCCAP climate with land-use change scenarios; median percent
change relative to current conditions)
Study Area
ACF
Ariz-Salt
Ariz-San Pedro
Ariz-Verde
Cook
GaFla-North
GaFla-Tampa
Illin
LErie
LPont
Minn
Neb-Elkhorn
Neb-Loup
NewEng
PowTon-Powder
PowTon-Tongue
RioGra
Sac
SoCal
SoPlat
Susq
TarNeu
Trin
UppCol
Willa
Dryness Ratio
(fraction of
precipitation lost to
ET) (%)
0
1
-1
-2
-8
-10
-6
-1
-3
-10
-5
0
o
— 3
-i
-7
-6
2
0
-2
-1
0
-8
-4
1
-11
Low Flow Sensitivity
(baseflow generation,
cfs/mi2) (%)
-16
-10
-7
-3
22
47
8
-7
22
59
28
3
12
-6
18
5
-28
-4
-5
-6
-6
15
-1
-8
5
Surface Runoff
Fraction of
Flow (%)
22
-5
23
7
4
-8
11
15
-4
-14
49
16
-1
1
-1
6
o
J
4
7
1
16
5
2
-4
1
Snowmelt
Fraction of
Flow (%)
-57
-46
-52
-50
-12
-32
-72
-39
-32
-22
-24
-24
-33
-82
-18
-17
-1
-45
-54
-17
-31
-49
-43
-15
-68
Deep Recharge
Rate
(depth) (%)
-14
-15
-12
4
-43
39
7
-6
20
47
24
1
13
-5
NA
-8
-28
-6
1
NA
-5
15
0
-16
6
7-54
-------
0 200 400 800
Kilometers
Figure 7-20. Median simulated percent changes in watershed Dryness Ratio for six NARCCAP scenarios
relative to current conditions (median of NARCCAP climate scenarios with urban development).
Note: Dryness ratio is the fraction of input precipitation lost to ET. Cook Inlet results do not include land-use change.
-------
*^-v
Ar'T
^k ex
0 200 400 800
Kilometers
Figure 7-21. Median simulated percent changes in watershed Low Flow Sensitivity for six NARCCAP scenarios
relative to current conditions (median of NARCCAP climate scenarios with urban development).
Note: Low Flow Sensitivity is the rate of streamflow generation by baseflow (cfs/mi2). Cook Inlet results do not include land-use change.
-------
0 200 400 800
Kilometers
Surface Runoff Fraction of Flow
Median Change (%)
Figure 7-22. Median simulated percent changes in watershed Surface Runoff Fraction for six NARCCAP
scenarios relative to current conditions (median of NARCCAP climate scenarios with urban development).
Note: Surface Runoff Fraction is the fraction of streamflow contributed by overland flow pathways. Cook Inlet results do not include land-use change.
-------
oo
0 200 400 800
Kilometers
Snow Fraction
Median Change (%)
Figure 7-23. Median simulated percent changes in watershed Snowmelt Fraction for six NARCCAP scenarios
relative to current conditions (median of NARCCAP climate scenarios with urban development).
Note: Snowmelt Fraction is the fraction of streamflow contributed by snowmelt. Cook Inlet results do not include land-use change.
-------
VO
0 200 400 800
Kilometers
Figure 7-24. Median simulated percent changes in watershed Deep Recharge for six NARCCAP scenarios
relative to current conditions (median of NARCCAP climate scenarios with urban development).
Note: Deep Recharge is the depth of water recharging deep aquifers per unit time. Cook Inlet results do not include land-use change. Areas shown in black have no deep
recharge simulated.
-------
The water balance summaries are presented as averages over whole watersheds. These are
generally consistent with the project study areas, except that several study areas (e.g., Central
Nebraska) were simulated using more than one SWAT model and thus show multiple results.
Figure 7-20 shows the change in the Dryness Ratio, expressed as the ratio of ET to precipitation.
The central tendency of the Dryness Ratio is estimated to increase in the southern Rocky
Mountains and adjacent parts of Arizona, consistent with median decreases in simulated mean
annual streamflow (see Figure 7-4).
Another aspect of low flows is shown by the Low Flow Sensitivity metric—the average rate of
baseflow generation per square mile of watershed area. This metric (see Figure 7-21) decreases
in areas for which the Dryness Ratio increases. However, it also decreases in various other
watersheds (such as SoCal and ACF) for which there is little change in the Dryness Ratio. Areas
where the Low Flow Sensitivity metric decreases may be expected to experience difficulties in
maintaining minimum streamflow for aquatic life support or for meeting wasteload dilution
expectations.
The Surface Runoff Fraction (the fraction of streamflow contributed by overland flow pathways)
increases strongly for various study areas on the east coast and some other areas, mostly due to
intensification of rainfall events in climate models (see Figure 7-22). Study areas for which the
Surface Runoff Fraction strongly increases, such as ACF and Ariz-San Pedro, are those where
the Low Flow Sensitivity decreases despite relatively small changes in the Dryness Ratio.
Snowmelt Fraction, the fraction of runoff that is due to melting snow (see Figure 7-23) declines
in all watersheds. The strongest percentage declines (in southern and coastal areas) are
somewhat misleading, as these watersheds generally have small amounts of snow. The lesser
percentage declines throughout the Rockies are of greater concern to water management in the
west.
The combination of a greater fraction of surface runoff in many watersheds coupled with
increased dryness and reduced total streamflow in many western watersheds leads to a reduction
in projected Deep Recharge (rates of recharge to deep aquifers) in many study areas (see Figure
7-24). The risks are estimated to be particularly acute in the Rockies and the ACF basins. In
other areas, increased precipitation in the models counteracts other forces through mid-century,
including the critical recharge areas in central Nebraska.
7.7. MODELING ASSUMPTIONS AND LIMITATIONS
Model simualtions in the study provide an improved understanding of streamflow and water
quality sensitivity in different regions of the United States to a range of plausible mid-21st
century climate change and urban development scnearios. The study also illustrates certain
challenges associated with the use of watershed models for conducting scenario-based studies of
climate change impacts. In the process, this study adds to our knowledge of how to implement
such investigations.
A number of sources of uncertainty must be considered in interpreting results from watershed
hydrologic and water quality simulations of response to climate change—including uncertainty
in the emissions scenario, uncertainty in the GCM simulations of future climate, uncertainty in
the downscaling of these GCM outputs to the local scale, and uncertainty in the watershed
7-60
-------
models used to translate potential changes in local climate to watershed response. The strong
dependence of streamflow and water quality on climate drivers (e.g., temperature, precipitation,
etc.) means that accurate weather data is necessary to generate accurate estimates of future flow
and water quality conditions. Inherent in the scenario approach to modeling climate futures is
uncertainty in knowledge of future climate conditions. It is therefore necessary to choose a range
of scenarios that reflect the full, plausible set of future conditions.
Simulation results showed a wide range of watershed responses to differences in climatic
forcing. Results suggest the variability resulting from scenarios based on different methods of
downscaling with a single GCM can be of the same order of magnitude as the variability among
GCMs. In many cases, simulations for scenarios based on different downscaling approaches
with a single GCM do not agree even in the direction of projected changes relative to current
values. In part, this issue reflects the skill associated with RCM simulations. A recent study by
Racherla et al. (2012) investigated the value added by using an RCM (the Weather Research and
Forecasting or WRF model) with the GCM GISS-ModelE2 and concluded that the RCM does
not achieve holistic improvement in the simulation of seasonally and regionally averaged surface
temperature or precipitation for historical data. They further suggested that no strong
relationship exists between skill in capturing climatological means and skill in capturing climate
change. If RCMs do not add considerable value to the global simulation, the underlying
uncertainties can only be reduced by improving the global-scale climate simulations.
As with any study of this type, simulation results are conditional upon the specific methods,
models, and scenarios used. The simulated range of response in this study is limited by the
particular set of climate model projections available in the NARCCAP archives (the subset of
BCSD projections was selected to match those in the NARCCAP set). For example, all climate
change scenarios evaluated in this study are based on the IPCC A2 greenhouse gas emissions
storyline. While simulations in this study represent a credible set of plausible future climatic
conditions, the scenarios evaluated should not be considered comprehensive of all possible
futures. A recent summary by Mote et al. (2011) concludes that ensemble scenarios with a
limited number of projections taken from the full set of available climate models yields results
that differ little from those achieved from larger sets given the current state of science;
furthermore, attempting to preselect the "best" models based on measures of model skill does
little to refine the estimate of central tendency of projected change. Mote et al. recommend a
sample size of approximately ten climate scenarios, which is greater than the six used in this
study. Inclusion of additional sources and types of scenarios could alter the ranges of change
simulated in this study. Similarly, alternative urban and residential development scenarios would
also expand the ensemble range of future responses.
Watershed model simulations developed here also do not consider feedback effects of human
and ecological adaptation to change. In essence, the climate-land use-watershed system is
considered independent of management and adaptation in this study. At the most direct level,
various aspects of human water management such as operation of dams, water use,
transboundary water inputs, and point source discharges are considered fixed at present levels.
In fact, we know these will change. For instance, a warmer climate is likely to result in
increased irrigation withdrawals for crops, while more intense precipitation is likely to result in
changes in operating rules for dams. In some cases, the models are driven by fixed upstream
boundary conditions (e.g., the Sacramento River model). There was, however, insufficient
7-61
-------
knowledge of these changes to incorporate them into the scenarios. The analyses thus provide an
increased understanding of the marginal changes in watershed responses due to potential changes
in climate and urban and residential development, but do not account for the net changes from all
factors, including human use and management of water.
At a more sophisticated level, both natural and human communities are likely to adapt to climate
changes, influencing the watershed response. The SWAT plant growth model takes into account
the effects of changed climate on plant growth as a function of CO2, temperature, water stress,
and nutrient availability. However, it does not take into account changes in the type of land
cover that may occur as a result of such stresses—either slowly, as through a gradual shifting of
ecological niches, or catastrophically, as might occur through drought-induced forest fires.
Human adaptations that affect watershed processes will also occur. For example, crop types (or
total area in crops) are likely to change as producers respond to changes in growing season
length and water availability (e.g., Polsky and Easterling, 2001). Simulation models are not yet
available to provide a credible analysis of such feedback loops at the scale necessary for
evaulating watershed responses.
In addition, many of the modeled study areas are highly managed systems influenced by dams,
water transfers and withdrawals, and point and nonpoint pollution sources. Given the difficulty
inherent in modeling watershed response at the large spatial scale used in this study, detailed
representation of all management and operational activities was not possible. Results therefore
represent the potential response of watersheds to different change scenarios, but should not be
considered quantitative forecasts of future conditions.
7.7.1. Model Calibration
Reliably reproducing the baseline period is important for any study of watershed response to
climate change because any biases present in the model calibration are likely to also affect the
future simulations of streamflow (Prudhomme and Davies, 2009), possibly with nonlinear
amplification. The experiences of this project emphasize the importance (and challenges) of
calibration and validation for watershed models. Water quality calibration is particularly
challenging due to limited amounts of readily available monitoring data. Additional efforts
similar to the one presented here should either focus on watersheds for which well-calibrated
models already exist (and the effort of assembling water quality input and monitoring data from
multiple sources has already been completed) or allocate sufficient time and budget to conduct
detailed, site-specific calibration.
The calibration process can introduce modeler bias, which could be mitigated through use of an
automated model calibration scheme. We avoided this option based on past experience with the
SWAT and HSPF models in which automated calibration often converges to physically
unrealistic model parameter sets. It may, however, be advisable to pursue stepwise, guided
model calibration with carefully specified parameter constraints to avoid the effects of user bias,
as was done, for example, in recent USGS simulations of watershed-scale streamflow response
to climate change using the PRMS model (Hay et al., 2011). PRMS, however, only addresses
streamflow and has a much more parsimonious data set than does SWAT or HSPF. Nonetheless,
the advantages of controlling for modeler bias may make use of a semiautomated calibration
procedure desirable.
7-62
-------
The significance of calibration bias is mitigated by focusing on projected changes relative to
baseline conditions as compared to actual future values. If biases are consistent and linear
between the baseline and future condition, the effect of such biases will tend to cancel out when
relative change is calculated. There is, however, no guarantee that biases will be linear. Further
testing to evaluate the effects of alternative model calibrations on the simulated response of
different study areas would be desirable.
7.7.2. Watershed Model Selection
Simulation results are sensitive to the watershed model applied. In the pilot studies, both HSPF
and SWAT appeared capable of providing similar quality of fit to observed streamflow at the
large basin scale and to pollutant loads at the monthly scale, while HSPF, using a shorter time
step, was better able to resolve streamflow at smaller spatial scales and better able to match
observed concentrations when fully calibrated. An important result of model comparisons
conducted in this study is the significant effect that increased atmospheric CO2 concentrations
(effects of reduced stomatal conductance that decrease ET) appeared to have on the water
balance. SWAT's integrated plant growth model takes this effect into account, whereas HSPF
does not.
It is unclear, however, how well SWAT is able to represent the complex processes affecting
plant growth, nutrient dynamics, and water budgets under changing climate. For example, as
CC>2 levels increase, leaf level reductions in stomatal conductance and evapotranspiration may be
offset by increased plant growth and leaf area. The effects of CC>2 on plant growth may also be
altered over time due to nutrient limitation (Reich et al., 2006). Further study is required to
better understand how climate change will affect these processes. It should also be noted that
SWAT (as implemented here, using version SWAT2005) has limitations in its representation of a
number of important watershed processes, including simplified simulation of direct runoff using
a curve number approach, erosion prediction with MUSLE that does not fully incorporate
changes in energy that may occur with altered precipitation regimes, and a simplistic
representation of channel erosion processes that appears unlikely to provide a firm foundation for
simulating channel stability responses to climate change. More recent versions of SWAT
considerably expand the options for simulating channel erosion, but do not appear to be fully
validated at this time and are limited by the model's use of a daily time step for hydrology.
These considerations suggest that a more sophisticated watershed model formulation, combining
a plant growth model (as in SWAT) with a more detailed hydrologic simulation would be
preferable for evaluating watershed responses to climate change. However, even if such a model
was available, fully validated, and ready for use, it would likely require a significantly higher
level of effort for model implementation and calibration.
Comparison of change scenarios using HSPF and SWAT suggests one must proceed with
caution when attempting to estimate even relative aggregate impacts at a national scale through
use of watershed models with different underlying formulations. For example, a national
synthesis that drew conclusions from a mix of models, some of which did and others of which
did not include explicit simulation of effects of increased CC>2 on evapotranspiration, could reach
erroneous conclusions regarding the relative intensity of impacts in different geographical areas.
7-63
-------
8. SUMMARY AND CONCLUSIONS
This report describes watershed modeling in 20 large, U.S. drainage basins (6,000-27,000 mi2 or
15,000-60,000 km2) to characterize the sensitivity of streamflow, nutrient (nitrogen and
phosphorus) loading, and sediment loading to a range of potential mid-21st century climate
futures, to assess the potential interaction of climate change and urbanization in these basins, and
to improve our understanding of methodological challenges associated with integrating existing
tools (e.g., climate models, downscaling approaches, and watershed models) and data sets to
address these scientific questions. Study areas were selected to represent a range of geographic,
hydroclimatic, physiographic, land use, and other watershed attributes. Other important criteria
used in site selection included the availability of necessary data for calibration and validation of
watershed models, and opportunities for leveraging the availability of preexisting watershed
models.
Models were configured by subdividing study areas into modeling units, followed by continuous
simulation of streamflow and water quality for these units using meteorological, land use, soil,
and stream data. A unique feature of this study is the use of a consistent watershed modeling
methodology and a common set of climate and land-use change scenarios in multiple locations
across the nation. Models in each study area are developed for current (1971-2000) observed
conditions, and then used to simulate results under a range of potential mid-21st century
(2041-2070) climate change and urban development scenarios. Watershed modeling was
conducted at each study location using the SWAT model and six climate change scenarios based
on dynamically downscaled (50 x 50 km2) output from four of the GCMs used in the IPCC 4
Assessment Report for the period 2041-2070 archived by the NARCCAP. Scenarios were
created by adjusting historical weather series to represent projected changes in climate using a
change factor approach. To explore the potential interaction of climate change and urbanization,
simulations also include urban and residential development scenarios for each of the 20 study
watersheds. Urban and residential development scenarios were acquired from EPA's national-
scale ICLUS project.
In a subset of five study areas (the Minnesota River, the Susquehanna River, the
Apalachicola-Chattahoochee-Flint, the Salt/Verde/San Pedro, and the Willamette River Basins),
additional simulations were conducted to assess the variability in simulated watershed response
resulting from use of different watershed models and different approaches for downscaling GCM
climate change scenarios. In these study areas, watershed simulations were also run with eight
additional scenarios derived from the same four GCMs used in NARCCAP: four scenarios
interpolated to station locations directly from the GCM output, and four scenarios based on
BCSD statistically downscaled climate projections described by Maurer et al. (2007). In
addition, in these five study areas, all scenario simulations were run independently with a second
watershed simulation model, the HSPF.
Given the large size of study areas, calibration and validation of all models was completed by
first focusing on a single HUC-8 within the larger study area (preferably one with a good record
of streamflow gaging and water quality monitoring data), and then extending the calibration to
adjacent areas with modifications as needed to achieve a reasonable fit at multiple spatial scales.
3-1
-------
Large-scale GCM projections are generally consistent in showing a continued warming trend
over the next century (although with sometimes significant regional-scale disagreements in the
magnitude of this warming), but offer a much wider range of plausible outcomes in other aspects
of local climate—particularly the timing and intensity of precipitation and the energy inputs (in
addition to air temperature) that determine potential evapotranspiration—that interact to create
watershed responses.
The simulated watershed responses to these changes provide an improved understanding of
system sensitivity to potential climate change and urban development scenarios in different
regions of the country and provide a range of plausible future hydrologic and water quality
change scenarios that can be applied in various planning and scoping frameworks. The results
illustrate a high degree of regional variability in the response of different streamflow and water
quality endpoints to a range of potential mid-21st century climatic conditions in different regions
of the nation. Watershed hydrologic response is determined by the interaction of precipitation
and evapotranspiration, while water quality response is largely dependent on hydrology.
Comparison of simulations in all 20 study areas for the 2041-2070 time horizon suggest
potential streamflow volume decrease in the Rockies and interior southwest, and increases in the
east and southeast coasts. Wetter winters and earlier snowmelt are likely in many of the northern
and higher elevation watersheds. Higher peak flows will also increase erosion and sediment
transport; nitrogen and phosphorus loads are also likely to increase in many watersheds.
Both the selection of an underlying GCM and the choice of downscaling method have a
significant influence on the streamflow and water quality simulations. In many cases, the range
of simulated responses across the different climate models and downscaling methodologies do
not agree in direction. The ultimate significance of any given simulation of future change will
depend on local context, including the historical range of variability, thresholds and management
targets, management options, and interaction with other stressors. The simulation results in this
study do, however, clearly illustrate that the potential streamflow and water quality response in
many areas could be large.
Watershed simulations were run in all study areas with and without projected mid-21st century
changes in urban and residential development. These results suggest that at the HUC-8 spatial
scale evaluated in this study, watershed sensitivity to projected urban and residential
development will be small relative to the changes resulting from climate change. It is important,
however, to qualify this result. The finest spatial scale reported in this study is that of an 8-digit
HUC, and most urbanized areas are located on larger rivers downstream of multiple 8-digit
HUCs. Over the whole of individual study areas, urban and residential growth scenarios
represented changes in the amount of developed land on the order of <1 to about 12% of total
watershed area and increases in impervious surfaces on the order of 0 to 5% of total watershed
area. The effects of urban development on adjacent water bodies at higher levels of development
are well documented. It is thus likely that at smaller spatial scales within study areas where the
relative fraction of developed land is greater, the effects of urbanization will be greater.
Identifying the scale at which urbanization effects become comparable to the effects of a
changing climate is an important topic for future research.
The simulation results also illustrate a number of methodological issues related to impacts
assessment modeling. These include the sensitivities and uncertainties associated with use of
8-2
-------
different watershed models, different approaches for downscaling climate change simulations
from global models, and the interaction between climate change and other forcing factors, such
as urbanization and the effects of changes in atmospheric CO2 concentrations on
evapotranspiration. Uncertainty associated with differences in emission scenarios and climate
model sensitivities is well known and widely discussed in previous assessments of climate
change impacts on water (e.g., IPCC, 2007; Karl et al., 2009). This study illustrates a potentially
significant additional sensitivity of watershed simulations to the method selected for
downscaling GCM model output. Results of the intercomparison of climate change data sets
suggest that the variability between downscaling of a single GCM with different RCMs can be of
the same order of magnitude as the ensemble variability between GCMs.
This study also suggests potentially important sensitivity of results to the use of different
hydrologic models (HSPF and SWAT in this study), associated with differences in process
representation, such as accounting for the influence of increased atmospheric CO2 on
evapotranspiration. One notable insight from these results is that, in many watersheds, climate
change (when precipitation amount and/or intensity is altered), increasing urbanization, and
increasing atmospheric CC>2 can have similar or additive effects on streamflow and pollutant
loading (e.g., a more flashy runoff response with higher high flows and lower low flows). The
results, while useful as guidance for designing and conducting similar impacts assessment
studies, are only a first step in understanding what are likely highly complex and context-
dependent relationships. Further study and evaluation of the implications of these and other
questions is necessary for improving the plausibility and relevance of coupled climate-hydrology
simulations, and ultimately for informing resource managers and climate change adaptation
strategies.
The model simulations in this study contribute to a growing understanding of the complex and
context-dependent relationships between climate change, land development, and water in
different regions of the nation. As a first order conclusion, results indicate that in many locations
future conditions are likely to be different from past experience. In the context of decision
making, being aware and planning for this uncertainty is preferable to accepting a position that
later turns out to be incorrect. Results also provide a plausible envelope on the range of
streamflow and water quality responses to mid-21st century climate change and urban
development in different regions of the nation. In addition, in many study areas the simulations
suggest a likely direction of change of streamflow and water quality endpoints. This information
can be useful in planning for anticipated but uncertain future conditions. The sensitivity studies
evaluating different methodological choices help to improve the scientific foundation for
conducting climate change impacts assessments, thus building the capacity of the water
management community to understand and respond to climate change.
Understanding and responding to climate change is complex, and this study is only an
incremental step towards fully addressing these questions. It must be stressed that results are
conditional upon the methods, models, and scenarios used in this study. Scenarios represent a
plausible range but are not comprehensive of all possible futures. Several of the study areas are
also complex, highly managed systems; all infrastructure and operational aspects of water
management are not represented in full detail. Finally, changes in agricultural practices, water
demand, other human responses, and natural ecosystem changes such as the prevalence of forest
fire (e.g., Westerling et al., 2006) or plant disease that will influence streamflow and water
8-3
-------
quality are not considered in this study. Further study is required to continue to build the
scientific foundation for assessing these and other questions relevant to the scientific and
watershed management communities.
Successful climate change adaptation strategies will need to encompass practices and decisions
to reduce vulnerabilities across a wide range of plausible future climatic conditions. Where
system thresholds are known, knowledge of the range of potential changes can help to identify
the need to consider future climate change in water planning. Many of these strategies might
also help reduce the impacts of other existing stressors. It is the ultimate goal of this study to
build awareness of the potential range of future watershed response so that where simulations
suggest large and potentially disruptive changes, the management community will respond to
build climate resiliency.
8-4
-------
REFERENCES
Ainsworth, EA; Rogers, A. (2007) The response of photosynthesis and stomatal conductance to
rising [CC>2]: mechanisms and environmental interactions. Plant Cell Environ
30:258-270.
Alexander, LV; Zhang, X; Peterson, TC; et al. (2006) Global observed changes in daily climate
extremes of temperature and precipitation. J Geophys Res 111(D05109). doi:
10:1029/2005JD006290.
Alexander, RB; Smith, RA; Schwarz, GE; et al. (2008) Differences in phosphorus and nitrogen
delivery to the Gulf of Mexico from the Mississippi River Basin. Environ Sci Technol
42(3):822-830.
Allan, RP; Soden, BJ. (2008) Atmospheric warming and the amplification of precipitation
extremes. Science 321(5895): 1481-1484.
Allan, RJ. (1986) The role of particulate matter in the fate of contaminants in aquatic
ecosystems. Scientific Series 142. Inland Waters Directorate, Environment Canada,
Ottawa.
Allen, RG; Pereira, LS; Raes, D; et al. (1998) Crop evapotranspiration: guidelines for computing
crop water requirements. Irrigation and Drainage Paper 56. Food and Agriculture
Organization of the United Nations, Rome. Available online at http://www.engr.scu.edu/
~emaurer/classes/ceng!40_watres/handouts/F AO_56_Evapotranspiration.pdf
Allen, RG; Walter, IA; Elliott, RL; et al. (2005) The ASCE standardized reference
evapotranspiration equation. Reston, VA: American Society of Civil Engineers.
Anandhi, A; Frei, A; Pierson DC; et al. (2011) Examination of change factor methodologies for
climate change impact assessment. Water Resour Res
47(W03501):doi:10.1029/2010WR009104.
Baker, DB; Richards, P; Loftus, TT; et al. (2004) A new flashiness index: characteristics and
applications to Midwestern rivers and streams. J Am Water Resour Assoc
40(2):503-522.
Berg, EE; Henry, JD; Fastie, CL; et al. (2006) Spruce beetle outbreaks on the Kenai Peninsula,
Alaska, and Kluane National Park and Reserve, Yukon Territory: relationship to summer
temperatures and regional differences in disturbance regimes. For Ecol Manage
227:219-232.
Bernacchi, CJ; Kimball, BA; Quarles, DR; et al. (2007) Decreases in stomatal conductance of
soybean under open-air elevation of [CO2] are closely coupled with decreases in
ecosystem evapotranspiration. Plant Physiol 143:134-144.
R-l
-------
Bicknell, BR; Imhoff, JC; Kittle JL, Jr; et al. (2001) Hydrological simulation program—Fortran
(HSPF). User's manual for release 12. U.S. Environmental Protection Agency, National
Exposure Research Laboratory, Athens, GA, in cooperation with U.S. Geological Survey,
Water Resources Division, Reston, VA.
Bicknell, BR; Imhoff, JC; Kittle JL, Jr; et al. (2005) HSPF version 12.2 user's manual. U.S.
Environmental Protection Agency, National Exposure Research Laboratory, Office of
Research and Development, Athens, GA.
Bosch, NS; Allan, JD; Dolan, DM; et al. (2011) Application of the Soil and Water Assessment
Tool for six watersheds of Lake Erie: model parameterization and calibration. J Great
Lakes Res 37:263-271.
Bouraoui, F; Galbiati, L; Bidoglio, G. (2002) Climate change impacts on nutrient loads in the
Yorkshire Ouse catchment (UK). Hydrol Earth Syst Sci 6:197-209.
Burwell, RE; Timmons, DR; Holt, RF. (1975) Nutrient transport in surface runoff as influenced
by soil cover and seasonal periods. Soil Sci Soc Am J 39(3):523-528.
Cao, L; Bala, G; Caldeira, K; Nemani, R; Ban-Weiss, G. (2010) Importance of carbon dioxide
physiological forcing to future climate change. PNAS 107(21):9513-9518.
CECWG (Commission for Environmental Cooperation Working Group). (1997) Ecological
regions of North America—toward a common perspective. Commission for
Environmental Cooperation, Montreal. Available online at http://www.calepa.ca.gov/
education/eei/Curriculum/Gradel 1/1186/1186EnvGuide.pdf
Chang, H; Evans, BM; Easterling, DR. (2001) The effects of climate change on streamflow and
nutrient loading. J Am Water Resour Assoc 37(4):973-985.
Cox, P; Stephenson, D. (2007) A changing climate for prediction. Science 317(5835):207-208.
Crawford, NH; Linsley, RK. (1966) Digital simulation in hydrology: Stanford watershed model
IV. Technical Report 39. Department of Civil Engineering, Stanford University, CA.
Available online at http://www.hydrocomp.com/publications/StanfordModelIV.PDF
Dai, A. (2006) Precipitation characteristics in eighteen coupled climate models. J Clim
19:4605-4630.
Donigian, AS, Jr. (2000) HSPF training workshop handbook and CD. Lecture #19: calibration
and verification issues. U.S. Environmental Protection Agency Headquarters,
Washington Information Center, 10-14 January, 2000. Prepared for U.S. Environmental
Protection Agency, Office of Water, Office of Science and Technology, Washington, DC.
Duda, P; Kittle, J, Jr; Gray, M; et al. (2001) WinHSPF Version 2.0: an interactive Windows
interface to HSPF (WinHSPF) user's manual. AQUA TERRA Consultants, Decatur, GA.
Available online at http://www.epa.gov/athens/research/ftable/toc_intr.pdf
R-2
-------
Easterling, WE; Rosenberg, NJ; McKenney, MS; et al. (1992) Preparing the erosion productivity
impact calculator (EPIC) model to simulate crop response to climate change and the
direct effects of CO2. Agric For Meteorol 59(1-2): 17-34.
Emori, S; Brown, SJ. (2005) Dynamic and thermodynamic changes in mean and extreme
precipitation under changed climate. Geophys Res Lett 32(L17706)
doi: 10.1029/2005GL023272
Ficklin, DL; Luo, Y; Luedeling, E; et al. (2009) Climate change sensitivity assessment of a
highly agricultural watershed using SWAT. J Hydrol 374(1/2): 16-29.
Follett, RF. (1995) Fate and transport of nutrients: nitrogen. Working Paper No. 7. U.S.
Department of Agriculture, Agricultural Research Service, Soil-Plant-Nutrient Research
Unit, Fort Collins, CO. Available online at http://www.nrcs.usda.gov/wps/portal/nrcs/
detail/national/technical/? cid=nrcs!43_014202
Garen, DC; Moore, DS. (2005) Curve number hydrology in water quality modeling: uses,
abuses, and future directions. J Am Water Resour Assoc 41(2):377-388.
Garrick, M; Cunnane, C; Nash, JE. (1978) A criterion of efficiency for rainfall-runoff models. J
Hydrol 36(3-4):375-381.
Gassman, PW; Reyes, MR; Green, CH; et al. (2007) The Soil and Water Assessment Tool:
historical development, applications, and future research directions. Trans Am Soc Agri
BiolEng 50(4): 1211-1250.
Gesch, D; Oimoen, M; Greenlee, S; Nelson, C; Steuck, M; Tyler, D. (2002) The national
elevation dataset. Photogramm Eng Remote Sens 68(1):5-11.
Gleckler, PJ; Taylor, KE; Doutriaux, C. (2008) Performance metrics for climate models. J
Geophys Res 113(D06104): doi:10.1029/2007JD008972.
Groisman, PY; Knight, RW; Karl, TR. (2012) Changes in intense precipitation over the central
United States. J Hydrometeorol 13:47-66.
Gutowski, WJ; Hegerl, GC; Holland, GJ; et al. (2008) Causes of observed changes in extremes
and projections of future changes. In: Karl, TR; Meehl, GA; Miller, CD; et al; eds.
Weather and climate extremes in a changing climate. Regions of focus: North America,
Hawaii, Caribbean, and U.S. Pacific Islands. Report by the U.S. Climate Change Science
Program. Global Change Research, Washington, DC; p. 81-116. Available online at
http://www.agci.org/dB/PDFs/Publications/07Sl_USCCSP.pdf
Hawkins, E; Sutton, R. (2009) The potential to narrow uncertainty in regional climate
predictions. Bull Am Meteor Soc 90:1095-1107.
Hay, LE; Markstrom, SL; Ward-Garrison, C. (2011) Watershed-scale response to climate change
through the twenty-first century for selected basins across the United States. Earth
Interact 15:1-37.
-------
Homer, C; Huang, C; Yang, L; et al. (2004) Development of a 2001 national landcover database
for the United States. Photogramm Eng Remote Sens 70(7):829-840.
Homer, C; Dewitz, J; Fry, J; et al. (2007) Completion of the 2001 national land cover database
for the conterminous United States. Photogramm Eng Remote Sens 73(4):337-341.
Available online at http://www.epa.gov/mrlc/pdf/april-07-highlight.pdf
Kurd, B; Leary, N; Jones, R; et al. (1999) Relative regional vulnerability of water resources to
climate change. J Am Water Resour Assoc 35(6): 1399-1409.
IPCC (Intergovernmental Panel on Climate Change). (2001) Climate change 2001: the scientific
basis. In: Houghton, JT; Ding, Y; Griggs, DJ; et al.; eds. Intergovernmental panel on
climate change: working group I. Cambridge, UK: Cambridge University Press.
IPCC (Intergovernmental Panel on Climate Change). (2007) Climate change 2007: synthesis
report—summary for policymakers. Available online at:
http://www.ipcc.ch/pdf/assessment-report/ar4/syr/ar4_syr_spm.pdf
Jensen, ME; Burman, RD; Allen, RG. (1990) Evapotranspiration and irrigation water
requirements. ASCE Manuals and Reports on Engineering Practice No. 70. ASCE, NY.
Johnson, T; Weaver, C. (2009) A framework for assessing climate change impacts on water and
watershed systems. Environ Manage 43:118-134.
Karl, TR; Knight, RW. (1998) Secular trends of precipitation amount, frequency, and intensity in
the United States. Bull Am Meteorol Soc 79(2):231-241.
Karl, TR; Melillo, JM; Peterson, TC; eds. (2009) Global climate change impacts in the United
States. Cambridge, UK: Cambridge University Press.
Kharin, VV; Zwiers, FW; Zhang, X; et al. (2013) Changes in temperature and precipitation
extremes in the CMIP5 ensemble. Clim Change 119(2):345-357. DOI 10.1007/sl0584-
013-0705-8
Kundzewicz, ZW; Mata, LJ; Arnell, NW; et al. (2007) Freshwater resources and their
management. In: Parry, ML; Canziani, OF; Palutikof, JP; et al; eds. Climate change
2007: impacts, adaptation and vulnerability. Contribution of Working Group II to the
Fourth Assessment Report of the Intergovernmental Panel on Climate Change.
Cambridge, UK: Cambridge University Press; pp. 173-210.
Leakey, ADB; Ainsworth, EA; Bernacchi, CJ; et al. (2009) Elevated CC>2 effects on plant
carbon, nitrogen, and water relations: six important lessons from FACE. J Exp Bot
60(10):2859-2876.
Legates, DR; McCabe, GJ, Jr. (1999) Evaluating the use of "goodness-of-fit" measures in
hydrologic and hydroclimatic model validation. Water Resour Res 35(1):233-241.
R-4
-------
Lempert, RJ; Groves, DG; Popper, SW; et al. (2006) A general, analytic method for generating
robust strategies and narrative scenarios. Manage Sci 52(4):514-528.
Lumb, AM; McCammon, KB; Kittle, JL, Jr. (1994) User's manual for an expert system
(HSPEXP) for calibration of the hydrological simulation program-FORTRAN. U.S.
Geological Survey Water Resources Investigation Report 94-4168. U.S. Geological
Survey, Reston, VA.
Luo, Y; Ficklin, DL; Liu, X; et al. (2013) Assessment of climate change impacts on hydrology
and water quality with a watershed modeling approach. Sci Total Environ
450-451:72-82.
Marshall, E; Randhir, T. (2008) Effect of climate change on watershed system: a regional
analysis. Clim Change 89(3/4):263-280.
Maurer, EP; Brekke, L; Pruitt, T; et al. (2007) Fine-resolution climate projections enhance
regional climate change impact studies. Eos Trans Am Geophys Union 88(47):504.
Mearns, L. (2009) The North American regional climate change assessment program
(NARCCAP): overview of phase II results. Earth Environ Sci Trans 6(022007):
doi: 10.1088/1755-1307/6/2/022007.
Mesinger, F; DiMego, G; Kalnay, E; et al. (2006) North American regional reanalysis. Bull Am
Meteorol Soc 87(3):343-360.
Milly, PCD; Dunne, K; Vecchia, A. (2005) Global pattern of trends in streamflow and water
availability in a changing climate. Nature 438:347-350.
Milly, PCD; Betancourt, J; Falkenmark, M; et al. (2008) Stationarity is dead: Whither water
management? Science 319:573-574.
Monteith, DT; Evans, CD; Reynolds, B. (2000) Are temporal variations in the nitrate content of
UK upland freshwaters linked to the North Atlantic Oscillation? Hydrol Processes
14(10):1745-1749.
Monteith, JL. (1965) Evaporation and the environment. In: The state and movement of water in
living organisms. XlXth Symposium, Swansea. Soc Exp Biology, Vol 19. Cambridge,
UK. Cambridge University Press; pp. 205-234.
Moriasi, DN; Arnold, JG; Van Liew, MW; et al. (2007) Model evaluation guidelines for
systematic quantification of accuracy in watershed simulations. Trans Am Soc Agri Biol
Eng 50(3):885-900.
Mote, P; Brekke, L; Duffy, PB; et al. (2011) Guidelines for constructing climate scenarios. Eos
Trans Am Geophys Union 92(31):257-258.
Murdoch, PS; Baron, JS; Miller, TL. (2000) Potential effects of climate change on surface-water
quality in North America. J Am Water Resour Assoc 36(2):347-366.
R-5
-------
Nash, JE; Sutcliffe, JV. (1970) River flow forecasting through conceptual models part I—a
discussion of principles. JHydrol 10(3):282-290.
Neitsch, SL; Arnold, JG; Kiniry, JR; et al. (2005) Soil and water assessment tool, theoretical
documentation. Grassland, Soil and Water Research Laboratory, U.S. Department of
Agriculture, Agricultural Research Service, Temple, TX. Available online at
http://swat.tamu.edu/media/1292/swat2005theory.pdf
Paul, MJ; Meyer, JL. (2001) Streams in the urban landscape. Annu Rev Ecol Syst 32:333-365.
Poff, NL; Tokar, S; Johnson, P. (1996) Stream hydrological and ecological responses to climate
change assessed with an artificial neural network. Limnol Oceanogr 41(5): 857-863.
Polsky, C; Easterling, WE. (2001) Adaptation to climate variability and change in the U.S. Great
Plains: a multi-scale analysis of Ricardian climate sensitivities. Agric Ecosyst Environ
85(1-3):133-144.
Preston, SD; Bierman, VJ, Jr; Silliman, SE. (1989) An evaluation of methods for the estimation
of tributary mass loads. Water Resour Res 25(6): 1379-1389.
Prudhomme, C; Davies, H. (2009) Assessing uncertainties in climate change impact analyses on
the river flow regimes in the UK. Part 1: baseline climate. Clim Change
93(1-2):177-195.
Pyke, C; Warren, M; Johnson, T; et al. (2011) Assessment of low impact development for
managing stormwater with changing precipitation due to climate change. Landsc Urban
Plan 103(2): 166-173.
Racherla, PN; Shindell, DT; Faluvegi, GS. (2012) The added value to global model projections
of climate change by dynamical downscaling: a case study over the continental U.S.
using the GISS-ModelE2 and WRF models. J Geophys Res-Atmos 117(D20118):
doi:10.1029/2012JD018091.
Raisanen, J. (2007) How reliable are climate models? Tellus A 59:2-29.
Reich, PB; Hungate, BA; Luo, Y. (2006) Carbon-nitrogen interactions in terrestrial ecosystems
in response to rising atmospheric carbon dioxide. Annu Rev Ecol Evol Syst 37:611-636.
Sarewitz, D; Pielke, RA, Jr; Byerly, R, Jr. (2000) Introduction: death, taxes, and environmental
policy. In: Sarewitz, D; Pielke, RA, Jr; Byerly, R, Jr; eds. Prediction: science, decision
making, and the future of nature. Washington, DC: Island Press.
SCS (Soil Conservation Service). (1972) Hydrology guide for use in watershed planning.
National engineering handbook, section 4: hydrology, supplement A. U.S. Department of
Agriculture, Natural Resources Conservation Service, Washington, DC.
Seaber, PR; Kapinos, FP; Knapp, GL. (1987) Hydrologic unit maps. Water-Supply Paper 2294.
U.S. Geological Survey, Denver, CO.
R-6
-------
Sharpley, AN; Williams, JR; eds. (1990) EPIC—erosion productivity impact calculator, 1.
Model documentation. U.S. Department of Agriculture, Agricultural Research Service.
Tech Bull 1768.
Stainforth, DA; Allen, MR; Tredger, ER; et al. (2007) Confidence, uncertainty, and decision-
support relevance in climate predictions. Phil Trans R Soc A 365:2145-2161.
Stockle, CO; Williams, JR; Rosenberg, NJ; et al. (1992) A method for estimating the direct and
climatic effects of rising atmospheric carbon dioxide on growth and yield of crops: part
1—modification of the EPIC model for climate change analysis. Agric Syst 38:225-238.
Suddick, EC; Davidson, EA; eds. (2012) The role of nitrogen in climate change and the impacts
of nitrogen-climate interactions on terrestrial and aquatic ecosystems, agriculture, and
human health in the United States. Technical report submitted to the U.S. National
Climate Assessment. North American Nitrogen Center of the International Nitrogen
Initiative (NANC-INI), Woods Hole Research Center, Falmouth, MA. Available online at
http://www.whrc.org/resources/publications/pdf/SuddicketalWHRC.12.pdf
Sun, Y; Solomon, S; Dai, A; et al. (2006) How often does it rain? J Clim 19:916-934.
SWCS (Soil and Water Conservation Society). (2003) Conservation implications of climate
change: soil erosion and runoff from cropland. Ankeny, IA: Soil and Water Conservation
Society.
TAMU (Texas A&M University). (2010) SWAT—soil and water assessment tool. Available
online at http://swatmodel.tamu.edu/software/arcswat.
Taner, MU; Carleton, JN; Wellman, M. (2011) Integrated model projections of climate change
impacts on a North American lake. Ecol Model 222(18):3380-3393.
Tetra Tech. (2008a) Quality assurance project plan for watershed modeling to evaluate potential
impacts of climate and land-use change on the hydrology and water quality of major U.S.
drainage basins. Prepared for the Office of Research and Development Global Change
Research Program, U.S. Environmental Protection Agency, Washington, DC.
Tetra Tech. (2008b) Minnesota River basin turbidity TMDL and Lake Pepin excessive nutrient
TMDL, model calibration and validation report. Prepared for Minnesota Pollution
Control Agency, St. Paul, MN.
Tong, STY; Sun, Y; Ranatunga, T; et al. (2011) Predicting plausible impacts of sets of climate
and land-use change scenarios on water resources. Appl Geogr 32(2):477-489.
Trenberth, KE; Dai, A; Rasmussen, RM; et al. (2003) The changing character of precipitation.
Am Meteorol Soc 84:1205-1217.
Trenberth, KE; Jones, PD; Ambenje, P; et al. (2007) Observations: surface and atmospheric
climate change. In: Solomon, S; Qin, D; Manning, M; et al; eds. Climate change 2007:
the physical science basis. Contribution of working group I to the fourth assessment
R-7
-------
report of the intergovernmental panel on climate change. Cambridge, UK and New York:
Cambridge University Press.
Tu, J. (2009) Combined impact of climate and land-use changes on streamflow and water quality
in eastern Massachusetts, USA. J Hydrol 379(3-4):268-283.
USDA (United States Department of Agriculture). (1991) State soil geographic (STATSGO) data
base; data use information. Miscellaneous Publication 1492. National Soil Survey
Center, Natural Resources Conservation Service, Fort Worth, TX.
U.S. EPA (Environmental Protection Agency). (1984) Report to congress: nonpoint source
pollution in the U.S. Office of Water Program Operations, Water Planning Division,
Washington, DC.
U.S. EPA (Environmental Protection Agency). (2000) Estimating hydrology and hydraulic
parameters for HSPF. BASINS Technical Note 6. Office of Water, Washington, DC;
EPA-823-ROO-012. Available online at http://water.epa.gov/scitech/datait/models/basins/
upload/2000_08_14_BASINS_tecnote6.pdf
U.S. EPA (Environmental Protection Agency). (2001) BASINS version 3.0 user's manual.
Office of Water, Washington, DC; EPA-823-B-01-001. Available online at
http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_03_BASINS_b3docs_
usermanual.pdf
U.S. EPA (Environmental Protection Agency). (2002) Nitrogen: multiple and regional impacts.
Clean Air Markets Division, Washington, DC; EPA-430-R-01-006. Available online at
http ://www. epa.gov/airmarkets/resource/docs/nitrogen.pdf
U.S. EPA (Environmental Protection Agency). (2008) Using the BASINS meteorological
database—version 2006. BASINS Technical Note 10. Office of Water, Washington,
DC. Available online at http://water.epa.gov/scitech/datait/models/basins/upload/
2009_04_13_BASINSs_tecnotel0.pdf
U.S. EPA (Environmental Protection Agency). (2009a) BASINS 4.0—fact sheet. Office of
Water, Wasthingtion, DC. Available online at
http://www.epa.gov/waterscience/BASINS/fs-basins4.html (accessed January 27, 2010).
U.S. EPA (Environmental Protection Agency). (2009b) BASINS 4.0 climate assessment tool
(CAT): supporting documentation and user's manual. Global Change Research Program,
National Center for Environmental Assessment, Office of Research and Development,
Washington, DC; EPA-600-R-08-088F. Available online at http://cfpub.epa.gov/ncea/
cfm/recordisplay.cfm?deid=203460
U.S. EPA (Environmental Protection Agency). (2009c) ICLUS V1.2 user's manual: ArcGIS
tools and datasets for modeling US housing density growth. Global Change Research
Program, National Center for Environmental Assessment, Office of Research and
Development, Washington, DC; EPA-600-R-09-143A.
R-8
-------
U.S. EPA (Environmental Protection Agency). (2010) NHDPlus user guide. Office of Water,
Washington, DC. Available online at ftp://ftp.horizon-systems.com/NHDPlus/
NHDPlus V1 /documentation/NHDPLUS V1 JJserGuide. pdf
USGS (United States Geological Survey). (1982) Guidelines for determining flood flow
frequency. Bulletin #17B of the Hydrology Subcommittee, Interagency Advisory
Committee on Water Data, Reston, VA. Available online at http://water.usgs.gov/
osw/bulletin!7b/dl_flow.pdf
Volkery, A; Ribeiro, T. (2009) Scenario planning in public policy: understanding use, impacts
and the role of institutional context factors. Technol Forecast Soc Change
76(9): 1198-1207.
Walsh, CJ; Roy, A; Feminella, J; et al. (2005) The urban stream syndrome: current knowledge
and the search for a cure. J N Am Benthol Soc 24(3):706-723
Wang, S-Y; Gillies, RR; Takle, ES; et al. (2009) Evaluation of precipitation in the Intermountain
Region as simulated by the NARCCAP regional climate models. Geophys Res Lett
36(L11704)doi:10:1029/2009GL037930.
Westerling, AL; Hidalgo, HG; Cayan, DR; et al. (2006) Warming and earlier spring increases
western U.S. forest wildfire activity. Science 313(5789):940-943.
Whitehead, PG; Wilby, RL; Battarbee, RW; et al. (2009) A review of the potential impacts of
climate change on surface water quality. Hydrol Sci 54:101-123.
Wilcox, BP; Rawls, WJ; Brakensiek, DL; et al. (1990) Predicting runoff from rangeland
catchments: a comparison of two models. Water Resour Res 26(10):2401-2410.
Williams, J.R. (1975) Sediment-yield prediction with universal equation using runoff energy
factor. In: Present and prospective technology for predicting sediment yield and sources:
proceedings of the sediment-yield workshop, November 28-30, 1972. U.S. Department
of Agriculture Sedimentation Lab, Oxford, MS. ARSS-40; pp. 244-252. .
Williams, MW; Losleben, M; Caine, N; et al. (1996) Changes in climate and hydrochemical
responses in a high-elevation catchment in the Rocky Mountains, USA. Limnol
Oceanogr 41(5):939-946.
Wilson, CO; Weng, Q. (2011) Simulating the impacts of future land use and climate changes on
surface water quality in the Des Plaines River watershed, Chicago Metropolitan
Statistical Area, Illinois. Sci Total Environ 409(20):4387-4405.
Winchell, M; Srinivasan, R; DiLuzio, M; et al. (2008) ArcSWAT 2.1 interface for SWAT 2005,
user's guide. U.S. Department of Agriculture, Agricultural Research Service, Temple,
TX. Available online at http://www.geology.wmich.edu/sultan/5350/Labs/
ArcSWAT_Documentation.pdf
R-9
-------
United States
Environmental Protection
Agency
National Center for Environmental Assessment (8601)
Office of Research and Development
Washington, DC 20460
-------
Appendix A
Model Setup Process
1 Overview
This memorandum describes the protocol to ensure an efficient and consistent model setup process
to implement the SWAT and HSPF models for the 20 Watersheds study areas.
Modeling in this study addresses large study areas, with an emphasis on relative change for future
conditions. A simplified approach is used for land use and soils coverages to the extent possible to make
this efficient. SWAT setup and calibration starts from a common land use platform representing current
(calibration) conditions.
Simplifying principles include the following:
Optimize for automated processing, taking advantage of features already built into
ArcSWAT.
Calibrate first for one 8-digit HUC, then extend to whole study area. The calibration
HUC should be selected in an area with the greatest availability of hydrology and water
quality calibration data. Avoid selecting a calibration HUC8 with complicating features,
such as large reservoirs.
SWAT is set up in the usual way, using an HRU overlay of land use and STATSGO
soils.
Use only weather stations already processed for BASINS 4 and supplied to the team.
Account for sparse coverage by using elevation bands in areas of high relief.
Existing impervious area is identified based on NLCD products.
HSPF setup is based on the same spatial coverages as SWAT but requires additional
processing in the WinHSPF interface.
HSPF is developed only for the five pilot watersheds.
A-l
-------
2 Processing Steps
2.1 INITIAL SETUP
Initial GIS processing has been performed by Tt in ArcGIS for the whole study area. Processed GIS
inputs are then used in ArcSWAT (which runs as an extension in ArcGIS).
Watershed Boundaries and Reach Hydrography
Subbasin boundaries and reach hydrography have been created from NHDPlus. In general these
delineations should be sufficient for application and isolation of appropriate calibration points. Note that
many river "basins" have multiple outlets, notably those adjacent to the Great Lakes or ocean shorelines.
DEM
Mosaic-ed DEMs are supplied for the full extent of the model watershed area. ArcSWAT will create a
slope grid during model setup. For some of the study areas (notably those with significant shoreline
adjacent to an ocean or Great Lake), the DEM extent does not fully overlap the subbasin extent, due to
differences in shoreline representation in the parent spatial files. If the study area has shoreline, or if the
SWAT "subbasin delineation" process fails, we recommend the following ArcGIS procedure to fill the
DEM NoData "holes".
• In the Single Output Map Algebra tool, use the following statement:
con(grid>0,grid,0)
where grid is the name of the DEM.
Before executing the tool, in the Environments section set the extent to the subbasin shapefile.
The procedure will replace NoData cells with 0 inside the subbasin extent, but not beyond. The
error associated with using 0 is likely to be minimal and not adversely affect model setup.
Land Cover and Soils
1. Clipped NLCD 2001 Land Cover is provided for the extent of the model watershed area. The
grid files have been modified to include four reclassified cells in each subbasin. These cells and
their values are used as placeholders for updating developed land cover in future scenarios. You
must use the land cover grid ending in "_sw" for SWAT applications.
2. NLCD 2001 Urban Imperviousness bas been analyzed to calculate developed class impervious
area. This is provided in an Excel spreadsheet.
3. STATSGO soils grids are provided with ArcSWAT.
A-2
-------
Other Data Provided
4. BASINS4 weather stations in proximity of model watersheds. Weather stations from this set
must be used to enable climate updates for the scenarios. The locations are provided within the
SWAT precipitation and air temperature input files on the project FTP site. ArcSWAT will
create shapefiles for the weather stations during model setup.
5. Locations of major point sources are provided in an Excel file.
6. Shapefiles with locations of long-term gaging stations and water quality monitoring stations.
7. Area-averaged nitrogen wet atmospheric deposition concentration (mg/L as N) for the study area.
8. Area-averaged percent impervious values for each of the NLCD developed classes. The values
were developed from an analysis specific to each study area. Both total (FIMP) and directly
connected (FCIMP) values are provided.
Other Setup Tasks (to be undertaken by modeler)
1. Identify locations and characteristics of any major reservoirs. Reservoirs included in the model
should be kept to the essential minimum of those that are sufficiently significant to the water
balance of the simulated area (at the HUC-8 scale or greater) to include explicitly.
2. Identify locations and characteristics of any major features of the watershed affecting water
balance (e.g., diversions, upstream areas not modeled, reaches that lose flow to groundwater).
Some of these features may best be represented with observed flow series as boundary conditions.
Irrigation should be explicitly considered only where needed as a significant part of the basin-
scale water balance, e.g. Rio Grande.
Special notes for the Cook Inlet study areas
a. The projection is different from the one used for the lower 48 states. Be sure that all input
data used in building the model is in the same projection as the DEM and land cover grid.
b. The DEM is in meters, not cm.
c. No wet atmospheric deposition rates for nitrogen were available, so it can be omitted
from the model.
d. The SWAT soils database does not include Alaska, so the user will need to obtain
appropriate STATSGO data for the extent of the model area.
2.2 SWAT MODEL SETUP
SWAT model setup follows directly from the initial setup, using the ArcSWAT extension in ArcGIS.
The following items should be noted and/or followed during SWAT model setup:
Watershed Delineation
Use the Automatic Watershed Delineation option with the following steps:
1. Open the Watershed Delineation window.
2. Import the DEM for the watershed and choose the z unit as cm.
3. Select the option for using user defined watersheds.
4. Import the subbasin and reach shapefiles subsequently. Create outlets.
5. Click the calculate subbasin parameters button. Please check "Skip Longest flow path
calculation" option before calculating subbasin parameters.
-------
6. Once the subbasin parameters have been calculated, add reservoirs, if any. It is desirable to avoid
inclusion of reservoirs where possible due to the difficulty in representing operating rules
adequately. Therefore, only include significant/major reservoirs that have a major impact on
flows. If a reservoir is at the terminus of the model area it may be ignored so that the model
represents input to, rather than output from the terminal reservoir.
7. For point sources, only those identified as majors and supplied by Tt are included in the model.
Create a GIS coverage of the point sources in the subbasin from the list of majors supplied by
Tetra Tech. Bring in the point sources layer to aid manual addition of point sources using the
ArcSWAT interface. Define the major point sources at this stage. For some watersheds it will
also be necessary to define an upstream boundary condition "point source".
8. Run Watershed Topographic Report for later use.
HRU Analysis
Start with the Land Use/Soils/Slope Definition
1. Save the model often, and make complete backups as you finish major tasks.
2. Use SWAT to classify the slopes into two categories with a breakpoint at 10%.
3. Bring in STATSGO soils with the MUID option.
4. NLCD 2001 land use coverage that is supplied can be loaded directly into ArcSWAT without
modification. The default NLCD class to SWAT class mapping is appropriate for most areas;
however, there are added future urban land uses (codes 121, 122, 123, and 124). Use the supplied
luc.txt file to ensure correct mapping. Adjustments to the land use assignment can be made
during cover setup, or parameters for SWAT classes adjusted at a later time. The clipped grids
distributed for a project include a nominal representation of all potential future developed classes
in each subwatershed, with near-zero area (e.g., 0.001 ha). This will provide a basis for ready
modification to address future land use scenarios.
5. Assign impervious percentage to developed land use classes in the SWAT urban database using
the values provided for the study area. The same assumptions must be applied for the future
developed land use classes UFRL, UFRM, UFRH, and UFHI, i.e., the future classes will have
the same total and connected impervious fractions as the corresponding URLD, URMD, URHD,
and UIDU urban land uses. Within ArcSWAT, the impervious values are saved to the main
program geodatabase at C:\Program Files\SWAT\SWAT2005.mdb. The corresponding
urban.dat file is regenerated from the geodatabase each time the model is run from the ArcGIS
interface. This has another important implication: If a given model is ported to a different
machine, the SWAT2005.mdb file must also be ported.
6. Proceed to the HRU Definition tab. Create HRUs by overlaying land use, soil, and slope at
appropriate cutoff tolerance levels (usually 5% for land use, 10% for soil, and 5% for slope). BE
SURE to EXCLUDE all 8 urban land use classes (URLD, URMD, URHD, UIDU, UFRL,
UFRM, URFH, and UFHI) from the threshold criteria. This is done on the Land Use Refinement
tab.
7. Proceed with standard SWAT model generation ("Write Input Tables") using met data provided
on the project FTP site (processed weather series and station locations files). Precipitation and
temperature use observed series; other weather data are simulated with the weather generator. It
is advisable to screen the precipitation and air temperature files for any gross errors during the
simulation time period. While errors are uncommon in the BASINS dataset from which these
were derived, they do occur. Outliers and periods of flatlined values have been discovered in the
pilot phase of the project, and the met data should be corrected and/or stations removed if gross
errors are found.
8. Specify PET option as 1 (Penman/Monteith) in General Watershed Parameters.
-------
9. Turn on elevation bands if necessary to account for orographic effects in areas with a sparse
precipitation network and significant elevation changes. This will generally be appropriate where
elevations within subbasins span a range of 250 m or more (see Watershed Topographic Report).
10. Assign management operations. The simulation option IURBAN in the mgt files associated with
urban land use classes should be left at the default value of 1 (use USGS regression equations).
11. Set instream water quality options; IWQ = 1, and start with program defaults.
12. Use (daily) Curve Number hydrology with observed precipitation and air temperature.
Remaining meteorological variables are simulated using the weather generator.
13. Specify atmospheric N wet deposition concentrations.
14. The time period for simulation should be 31 water years. The first year will be dropped from
analysis to account for model spinup. The remaining 30 years span a period for which the
supplied weather data are complete and include the year 2000 (with the exception of the
Nebraska (Loup and Elkhorn River) basins, where the weather data are complete only through
1999). Note that the start of simulation for some of the non-pilot study areas may be one year
prior to the complete weather data period. For this spinup year, some weather stations may be
absent, but SWAT will fill in the missing records using the weather generator. Save and
backup the model at this point.
15. Run the model for the full 31 year period. Due to spin up effects and interaction with the weather
generator random number processing, all model runs (calibration, validation, and scenario
application) should use the entire model network and the entire simulation time period. (Initial
testing can be done on a subset of the model or a reduced time period; however, it is necessary to
run the full model extent and time period to obtain valid final results.)
16. Undertake calibration for target HUC8. After calibration, repeat for remainder of study area.
Calibration should first be performed for hydrology, then sediment, then nutrients. Calibration
spreadsheet templates are distributed for Hydrology and Water Quality, as described in the next
section.
NOTE: When pursuing calibration through the Arc SWAT interface be sure to use the option to
REwrite SWAT input files (and not "Write Input Files"). The latter option will cause default
parameters to be reloaded from the geodatabase.
2.2 HSPF MODEL SETUP
BASINS4/HSPF uses primarily two applications - Map Window GIS and WinHSPF to create, modify and
run HSPF UCI files. The following steps should be implemented first for the Calibration HUC8 subbasins,
then repeated for the entire model watershed.
1. Prepare a starter.uci file defining default values for PERLND/IMPLND base numbers (see below).
Where previous modeling is available, the initial parameter values will be based on that earlier
modeling. For areas without previous modeling, hydrologic parameters will be based on
recommended ranges in BASINS Technical Note 6 and related to soil and meteorological
characteristics where appropriate.
2. Load HSPF land cover/soils grid (as discussed in the GIS Processing Memorandum), DEM,
subbasin, and reach file into BASINS4 Map Window interface.
3. Use Manual Delineation to calculate subbasin and reach parameters.
4. Assign subbasins to model segment groups using the Model Segmentation Specifier Tool. (A
segment is a group of subbasins with a unique set of PERLNDs and IMPLNDs. Segments are used
primarily for weather station assignment, but may also be used for other factors such as differences
in soils, geology, etc.) Assignment will be based on proximity to weather stations, elevation bands,
^^^^^^^^^^^^^^^^^^^——^^^^^^^^^^^^^^^^^^—
-------
and any other unique watershed characteristics identified previously. The Tool will generate unique
model segments within the HSPF model, using the PERLND/IMPLND numbering scheme shown
below.
5. Run the BASINS4 HSPF tool.
a. General Tab
i. Land Use Type, assign to LU+HSG grid
ii. Subbasins and Streams, use subbasin and reach shapefiles developed previously.
iii. Point sources, using input shapefile developed for watershed.
iv. Met stations, using input shapefile developed for watershed.
b. Land Use Tab: assign imperviousness to the developed land use classes using values
developed from the impervious area analysis.
c. Streams Tab: uses default fields
d. Subbasins Tab: uses default fields.
e. Point Sources Tab: uses default fields.
f Met Stations Tab: uses default fields.
6. Use separate automated processing tool to do the following:
a. Lump area from the four developed classes into one IMPLND and four PERLND categories.
b. Assign PERLND/IMPLND model segmentation using a set numbering scheme for the land
use classes. Model segments will be implemented in groups of 25 (i.e., PERLND 1, 26, 51,
76, and 101 are all WATER). Ensure that all combinations of base number and segment are
represented throughout the model. If necessary, define a nominal area for missing
PERLND/IMPLND values in the SCHEMATIC block. Assignments are shown below, first
for PERLND, and then IMPLND.
LC_HSG Class
WATER
BARREN_D
WETL_D
FOREST_A
FOREST_B
FOREST_C
FOREST_D
SHRUB_A
SHRUB_B
SHRUB_C
SHRUB_D
GRASS_A
GRASS_B
GRASS_C
GRASS_D
AGRI_A
AGRI_B
LC_HSG
Value
101
4
14
21
22
23
24
31
32
33
34
41
42
43
44
51
52
HSPF PERLND
Name
WATER
BARREN_D
WETL_D
FOREST_A
FOREST_B
FOREST_C
FOREST_D
SHRUB_A
SHRUB_B
SHRUB_C
SHRUB_D
GRASS_A
GRASS_B
GRASS_C
GRASS_D
AGRI_A
AGRI_B
HSPF PERLND
Base Number
1
2
3
5
6
7
8
9
10
11
12
13
14
15
16
17
18
A-6
-------
LC_HSG Class
AGRI_C
AGRI_D
DEVO_A
DEVO_B
DEVO_C
DEVO_D
DEVL_A
DEVL_B
DEVL_C
DEVL_D
DEVM_A
DEVM_B
DEVM_C
DEVM_D
DEVH_A
DEVH_B
DEVH_C
DEVH_D
LC_HSG
Value
53
54
61
62
63
64
71
72
73
74
81
82
83
84
91
92
93
94
HSPF PERLND
Name
AGRI_C
AGRI_D
DEVPERV_A
DEVPERV_B
DEVPERV_C
DEVPERV_D
DEVPERV_A
DEVPERV_B
DEVPERV_C
DEVPERV_D
DEVPERV_A
DEVPERV_B
DEVPERV_C
DEVPERV_D
DEVPERV_A
DEVPERV_B
DEVPERV_C
DEVPERV_D
(all imperv)
HSPF PERLND
Base Number
19
20
21
22
23
24
21
22
23
24
21
22
23
24
21
22
23
24
25
7. In WinHSPF, define hydraulic characteristics for major reservoirs and flow/load characteristics for
major point sources. This step can be done in common with the corresponding step for SWAT.
8. FTABLES will be generated automatically during model creation. FTABLES can be easily adjusted
in WinHSPF if specific information is available to the modeler. The WinHSPF FT ABLE tool also
includes a way to recalculate FTABLES using relationships developed for three regions in the
Eastern United States.
9. Adjust lapse rates as needed to account for elevation bands associated with model segments. Lapse
rates are needed for precipitation and air temperature only. For snowmelt, the simplified degree-day
method for snowmelt should be employed.
10. Add additional UCI tables as needed, using the WinHSPF interface.
11. Undertake calibration for target HUC8. After calibration, repeat for remainder of study area.
Calibration should first be performed for hydrology, then sediment, then nutrients.
12. Nutrients will be modeled as inorganic N, inorganic P, and organic matter. The latter will be
transformed to organic N and organic P in the MASS-LINK to the stream. The buildup-washoff
approach should be used to simulate land surface processes; it is easy to implement and tools are
available to translate storm Event Mean Concentrations (EMCs) to model inputs. The instream
simulation should use GQUALS with exponential decay. The project does not have sufficient
resources to develop models with a full algal simulation.
3 Calibration and Reporting Procedures
Calibration will be pursued for flow, TSS, TN, and TP sequentially. The water quality calibration focuses
on replicating inferred monthly loads, although some attention must be paid to the calibration of
concentration as well.
A-7
-------
As noted above, calibration starts with a single HUC8. Parameters derived for this site are then extended
to the remainder of the watershed. However, experience is that this procedure alone is not adequate to
provide reasonable results for the entire study area. Therefore, secondary calibration adjustments should
be made at 2-3 additional stations in the watershed, at least one of which should be at a larger spatial scale
than HUC8 (if available).
The time period for calibration should generally be the last 10 water years of the simulation. Validation
will be applied to the preceding 10 water years (for which fit may be less good due to changes in land use
and management relative to the 2001 NLCD). Adjustments can be made based on data availability if
necessary.
The hydrologic calibration uses the Hydrocal spreadsheet. A version of this spreadsheet will be provided
for the assigned non-pilot study areas. The spreadsheet is already loaded with a list of USGS gages with
sufficiently long periods of record. The spreadsheet will automatically download USGS flows from
NWIS and will load SWAT simulated flows from the program output. Note that this spreadsheet should
be opened in Excel 2003 to ensure proper operation. Most of the functions can be run in Excel 2007 if
needed, but, if so, it should be saved in compatibility mode for Excel 2003. The program uses macros,
and macro security will need to be set to "Medium" for the program to operate.
User controls for the Hydrocal spreadsheet are located on the Data Management tab. Key results are
provided on the Analysis tab. In general, the user should strive to meet the "Recommended Criteria" on
the Analysis tab.
Water quality calibration uses the WQUAL-GCRP spreadsheet. This program also uses macros and is
now designed for Excel 2007. The water quality calibration takes place at stations for which there are
both flow and concentration data and comes pre-loaded with a list of USGS gaging stations. Note that
one spreadsheet covers both the calibration and validation periods.
The modeler is responsible for preparing and loading observed water quality data for TSS, TN, and TP on
the "Obs" tab. At a minimum use the data available on the USGS NWIS system. (In some cases,
previous studies have resulted in creation of more extensive, QA'd data sets that may be used instead.)
Pre-processing is required. Duplicate observations on the same day should be pre-averaged. The 'TSS"
column can be used to store both SSED and TSS data; when both are present on a given day use the
SSED results preferentially. A special note is required regarding TN. TN is a calculated value, and
USGS reports it directly only in some years. In many other cases the user can construct TN observations
by adding TKN plus NO2+NO3 observations. Note that the "Obs" tab contains a column for denoting
samples below detection limit (using symbol "<") for each parameter. For TN, observations should be
flagged as non-detect only when both TKN and NO2+NO3 are non-detect. If only one of these is non-
detect use one-half the detection limit to create the sum.
After entering the observed data, the main user controls are on the "Interface" tab. After running the
macros, key results are shown on the GCRP tab at BH22+. The key interest of EPA is in the prediction of
loads. Loads are, of course, not observed, but are estimated using a stratified regression (TSS, TP) or an
averaging estimator (TN). The first objective of calibration is to reduce the relative percent deviation
between simulated and estimated loads - to below 25% if possible. (This will sometimes not be
possible.) However, the user should also examine the diagnostic plots on the TSS, TP, and TN
worksheets.
A-8
-------
4 Implementing Future Scenarios
Both climate and land use scenarios will be implemented by working directly with the model input files,
rather than returning to the GIS interface.
4.1 CLIMATE SCENARIOS
In both HSPF and SWAT, climate scenarios are readily implemented simply by substituting new
meteorological series. We will not assume any feedback between climate changes and HSPF parameters.
4.2 LAND USE SCENARIOS
ICLUS will be used to estimate a change table by subwatershed. In SWAT, the HRU fractions in each
HRU file are changed using automation scripts.
A-9
-------
(This page intentionally left blank.)
A-10
-------
Appendix B
Project Quality Assurance Project Plan
Section 8: Model Calibration
This appendix reproduces Section 8.0 of the QAPP which describes model calibration
requirements; taken from:
Tetra Tech. (2008) Quality assurance project plan for watershed modeling to evaluate potential
impacts of climate and land use change on the hydrology and water quality of major U.S.
drainage basins. Prepared for the Office of Research and Development, Global Change Research
Program, U.S. Environmental Protection Agency, Washington, DC.
8.0 MODEL SETUP/CALIBRATION
8.1 PROJECT QUALITY OBJECTIVES
EPA emphasizes (USEPA 2000, 2002) a systematic planning process to determine the type and
quality of output needed from modeling projects. This begins with a Modeling Needs and
Requirements Analysis, which includes the following components:
• Assess the need(s) of the modeling project
• Define the purpose and objectives of the model and the model output specifications
• Define the quality objectives to be associated with model outputs
The first item (needs assessment) is covered in EPA's task order. In essence, simulation models
are needed to predict future responses to changes in climate and land use. The existing
simulation models HSPF and SWAT are believed to be sufficient to this purpose, and creation of
new models is not required.
The second item (define purpose and objectives) is the subject of EPA's Draft Analysis Plan.
This proposes both the purpose of the modeling and the specific endpoints to be evaluated as a
result of the modeling. At a general level, the objective of this modeling project is to assess the
potential effects of climate and land use change on the hydrology and water quality of major U.S.
drainage basins; however, this general objective will need to be made more specific to guide
development of the modeling effort. The Tt team is tasked with reviewing and commenting on
the Analysis Plan as part of this work—and revisions to the existing Analysis Plan could arise as
a result of these recommendations. At the end of this review, the Tt team and the EPA COR must
agree on the principal study questions to be addressed through the modeling.
B-l
-------
The quality objectives for the model(s) follow directly from the purposes and objectives—and
can be refined in conjunction with the review of the Analysis Plan. In general, the modeling
effort needs to be designed to achieve an appropriate level of accuracy and certainty in
answering the principal study questions. This process takes into account the following elements:
• The accuracy and precision needed for the models to predict a given quantity at the
application site of interest to satisfy study questions
• The appropriate criteria for making a determination of whether the models are accurate
and precise enough on the basis of past general experience combined with site-specific
knowledge and completeness of the conceptual models
• How the appropriate criteria would be used to determine whether model outputs achieve
the needed quality
EPA's Draft Analysis Plan suggests that the principal study questions to be addressed by the
models are changes in (defined on the basis of modeling at a daily time step): (1) the 100-year
flood, (2) 7Q10 low flow, (3) runoff center of mass, (4) monthly sediment loads, (5) monthly
total nitrogen loads, and (6) monthly total phosphorus loads. This list could be expanded or
modified on the basis of the review of the Draft Analysis Plan.
The models will be calibrated and validated to existing (1970-2000) data to establish their
credibility for use in forecasting responses to future change. Specific calibration and validation
targets for model acceptability (see Sections 8.2 and 8.3) will be selected in light of the intended
uses of the model, as identified in the final revisions to the Analysis Plan.
8.2 MODEL CALIBRATION AND VALIDATION
Model calibration is the process of adjusting model inputs in acceptable limits until the resulting
predictions give good correlation with observed data. Commonly, calibration begins with the
best estimates for model input on the basis of measurements and subsequent data analysis.
Results from initial simulations are then used to improve the concepts of the system or to modify
the values of the model input parameters. The use of calibrated models, the scientific veracity of
which is well defined, is of paramount importance to this project. Because the goal is to be able
to assess the potential effects of climate and land use change on the hydrology and water quality
of major U.S. drainage basins, model calibration and validation should strive to minimize errors
(deviations between model predictions and observed measurement data.).
The Tt Co-TOLs or lead modeler will direct the model calibration efforts. Models are often
calibrated through a subjective trial-and-error adjustment of model input data because a large
number of interrelated factors influence model output. However, the experience and judgment of
the modeler are a major factor in calibrating a model accurately and efficiently. Further, the
model should meet pre-specified quantitative measures of accuracy to establish its acceptability
in answering the principal study questions.
The model calibration process proceeds through both qualitative and quantitative analyses.
Qualitative measures of calibration progress are commonly based on the following:
• Graphical time-series plots of observed and predicted data
______^^
-------
• Graphical transect plots of observed and predicted data at a given time interval
• Scatter plots of observed versus predicted values in which the deviation of points from a
45-degree straight line gives a sense of fit
• Tabulation of measured and predicted values and their deviations
After initially configuring the modeling systems, the Tt team will perform model calibration and
validation. The watershed models will be calibrated to the best available data, including literature
values, and interpolated or extrapolated values using existing field data. If multiple data sets are
available, an appropriate time period and corresponding data set will be chosen on the basis of
factors characterizing the data set, such as corresponding weather conditions, amount of data,
and temporal and spatial variability of data.
A model is considered calibrated when it reproduces data within an acceptable level of accuracy,
as described in Section 8.3 and itemized in Table 4 (quantitative measures). A set of parameters
used in a calibrated model might not accurately represent field values, and the calibrated
parameters might not represent the system under a different set of boundary conditions or
hydrologic stresses. Therefore, a model validation period helps establish greater confidence in
the calibration and the predictive capabilities of the model. A site-specific model is considered
validated if its accuracy and predictive capability have been proven to be within acceptable limits
of error independently of the calibration data.
Table 4. General percent error calibration/validation targets for watershed
models (applicable to monthly, annual, and cumulative values)
Hydrology/Flow
Sediment
Water Quality/Nutrients
Relative percent error
Very good
<10
<20
<15
Good
10-15
20-30
15-25
Fair
15-25
30-45
25-35
In general, model validation is performed using a data set separate from the calibration data. If
only a single time series is available, the series could be split into two subseries, one for
calibration and another for validation. If the model parameters are changed during the validation,
this exercise becomes a second calibration, and the first calibration needs to be repeated to
account for any changes. Representative stations will be used to guide parameter adjustment to
get an accurate representation of the conditions of the individual subwatersheds and streams. The
calibration and validation process will be documented for inclusion in the technical reports.
8.3 SPECIFIED PERFORMANCE AND ACCEPTANCE CRITERIA
Model Testing
Model testing includes calibration, verification, and validation. The previous section described
model calibration and validation. Model verification is the process of testing the model code,
including program debugging, to ensure that the model implementation has been done correctly.
Testing usually begins with the best estimates for model input on the basis of measurements and
subsequent data analyses. Results from initial simulations are then used to improve the concepts
of the system or to modify the values of the model input parameters.
B-3
-------
For this project, existing tested model code will be used (HSPF and SWAT). Therefore, model
verification is required only for new bridge code, such as that required to translate climate
scenarios into model input.
The Tt team will calibrate the project models using the best available data, including literature
values and interpolated or extrapolated existing field data. The model will be considered
successfully tested when it reproduces data at an acceptable level of accuracy.
The work proposed for this project, as defined in the Draft Analysis Plan, differs from other,
more common applications of watershed models (e.g., for TMDLs) in several ways that affect
the calibration strategy:
• Models will be developed at a very large spatial scale (i.e., HUC4 scale) and will be
calibrated at a limited number of points, most of which will likely be at the HUC8 scale.
• Models will be developed for multiple watersheds, and calibration will be done by
multiple teams of modelers. The different teams should all apply the same calibration
metrics.
• Two separate models (HSPF and SWAT) will be developed for some or all the
watersheds. A common set of calibration criteria should be applied to both models to
facilitate comparison.
• Models are proposed to be developed using a daily time step (based on the scale of the
analysis), which will limit the ability to resolve extreme flows.
• Model application is not for regulatory purposes but to inform possible long-term effects
of different change scenarios. While calibration to establish model credibility is essential,
the ability to correctly simulate relative changes is most important.
• Comparison of observed and predicted values on a frequency-duration plot.
Quantitative acceptance criteria for the models will be selected to reflect the final set of principal
study questions in the revised Analysis Plan and incorporated into the QAPP. Given the
considerations listed above, quantitative acceptance criteria will be expressed in relative, rather
than absolute form. That is, relevant calibration outputs will be ranked on a scale ranging from
poor to very good. Calibration will strive to obtain the best fit possible; however, specific values
of quantitative measures will not be proposed to define whether results should be accepted or
rejected. Rather, the level of uncertainty determined in calibration and validation will be
documented to decision makers to aid in interpretation of results.
The current Draft Analysis Plan references only three measures related to hydrology (100-year
flood, 7Q10 low flow, and runoff center of mass); however, accurate representation of the
general water balance is required to demonstrate that the model provides a reasonable
representation of reality that can serve as a foundation for water quality simulation. Therefore,
commonly accepted measures of model hydrologic fit will be applied.
Model simulation of water quality is, in general, more difficult than simulating hydrology, in part
because any uncertainty in the hydrologic simulation will propagate into the water quality
______^^
-------
simulation. In addition, the principal study questions related to water quality contained in the
Draft Analysis Plan address loads. Loads are not directly observed but are inferred from point-in-
time concentration data and continuous flow data. As a result, observed load estimates are
subject to considerable uncertainty.
Quantitative measures, sometimes referred to as calibration criteria, include the relative percent
error between model predictions and observations as defined generally below:
Z O-P
Q
lOO,
where Erei = relative error in percent. The relative error is the ratio of the absolute mean error to
the mean of the observations and is expressed as a percent. A relative error of zero is ideal.
Additional statistics that will be applied include the correlation coefficient (R) and its squared
value, the coefficient of determination (R2), where
O - P
where the overbar indicates the sample mean.
For hydrology and the water balance, percent error tests will be applied to the following
components:
• Total flow volume
• 10 percent high flows
• 50 percent low flows
• Seasonal flow volumes
For water quality, the outcomes of interest defined in the current Draft Analysis Plan are monthly
loads. Therefore, similar calculations of relative percent error will be applied to the series of
predicted and observed monthly loads (where the observed monthly loads will need to be
estimated from observed flow and concentration data using an appropriate estimation technique,
such as those described in Preston et al. 1989).
These tests are relevant to monthly and annual values. General calibration/validation targets for
percent error consistent with current best modeling practices (Donigian 2000) are shown in Table
4.
For hydrology, there is also an interest in extreme high and low flows. Answering this study
question requires calibration to daily flows, rather than just monthly and annual values. Figure 3
(also from Donigian 2000) summarizes R and R2 ranges for the evaluation of daily and monthly
flows:
_____^^
-------
I
GL95
Q8 0.9
Very Good
Good VeryGood
Figure 3. R and R value ranges for model performance
In addition, the Nash-Sutcliffe coefficient of model fit efficiency (COE) will be reported for all
calibration and validation runs — although no specific criteria are proposed. This is calculated as
A COE value of one indicates a perfect fit between measured and predicted values for all events.
A value of zero indicates that the model fit is not better than using the average value of all the
measured data.
Following model calibration, model validation will be conducted using separate, independent
portions of the available time series at the calibration stations. Because the Analysis Plan calls
for simulating the period 1970-2000, while land use will be based on 2001 NLCD information,
the 10-year period from 1991 through 2000 will generally be proposed for calibration, while an
earlier period (dependent on data availability) will be used for validation tests. Because the land
use distribution during the 1970-1991 period could be different in some regions than during the
1991-2000 period, it is important to note that validation results might not achieve the same
quantitative acceptance levels as for calibration.
The Tt team will document model performance over both the calibration and validation period in
the technical reports, using the quantitative measures of accuracy documented above (or any
additional measures that could be identified in modifications to this QAPP). In addition to
measures of accuracy, additional acceptance criteria will include modeling result precision and
representativeness:
• Precision of model results: Precision of generated data produced by the model will be
examined by performing replicate runs. By confirming that an identical data set is
generated when a replicate of the previous model run will rule out numerical instability
issues and verify the precision of the model.
• Representativeness of model results: The Tt team technical staff will compare the
loadings data and measured environmental concentrations to examine sources and sinks
of materials.
An overall assessment of the success of the calibration can be expressed using calibration levels.
B-6
-------
Level 1: Quantitative performance measures fall within the very good range (highest degree
of calibration).
Level 2: Quantitative performance measures fall within the good range.
Level 3: Quantitative performance measures fall within the fair range.
Level 4: Quantitative performance measures fall within the poor range (lowest degree of
calibration).
Model Sensitivity Analysis
The sensitivity to variations or uncertainty in input parameters is an important characteristic of a
model. Sensitivity analysis is used to identify the most influential parameters in determining the
accuracy and precision of model predictions. This information is of importance to the user who
must establish required accuracy and precision in model application as a function of data
quantity and quality. Sensitivity analysis quantitatively or semi-quantitatively defines the
dependence of the model's performance assessment measure on a specific parameter or set of
parameters. Sensitivity analysis can also be used to decide how to simplify the model simulation
and to improve the efficiency of the calibration process.
Model sensitivity can be expressed as the relative rate of change of selected output caused by a
unit change in the input. If the change in the input causes a large change in the output, the model
is considered to be sensitive to that input parameter. Sensitivity analysis methods are mostly
nonstatistical or even intuitive by nature. Sensitivity analysis is typically performed by changing
one input parameter at a time and evaluating the effects on the distribution of the dependent
variable. Nominal, minimum, and maximum values are specified for the selected input
parameter.
Sensitivity analysis is performed at the beginning of the calibration process to design a
calibration strategy. After calibration is completed, a more elaborate sensitivity analysis is
performed to quantify the uncertainty in the calibrated model caused by uncertainty in the
estimates of the model input parameters.
Informal sensitivity analyses (iterative parameter adjustments) are generally performed during
model calibration to ensure that reasonable values for model parameters will be obtained,
resulting in acceptable model results. The degree of allowable adjustment of any parameter is
usually directly proportional to the uncertainty of its value and is limited to its expected range of
values.
8.4 ASSESSMENT AND RESPONSE ACTIONS
The ability of computer code to represent model theory accurately will be ensured by following
rigorous programming protocols, including documentation within the source code. Specific tests
will be required of all model revisions to ensure that fundamental operations are verified to the
extent possible, including testing numerical stability and convergence properties of the model
code algorithms, if appropriate. Model results will generally be checked by comparing results to
those obtained by other models or by comparing them to hand calculations. Visualization of
model results will help determine whether model simulations are realistic. Model calculations
______^^
-------
will be compared to field data. If adjustments to model parameters are made to obtain a fit to the
data, the modelers will provide an explanation and justification that must agree with scientific
knowledge and fit within reasonable ranges of process rates as found in the literature.
As described in Section 5.1, non-project-generated data will be used for model development and
calibration. The model calibration procedure is discussed in Section 8.2. The DQOs were
discussed in Section 7.0 and 8.0 of this document. Modelers will cross-check data for bias,
outliers, normality, completeness, precision, accuracy, and other potential problems.
Data generated outside the project will be obtained primarily from quality assured databases
maintained by EPA, USGS, and other entities. Additional data may be obtained from either
published or nonpublished sources. The published data will have some degree or form of peer
review. Typically, modelers examine these data as part of a data quality assessment.
Unpublished databases are also examined in light of a data quality assessment. Data provided by
EPA or other sources will be assumed to meet precision objectives established by those entities.
The QA program under which this task order will operate includes surveillance, with
independent checks of the data obtained from sampling, analysis, and data gathering activities.
This process is illustrated in Figure 4.
B-8
-------
Problem Identified
QC Officer
TOL
Other Tetra Tech Staff
QA Officer
(System Audit [Audit Report Form
Completed])
EPATOM
Other non-Tetra Tech Personnel
Tt TOL Informed
I
Problem Defined
Tt TOL Confirms Problem and
Reviews Nature of the Problem
Yes
Significant Corrective Action
Needed
No
QAO Informed
Documented on
Corrective Action
Request and Response
Verification Form
Tt TOL Works with Task
Leader and Technical
Staff to Correct the
Problem
Yes
Stop Work Decision or
Recommendation
Made by QAO
Stop Work Order Issued
Tt PIC and PM Informed by
QAO; Tt TOL and technical staff
informed by Tt PM (client
notified as appropriate)
Problem and Corrective
Action Recorded in
Project Documents as
Appropriate
No
Tt TOL Ensures That
Corrective Action Has
Been Implemented
No
Problem Investigated
TtTOL Determines Cause, Assigns
->• Responsibility, Develops Procedure
to Correct the Problem, Implements
the Procedure
Corrective Action
Implementation Verified
By QAO
No
•Yes I
Start Work Order Issued
By QAO; PIC and PM Informed
by QAO; Tt TOL and Technical
Staff Informed by Tt PM (client
notified as appropropriate)
Revisions Made
to QA Documents, if Necessary
Yes
Revisions Distributed
to Client and All Staff Involved
Work Proceeds
Corrective Action Form
Completed
Filed by QAO, PM, and Tt TOL
Work Proceeds
Tt = Tetra Tech
PIC = Principal-in-Charge
Figure 4. Problem assessment and correction operations
B-9
-------
The essential steps in the QA program are as follows:
• Identify and define the problem
• Assign responsibility for investigating the problem
• Investigate and determine the cause of the problem
• Assign and accept responsibility for implementing appropriate corrective action
• Establish the effectiveness of and implement the corrective action
• Verify that the corrective action has eliminated the problem
Many of the possible technical problems can be solved on the spot by staff, for example, by
modifying the Initial Technical Approach memorandum or correcting errors or deficiencies in
implementation of the approach. Immediate corrective actions are considered SOPs, and they are
noted in records for the project. Problems that cannot be solved in this way require more
formalized, long-term corrective action.
If quality problems that require attention are identified, Tt will determine whether attaining
acceptable quality requires either short- or long-term actions. If a failure in an analytical system
occurs (e.g., performance requirements are not met), the Tt team modeling QC officers will be
responsible for corrective action and will immediately inform the Tt Co-TOLs or the QAO, as
appropriate. Subsequent steps taken will depend on the nature and significance of the problem, as
illustrated in Figure 4.
Crat™ct(™nie) The Tt Co-TOLs have primary
responsibility for monitoring the
activities of this project and
~ identifying or confirming any quality
problems. The Co-TOLs will also
bring these problems to the attention
of the Tt QAO, who will initiate the
corrective action system described
above, document the nature of the
problem (using a form such as that
shown in Figure 5), and ensure that the
recommended corrective action is
carried out. The Tt QAO has the
authority to stop work on the project if
problems affecting data quality that
will require extensive effort to resolve
are identified.
Date of Assessment _
Title (of project or other) _
Project Leader
Other Responsible Personnel
Auditor or Initiator of This Corrective Action Request _
Problem Description:
Recommended Action:
Quality Assurance Officer
Principal-in-Charge or Program Manager
Action Taken:
Verification of Completion of Corrective Action:
Date to Be Completed:
Quality Assurance Officer
Principal-in-Charge or Program Manager Date
Original form to be filed in QAO File; one copy to be filed in Project File and one copy in Contract File (if corrective
action pertains to a project), or one copy to be filed in Contract File (if corrective action pertains to a contract).
Figure 5. Example corrective
action request and response
verification form
B-10
-------
The EPA COR, Tt PGM and Tt Co-TOLs will be notified of major corrective actions and stop
work orders. Corrective actions can include the following:
• Reemphasizing to staff the project objectives, the limitations in scope, the need to adhere
to the agreed-upon schedule and procedures, and the need to document QC and QA
activities
• Securing additional commitment of staff time to devote to the project
• Retaining outside consultants to review problems in specialized technical areas
• Changing procedures
The Tt Co-TOLs may replace a staff member, as appropriate, if it is in the best interest of the
project to do so.
Performance audits are quantitative checks on different segments of project activities; they are
most appropriate for sampling, analysis, and data-processing activities. The Tt modeling QC
officer is responsible for overseeing work as it is performed and periodically conducting internal
assessments during the data entry and analysis phases of the project. As data entries, model
codes, calculations, or other activities are checked, the Tt modeling QC officer will sign and date
a hard copy of the material or complete Tt's standard Technical/Editorial Review Form, as
appropriate, and provide it to the Tt Co-TOLs for inclusion in the administrative record.
Performance audits will consist of comparisons of model results with observed historical data.
Performing control calculations and post-simulation validation of predictions are major
components of the QA framework.
The Tt Co-TOLs will perform or oversee the following qualitative and quantitative assessments
of model performance periodically to ensure that the model is performing the required task while
meeting the quality objectives:
• Data acquisition assessments
• Model calibration studies
• Sensitivity analyses
• Uncertainty analyses
• Data quality assessments
• Model evaluations
• Internal peer reviews
Internal peer reviews will be documented in the project and QAPP files. Documentation will
include the names, titles, and positions of the peer reviewers; their report findings; and the
project management's documented responses to their findings.
The Tt Co-TOLs will perform surveillance activities throughout the duration of the project to
ensure that management and technical aspects are being properly implemented according to the
__^^
-------
schedule and quality requirements specified in this QAPP. These surveillance activities will
include assessing how project milestones are achieved and documented; corrective actions
implemented; budgets adhered to; peer reviews performed; data managed; and whether
computers, software, and data are acquired in a timely manner.
System audits are qualitative reviews of project activity to check that the overall quality program
is functioning and that the appropriate QC measures identified in the QAPP are being
implemented. If requested by the EPA COR, and EPA provides additional funding, the Tt QAO
or designee will conduct an internal system audit of the project and report the results to the EPA
COR and the Tt Co-TOLs.
8.5 DOCUMENTATION AND RECORDS
Thorough documentation of all modeling activities is necessary for interpreting study results. Tt
will prepare monthly progress reports that will address task and subtask milestones, deliverables,
adherence to schedule, and financial progression at the end of each full month while the task
order for this project is still open. Data needs and deadlines for Tt's receipt of information
needed to meet the project schedule will also be included in the progress reports and Gantt chart.
The progress in meeting modeling QA targets (QA reports) will also be included in the progress
reports. Other deliverables will be distributed to project participants as indicated by the EPA
COR. Data tables, assumptions and analyses used to develop the models will be recorded and
provided to EPA as a separate deliverable. The format of the raw data to be used for model
parameters, model input, model calibration, and model output will be converted to the
appropriate units, as necessary.
The Tt team will save on an external hard drive all modeling output data from all 20 watersheds
as digital computer files in a file directory using a file-naming convention specified by the EPA
COR. In addition, the Tt team will save on an external hard drive all scripts, project files,
calibration data, and other information used to conduct watershed modeling at each of the 20
study watersheds. Tt will deliver these external hard drives to EPA within 2 weeks of the EPA
COR's approval of the final report presenting and discussing the goals, methods, results and
conclusions of watershed modeling in all 20 study watersheds (see the schedule in Table 2). Tt
will maintain a copy of the project files at the Cincinnati, Ohio and/or Fairfax, Virginia, office
for at least 3 years (unless otherwise directed by the EPA COR). The EPA COR and Tt Co-TOLs
will maintain files, as appropriate, as repositories for information and data used in models and
for preparing any reports and documents during the project. Electronic project files are
maintained on network computers and are backed up periodically. The Tt Co-TOLs will
supervise the use of project materials. The following information will be included in the
electronic project files within Tt and on the external hard drives:
• Any reports and documents prepared
• Contract and task order information
• Electronic copies of model input/output (for model calibration and allocation scenarios)
• Results of technical reviews, model tests, data quality assessments of output data, and
audits
B-12
-------
• Documentation of response actions during the project to correct model development or
implementation problems
• Assessment reports for acquired data
• Statistical goodness-of-fit methods and other rationale used to decide which statistical
distributions should be used to characterize the uncertainty or variability of model input
parameters
• Communications (memoranda; internal notes; telephone conversation records; letters;
meeting minutes; and all written correspondence among the project team personnel,
subcontractors, suppliers, or others)
• Maps, photographs, and drawings
• Studies, reports, documents, and newspaper articles pertaining to the project
• Spreadsheet data files including physical measurements, analytical chemistry data, and
microbiological data (hard copy and on diskette)
The model application will include complete record keeping of each step of the modeling
process. The documentation will consist of reports and files addressing the following items:
• Selection of study watersheds and model calibration points
• Assumptions
• Adjustments
• Parameter values and sources
• Nature of grid, network design, or subwatershed delineation
• Changes and verification of changes made in code
• Actual input used
• Output of model runs and interpretation
• Sensitivity analyses results
• Calibration and validation of the models
Formal reports submitted to EPA that are generated from the data will be maintained in the
central file (diskette and hard copy) at Tt's Cincinnati, Ohio, and Fairfax, Virginia, offices. The
data reports will include a summary of the types of data collected, sampling dates, and any
problems or anomalies observed during sample collection.
8.6 OUTPUT ASSESSMENT AND MODEL USABILITY
Tt team technical staff will review model predictions for reasonableness, relevance, and
consistency with the requirements of the model development process through model calibration
as described in Section 8.0 of this QAPP. Tt team modeling experts will also determine
consistency with the acceptance criteria described in Sections 7.0 and 8.0 of this QAPP. The Tt
modeling QC officer will ensure that all steps of the modeling process are performed correctly.
_____^^
-------
Electronic copies of model input/output for model calibration, data quality assessments of output
data, and QA reports will be maintained as part of the project files.
LITERATURE CITED
Donigian, A.S., Jr. 2000. HSPF Training Workshop Handbook and CD. Lecture #19. Calibration
and Verification Issues, Slide #L19-22. EPA Headquarters, Washington Information Center,
January 10-14, 2000. Presented and prepared for U.S. Environmental Protection Agency,
Office of Water, Office of Science and Technology, Washington, DC.
Preston, S.D., V.J. Bierman, Jr., and S.E. Silliman. 1989. An evaluation of methods for the
estimation of tributary mass loads. Water Resources Research, 25(6): 1379-1389.
USEPA (U.S. Environmental Protection Agency). 2000. Guidance for the Data Quality
Objectives Process (G-4). EPA 600-R-96-055. U.S. Environmental Protection Agency,
Office of Environmental Information, Washington, DC.
USEPA (U.S. Environmental Protection Agency). 2002. Guidance for Quality Assurance Project
Plans for Modeling (G-5M). EPA 240-R-02-007. U.S. Environmental Protection Agency,
Office of Environmental Information, Washington, DC.
B-14
-------
Appendix C
Climate Change and the Frequency and
Intensity of Precipitation Events
Technical Note
By: Charles Rodgers, Stratus Consulting Inc.
This review has been prepared to address issues raised in the context of the preparation of
meteorological data used as input to the SWAT and HSPF watershed models. The data
preparation process is summarized as follows: The approach selected for this project is to use the
U.S. Environmental Protection Agency's (EPA's) BASINS Climate Assessment Tool (CAT;
U.S. EPA, 2009) to modify historical meteorological records to reflect the projected impacts of
climate change on important meteorological variables. Temperature and precipitation records
from the National Oceanic and Atmospheric Administration (NOAA) National Climatic Data
Center (NCDC) stations in and surrounding each pilot watershed have been identified, and
hourly data covering the period 1970-2000 is being used in the calibration of models and the
simulation of historical patterns of discharge. The projected regional impacts of climate change
will be obtained from the North American Regional Climate Change Assessment Program
(NARCCAP) dynamically downscaled 50 kilometer Regional Climate Model (RCM) output,
primarily at monthly resolution. The original proposal was to use 15 General Circulation Model
(GCM)-RCM combinations to simulate a range of future projections, although due to restrictions
on the likely availability of NARCCAP downscaled data, a combination of NARCCAP,
statistically downscaled CMIP3 projections and direct GCM outputs will likely be used. In each
instance, the model-projected changes in temperature and precipitation patterns will be used to
modify the historical climate records using CAT. The advantages to this approach include the
preservation of short-timescale variability and other aspects of time series behavior, and the
preservation of inter-site variability and correlation patterns, none of which are feasible using
downscaled GCM outputs directly as model inputs.
CAT permits the sequential modification of weather records to introduce a number of alterations,
each reflecting various assumptions concerning the regional manifestations of climate change.
Precipitation records can be modified by (1) multiplying all records by an empirical constant
reflecting projected climate change to simulate a shift in total precipitation, applied uniformly to
all periods and intensity classes, (2) selective application of such a multiplier to specific seasons
or months, (3) selective application of the multiplier to a range of months or years within the
record, (4) selective application of the multiplier to storm events of a specific size or intensity
class; and (5) addition or removal of storm events to simulate changes in the frequency of
C-l
-------
precipitation events (U.S. EPA, 2009). Modification (4) can be iteratively applied to more than
one event size class. In summary, changes in frequency and intensity as well as changes in overall
precipitation accumulation can be represented using CAT and historical records.
Relative changes in the frequency and intensity of precipitation events associated with climate
change may prove to be more influential in determining future patterns of discharge than
projected changes in overall (annual, seasonal) precipitation. In particular, the partitioning of
precipitation into re-evaporation, runoff and percolation to groundwater is understood to be
sensitive to the intensity and timing of precipitation events. Thus, to ensure that model
simulations embody the most important dimensions of projected climate change, particular
attention should be paid to precipitation intensity-frequency-duration (IFD) relationships. As a
general pattern, the warming of the lower atmosphere is projected to lead to a more vigorous
hydrologic cycle, characterized by increases in global precipitation, and proportionally larger
increases in high-intensity events (Trenberth et al., 2007). This memorandum is intended to
provide a summary review of recent literature to address the following questions: (1) How
should precipitation change as a consequence of lower atmosphere warming? (2) What is the
historical evidence for increases in precipitation intensity over the United States? (3) What do
climate models project with respect to precipitation frequency and intensity? (4) What are the
important limitations in these projections? and (5) What are the implications for the development
of meteorological time series used in the modeling study?
1. How should precipitation change as a consequence of lower atmosphere warming?
Physical arguments predicting increases in precipitation intensity as a consequence of the
warming of the lower atmosphere are presented by Trenberth et al. (2003). The basic argument
can be summarized as follows: (1) The primary conditions for precipitation to occur include
(a) availability of precipitable moisture in the atmosphere and (b) a mechanism for lifting and
cooling parcels of air, leading to condensation and precipitation. (2) Progressive warming of the
land surface and lower atmosphere (i.e., climate change) will lead to increases in atmospheric
(precipitable) moisture through the positive relationship between air temperature and saturation
vapor pressure (moisture-holding capacity). The Clausius-Clapeyron equation quantifies this
relationship, and can be used to predict an increase in atmospheric water holding capacity of
around 7% per °C at current global mean temperatures.l (3) Precipitation, when it occurs, often
exceeds the extractable fraction (typically below 30%) of available moisture in the immediate
zone of precipitation. This reflects the role of low-level convergence in drawing moist air into
convective zones from surrounding areas. Trenberth et al. (2003) calculate that as an
approximate global average, a zone of precipitation is supported by a larger region - roughly three
to five times the radius of the precipitation zone - from which it draws moisture. Assuming no
significant change in the efficiency of precipitation generation (i.e., maximum rate of extraction of
water from the atmosphere), the intensity of such events should therefore increase as a function of
mean temperature at roughly the same rate as atmospheric water-holding capacity (i.e., around 7%
per °C). Finally (4), the increased atmospheric moisture supply also provides additional latent heat
C-2
-------
to drive the convective process, further enhancing low-level convergence. Thus, increases in
atmospheric temperature lead to proportional increases in precipitation intensity when other
conditions required for initiating convective precipitation are present. Trenberth et al. (2003) note
that the GCMs supporting the Intergovernmental Panel on Climate Change (TPCC) assessment
simulate increases in precipitation of around 1% to 2% per °C.2 This suggests (assuming these
GCMs accurately simulate the atmospheric water balance) that the increase in precipitation
intensity predicted by Clausius-Clapeyron (7% per °C) for intense convective events must be
compensated for by reductions in the frequency and/or intensity of light to moderate intensity
events. The latter hypothesis assumes that the durations of intense precipitation events do not also
change significantly as a function of temperature.
2. Historical evidence of increases in precipitation intensity over the United States
If the proposed relationship between increasing air temperatures and precipitation intensity is
theoretically sound, then the predicted changes should already be evident due to observed
increases in global and U.S. air temperatures. Analysis of instrumental records from 1850-2005
indicates that globally-averaged temperatures have increased by 0.76°C (+/- 0.19°C) over this
period, with the most rapid warming occurring in the last 50 years and the steepest increase in
global temperatures, equivalent to changes of+0.177°C per decade, occurring over the last
25 years (Trenberth et al., 2007). Within the United States, temperature increases have also been
observed at rates exceeding the global average. Present (1993-2008) U.S. temperatures are on
average over 1.1°C warmer than during the 1961-1979 period (Karl et al., 2009). Corresponding,
increasing trends in evaporation, atmospheric moisture and precipitation, particularly high-
intensity precipitation, should thus be in evidence. However, it is not necessarily the case that all
of these increases (if observed) should be of the magnitude predicted by Clausius-Clapeyron (7%
per °C) since globally, evaporation is controlled by the availability of surface moisture (over
land) and by the availability of energy at the earth's surface to drive evaporation and
transpiration (Allen and Ingram, 2002).
Evaporation: Among the predicted impacts of a warming lower atmosphere, increases in actual
evaporation and transpiration (evapo-transpiration, or ET) have been the most difficult to
demonstrate, largely due to the relative absence of long-term records of direct ET measurements
(Lettenmaier et al., 2008). Physical theory (the Clausius-Clapeyron equation) predicts an
increase in potential ET., since a supply of moisture available for ET cannot be assumed. A
relatively small number of recent land-based studies in the United States, India, China and
Australia that make use of long-term evaporation pan data conclude that actual evaporation rates
2. Trenberth et al. (2003) refer to the GCMs supporting the Third IPCC Assessment (2001).
C-3
-------
have decreased. One proposed explanation for this paradox is a reduction in incoming solar
radiation due to increases in aerosols associated with air pollution (Trenberth et al., 2007).
Alternatively, Brutsaert and Parlange (1998) conjecture that as humidity supplied by the
surrounding landscape increases, pan evaporation will decrease (a reverse of the "oasis effect").
A synthesis of water balance studies of several major North American watersheds (Walter et al.,
2004), in which ET was estimated as the residual of precipitation and discharge, concludes that
actual ET has increased over the last 50 years. More direct evidence of a temperature-induced
increase in actual evaporation is provided by Yu and Weller (2007). These authors utilized
satellite remote sensing and atmospheric model re-analysis to estimate trends in evaporation over
the ocean surface, where moisture supply is not limited. They estimate that globally averaged
ocean latent heat flux (evaporation) has increased by approximately 10% over the 25-year period
1981-2005. This reflects increases in both atmospheric moisture capacity (Clausius-Clapeyron)
and sea surface temperature (SST).
Atmospheric moisture: Clausius-Clapeyron predicts an increase in absolute or specific humidity
(q) with increasing temperature, as distinct from relative humidity (RH). Climate model
simulations tend to indicate that temperature-related changes in RH are small (Trenberth et al.,
2003). Balloon-borne radiosonde has been used to estimate altitude-integrated RH, although time
series analysis based on radiosonde is subject to a number of constraints. Specifically, the density
of radiosonde observations is low, observations are unavailable over the open ocean and
radiosonde sensors have changed over time, confounding efforts to measure decadal-scale
changes in atmospheric water content (Dai, 2006b), effectively limiting analysis to the mid-
1970s and onward. Nevertheless, Ross and Elliott (1996; 2001) used radiosonde time series
records to estimate changes in RH and precipitable water (up to 500 mb) over North America in
recent decades. They found that precipitable water has increased by 3% to 7% per decade
between 1973 and 1995 over the area ranging from the Caribbean to 45°N, with greater increases
in the south and smaller increases in the north. Above 45°N, changes were either uncertain or
negative over this period. Ross and Elliott (2001) note also that these changes appear greater and
more uniform over North America than over Eurasia.
Dai (2006b) evaluated changes in surface specific q and RH using a much wider range of
sensors, located both over land and over ocean. Near-surface measurements do not provide
altitude-integrated estimates of q and RH, although the spatial sampling is greatly improved
relative to Ross and Elliott (1996; 2001) due to the large number of records (over 15,000 surface
and ocean weather stations), and the sensor technology is more consistent over the period of
record. Dai (2006b) found that globally averaged specific humidity (q) increased by around
0.06 g kg"1 per decade over the 1976-2004 period. This corresponds to roughly 4.9% per degree
(°C) of warming over that period, globally averaged. The response of q to temperature increases
over water (i.e., not source-limited) was found to be around 5.7% per °C of warming, reasonably
consistent with the predictions of Clausius-Clapeyron (7% per °C) at constant RH. By contrast,
the response over land is around 4.3% per °C of warming, presumably reflecting spatio-temporal
C-4
-------
limitations in water available for evaporation. Finally, column-integrated estimates of
atmospheric water vapor have been available since 1988 from the special sensor microwave
imager (SSM/I). On the basis of SSM/I, Trenberth et al. (2005) estimate that over the period
1988-2003, altitude-integrated atmospheric precipitable water over oceans has increased by
around 0.40 mm per decade (1.3%). Variability over the period of analysis was found to reflect
variations in SST. Assuming relatively constant RH, the observed trend is reasonably close to the
7% per °C predicted by Clausius-Clapeyron.
Precipitation: Increases in the frequency and intensity of heavy precipitation events over the last
several decades are among the most clearly documented changes in recent U.S. climate (Kunkel,
2008). The following studies are representative of several recent studies examining trends in
precipitation in the United States and globally. Karl and Knight (1998) found that precipitation
over the U.S. increased by around 10% between the 1910s and the 1980s. These authors
examined the respective contributions of changes in both frequency and intensity of precipitation
to changes in total precipitation. Precipitation events were disaggregated into 20 intensity
classes, each encompassing 5% of observed events; and extreme intensity events, defined as
precipitation exceeding 2 inches (50.4) mm per day, were also examined. Among their
conclusions, Karl and Knight (1998) found that observed increases reflect both increased
frequency and intensity of rainfall events. While the frequency of events increased for all
intensity (percentile) classes, intensity increased for heavy and extreme precipitation days only,
and the proportion of total annual precipitation attributable to these heavy and extreme events
has increased relative to more moderate events. Specifically, over half (53%) of the observed
increase was due to increases in the upper 10% of events. Karl and Knight (1998) also found that
the percentage of total area within the U.S. experiencing extreme precipitation events
(> 50.4 mm/day) had increased by roughly 20% between 1910 and the mid-1990s. Kunkel et al.
(1999) found statistically significant increasing trends in 1-year and 5-year return period 7-day
precipitation events in the United States. However, subsequent work (Kunkel et al., 2003)
extended the period of record back to 1895, and the frequency of extreme events in the late 19th-
early 20th century was found to be similar to the late 20th Century, suggesting that natural
variability cannot be ruled out as an additional factor contributing to the observed late
20th century increases in intensity.
Groisman et al. (2004) examined trends in several climatologic and hydrologic variables for the
conterminous U.S. potentially influenced by climate change, including total precipitation,
precipitation intensity, temperature and streamflow. Heavy precipitation events, defined as the
upper 5% of daily events, increased by 14% over the period 1908-2002. Very heavy events
(upper 1 %) increased by 20% over this period, and extreme events (upper 0.1 %) by 21%. The
most significant increases occurred in the upper and lower Midwest for annual events, the upper
Midwest and Great Lakes areas for summer events, and in New England for winter events.
Similar results are presented in the global analysis of Groisman et al. (2005) for the period 1910-
1999, who found that while total annual precipitation volumes over the United States increased
C-5
-------
by 1.2% per decade over the period 1970-1999, the share of annual precipitation associated with
extreme events (defined as above) increased by 14% per decade over this period. These authors
note that " ...practically the entire nationwide increase in heavy and very heavy precipitation
occurred during the last three decades" (p. 1328). Alexander et al. (2006), examining global
precipitation statistics for the period 1951-2003, reached similar conclusions, specifically, that
the contribution to total annual precipitation from very wet days, defined as the upper 5% of
daily precipitation events, has increased over this period, even in many areas where total
precipitation has decreased.
3. Model projections of trends in precipitation intensity
Evaporation and transpiration are in many circumstances controlled by factors other than the
moisture-holding capacity of the atmosphere, including availability of moisture supply over land
areas and energy available to drive the ET process (e.g., Allen and Ingram, 2002). Thus, GCMs
generally predict increases in the global hydrologic cycle that are more modest than the 7% per
°C predicted by Clausius-Clapeyron (Trenberth et al., 2003). Sun et al. (2007) have summarized
changes in total global precipitation, precipitation frequency, intensity, fraction of precipitation
from convective events and other related variables as projected by 17 of the most recent
generation of GCMs from the Program for Climate Model Diagnosis and Intercomparison
(PCMDI), used in the IPCC AR4 (2007), for emissions [Special Report on Emissions Scenarios
SRES] scenarios Bl (low), A1B (medium) and A2 (high). Ensemble results, averaged over
models and scenarios, indicate that global mean precipitation is projected to increase by around
1.2% per °C, and latent heat flux (evaporation) by a comparable amount, although global
precipitable water is projected to increase by around 9.1% per °C. These results indicate that the
atmospheric state variable (atmospheric precipitable water) responds approximately as predicted
by Clausius-Clapeyron (consistent with relatively small increases in average RH), while
atmospheric water fluxes (ET, precipitation) are constrained by other factors. Sun et al. (2007)
report that overall, the frequency of (daily) precipitation events is projected to decrease, and the
intensity of events to increase on average, consistent with Trenberth et al. (2003). However, the
frequency of heavy precipitation events is projected to increase, indicating a more dramatic
reduction in the frequency of light precipitation events. Thus, heavy (20-50 mm day"1) and very
heavy (> 50 mm day"1) precipitation events are projected to contribute a disproportionately
increasing share of total precipitation, through the combined effects of increased frequency and
increased magnitude, with frequency effects more influential than intensity effects. These
projected impacts are most pronounced under SRES scenario A2 (high emissions). These authors
acknowledge that the (simulated) increases in intensity may not be fully captured in an analysis
based on daily precipitation totals, since high-intensity events are often of shorter duration.
Tebaldi et al. (2006) have also examined model-simulated changes in extreme events,
encompassing both temperature and precipitation events, on the basis of nine GCMs included in
the IPCC AR4 (2007). These authors use a set of indicators of precipitation intensity proposed by
C-6
-------
Frich et al. (2002) including the following: (1) frequency of days with precipitation exceeding
10 mm, (2) maximum 5-day precipitation total, (3) mean precipitation intensity (total
precipitation divided by number of days with precipitation exceeding 1 mm) and (4) fraction of
total precipitation due to events exceeding the 95th percentile. Significant increases in each of
these four indices were projected by the GCMs evaluated, although not all trends were
statistically significant. Significant (increasing) trends were associated with mid- to high
latitudes in the Northern Hemisphere as well as some tropical areas within South America and
Africa. Tebaldi et al. (2006) conclude that "Models (also) agree with observations over the
historical period that there is a trend towards a world characterized by intensified precipitation,
with a greater frequency of heavy-precipitation and high-quantile events, although with
substantial geographic variability" (p. 206).
4. How well do GCMs capture the frequency-intensity relationships observed in actual
precipitation?
In evaluating the model-generated evidence that precipitation intensity is likely to increase as a
consequence of increasing tropospheric temperatures, it is important to recognize that many
GCMs display well-documented biases with regard to precipitation frequency and intensity.
Specifically, there is a tendency for GCMs to generate too many low-intensity events, and to
under-simulate the intensity of heavy events. There are several possible reasons for this (since
there are several convective precipitation parameterization schemes in use) although problems
associated with the simulation of the diurnal cycle appear to play an important role (Dai, 1999;
Trenberth et al., 2003). If conditions for the onset of moist convection in models are biased or
poorly specified, convection occurs too early in the diurnal cycle, weaker convection results in
less vigorous precipitation, and the removal of atmospheric moisture reduces the likelihood of
more intense convective events subsequently (Trenberth et al., 2003).
Sun et al. (2006) compared the performance of 18 coupled Atmosphere-Ocean General
Circulation Models (AOGCMs) used in the IPCC AR4 (2007) in simulating precipitation with
historic observational data. Most of the models were found to greatly overestimate the frequency
of summer (June-August) light precipitation events in the Northern Hemisphere, although the
frequency of heavy precipitation events was simulated with greater skill, each subject to regional
variations. Sun et al. (2006) summarize their observations as follows: "For light precipitation,
most of the models greatly overestimate the frequency but reproduce the observed patterns of
intensity relatively well. For heavy precipitation, most of the models roughly reproduce the
observed frequency but underestimate the intensity" (p. 928, emphasis in original). Light
precipitation is defined as 1-10 mm day"1 and heavy precipitation as >10 mm day"1. These
authors emphasize the importance of getting precipitation "right for the right reasons" since
surface runoff, evaporation and soil moisture are all highly sensitive to precipitation IFD
relationships. Dai (2006a) also examines the performance of 18 models from the PCMDI (AR4)
ensemble with respect to the characterization of precipitation. Dai's (2006a) study emphasizes
C-7
-------
model skill in simulating precipitation via the distinct convective and stratiform mechanisms,
noting that (in GCMs) stratiform precipitation is a grid-scale process while convective
precipitation is a subgrid-scale process. Model performance is compared to the Tropical Rainfall
Measuring Mission (TRMM) observational data. Among other results, Dai (2006a) found that
the models examined derived too much of total precipitation (28%-38%) from light precipitation
(1-5 mm day"1) relative to TRMM data (19%), and far too little (0-2%) from very heavy
precipitation (> 50 mm day"1) (7% for TRMM). Model replication of TRMM results were best
for moderate events, defined as 10-20 mm day"1. Dai (2006a) concludes that".. .the newest
generation of Coupled Ocean-Atmosphere General Circulation Models (CGCMs) still rains too
frequently, as in previous generations (...), mostly within the 1-10 mm day"1 categories, while
heavy precipitation (> 20 mm day"1) occurs too rarely" (p. 4622).
5. Implications for the ORD modeling study
On the basis of literature reviewed here, several observations can be made. First, the importance
of getting IFD relationships right cannot be over-emphasized. Analysis of historical data
indicates that changes in the distribution of precipitation between light- and heavy-intensity
events are quantitatively greater than changes in overall precipitation at annual or seasonal levels
in many regions, and projected runoff estimates are likely to be quite sensitive to these IFD
relationships. Second, model-generated projections of precipitation are characterized by
documented biases with respect to precipitation intensity and frequency. This suggests that the
relative changes in precipitation IFD relationships should be used as the basis for adjusting
historical precipitation records in CAT rather than their absolute levels.
NARCCAP has to date provided summaries of three GCM-RCM downscaled products intended
for use in modifying the historical gauge records. These datasets include changes in monthly
precipitation accumulation, and changes in the contributions of precipitation by intensity class
for 25%, 50%, 70%, and 90% percentile classes. While this data is extremely useful, it is
recommended that we obtain a number of representative output series at finer time resolution -
down to 3-hourly as output by NARCCAP RCMs. Ideally, we would obtain series that sample
from at least three of the climatic zones associated with the pilot watersheds, and including the
upper Midwest (Minnesota) in particular, since many of the greatest observed changes in
precipitation intensity have occurred in this region. These time series would be processed to
obtain estimates of the change in frequency of events, by event size class, to support the
appropriate use of CAT. As emphasized above, the "deltas" (changes in projection period
relative to base period) would be the basis for CAT transformations rather than the projections
themselves, which potentially contain biases, as discussed.
References
Alexander, L.V., X. Zhang, T.C. Peterson, J. Caeser, B. Gleason, A.M.G. Klein Tank, M.
Haylock, D. Collins, B. Trewin, F. Rahimzadeh, A. Tagipour, K. Rupa Kumar, J. Revadekar, G.
C-8
-------
Griffiths, L. Vincent, D.B. Stephenson, J. Burn, E. Aguilar, M. Brunet, M. Taylor, M. New, P.
Zhai, M. Rusticucci, and J.L. Vazquez-Aguirre. 2006. Global observed changes in daily climate
extremes of temperature and precipitation. Journal of Geophysical Research 111(D05109). doi:
10:1029/2005JD006290, 22 pp.
Allen, M.R. and WJ. Ingram. 2002. Constraints on future changes in climate and the hydrologic
cycle. Nature 419:224-232.
Brutsaert, W. and M.B. Parlange. 1998. Hydrologic cycle explains the evaporation paradox.
Nature 396:29-30.
Dai, A. 1999. Recent changes in the diurnal cycle of precipitation over the United States.
Geophysical Research Letters 26:341-344.
Dai, A. 2006a. Precipitation characteristics in eighteen coupled climate models. Journal of
Climate 19:4605-4630.
Dai, A. 2006b. Recent climatology, variability, and trends in global surface humidity. Journal of
Climate 19:3589-3606.
Frich, P., L.V. Alexander, P. Della-Marta, B. Gleason, M. Haycock, A.M.G.K. Tank, and T.
Peterson. 2002. Observed coherent changes in climatic extremes during the second half of the
twentieth century. Climatic Research 19:193-212.
Groisman, P.Y., R.W. Knight, D. Easterling, T.R. Karl, G.C. Hegerle, and V.N. Razuvaev. 2005.
Trends in intense precipitation in the climate record. Journal of Climate 18:1326-1350.
Groisman, P.Y., R.W. Knight, T.R. Karl, D. Easterling, B. Sun, and J.H. Lawrimore. 2004.
Contemporary changes of the hydrologic cycle over the contiguous United States: Trends
derived from in situ observations. Journal of Hydrometeorology 5:64-85.
Karl, T.R. and R.W. Knight. 1998. Secular trends of precipitation amount, frequency, and
intensity in the United States. Bulletin of the American Meteorological Society 79(2): 231-241.
Karl, T.R., J.M. Melillo, and T.C. Peterson (eds.). 2009. Global Climate Change Impacts in the
United States. Cambridge University Press, New York.
Kunkel, K.E. (Convening Lead Author). 2008. Observed changes in weather and climate
extremes. Chapter 2 in Weather and Climate Extremes in a Changing Climate. Regions of Focus:
North America, Hawaii, Caribbean, and U. S. Pacific Islands. U.S. Climate Change Science
Program Synthesis and Assessment Product 3.3.
C-9
-------
Kunkel, K.E., K. Andsager, and D.R. Easterling. 1999. Long-term trends in extreme precipitation
events over the conterminous United States and Canada. Journal of Climate 12:2515-2527.
Kunkel, K.E., D.R. Easterling, K. Redmond, and K. Hubbard. 2003. Temporal variations of
extreme precipitation events in the United States: 1895-2000. Geophysical Research Letters
30(17):1900. doi:10.1029/2003GL018052.
Lettenmaier, D., D. Major, L. Poff, and S. Running. 2008. Water resources. In The effects of
climate change on agriculture, land resources, water resources, and biodiversity in the United
States. (Synthesis and Assessment Product 4.3.) A Report by the U.S. Climate Change Science
Program and the Subcommittee on Global Change Research. Washington, DC. 362 pp.
Ross, RJ. and W.P. Elliott. 1996. Tropospheric water vapor climatology and trends over North
America: 1973-93. Journal of Climate 9:3561-3574.
Ross, RJ. and W.P. Elliott. 2001. Radiosonde-based northern hemisphere tropospheric water
vapor trends. Journal of Climate 14:1602-1611.
Sun, Y., S. Solomon, A. Dai, and R.W. Portmann. 2006. How often does it rain? Journal of
Climate 19:916-934.
Sun, Y., S. Solomon, A. Dai, and R.W. Portmann. 2007. How often will it rain? Journal of
Climate 20:4801-4818.
Tebaldi, C., K. Hayhoe, J.M. Arblaster, and G.A. Meehl. 2006. Going to the extremes: An
intercomparison of model-simulated historical and future changes in extreme events. Climatic
Change 79:185-211.
Trenberth, K.E., J. Fasullo, and L. Smith. 2005. Trends and variability in column-integrated
atmospheric water vapor. Climate Dynamics 24:741-758.
Trenberth, K.E., A. Dai, R.M. Rasmussen, and D.B. Parsons. 2003. The changing character of
precipitation. Bulletin of the American Meteorological Society (September 2003): 1205-1217.
Trenberth, K.E., P.O. Jones, P. Ambenje, R. Bojariu, D. Easterling, A. Klein Tank, D. Parker, F.
Rahimzadeh, J.A. Renwick, M. Rusticucci, B. Soden, and P. Zhai. 2007. Observations: Surface
and atmospheric climate change. In Climate Change 2007: The Physical Science Basis,
Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental
Panel on Climate Change, S. Solomon, D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt,
M. Tignor, and H.L. Miller (eds.). Cambridge University Press, Cambridge, UK and New York.
C-10
-------
U.S. EPA. 2009. BASINS 4.0 Climate Assessment Tool (CAT): Supporting Documentation and
User's Manual. U.S. Environmental Protection Agency, Washington, DC. National Center for
Environmental Assessment, Office of Research and Development. 191 pp.
Walter, M.T., D.S. Wilks, J-Y. Parlange, andR.L. Schneider. 2004. Increasing
evapotranspiration from the coterminous United States. Journal of Hydrometeorology 5:405-
408.
Yu, L. and R.A. Weller. 2007. Objectively analyzed air-sea heat fluxes for the global ice-free
oceans (1981-2005). Bulletin of the American Meteorological Association doi: 10.1175/BAMS-
88-4-527. 527-539.
C-ll
-------
Appendix D
Model Configuration, Calibration and
Validation
Basin: Apalachicola-Chattahoochee-Flint (ACF)
D-l
-------
Contents
Watershed Background D-8
Water Body Characteristics D-8
Soil Characteristics D-11
Land Use Representation D-11
Point Sources D-16
Meteorological Data D-19
Watershed Segmentation D-22
Calibration Data and Locations D-24
HSPF Modeling D-26
Changes Made to Base Data Provided D-29
Assumptions D-29
Hydrology Calibration D-31
Hydrology Validation D-36
Hydrology Results for Larger Watershed D-41
Water Quality Calibration and Validation D-53
Water Quality Results for Larger Watershed D-63
SWAT Modeling D-66
Changes Made to Base Data Provided D-66
Assumptions D-66
Hydrology Calibration D-66
Hydrology Validation D-72
Hydrology Results for Larger Watershed D-77
Water Quality Calibration and Validation D-84
Water Quality Results for Larger Watershed D-93
References D-96
D-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes D-14
Table 2. Land use distribution forthe ACF River basin (2001 NLCD) (mi2) D-15
Table 3. Major point source discharges in the ACF River basin D-16
Table 4. Precipitation stations forthe ACF River basin model D-19
Table 5. Calibration and validation locations in the ACF River basin D-25
Table 6. Reservoirs represented in the ACF basin model D-29
Table 7. Seasonal summary at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - calibration
period (HSPF) D-34
Table 8. Summary statistics at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - calibration
period (HSPF) D-36
Table 9. Seasonal summary at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - validation
period (HSPF) D-39
Table 10. Summary statistics at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - validation
period (HSPF) D-41
Table 11. Seasonal summary: Model DSN 9001 vs. USGS 02358000 Apalachicola River at
Chattahoochee, FL- calibration period (HSPF) D-44
Table 12. Summary statistics: Model DSN 9001 vs. USGS 02358000 Apalachicola River at
Chattahoochee, FL- calibration period (HSPF) D-46
Table 13. Summary statistics (percent error) for all stations - calibration period 1993-2002 (HSPF) D-47
Table 14. Summary statistics (percent error) for all stations - validation period 1983-1992 (HSPF) D-48
Table 15. Model fit statistics (observed minus predicted) for monthly TSS loads using stratified
regression D-55
Table 16. Relative errors (observed minus predicted), TSS concentration at USGS 02349605 Flint River
at Ga 26, near Montezuma, GA (HSPF) D-57
Table 17. Model fit statistics (observed minus predicted) for monthly total phosphorus loads using
stratified regression D-58
Table 18. Relative errors (observed minus predicted), total phosphorus concentration at USGS 02349605
Flint River at Ga 26, near Montezuma, GA (HSPF) D-60
Table 19. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using a
veraging estimator (HSPF) D-61
Table 20. Relative errors (observed minus predicted), total nitrogen concentration at USGS 02349605
Flint River at Ga 26, near Montezuma, GA (HSPF) D-63
Table 21. Summary statistics for water quality (observed minus predicted) for all stations - calibration
period 1999-2002 (HSPF) D-64
Table 22. Summary statistics for water quality (observed minus predicted) for all stations - validation
period 1986-1998 (HSPF) D-65
Table 23. Seasonal summary at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - calibration
period (SWAT) D-70
Table 24. Summary statistics at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - calibration
period (SWAT) D-72
Table 25. Seasonal summary at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - validation
period (SWAT) D-75
Table 26. Summary statistics at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - validation
period (SWAT) D-77
Table 27. Seasonal summary at USGS 02358000 Apalachicola River at Chattahoochee, FL- calibration
period (SWAT) D-80
Table 28. Summary statistics at USGS 02358000 Apalachicola River at Chattahoochee, FL- calibration
period (SWAT) D-82
Table 29. Summary statistics (percent error) for all stations - calibration period 1993-2002 (SWAT) D-83
Table 30. Summary statistics (percent error) for all stations - validation period 1983-1992 (SWAT) D-84
D-3
-------
Table 31. Model fit statistics (observed minus predicted) for monthly TSS loads using stratified
regression at USGS 02349605 Flint River at Ga 26, near Montezuma, GA (SWAT) D-85
Table 32. Relative errors (observed minus predicted), TSS concentration at USGS 02349605 Flint River
at Ga 26, near Montezuma, GA (SWAT) D-87
Table 33. Model fit statistics (observed minus predicted) for monthly total phosphorus loads using
stratified regression at USGS 02349605 Flint River at Ga 26, near Montezuma, GA (SWAT).... D-88
Table 34. Relative errors (observed minus predicted), total phosphorus concentration, USGS 02349605
Flint River at Ga 26, near Montezuma, GA (SWAT) D-90
Table 35. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 02349605 Flint River at Ga 26, near Montezuma, GA (SWAT).... D-91
Table 36. Relative errors (observed minus predicted), total nitrogen concentration, USGS 02349605
Flint River at Ga 26, near Montezuma, GA (SWAT) D-93
Table 37. Summary statistics (observed minus predicted) for water quality for all stations - calibration
period 1999-2002 (SWAT) D-94
Table 38. Summary statistics (observed minus predicted) for water quality for all stations - validation
period 1986-1998 (SWAT) D-95
Figures
Figure 1. Location of the Apalachicola-Chattahoochee-Flint (ACF) River basin D-10
Figure 2. Land use in the ACF River basin D-13
Figure 3. Major point sources in the ACF River basin D-18
Figure 4. Weather stations for the ACF River basin model D-21
Figure 5. Model segmentation and USGS stations utilized for the ACF River basin D-23
Figure 6. Parameter mapping utilized in the HSPF ACF River basin model D-28
Figure 7. Mean daily flow at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - calibration
period (HSPF) D-32
Figure 8. Mean monthly flow at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - calibration
period (HSPF) D-32
Figure 9. Monthly flow regression and temporal variation at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - calibration period (HSPF) D-33
Figure 10. Seasonal regression and temporal aggregate at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - calibration period (HSPF) D-33
Figure 11. Seasonal medians and ranges at USGS 02349605 Flint River at Ga 26, near Montezuma, GA -
calibration period (HSPF) D-34
Figure 12. Flow exceedence at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - calibration period
(HSPF) D-35
Figure 13. Flow accumulation at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - calibration
period (HSPF) D-35
Figure 14. Mean daily flow at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - validation period
(HSPF) D-37
Figure 15. Mean monthly flow at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - validation
period (HSPF) D-37
Figure 16. Monthly flow regression and temporal variation at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - validation period (HSPF) D-38
Figure 17. Seasonal regression and temporal aggregate at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - validation period (HSPF) D-38
Figure 18. Seasonal medians and ranges at USGS 02349605 Flint River at Ga 26, near Montezuma, GA -
validation period (HSPF) D-39
Figure 19. Flow exceedence at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - validation period
(HSPF) D-40
Figure 20. Flow accumulation at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - validation period
(HSPF) D-40
D-4
-------
Figure 21. Mean daily flow: Model DSN 9001 vs. USGS 02358000 Apalachicola River At Chattahoochee, FL-
calibration period (HSPF) D-42
Figure 22. Mean monthly flow: Model DSN 9001 vs. USGS 02358000 Apalachicola River At Chattahoochee,
FL-calibration period (HSPF) D-42
Figure 23. Monthly flow regression and temporal variation: Model DSN 9001 vs. USGS 02358000 Apalachicola
River At Chattahoochee, FL- calibration period (HSPF) D-43
Figure 24. Seasonal regression and temporal aggregate: Model DSN 9001 vs. USGS 02358000 Apalachicola
River At Chattahoochee, FL- calibration period (HSPF) D-43
Figure 25. Seasonal medians and ranges: Model DSN 9001 vs. USGS 02358000 Apalachicola River At
Chattahoochee, FL- calibration period (HSPF) D-44
Figure 26. Flow exceedence: Model DSN 9001 vs. USGS 02358000 Apalachicola River At Chattahoochee,
FL-calibration period (HSPF) D-45
Figure 27. Flow accumulation: Model DSN 9001 vs. USGS 02358000 Apalachicola River At Chattahoochee,
FL-calibration period (HSPF) D-45
Figure 28. Histogram of simulated and measured elevation for Lake Lanier from 1/1/1993 to 12/31/2002 D-50
Figure 29. Histogram of simulated and measured elevation for West Point Lake from 1/1/1993 to 12/31/
2002 D-51
Figure 30. Histogram of simulated and measured elevation for Lake Walter F. George from 1/1/1993 to
12/31/2002 D-52
Figure 31. Histogram of simulated and measured elevation for Lake Seminole from 1/1/1993 to 12/31/2002.. D-53
Figure 32. Fit for monthly load of TSS USGS 02349605 Flint River at Ga 26, near Montezuma, GA
(HSPF) D-54
Figure 33. Power plot for observed and simulated TSS at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - calibration period (HSPF) D-55
Figure 34. Power plot for observed and simulated TSS at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA- validation period (HSPF) D-56
Figure 35. Time series plot of TSS concentration at USGS 02349605 Flint River at Ga 26, near Montezuma,
GA - calibration period (HSPF) D-57
Figure 36. Fit for monthly load of total phosphorus at USGS 02349605 Flint River at Ga 26, near Montezuma,
GA(HSPF) D-58
Figure 37. Power plot for observed and simulated total phosphorus at USGS 02349605 Flint River at Ga 26,
near Montezuma, GA - calibration period (HSPF) D-59
Figure 38. Power plot for observed and simulated total phosphorus at USGS 02349605 Flint River at Ga 26,
near Montezuma, GA - validation period (HSPF) D-59
Figure 39. Time series plot of total phosphorus concentration at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - calibration period (HSPF) D-60
Figure 40. Fit for monthly load of total nitrogen at USGS 02349605 Flint River at Ga 26, near Montezuma,
GA(HSPF) D-61
Figure 41. Power plot for observed and simulated total nitrogen at USGS 02349605 Flint River at Ga 26,
near Montezuma, GA - calibration period (HSPF) D-62
Figure 42. Power plot for observed and simulated total nitrogen at USGS 02349605 Flint River at Ga 26,
near Montezuma, GA - validation period (HSPF) D-62
Figure 43. Time series plot of total nitrogen concentration at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - calibration period (HSPF) D-63
Figure 44. Mean daily flow at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - calibration
period (SWAT) D-68
Figure 45. Mean monthly flow at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - calibration
period (SWAT) D-68
Figure 46. Monthly flow regression and temporal variation at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA- calibration period (SWAT) D-69
Figure 47. Seasonal regression and temporal aggregate at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA- calibration period (SWAT) D-69
D-5
-------
Figure 48. Seasonal medians and ranges at USGS 02349605 Flint River at Ga 26, near Montezuma, GA -
calibration period (SWAT) D-70
Figure 49. Flow exceedence at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - calibration
period (SWAT) D-71
Figure 50. Flow accumulation at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - calibration
period (SWAT) D-71
Figure 51. Mean daily flow at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - validation
period (SWAT) D-73
Figure 52. Mean monthly flow at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - validation
period (SWAT) D-73
Figure 53. Monthly flow regression and temporal variation at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA- validation period (SWAT) D-74
Figure 54. Seasonal regression and temporal aggregate at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - validation period (SWAT) D-74
Figure 55. Seasonal medians and ranges at USGS 02349605 Flint River at Ga 26, near Montezuma, GA -
validation period (SWAT) D-75
Figure 56. Flow exceedence at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - validation
period (SWAT) D-76
Figure 57. Flow accumulation at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - validation
period (SWAT) D-76
Figure 58. Mean daily flow at USGS 02358000 Apalachicola River at Chattahoochee, FL- calibration period
(SWAT) D-78
Figure 59. Mean monthly flow at USGS 02358000 Apalachicola River at Chattahoochee, FL- calibration
period (SWAT) D-78
Figure 60. Monthly flow regression and temporal variation at USGS 02358000 Apalachicola River at
Chattahoochee, FL- calibration period (SWAT) D-79
Figure 61. Seasonal regression and temporal aggregate at USGS 02358000 Apalachicola River at
Chattahoochee, FL- calibration period (SWAT) D-79
Figure 62. Seasonal medians and ranges at USGS 02358000 Apalachicola River at Chattahoochee, FL-
calibration period (SWAT) D-80
Figure 63. Flow exceedence at USGS 02358000 Apalachicola River at Chattahoochee, FL- calibration
period (SWAT) D-81
Figure 64. Flow accumulation at USGS 02358000 Apalachicola River at Chattahoochee, FL- calibration
period (SWAT) D-81
Figure 65. Fit for monthly load of TSS at USGS 02349605 Flint River at Ga 26, near Montezuma, GA -
calibration period (SWAT) D-85
Figure 66. Power plot for observed and simulated TSS at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA- calibration period (SWAT) D-86
Figure 67. Power plot for observed and simulated TSS at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - validation period (SWAT) D-86
Figure 68. Time series plot of TSS concentration at USGS 02349605 Flint River at Ga 26, near Montezuma,
GA(SWAT) D-87
Figure 69. Fit for monthly load of total phosphorus at USGS 02349605 Flint River at Ga 26, near Montezuma,
GA(SWAT) D-88
Figure 70. Power plot for observed and simulated total phosphorus at USGS 02349605 Flint River at Ga 26,
near Montezuma, GA- calibration period (SWAT) D-89
Figure 71. Power plot for observed and simulated total phosphorus at USGS 02349605 Flint River at Ga 26,
near Montezuma, GA - validation period (SWAT) D-89
Figure 72. Time series plot of total phosphorus concentration at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA (SWAT) D-90
Figure 73. Fit for monthly load of total nitrogen at USGS 02349605 Flint River at Ga 26, near Montezuma,
GA(SWAT) D-91
D-6
-------
Figure 74. Power plot for observed and simulated total nitrogen at USGS 02349605 Flint River at Ga 26,
near Montezuma, GA - calibration period (SWAT) D-92
Figure 75. Power plot for observed and simulated total nitrogen at USGS 02349605 Flint River at Ga 26,
near Montezuma, GA - validation period (SWAT) D-92
Figure 76. Time series plot of total nitrogen concentration at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA (SWAT) D-93
D-7
-------
The Apalachicola-Chattahoochee-Flint (ACF) River basin lies in Georgia, Alabama, and Florida, empties into the
Gulf of Mexico at Apalachicola Bay (Figure 1). It is comprised of 12 HUC8 cataloging units, and stretches across
parts of three geological physiographic provinces. The ACF basin along with the Alabama-Coosa-Tallapoosa
(ACT) River basin, are the central focus of water war that has been ongoing for over 20 years. The states of
Georgia, Alabama, and Florida have been involved in a legal controversy over the fair management of the waters
that these states share (Alabama River Alliance 2007).
Approximately 64 percent of the basin is forested. Approximately 25 percent of these forests are timberlands used
for manufacturing wood products. Agricultural land represents a mix of cropland, pasture, orchards, and areas of
confined feeding for poultry and livestock production. The dominant agricultural land use in the Piedmont
Province is pasture and confined feeding for dairy or livestock production. Most of the poultry operations in the
ACF River basin are concentrated in the upper part of the Chattahoochee River basin. Row-crop agriculture,
orchards, and silviculture are most common in the Coastal Plain areas. Common crops in the watershed include
peanuts, corn, soybeans, wheat, and cotton. The largest concentration of urban land in the basin is in the Atlanta
area. Nearly 90 percent of the total population in the basin lives in Georgia, and nearly 75 percent live in the
Atlanta metropolitan area.
The ACF River basin is characterized by a warm and humid, temperate climate. Precipitation is greatest in the
mountains and near the Gulf of Mexico, lowest in the center of the basin. Average annual precipitation in the
basin is about 55 inches, but ranges from a low of 45 inches in the east-central part of the basin to a high of 60
inches in the Florida panhandle. Throughout the ACF River basin, low flows usually occur from September to
November and peak flows usually occur from January to April when rainfall is high and evapotranspiration is
low.
The watershed is underlain by five major aquifer systems: crystalline rock aquifers in the Blue Ridge and
Piedmont physiographic provinces, and four aquifer systems in the Coastal Plain physiographic province.
Watershed hydrology is influenced by 16 reservoirs, 13 of which are on the Chattahoochee River. These
reservoirs play a major role in controlling flow and influencing the quality of water in the watershed (Couch
1993).
Water Body Characteristics
Chattahoochee River
The Chattahoochee River is 430 miles long, drains an area of 8,770 mi2, and has an average discharge of 11,500
cubic feet per second (cfs). The river begins in the Blue Ridge Province in the mountainous region of northeast
Georgia, which is characterized by steep topography and relatively high precipitation and runoff. Annual
precipitation ranges from 53 to 70 inches and annual runoff from 27 to 37 inches. The part of the Chattahoochee
River watershed in the Blue Ridge Province is underlain by crystalline rock, and surface water in the area is
siliceous and low in natural mineral content (Couch 1993).
Thirteen of 16 dams on mainstem locations in the ACF River basin are on the Chattahoochee River. Dam
construction in the watershed began in the early 1800s on the Chattahoochee River above the Fall Line at
Columbus, Georgia, to take advantage of natural gradients for power production. Pronounced decreases in the
frequency of high and low flows have occurred since the start of operation of Buford Dam, which forms Lake
Sidney Lanier. Lake Sidney Lanier, West Point Lake, and Lake Walter F. George provide most water storage
available to regulate flows in the watershed. Lake Sidney Lanier alone provides 65 percent of conservation
D-8
-------
storage, although it drains only 5 percent of the ACF River basin. In addition, West Point Lake and Lake Walter
F. George provide 18 and 14 percent, respectively, of the watershed's conservation storage (USGS 2008).
Throughout most of its length, the Chattahoochee River is controlled by hydropeaking hydroelectric plants, which
contribute to power supply during peak periods of electric demand. From Cornelia, Georgia all the way down to
Lake Seminole, the hydrograph shows the influence of hydropeaking operations and these operations can result in
daily stage fluctuations of 4 feet or more (USGS 2008).
In contrast to the mainstem Chattahoochee River, many tributaries remain free flowing. Flows of tributaries in
forested watersheds are represented by Snake Creek and flows typical of urban watersheds are represented by
Peachtree Creek. Similar to most Piedmont streams, both streams have higher sustained flows during winter
months and show response to storm events throughout the year. However, sharper peaks in the hydrography of
Peachtree Creek reflect greater influence of impervious land cover in the urban watershed (USGS 2008).
Flint River
The Flint River, which is 340 miles long and drains an area of 8,460 mi2, has an average discharge of 9,800 cfs
and begins in the Piedmont Province near Atlanta's Hartsfield International Airport. In the upper part of the Flint
River watershed annual precipitation ranges from 44 to 59 inches, and annual runoff ranges from 10 to 39 inches.
The upper part of the Flint River watershed is characterized by both broad and narrow ridges separated by narrow
valleys. Natural surface water quality in the part of the watershed in the Piedmont Province is similar to that in the
Blue Ridge Province, but the water generally has higher concentrations of dissolved minerals and higher turbidity
(Couch 1993, Cherry 1961).
Most of the larger tributaries in the ACF River basin are located in the Coastal Plain Province part of the Flint
River watershed. These tributaries include Ichawaynochaway Creek, Chickasawhatchee Creek, Kinchafoonee
Creek , and Muckalee Creek.
Apalachicola River
The Chattahoochee and Flint Rivers flow through the Piedmont and Coastal Plain Provinces to their confluence at
Lake Seminole where they form the Apalachicola River. The parts of these river watersheds that lie in the Coastal
Plain Province are underlain by unconsolidated sediments consisting of sand, gravel, and clay. Surface water
tends to be siliceous in the upper part of the Coastal Plain Province but is predominantly carbonate in
southwestern Georgia where it is in contact with limestone. Rainfall in the lower Chattahoochee and Flint River
watersheds ranges from 43 to 55 inches, annually. Rainfall in this area is rapidly absorbed by the permeable soils,
and annual runoff ranges from 12 to 28 inches (Couch 1993).
The Apalachicola River is 106 miles long and drains an area of about 2,400 mi2 in the lower Coastal Plain
Province. Because of the low gradient of the lower Coastal Plain Province, the channel of the Apalachicola River
meanders through a wide, swampy floodplain. The floodplain ranges in width from 0.6 miles below Lake
Seminole to 5 miles near its mouth, where the Apalachicola River flows through a system of distributaries to the
Apalachicola Bay. The Apalachicola River has an average discharge of 26,000 cfs (Couch 1993).
D-9
-------
Hydrography
Water (Nat Atlas Dataset)
US Census Populated Places
Municipalities (pop 2 50.000)
County Boundaries
1 I Watershed with HUCSs
Athenslarke
County
iddle
Chattahoochee-
take Harding
Upper
Flint
03130005 \
Ichawaynocha
3130009
Chattahoochee
004
Lower Flint
03130008
GCRP Model Areas - ACF River Basin
Base Map
Figure 1. Location of the Apalachicola-Chattahoochee-Flint (ACF) River basin.
D-10
-------
Soil Characteristics
The ACF River basin contains parts of the Blue Ridge, Piedmont, and Coastal Plain physiographic provinces that
extend throughout the southeastern United States. Similar to much of the Southeast, the watershed's physiography
reflects a geologic history of mountain building in the Appalachian Mountains, and long periods of repeated land
submergence in the Coastal Plain Province. Physiography within the major provinces is not homogeneous and has
been subdivided by the states of Alabama, Florida, and Georgia. Although similar physiography may extend
across state boundaries, districts may be assigned different names by state geologists in each state (USGS 2008).
Three major soil orders, ultisols, entisols, and spodosols, and more than 50 soil series are present in the ACF
River basin. Ultisols are characterized by sandy or loamy surface horizons and loamy or clayey subsurface
horizons. These deeply weathered soils are derived from underlying acid crystalline and metamorphic rocks.
Entisols are young soils with little or no change from parent material and with poorly developed subhorizons.
These soils are frequently infertile and droughty because they are deep, sandy, well-drained, and subject to active
erosion. Spodosols are characterized by a thin sandy subhorizon underlaying the A horizon. This sandy
subhorizon is cemented by organic matter and aluminum. The ACF River basin is similar to much of the
southeastern coastal plain in the dominance of ultisols. Entisols are found at and below the Fall Line and in the
Dougherty Plain; and spodosols are found in the Gulf Coast Lowlands (USGS 2008).
The 20 Watershed study utilized STATSGO soil survey hydrologic soil group (HSG) information during model
set-up. The descriptions of each hydrologic soil group are provided below.
Group A Soils Have low runoff potential and high infiltration rates even when thoroughly wetted. They
consist chiefly of deep, well to excessively drained sands or gravels and have a high rate
of water transmission.
Group B Soils Have moderate infiltration rates when wet and consist chiefly of soils that are moderately
deep to deep, moderately well to well drained, and moderately fine to moderately course
textures.
Group C Soils Have low infiltration rates when thoroughly wetted and consist chiefly of soils having a
layer that impedes downward movement of water with moderately fine to fine structure.
Group D Soils Have high runoff potential, very low infiltration rates and consist chiefly of clay soils
with high swelling potential, soils with a permanent water table, soils with a claypan or
clay layer at or near the surface and shallow soils over nearly impervious material.
The ACF basin has all four HSGs in the watershed. The Upper and Middle Chattahoochee and most of the Upper
Flint watersheds are dominated by hydrologic type B soils. As both rivers reach and cross over the Fall Line, the
boundary between the Piedmont and Coastal Plain physiographic provinces, they flow through an area dominated
by HSG A soils. As the two rivers come together near Lake Seminole the soil distribution is equally split between
HSG A and B soils. The southernmost extents of the Apalachicola River are dominated by hydrologic soil type D.
Land Use Representation
Land use/cover in the watershed is based on the 2001 NLCD coverage (Figure 2). The 2001 NLCD land cover
was used in order to generate consistency amongst all models for the 20 Watershed project.
Chattahoochee River Watershed
The Chattahoochee River watershed above Lake Sidney Lanier is dominated by forested land with a majority of
the remaining land being pasture. As the Chattahoochee River flows out of Buford Dam to the southwest the
dominant land use starts shifting to urban. As the Chattahoochee River nears and flows through the Atlanta metro
D-ll
-------
area, land use is almost entirely urban. After leaving the Atlanta Metro area, land use shifts back to forest
dominance, but with a greater amount of pasture than the area above Lake Sidney Lanier. Continuing down the
Chattahoochee and across the Fall Line, the dominant land use is still forest, but wetland areas begin to increase
while pasture areas begin to decrease. After the Chattahoochee leaves Lake Walter F. George, agriculture and
forest become equally dominant with wetlands still being prevalent. This land use/land cover pattern continues
until the Chattahoochee empties into Lake Seminole.
Flint River Watershed
The most northern portions of the Flint River watershed are almost entirely urban. As the Flint River flows south
the land use shifts to predominately pasture and then to forest. Once the Flint River crosses the Fall Line,
agriculture become increasingly prevalent but there is still a good portion of forest and an increase in wetlands.
The land uses of major tributaries to the Flint River are also chiefly comprised of agriculture, forest, and wetlands.
The Flint River immediately above Lake Seminole is mostly agriculture with a small portion of pasture and forest.
Apalachicola River Watershed
The Apalachicola River immediately below Lake Seminole is comprised of mostly forest, scrub land and pasture.
As the Apalachicola River flows south, the land use is almost entirely dominated by wetlands until the river
empties into the Gulf of Mexico. The Chipola, a major tributary to the Apalachicola River, is chiefly comprised of
pasture and barren lands to the north and forest and wetlands to the south.
D-12
-------
Interstate
^^— Hydrography
^H Water (Nat. Atlas Dataset)
I | County Boundaries
t State Boundaries
Watershed
2001 NLCD Land Use
I Open water
^ Developed, open space
| Developed, low intensity
I Developed, medium intensity
I Developed, high intensity
^ Barren land
I Deciduous forest
Jj Evergreen forest
| | Mixed forest
^ Scrub/shrub
^ Grassland/hertaceous
~j PastureAiay
| Cultivated crops
^ Woody wetlands
~j Emergent hertiaceous wetlands
GCRP Model Areas - ACF River Basin
Land Use Map
Figure 2. Land use in the ACF River basin.
D-13
-------
NLCD land cover classes were aggregated according to the scheme shown in Table 1 for representation in the 20
Watershed model, and then overlain with the soils HSG grid. Pervious and impervious lands are specified
separately for HSPF, so only one developed pervious class is used, along with an impervious class. HSPF
simulates impervious land areas separately from pervious land. Impervious area distributions were also
determined from the NLCD Urban Impervious data coverage. Specifically, percent impervious area was
calculated over the entire watershed for each of the four developed land use classes. These percentages were then
used to separate out impervious land. NLCD impervious area data products are known to underestimate total
imperviousness in rural areas. However, the model requires properly connected impervious area, not total
impervious area, and the NLCD tabulation is assumed to provide a reasonable approximation of connected
impervious area. Different developed land classes are specified separately in SWAT. The WATER, BARREN,
DEVPERV, and WETLAND classes are not subdivided by HSG in HSPF; SWAT uses the built-in HRU overlay
mechanism in the ArcSWAT interface. The distribution of land use in the watershed is summarized in Table 2.
Table 1. Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area
usually accounted for as
reach area
Deciduous
Evergreen
Mixed
Emergent & woody
wetlands
Aquatic bed wetlands (not
emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
HSPF (after processing)
WATER
BARREN, Assume HSG D
DEVPERV;
IMPERV
BARREN (D)
FOREST (A,B,C,D)
SHRUB (A,B,C,D)
GRASS (A,B,C,D), BARREN (D)
GRASS(A,B,C,D)
AGRI (A,B,C,D)
WETLAND, Assume HSG D
WATER
D-14
-------
Table 2. Land use distribution for the ACF River basin (2001 NLCD) (mi2)
HUC8
watershed
Upper
Chattahoochee
03130001
Middle
Chattahoochee
Lake Harding
03130002
Middle
Chattahoochee
Walter F. George
Reservoir
03130003
Lower
Chattahoochee
03130004
Upper Flint
03130005
Middle Flint
03130006
Kinchafoonee
Muckalee
03130007
Lower Flint
03130008
Ichawaynochaway
03130009
Spring"
03130010
Apalachicola
03130011
Chipola
03130012
Total
Open
water
66.6
82.8
75.9
8.7
28.3
17.3
4.0
6.1
4.6
36.6
14.8
10.4
356.1
Developed land9
Open
Space
226.5
218.6
116.5
41.4
143.7
62.6
40.0
62.5
31.4
33.5
25.9
68.4
1,071.0
Low
density
150.0
127.1
56.7
10.2
79.6
20.7
13.0
27.4
5.6
8.9
2.3
7.6
509.0
Medium
density
53.8
29.9
14.8
2.3
21.9
3.9
3.1
6.5
1.1
1.4
0.6
2.2
141.4
High
density
30.5
15.6
6.0
0.8
15.0
1.6
1.5
3.3
0.2
0.4
0.1
0.6
75.6
Barren
land
10.8
23.0
9.7
0.9
16.3
1.6
0.8
1.7
0.8
0.8
0.6
1.0
68.1
Forest
804.8
1,809.3
1,707.5
470.5
1,454.6
618.3
506.4
455.7
411.6
257.6
335.7
412.7
9,244.8
Shru bland
56.8
246.5
385.0
138.1
227.5
126.6
97.1
125.9
90.6
73.3
61.4
231.4
1,860.1
Pasture/Hay
176.0
388.1
157.9
94.8
376.2
120.4
83.9
82.2
82.0
70.1
25.7
100.7
1,758.2
Cultivated
0.1
1.3
142.8
294.0
79.6
416.2
232.7
348.5
303.9
347.5
35.8
194.2
2,396.6
Wetland
9.5
98.7
164.2
70.3
192.3
164.9
117.7
95.1
172.3
127.5
339.5
250.4
1,802.5
Total
1,585.5
3,040.9
2,837.0
1,132.1
2,635.0
1,554.1
1,100.1
1,214.9
1,104.0
957.7
842.2
1,279.6
19,283.2
The percent imperviousness applied to
density(89.9%).
each of the developed land uses is as follows: open space (8.04%), low density (30.16%), medium density (60.71%), and high
Delineation for Lake Seminole crossed HUC8 boundaries so whole watershed is represented in Spring HUC8.
D-15
-------
The HSPF model is set up on a hydrologic response unit (HRU) basis. For HSPF, HRUs were formed from an
intersection of land use and hydrologic soil group, and then further subdivided by precipitation gage. Average
slopes (which tend to correlate with soils) were calculated for each HRU. Slopes in most of the watershed are
relatively mild (1-5 percent), therefore HSPF HRU's were not further subdivided by slope. The three HRUs above
Lake Lanier have average slopes of 15-24 percent, but since there were already three HRU's for four delineated
subwatersheds it was not further divided. The water land use area was adjusted to prevent double counting with
area described in HSPF reaches. SWAT HRUs are formed from an intersection of land use and SSURGO major
soils.
Facilities permitted under the National Pollutant Discharge Elimination System (NPDES) are, by definition,
considered point sources. For all models in the 20 Watershed application, it was assumed that minor dischargers
(below 1.0 MOD) were insignificant and, therefore, not included in the model setup and simulation. Data were
sought from the PCS database for the major dischargers in the ACF River basin and reflect the time period from
1991-2006. Facilities that were missing total nitrogen, total phosphorus, ortotal suspended solids (TSS)
concentrations were filled with a typical pollutant concentration value from literature based on Standard Industrial
Classification (SIC) classification. For the 20 Watershed application, the assumption was to use constant point
source flows and concentrations, for the entire simulation period, for each major discharge facility in the
watershed. Figure 3 presents the locations of the major point sources included in the models.
During the water quality calibration, it was noticed that assumptions used for total phosphorus at some facilities
were too high. An investigation into the point sources that had assumed values for total phosphorus was
conducted. It was found that point sources with assumed values for total phosphorus, that were too high, were
water pollution control plants (WPCP) and the assumed total phosphorus concentration for those facilities was 7
mg/L. A new assumed value was needed for these facilities. The new assumed value was 1.5 mg/L, which is an
average of the total phosphorus concentration for WPCP's that do monitor for total phosphorus. It is assumed that
1.5 mg/L is a much better estimate of the true total phosphorus concentration coming out of WPCP's in the ACF
basin. The new assumed value was also applied to the SWAT simulation. Both the HSPF and SWAT models used
the exact same flows and concentrations for each of the major point sources included in the simulations for the
ACF basin.
Table 3. Major point source discharges in the ACF River basin
NPDES ID
AL0000817
AL0022209
AL0022764
AL0023159
AL0024619
AL0024724
AL0059218
AL0061671
AL0072737
FL0002283
FL0026867
FL0031402
GA0000973
GA0001112
GA0001198
GA0001201
Name
MEADWESTVACO COATED BOARD INC
PHENIXCITYWWTP
DOTHAN CITY OF OMUSSEE WWTP
LANETT CITY OF WWTP
SOUTHERN NUCLEAR OPERATNG CO
EAST AL WATER LOWER VALLEY WTP
OPELIKA CITY OF EASTSIDE WWTP
EUFAULA CITY OF
DOTHAN CITY CYPRESS WWTP NEW
GULF PWR SCHOLZ STEAM
BLOUNTSTOWN-STP
FL STATE HOSPITAL
COLUMBUS WATER WKS-FT.BENNING
SCOVILL FASTENERS, INC.
USAF PLT #6 - LOCKHEED MARTIN
GA. PACIFIC CORP (GREAT S.P)
Design flow
(MGD)*
40.00
7.75
7.12
5.00
0.16
4.00
1.00
2.70
3.00
129.60
1.50
1.30
4.60
Observed flow
(MGD)
(1991-2006 average)
22.34
3.32
3.97
1.90
0.52
2.83
0.70
1.75
1.32
3.24
0.57
0.71
11.71
0.26
1.63
32.60
D-16
-------
NPDES ID
GA0001619
GA0020052
GA0020079
GA0020168
GA0020486
GA0020516
GA0021156
GA0021326
GA0021458
GA0021482
GA0021504
GA0023167
GA0024040
GA0024333
GA0024503
GA0024678
GA0025381
GA0025585
GA0026077
GA0026140
GA0026158
GA0026433
GA0030121
GA0030341
GA0030686
GA0030791
GA0031721
GA0033511
GA0035777
GA0035807
GA0036951
GA0037222
GA0038369
GA0046019
GA0046655
GA0047201
GA0047767
Name
MERCK & CO -FLINT RVR PLNT
WEST POINT WPCP
THOMASTON-BELL CREEK WPCP
GAINESVILLE (LINWOOD DRIVE)
MONTEZUMAWPCP#2
COLUMBUS (SOUTH WPCP)
GAINESVILLE FLAT CR WPCP
DAWSON WPCP
ATLANTA (UTOY CREEK WRC)
ATLANTA (R.M. CLAYTON WPCP)
CORNELIA WPCP
BUFORD SOUTHSIDE WPCP
ATLANTA (SOUTH RIVER WRC)
FULTON CO-BIG CREEK WPCP
CORDELEWPCP
BAINBRIDGEWPCP
FULTON CO-CAMP CREEK WPCP
BLAKELYWPCP
DAHLONEGAWPCP
COBB CO-SUTTON WPCP
COBB CO.-SO. COBB WPCP
GWINNETT CO (CROOKED CRK WPCP)
THOMASTON-TOWN BRANCH WPCP
DOUGLASVILLE SOUTHSIDE WPCP
FULTON CO-JOHNS CREEK WPCP
GRIFFIN POTATO CR WPCP
NEWNAN WAHOO WPCP
DECATUR CO-IND. AIRPARK WPCP
PEACHTREE CTY (LINE CRK WPCP)
FAYETTEVILLE-WHITEWTER CR WPCP
LAGRANGE WPCP (LONG CANE CRK)
ALBANY (WPCP NO 2)
CLAYTON COUNTY (SHOAL CRK)
GUMMING WPCP
PEACHTREE CTY (ROCKAWAY WPCP)
DOUGLASVILLE (SWEETWATER CRK.)
AMERICUS MILL CRK, WPCP
Design flow
(MGD)*
1.00
2.00
2.70
1.95
42.00
10.20
2.50
40.00
100.00
3.00
2.00
48.00
24.00
5.00
2.50
13.00
1.32
1.44
40.00
40.00
36.00
2.00
3.25
7.00
2.00
3.00
1.50
2.00
3.75
12.50
32.00
4.40
2.00
4.00
3.00
4.40
Observed flow
(MGD)
(1991 -2006 average)
1.13
0.64
1.01
2.03
0.30
30.35
6.45
1.29
29.50
80.00
2.33
1.11
35.31
21.09
2.66
1.32
12.47
1.01
0.51
30.62
24.50
23.52
0.97
2.48
5.73
1.41
1.66
0.38
1.44
1.88
5.91
18.05
1.73
1.30
1.66
1.39
2.73
*Note: Facilities that
the state of Georgia
do not list a design flow are large industrial facilities.
and these permits do not report a design flow.
These industrial facilities have different permitting in
D-17
-------
SCQMLL-'FASTENERS
DAHLONEGA
CORNE.lilA
GAINESVILLE
f '\ COBBCO.
DOUG~LATSVILLE
FAYETTEVILLE
FULTON GO.
CLAYTON
V 1
PEACHTREE
GRIFFIN'POTATO CR.
NEWNAN WAHOO
LAGRANGE
WEST-POINT
THOMASTON
EAST AL WATER
^ PHENIX
COLUMBUS
MONTEZUMA
AMERICUSMILLCR-.
MEADWESTVACO
DAWSON
"\ 1 r*
ALBANY
MERCK
SOUTHERN NUCLEAR
D'OTHAN
. PACIFIC C.QRB
GULF PWR
FL STATE HOSPITAL
Point Sources
Hydrography
Interstate
Water (Nat. Atlas Dataset)
US Census Populated Places
Municipalities (pop > 50,000)
County Boundaries
Watershed with HUCSs
BLOUNTSTOWN STP
. v—
Florida
GCRP Model Areas - ACF River Basin
Major Point Sources
Figure 3. Major point sources in the ACF River basin.
D-18
-------
Meteorological Data
The required meteorological data series for the 20 Watershed study are precipitation, air temperature, and
potential evapotranspiration. The 20 Watershed model does not include water temperature or algal simulation and
uses a degree-day method for snowmelt. These are drawn from the BASINS4 Meteorological Database (USEPA
2008), which provides a consistent, quality-assured set of nationwide data with gaps filled and records
disaggregated. Scenario application will require simulation over 30 years, so the available stations are those with
a common 30-year period of record (or one that can be filled from an approximately co-located station). A total of
37 precipitation stations were identified for use in the ACF basin model with a common period of record of
10/1/1972-9/30/2002 (Table 4 and Figure 4). Temperature records were sparse; where these were absent,
temperature was taken from nearby stations with an elevation correction. For each weather station, Penman-
Monteith reference evapo-transpiration was calculated for use in HSPF using observed precipitation and
temperature coupled with SWAT weather generator estimates of solar radiation, wind movement, cloud cover,
and relative humidity.
For the 20 Watershed model applications, SWAT uses daily meteorological data, while HSPF requires hourly
data. It is important to note that a majority of the meteorological stations available for the ACF basin are
Cooperative Summary of the Day stations that do not report sub-daily data. The BASINS4 dataset already has
versions of the daily data that have been disaggregated to an hourly time step using template stations. For each
daily station, this disaggregation was undertaken in reference to a single disaggregation template. Occasionally,
this automated procedure provides undesirable results, particularly when the total rainfall for the day is very
different between the subject station and the disaggregation template. This yields a small number of hourly
precipitation intensity estimates that are unrealistically high (e.g., much greater than the 100-year 1-hour event for
the region). This has only a small impact on the watershed-scale hydrologic calibration as gages are influenced by
rainfall from multiple weather stations, but can introduce significant problems for the prediction of erosion and
sediment loads.
Table 4. Precipitation stations for the ACF River basin model
COOP ID
GA094230
GA092283
GA091998
GA092408
GA096407
GA090444
GA092791
GA090451
GA096335
GA093570
GA094949
GA099506
GA099291
GA098661
GA098535
GA091425
GA092166
GA091372
GA095979
AL015397
GA090253
GA095394
AL012730
Name
HELEN
CORNELIA
CLERMONT 4 WSW
GUMMING 1 ENE
NORCROSS
ATLANTA BOLTON
DOUGLASVILLE 4 S
ATLANTA HARTSFIELD
INTERNATIONAL
NEWNAN 4NE
FRANKLIN
LA GRANGE
WOODBURY
WEST POINT
THOMASTON 2 S
TALBOTTON
BUTLER
COLUMBUS METRO AP
BUENA VISTA
MONTEZUMA
MIDWAY
AMERICUS 3 SW
LUMPKIN2SE
EUFAULA WILDLIFE REF
Latitude
34.6997
34.5181
34.4503
34.2214
33.9483
33.8236
33.7006
33.63
33.4428
33.2758
33.065
32.9839
32.8694
32.8664
32.6856
32.6525
32.5161
32.3178
32.2903
32.0597
32.0503
32.0306
32.0086
Longitude
-83.7261
-83.5286
-83.855
-84.1222
-84.2219
-84.4983
-84.7303
-84.4417
-84.7886
-85.0992
-85.0294
-84.5889
-85.1892
-84.3175
-84.5192
-84.1858
-84.9422
-84.5203
-84.0314
-85.4953
-84.2753
-84.7753
-85.0919
Temperature
Yes
Yes
No
No
No
No
No
Yes
Yes
No
Yes
No
Yes
Yes
Yes
No
Yes
No
No
No
Yes
Yes
Yes
Elevation
(ft)
1440
1470
1281
1306
1030
885
1002
1010
920
790
715
790
575
672
686
446
392
646
327
556
490
485
215
D-19
-------
COOP ID
GA092266
GA092570
GA092450
AL010008
GA093028
GA090140
GA090979
AL012377
GA091500
GA092153
GA090586
FL081544
FL089795
FL089566
Name
CORDELE
DAWSON
CUTHBERT
ABBEVILLE
EDISON
ALBANY 3 SE
BLAKELY
DOTHAN
CAMILLA 3 SE
COLQUITT2W
BAINBRIDGE INT PAPER
CHIPLEY
WOODRUFF DAM
WEWAHITCHKA
Latitude
31.9847
31.7819
31.7672
31.5703
31.5664
31.5339
31.3811
31.1942
31.1903
31.1681
30.8228
30.7836
30.7219
30.1192
Longitude
-83.7758
-84.4497
-84.7931
-85.2483
-84.7339
-84.1489
-84.9508
-85.3708
-84.2036
-84.7664
-84.6175
-85.4847
-84.8742
-85.2042
Temperature
Yes
No
Yes
Yes
Yes
Yes
Yes
No
Yes
Yes
No
Yes
No
Yes
Elevation
(ft)
308
355
461
456
294
180
268
275
175
153
190
130
107
42
D-20
-------
Legend
Weather Stations
Interstate
Hydrography
Water (Nat. Atlas Dataset)
US Census Populated Places
^H Municipalities (pop > 50,000)
^] County Boundaries
State Boundaries
Watershed with HUCSs
GCRP Model Areas - ACF River Basin
Weather Stations
NAD_1983_Albers_meters
Map produced 2-10-2010 - B. Tucker
140
^•Kilometers
100
• Miles
TETRATECH
Figure 4. Weatherstations for the ACF River basin model.
D-21
-------
Watershed Segmentation
The ACF River basin was divided into 101 sub-watersheds for the purposes of modeling (Figure 5). The initial
calibration watershed (Upper Flint HUC) is highlighted. Each of the subwatershed delineations represents roughly
a HUC 10 scale watershed. Each of the major reservoirs in the ACF basin was delineated so that the each dam
outlet represents an individual watershed outlet. The delineations were done this way to ensure that any individual
lake was contained in one watershed and that the watershed was only represented by one outlet. The ACF 20
Watershed model encompasses the complete watershed and does not require specification of any boundary
conditions for application.
D-22
-------
Legend
A USGS Gages
Hydrography
Interstate
Water (Nat. Atlas Dataset)
US Census Populated Places
Municipalities (pop > 50,000)
] County Boundaries
I I Initial Calibration Watershed
] Model Subbasins
87 // n 53 VJ51
28 JL—02353000
GCRP Model Areas - ACF River Basin
Model Segmentation
Figure 5. Model segmentation and USGS stations utilized for the ACF River basin.
Note: SWAT subwatershed numbering is shown; the HSPF model for this watershed uses the same
subwatershed boundaries with an alternative internal numbering scheme.
D-23
-------
Calibration Data and Locations
The ACF basin was selected as an early pilot site application because of previous modeling experience in parts of
the watershed and the state of Georgia. The specific site chosen for initial calibration was the Flint River at GA
26, near Montezuma, Georgia (USGS 02349605) (Table 5). This is a flow and water quality monitoring location
that approximately coincides with the pour point of the Upper Flint (03130005) 8-digit HUC (Figure 5). This
location was selected for several reasons: 1) there is a good set of flow and water quality data available, 2)
previous modeling efforts in nearby HUCS's were successful, and 3) investigations of land use, drainage area, and
percentage of drainage area controlled by flow control structures, compared with other USGS gage locations in
the ACF basin, identified this gage as the best possible choice.
There were an additional eight sites chosen for the whole ACF basin to check the performance of the model.
These sites were chosen based on subwatershed delineation boundaries, land use, drainage area, flow control
structures, data completeness and location. The eight additional sites are in Table 5 and shown spatially in Figure
5. The idea was to have some locations that were un-impacted by upstream flow control structures and
additionally also have some locations downstream of major reservoirs to check the model performance of the
reservoir simulation.
Three of the chosen sites were located at the outfall of two delineated subwatersheds. In reality, these gages are
slightly downstream of a tributary joining the mainstem. These sites were Flint River at Montezuma, Georgia,
Chattahoochee River near Cornelia, Georgia, and Flint River at Newton, Georgia. It is easy to add two flows
together to get the theoretical flow at the sampling location but two water quality concentrations cannot be
summed to get the theoretical concentration. In order to generate a theoretical concentration, constituent masses
must be added together and then divide by the summed volumes to determine what the water quality
concentration would be. Accordingly, in the SWAT application, constituent masses from two reaches were added
together, and then divided by the summed volumes to determine constituent concentration. The HSPF application,
dealt with this by combining the watersheds internally and generating one time series that represented the
hydrology and water quality where these subwatersheds merge. This method makes the assumption that the main
stem and tributary waters have fully mixed at the sampling location.
The gage for Peachtree Creek is not at the outlet of the subwatershed for Peachtree Creek. Both HSPF and SWAT
applications utilized an area weighting approach for this gage. The USGS published drainage area was 66 percent
of the drainage area of the subwatershed delineated for Peachtree Creek, so a multiplier of 0.66 was applied to the
time series at the output of Peachtree Creek. Although this is not exact, theoretically it should be close to reality at
the sampling location, because land use differences are insignificant for this watershed (90 percent urban).
The water quality data found in the NAWQA database for the chosen calibration and validation locations were
limited in certain situations. Therefore, additional data were utilized from Georgia Environmental Protection
Division (EPD). Due to earlier modeling work done in the state, these data were readily available, vetted and
included in the water quality data for calibration and validation. While combining the NAWQA and EPD datasets
there was some overlap on a few dates. The data from both datasets were compared and they always agreed on
constituent values. It was decided to keep the NAWQA data and remove the EPD data. Georgia EPD submits
their monitoring data to NAWQA and in general the additional EPD data contributed to lengthening the period of
record available for calibration and validation.
Many of the locations chosen for calibration and validation did not specifically monitor for all constituents
making up total nitrogen. Many times reported values were only for ammonia and nitrate+nitrite. The sum of
those two constituents does not represent total nitrogen because it is missing the component for organic nitrogen.
Because of this, the data available for total nitrogen were very limited. An approach was developed to bolster the
amount of total nitrogen data available for calibration and validation. The NAWQA database was investigated for
sampling dates that reported total nitrogen, ammonia, and nitrate+nitrite. These sampling dates and data were
extracted and a regression analysis of total nitrogen vs. ammonia+nitrate+nitrite was performed. The regression
D-24
-------
had an R2 vale of 0.80. Because the fit was high, the regression was applied to the ammonia+nitrate+nitrite value
and the result was an estimated value of total nitrogen based on two of the three components making up total
nitrogen. For the 20 Watershed application, it was assumed that an estimated value for total nitrogen was better
than having no value at all.
A database containing NAWQA, EPD, and calculated total nitrogen values was compiled and used in both the
SWAT and HSPF modeling applications. This ensured the data that both models were calibrated to were
consistent.
Table 5. Calibration and validation locations in the ACF River basin
Station Name
FLINT RIVER AT MONTEZUMA, GA
CHATTAHOOCHEE RIVER NEAR
CORNELIA, GA
PEACHTREE CREEK AT ATLANTA, GA
CHATTAHOOCHEE RIVER AT ATLANTA,
GA
ICHAWAYNOCHAWAY CREEK AT
MILFORD, GA
CHATTAHOOCHEE RIVER AT WEST
POINT, GA
CHATTAHOOCHEE RIVER NEAR
COLUMBIA, AL
FLINT RIVER AT NEWTON, GA
APALACHICOLA RIVER AT
CHATTAHOOCHEE FLA
USGS ID
USGS02349605
USGS02331600
USGS02336300
USGS02336000
USGS02353500
USGS02339500
USGS02343801
USGS02353000
USGS02358000
Drainage area
(mi2)
2,900.00
315
86.8
1,450
620
3,550
8,210
5,740
17,200.00
Hydrology
calibration
X
X
X
X
X
X
X
X
X
Water quality
calibration
X
X
X
X
X
X
X
X
For hydrology, the model calibration period was set to calendar years 1993-2002 (from within the 30-year period
of record for modeling). The end date is constrained by the common period of the set of 20 Watershed
meteorological stations available for the watershed, and a ten year calibration period was desired. Hydrologic
validation was then performed on Calendar Years 1983-1992. Water quality calibration used calendar years 1999-
2002, because all gages had a decent set of data during that time period. Water quality validation was limited to
1986-1998, as very sparse data were available prior to 1986. Some of the stations didn't have observed water
quality data prior to 1991. In these situations, the validation period represents all data available prior to January 1,
1999.
D-25
-------
Initial hydrologic parameterization for the Upper Flint calibration focus area came from a Loading Simulation
Program - C++ (LSPC) model created for the Upper Oconee watershed in north central Georgia (HUC
03070101). LSPC is a comprehensive data management and modeling system that is capable of representing both
flow and water quality loading from nonpoint and point sources and simulating in-stream processes. It is capable
of simulating flow, sediment, metals, nutrients, pesticides, and other conventional pollutants, as well as
temperature and pH for pervious and impervious lands and waterbodies. LSPC and HSPF use the same
parameterization, therefore, an LSPC model was chosen for initial parameterization assignments.
The LSPC model used for initial parameterization and the HSPF model set up for the 20 Watershed study differed
in land use representation as well as soils/HRU representation. The LSPC models utilized a much more detailed
land use that was develop by the state of Georgia called Georgia Land Use Trends (GLUT). Additionally, in the
LSPC model, subwatersheds were assigned to hydrologic soil groups by utilizing the hydrologic soil group that
had the greatest area within the subwatershed. Using this method the Upper Oconee watershed did not have any
hydrologic soil groups A or D. Therefore, the LSPC model was investigated to ensure the most representative
land use and soil type parameterization was transferred to the HSPF 20 Watershed model. Technical Note 6
(USEPA 2000) was utilized to establish initial parameterization for infiltration rates in areas that had hydrologic
soil groups A and D in the 20 Watershed application that.
Upon initial hydrologic parameterization of the focus area, both a model for the Upper Oconee watershed and a
model for the Lake Lanier watershed were investigated as potential starting points. The Upper Oconee watershed
parameterization did a much better job of representing measured flow than did those for the Lake Lanier
watershed parameters. At this point it was realized that parameterization assigned to the calibration focus area
probably will not work for all areas in the ACF basin.
The calibration focus area represents 7 HRUs. After calibrating the 7 HRUs, the calibrated parameterization was
transferred to the remaining 30 HRUs. Three locations, un-affected by upstream impoundments, therefore only
affected by parameterization, were selected to check the results with the focus area parameterization. These three
locations were Chattahoochee River near Cornelia, Georgia (USGS 02331600), Peachtree Creek at Atlanta,
Georgia (USGS 02336300), and Ichawaynochaway Creek at Milford, Georgia (USGS 02353500). All three of
these locations had a poor simulation of the observed hydrology. Because of the poor hydrologic simulation,
additional calibration was completed at each of the three locations.
The area contributing to Peachtree Creek at Atlanta, Georgia (USGS 02336300) was utilized as an urban area
calibration. Land use at this location is roughly 79 percent urban. Since calibration at this location, entirely
revolved around the urban land use, the calibrated results were transferred to all other urban areas throughout the
ACF basin.
The area contributing to Chattahoochee River near Cornelia, Georgia (USGS 02331600) was parameterized with
the Lake Lanier TMDL LSPC model, and is represented by two HRUs in the HSPF 20 Watershed model. The
initial parameterization was adjusted slightly, to account for the indirect transfer of land use associated
parameters, from the more detailed LSPC model to the 20 Watershed model. The calibrated parameters at this
location were transferred to one more HRU, immediately downstream. The area represented by these parameters
is closely associated with the area of the ACF basin that is in the Blue Ridge Geographic Province (Figure 6.)
The area contributing to Ichawaynochaway Creek at Milford, Georgia (USGS 02353500) was used to represent
hydrologic conditions for the HRUs in the Coastal Plain Province. To calibrate the area contributing to this gage,
the calibration focus area parameterization was adjusted until the simulated hydrology closely resembled the
D-26
-------
observed hydrology. Since adjustment was needed for multiple parameters on multiple land uses it was decided
that this gage would represent the hydrology of the Coastal Plain Province (Figure 6.)
The initial calibration focus area parameters were supplied for all other HRUs and this represents the Piedmont
Province (Figure 6.)
In summary, after realizing that parameterization assigned to the calibration focus area will not work for all areas
in the ACF basin an approach needed to be developed for assigning parameters for each HRU. After each area
that wasn't influenced by major impoundments was calibrated separate of the others it was decided that each of
the calibration areas would represent either the geologic province that each was contained in or in the case of the
Peachtree Creek gage, the dominant land use. Essentially, there are three parameter groups assigned by geologic
province and the urban land use is parameterized the same throughout the model.
After the parameter mapping was complete for all three geologic provinces the calibration turned to reservoir
representation and operation. There is a more detailed discussion about the challenges faced during modeling
reservoirs in the HSPF Assumptions section of this report.
Once the hydrology calibration was complete for the whole ACF basin, the focus turned to sediment and water
quality representation. Initial parameterization for sediment and water quality simulation was taken from a LSPC
model developed for the Lake Allatoona watershed. The Lake Allatoona TMDL model was utilized rather than
the Lake Lanier TMDL or Upper Oconee watershed LSPC models because the Lake Allatoona model utilized the
same general water quality approach that is utilized for the 20 Watershed application. The water quality
simulation also generally reflected the need to assign parameters by geologic province, therefore, water quality
was calibrated at the same locations as hydrology and the parameterization was transferred to the same HRUs as
the hydrology parameters.
D-27
-------
Legend
— Hydrography
Interstate
^H Water (Nat. Atlas Dataset)
^B US Census Populated Places
| Municipalities (pop > 50,000)
I I County Boundaries
Model Subbasins
Parameter Mapping Region
GCRP Model Areas - ACF River Basin
Parameter Mapping Regions
Figure 6. Parameter mapping utilized in the HSPF ACF River basin model.
D-28
-------
to
No changes were made to the meteorological or land use base data. The impoundments of Lake Blackshear and
Lake Seminole created an odd subwatershed connectivity when developed by BASINS. These two lakes both
have large tributaries contributing to them and BASINS delineated the watersheds by having all of the individual
tributaries pouring into the next downstream watershed. This made it difficult to represent the dam operation. The
connectivity was modified so that there was only one watershed representing the outflow of each impoundment.
The upstream subwatersheds, for all of the tributaries entering the lakes, were adjusted to pour into the
subwatershed containing the lake. This change was also made in the SWAT model. Before this change was added,
the HSPF model was in operation with the original connectivity, and BASINS generated f-tables. The simulation
results below both of the lakes, for both original and updated connectivity, showed very similar results. This
suggests that the update to the connectivity should not pose any problems.
As discussed earlier in the Point Sources section, an amendment was made to WPCP's that had default values
assigned for total phosphorus. Additionally, as discussed earlier, GaEPD data were used to supplement the
observed water quality data found in the NAWQA database.
Reservoirs
The Chattahoochee and the Flint Rivers represent two very different types if rivers. The Chattahoochee River has
many impoundments while the Flint River has one of the longest unimpeded stretches of flow in the United
States. The base data supplied point coverage for nine dams in the ACF basin and each was at the outlet of a
delineated subwatershed. Three of those given dams were assumed to operate as run of the river (Oliver, Bartlett's
Ferry, and George W. Andrews) and they were not included in the simulation. Table 6 identifies the dams and
corresponding reservoirs represented in the ACF basin 20 Watershed model.
Table 6. Reservoirs represented in the ACF basin model
Dam Name
Buford
West Point
Walter F George Lock,
Dam, Powerhouse
Crisp County (Warwick)
Muckafoonee Creek
Dam
Jim Woodruff Dam
Other Name
Lake Sidney Lanier
West Point Lake
Eufala
Lake Blackshear
Lake Worth
Lake Seminole
River
Chattahoochee
Chattahoochee
Chattahoochee
Flint
Flint
Flint
Owner
USAGE
USAGE
USAGE
Crisp County Power
Commission
Georgia Power
Company
USAGE
All of the Army Corps of Engineers' (USAGE) lakes simulated in the model have data published including
elevation, inflow, discharge, and power generation, since the facility became operational (USAGE 2010). The
USAGE has also made available a graph of the area capacity curve for each of the lakes. For the 20 Watershed
application, it was assumed that the best representation of the reservoirs was to try to simulate them without
supplying time series operations or boundary conditions. If time series operations were supplied, it would be
difficult to predict what the boundary condition would be in the future. Therefore, the area-capacity curves were
developed into an f-table, consistent with Technical Note 1 (USEPA 2007) and supplied for the subwatershed
D-29
-------
containing the lake. By having the controlling feature on the lake be an f-table future climatic conditions will not
be affected by boundary conditions of the past.
Elevation-Storage relationships were not available for Lake Blackshear and Lake Worth. Research was done to
look for the average depth and average surface area for each of the lakes. This information was utilized to come
up with a reservoir of similar size and storage by using the stage-storage-discharge relationships tool in the HSPF
BMP Toolkit on the EPA website.
One of the biggest challenges was trying to represent reservoir operations with only an f-table. When the model
was first set up, the focus was on calibrating to areas un-impacted by flow control devices, and the reservoirs
simply had the f-table created by BASINS. The hydrology results at the gages used to check reservoir operations
actually looked pretty decent considering lake storage and dam releases were not accounted for. F-tables were
developed and inserted into the model for all reservoirs included in the simulation. The physical relationship
between surface area and storage was left unchanged and reservoir calibration focused on assigning a discharge to
a particular depth of water. When there were elevation and discharge data available, they were used as a guideline
for assignment of the outflow for a particular depth. The results on the mainstem Flint River changed very little
but the results on the mainstem Chattahoochee changed drastically and the simulation became very poor. Much
work was done on the Chattahoochee River simulation to try to represent the reservoirs properly. Since the
reservoirs are highly controlled by peaking hydro electric operations and targeted elevations based on the season,
the approach used for the 20 Watershed application did not do a very good job at representing observed flows
below the dams on the Chattahoochee River. Much of the error on the Chattahoochee can be attributed to the
improper operation of Buford Dam. The discharges at Buford Dam impact the discharges at all other dams on the
Chattahoochee and this is also the case in the 20 Watershed model.
Withdrawals
It is not known what water withdrawals by municipal and industrial facilities will look like in the future,
therefore, they were not included in the 20 Watershed model application. Recent court rulings suggest that current
withdrawals below Buford dam may change in as little as three years.
Irrigation
Irrigation in the Lower Flint, its tributaries, and the Lower Chattahoochee is used quite extensively when needed.
The model, for the 20 Watershed application, is not using the irrigation module. HSPF requires that a land use be
associated with irrigation and applying irrigation to all agricultural land may greatly over estimate the amount of
irrigation that is actually taking place. Additionally, no one knows what agricultural irrigation may look like in the
future. There have been numerous studies commissioned in the past decade to look into the amount of irrigation in
the state of Georgia. A majority of the irrigation is from groundwater sources and this would represent new water
to the HSPF model. It was assumed that an irrigation component would not benefit the model for the 20
Watershed application.
Snow Simulation
Previous modeling experience in Georgia did not utilize snow simulation. The model for the 20 Watershed
application is to include snow simulation using the degree-day method for snowmelt. With no previous models to
obtain initial parameters for snow simulation, the initial parameters needed to be developed. Technical Note 6
(USEPA 2000) was used as a guideline for parameterization. The parameters for the physical properties of each
HRU are assigned by HRU but all other snow simulation parameters are the same for each HRU. These values are
assumed to be appropriate and the initial parameterization was not adjusted.
D-30
-------
Hydrology Calibration
As explained above, the starting parameters for ACF 20 Watershed application came from an LSPC model for the
Upper Oconee watershed. Differences amongst the model set up between the LSPC application and HSPF 20
Watershed application meant that not all parameters in the LSPC application were directly transferable to the
HSPF 20 Watershed application. When it did not make sense to utilize parameters from the LSPC model,
Technical Note 6 (USEPA 2000) was utilized to determine a good starting value. The parameters from the Upper
Oconee watershed simulated flows in the range of the observed flows but minor adjustments needed to be made to
better fit the simulated flows to the observed flows. Calibration adjustments focused on the following parameters:
• INFILT (index to mean soil infiltration rate): The LSPC model did not represent hydrologic soil groups A
or D. A representative value was obtained from Technical Note 6 (USEPA 2000). Very minor
modifications were made to these estimated values because the calibration focus area is dominated by B
and C soils. Additionally, INFILT was adjusted slightly by land use to account for the land use
differences between the LSPC model and the HSPF 20 Watershed model. INFILT values for the
hydrologic soil groups are in the range of those stated in Technical note 6.
• AGWRC (Groundwater recession rate): Adjusted slightly in order to replicate groundwater recession in
the observed data.
• LZSN (lower zone nominal soil moisture storage): This was increased slightly upward due to baseflow
contributions severely tapering off during extreme dry weather. The changes to INFILT and AGWRC
resulted in small modifications in this parameter.
• BASETP (ET by riparian vegetation): Even with the modifications mentioned above, simulation of low
flows was not that good. It was assumed that the Flint River watershed has a greater amount of riparian
vegetation than that of the Upper Oconee watershed. Slightly increasing the BASETP value made the
simulation of low flows much better.
Initial calibrations were performed for the upper Flint River, comparing model results to data from USGS
02349605, and are summarized in Figures 7 through 13 and Tables 7 and 8. The model fit is of high quality but
always simulates a little bit high. This could be because municipal and industrial withdrawals were not included
in the in the simulation. None of the metrics fall out of those set for the 20 Watershed study. The model
calibration period was set to calendar years 1993-2002.
D-31
-------
Avg Monthly Rainfall (in)
-Avg Observed Flow (1/1/1993 to 12/31/2002)
•Avg Modeled Flow (Same Period)
i9nnnn
mnnnn
onnnn
finnnn
/nnnn
onnnn -
Hlf
J
II
pi
, A
T
Jan-93 Jul-94
in
r
i
1
Jan-96
ill
1
JL
Jul-97
J^
f'l
JL
Jan-99 Jul-00
•*^_
Jan-02
1
9
4 ^
R £*
b ^
'ro
Q ry
^s
m 'ra
1U ro
19
Date
Figure 7. Mean daily flow at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - calibration
period (HSPF).
i
o
30000
20000
10000 -
J-93
Avg Monthly Rainfall (in)
-Avg Observed Flow (1/1/1993 to 12/31/2002
•Avg Modeled Flow (Same Period)
J-94
J-02
Figure 8. Mean monthly flow at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - calibration
period (HSPF).
D-32
-------
Avg Flow (1/1/1993 to 12/31/2002 )
•Line of Equal Value
Best-Fit Line
30000
LJ- 20000
10000
O)
TO
0 10000 20000 30000
Average Observed Flow (cfs)
O
0)
o
ro
m
_
0)
TO
100%
90% -
80%
70% -
60%
50%
Avg Observed Flow (1/1/1993 to 12/31/2002 )
Avg Modeled Flow (1/1/1993 to 12/31/2002 )
-Line of Equal Value
40% -
30% -
20% -
10% -
0%
J-93 J-94 J-96
J-97 J-99
Month
J-00 J-02
Figure 9. Monthly flow regression and temporal variation at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - calibration period (HSPF).
8000
t
36000
o
Avg Flow (1/1/1993 to 12/31/2002)
•Line of Equal Value
Best-Fit Line
y = 1.0261x-
R- = 0.9946
2000 4000 6000
Average Observed Flow (cfs)
8000
Avg Monthly Rainfall (in)
-Avg Observed Flow (1/1/1993 to 12/31/2002)
Avg Modeled Flow (Same Period)
8000
6000
4000
2000
1 2 3 4 5 6 7 8 9 10 11 12
Figure 10. Seasonal regression and temporal aggregate at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - calibration period (HSPF).
D-33
-------
Average Monthly Rainfall (in)
-Median Observed Flow (1/1/1993 to 12/31/2002)
10000 n
I Observed (25th, 75th)
Modeled (Median, 25th, 75th)
10 11
12
Figure 11. Seasonal medians and ranges at USGS 02349605 Flint River at Ga 26, near Montezuma, GA
- calibration period (HSPF).
Table 7. Seasonal summary at USGS 02349605 Flint River At Ga 26, near Montezuma, GA
calibration period (HSPF)
MONTH
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
3955.13
5767.27
6976.42
3970.93
2277.06
1783.49
3617.54
1440.71
1260.28
1816.83
2297.06
3137.85
3665.00
3915.00
5150.00
3065.00
1810.00
1670.00
1290.00
1035.00
964.00
1105.00
1660.00
2240.00
2110.00
2572.50
2932.50
2210.00
1400.00
1180.00
977.25
752.00
770.50
775.75
1220.00
1490.00
5155.00
6887.50
8377.50
4297.50
2470.00
2170.00
1765.00
1420.00
1252.50
1855.00
2685.00
3497.50
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
4080.88
5840.28
7509.10
4138.24
2259.06
1976.13
3734.42
1646.45
1584.54
1920.32
2506.66
3202.84
3477.76
4036.36
5064.71
3192.51
2007.62
1497.68
1309.40
1108.31
917.93
1202.58
1807.51
2167.47
2037.87
2764.25
2921.96
2099.52
1330.76
1168.42
880.28
803.03
712.46
785.75
878.28
1344.21
5135.68
6514.59
8921.61
4715.87
2653.59
2453.92
2131.87
1567.80
1731.82
2195.35
3533.72
3347.46
D-34
-------
•Observed Flow Duration (1/1/1993 to 12/31/2002 )
Modeled Flow Duration (1/1/1993 to 12/31/2002 )
o
D)
ro
Q
1000000
100000
10000
1000
100
10%
20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90%
100%
Figure 12. Flow exceedence at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - calibration
period (HSPF).
o
o
ro
T3
I
.0
E
"o
^3
LL
T3
.N
(0
•Observed Flow Volume (1/1/1993 to 12/31/2002 )
Modeled Flow Volume (1/1/1993 to 12/31/2002 )
120%
100%
80%
60%
40%
20%
Jan-93
Jul-94
Jan-96
Jul-97
Jan-99
Jul-00
Jan-02
Figure 13. Flow accumulation at USGS 02349605 Flint River at Ga 26, near Montezuma, GA •
calibration period (HSPF).
D-35
-------
Table 8. Summary statistics at USGS 02349605 Flint River At Ga 26, near Montezuma, GA
calibration period (HSPF)
HSPF Simulated Flow
REACH OUTFLOW FROM DSN 1001
10-Year Analysis F^riod: 1/1/1993 - 12/31/2002
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9)
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Srjring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
NasjT^SuteliffeJDpjsfficj^^
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
15.61
6.10
2.71
2.73
2.98
6.67
3.23
4.86
1.00
Error Statistics
5.50
-1.49
6.32
10.14
USGS 02349605 FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA
Hydrologic Unit Code: 3130006
Latitude: 32.29305556
Longitude: -84.04361 1 1
Drainage Area (sq-rri): 2920
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
5.17 » | 30
4.48
4.21
1.25
-2.14
0.707
0.568
0.934
30
30
20
50
Model accuracy increases
as E or E' approaches 1 .0
14.80
5.74
2.75
2.48
2.83
6.39
3.10
4.80
1.02
Clear [
Hydrology Validation
Validation for the Upper Flint calibration focus area was performed at the same location but for calendar years
1983-1992. Results are presented in Figures 14 through 20 and Tables 9 and 10. Similarly to the calibration years,
the validation years' model fit is of high quality but always simulates a little bit high. None of the metrics fall out
of the range set for the 20 Watershed study. The model validation period was set to calendar years 1983-1992.
D-36
-------
Avg Monthly Rainfall (in)
-Avg Observed Flow (1/1/1983 to 12/31/1992)
•Avg Modeled Flow (Same Period)
70000
60000
_ 5000°
^, 40000
1 30000
LJ_
20000
10000
Jan-83 Jul-84 Jan-86 Jul-87 Jan-89
Date
Jul-90
Jan-92
Figure 14. Mean daily flow at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - validation
period (HSPF).
I
o
15000
10000
5000
J-83
Avg Monthly Rainfall (in)
-Avg Observed Flow (1/1/1983 to 12/31/1992 )
•Avg Modeled Flow (Same Period)
J-84
c
ro
J-92
Figure 15. Mean monthly flow at USGS 02349605 Flint River at Ga 26, near Montezuma, GA -
validation period (HSPF).
D-37
-------
• Avg Flow (1/1/1983 to 12/31/1992 )
• - - - • Line of Equal Value
Best-Fit Line
15000
10000
5000
O)
ro
0 5000 10000 15000
Average Observed Flow (cfs)
O
0)
o
ro
m
0)
to
100%
90% -
80%
70% -
60% -
Avg Observed Flow (1/1/1983 to 12/31/1992 )
Avg Modeled Flow (1/1/1983 to 12/31/1992 )
-Line of Equal Value
50%
0%
J-83 J-84 J-86
J-87 J-89
Month
J-90 J-92
Figure 16. Monthly flow regression and temporal variation at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - validation period (HSPF).
8000
Avg Flow (1/1/1983 to 12/31/1992)
•Line of Equal Value
Best-Fit Line
y = 0.981 6x + 228.77
R2 = 0.9888
2000 4000 6000
Average Observed Flow (cfs)
8000
Avg Monthly Rainfall (in)
-Avg Observed Flow (1/1/1983 to 12/31/1992)
Avg Modeled Flow (Same Period)
8000
6000
4000
2000
2 3
4 5 6 7 8 9 10 11 12
Month
Figure 17. Seasonal regression and temporal aggregate at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - validation period (HSPF).
D-38
-------
Average Monthly Rainfall (in)
•Median Observed Flow (1/1/1983 to 12/31/1992)
• Observed (25th, 75th)
Modeled (Median, 25th, 75th)
8000
1 23456789 10 11
12
Figure 18. Seasonal medians and ranges at USGS 02349605 Flint River at Ga 26, near Montezuma, GA
- validation period (HSPF).
Table 9. Seasonal summary at USGS 02349605 Flint River at Ga 26, near Montezuma, GA
validation period (HSPF)
MONTH
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
4168.00
5242.97
5976.65
4328.17
2451 .47
1753.24
1908.70
1632.54
1283.93
1425.06
2275.14
3667.95
3115.00
4150.00
4310.00
3275.00
2135.00
1300.00
1330.00
1150.00
1120.00
1020.00
1330.00
2275.00
2152.50
2850.00
2830.00
2277.50
1420.00
986.75
927.50
816.50
881 .75
869.50
1120.00
1602.50
5350.00
6430.00
6542.50
4880.00
2937.50
2072.50
2257.50
1830.00
1500.00
1390.00
2277.50
4277.50
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
4169.05
5353.14
6371 .53
4468.43
2557.55
2100.76
2323.54
1765.46
1577.80
1695.02
2252.76
3559.80
3287.48
5008.00
4667.85
3626.93
2256.24
1331.44
1314.50
1414.88
1249.76
1001.13
1293.90
2145.93
1922.32
3062.19
2944.79
2094.33
1533.87
985.52
866.44
867.56
876.51
717.97
680.19
1079.64
5410.41
6675.95
6916.63
5365.85
3083.22
2134.09
2921.29
2433.92
1816.29
1808.37
2259.65
5244.57
D-39
-------
•Observed Flow Duration (1/1/1983 to 12/31/1992 )
Modeled Flow Duration (1/1/1983 to 12/31/1992 )
100000
o
g
o
D)
ro
Q
1 0000
1000 =
100
10%
20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90%
100%
Figure 19. Flow exceedence at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - validation
period (HSPF).
O
o
ro
T3
I
.0
E
"o
^3
LL
T3
.N
(0
•Observed Flow Volume (1/1/1983 to 12/31/1992 )
Modeled Flow Volume (1 /1/1983 to 12/31/1992 )
120%
100%
80%
60%
40%
20%
Jan-83
Jul-84
Jan-8
Jul-87
Jan-89
Jul-90
Jan-92
Figure 20. Flow accumulation at USGS 02349605 Flint River at Ga 26, near Montezuma, GA •
validation period (HSPF).
D-40
-------
Table 10. Summary statistics at USGS 02349605 Flint River at Ga 26, near Montezuma, GA
validation period (HSPF)
HSPF Simulated Flow
REACH OUTFLOW FROM DSN 1001
10-Year Analysis F^riod: 1/1/1983 - 12/31/1992
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9)
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Srjring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
NasjT^SuteliffeJDpjsfficj^^
Baseline adjusted coefficient (Garrick), E':
Monthly NSE ~
14.76
5.05
2.70
2.22
2.94
6.09
3.52
4.30
0.55
Error Statistics
5.79
-3.16
3.69
17.40
USGS 02349605 FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA
Hydrologic Unit Code: 3130006
Latitude: 32.29305556
Longitude: -84.04361 1 1
Drainage Area (sq-rri): 2920
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
1 .92 » | 30
3.33
6.93
-2.74
9.14
0.651
0.551
0.899
30
30
20
50
Model accuracy increases
as E or E' approaches 1 .0
13.95
4.87
2.78
1.89
2.88
5.89
3.29
4.42
0.51
Clear [
Hydrology Results for Larger Watershed
As discussed above, the parameters from the calibration focus were not fully transferrable to other gages in the
ACF basin. Therefore, at each of the gages un-impacted by flow control devices, an additional level of calibration
was performed. Please refer to the discussion and Figure 6 for details on how the additional calibration areas'
parameterization was assigned to the other areas in the watershed.
As stated above, the Upper Chattahoochee borrowed parameterization from a model done for the Lake Lanier
TMDL. Due to careful transferring of parameters there wasn't any adjustment made to achieve calibration. The
model fit was of high quality in the TMDL model and also has a very high goodness of fit in the 20 Watershed
model. The statistics for Upper Chattahoochee gage are within the range defined for the 20 Watershed application.
The calibration and validation results for this region are shown in Tables 11 and 12 (station 02331600).
The Ichawaynochaway Creek subwatershed did not have a model to borrow parameterization from. The calibrated
results of the calibration focus area were assigned and then adjusted until the simulated flows closely matched the
observed flows. All of the parameters adjusted were for the baseflow component of the simulation. Baseflow was
being simulated too high so the goal was to lower the amount of baseflow reaching the stream. This was achieved
D-41
-------
through increasing the amount of ET satisfied by riparian vegetation and direct evaporation from groundwater.
The model fit is fairly good except for the extreme low flows. The high simulation of the extreme low flows may
be explained by not having simulated irrigation. The calibration and validation results for this region are shown in
Tables 13 and 14 (station 02353500).
As discussed in the Assumptions section for Reservoirs, the simulation at all gages on the mainstem
Chattahoochee, below Lake Lanier, is very poor. The approach taken to simulate the reservoirs in the ACF did not
simulate Lake Lanier very well, but simulation of the other reservoirs was acceptable. This could be because the
reservoirs other than Lake Lanier are mostly managed as inflow equals outflow and the discharges at Lake Lanier
usually control most of the inflow. Once the Chattahoochee and Flint converge and leave Lake Seminole, the
simulation and model fit is once again of high quality and shown for the calibration period in Figures 21 through
27 and Tables 11 and 12.
Avg Monthly Rainfall (in)
-Avg Observed Flow (1/1/1993 to 12/31/2002)
•Avg Modeled Flow (Same Period)
350000
300000
250000
200000
150000
100000
50000
Jan-02
14
Figure 21. Mean daily flow: Model DSN 9001 vs. USGS 02358000 Apalachicola River At
Chattahoochee, FL- calibration period (HSPF).
100000
80000
Avg Monthly Rainfall (in)
-Avg Observed Flow (1/1/1993 to 12/31/2002 )
•Avg Modeled Flow (Same Period)
J-:
J-02
c
'ro
OL
Figure 22. Mean monthly flow: Model DSN 9001 vs. USGS 02358000 Apalachicola River At
Chattahoochee, FL- calibration period (HSPF).
D-42
-------
JD
LL
0)
E
0)
0)
100000
80000
60000
40000
20000
Avg Flow (1/1/1993 to 12/31/2002 )
•Line of Equal Value
Best-Fit Line
0.9871X
R2 - Q
+ 1737
9334
=2
0 20000 40000 60000 80000 100000
Average Observed Flow (cfs)
O
£=
_
ro
m
y
"co
Avg Observed Flow (1/1/1993 to 12/31/2002 )
Avg Modeled Flow (1/1/1993 to 12/31/2002 )
-Line of Equal Value
J-93 J-94
J-96
J-97 J-99
Month
J-OO J-02
Figure 23. Monthly flow regression and temporal variation: Model DSN 9001 vs. USGS 02358000
Apalachicola River At Chattahoochee, FL- calibration period (HSPF).
Avg Flow (1/1/1993 to 12/31/2002)
Best-Fit Line
cnnnn
4nnnn -
onnnn
onnnn
10000
n
y = 1.
R
.'
OOOSx +
2 = 0 99'
*
1470
>o
s
£fr*
>
^f'
.•'
10000 20000 30000 40000 50000
Average Observed Flow (cfs)
Avg Monthly Rainfall (in)
-Avg Observed Flow (1/1/1993 to 12/31/2002)
•Avg Modeled Flow (Same Period)
50000
2 3
4 5 6 7 8 9 10 11 12
Month
Figure 24. Seasonal regression and temporal aggregate: Model DSN 9001 vs. USGS 02358000
Apalachicola River At Chattahoochee, FL- calibration period (HSPF).
D-43
-------
Average Monthly Rainfall (in)
•Median Observed Flow (1/1/1993 to 12/31/2002)
• Observed (25th, 75th)
Modeled (Median, 25th, 75th)
11
12
c
'co
>,
o
Figure 25. Seasonal medians and ranges: Model DSN 9001 vs. USGS 02358000 Apalachicola River At
Chattahoochee, FL- calibration period (HSPF).
Table 11. Seasonal summary: Model DSN 9001 vs. USGS 02358000 Apalachicola River At
Chattahoochee, FL- calibration period (HSPF).
MONTH
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
24730.16
34453.44
41056.97
25249.20
15972.58
12939.23
18873.90
12411.87
10850.80
12458.10
13896.10
19332.77
19950.00
29050.00
36850.00
19800.00
15250.00
12600.00
11950.00
11050.00
8235.00
10750.00
13500.00
14550.00
13500.00
18275.00
19300.00
16000.00
9570.00
8237.50
7665.00
7102.50
6530.00
6112.50
6557.50
9150.00
30475.00
43900.00
53300.00
30250.00
19675.00
17600.00
15575.00
14300.00
12900.00
13900.00
18250.00
23800.00
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
26069.98
35291 .43
43511.83
26617.77
16032.37
14124.72
20000.73
13379.10
12790.03
15607.91
16250.34
20267.76
20647.35
27706.10
32896.58
21242.91
14369.52
12931.98
11560.84
11102.23
9367.20
11368.60
12148.80
15105.64
14491.39
17766.08
19444.48
15995.22
12295.58
9530.52
9687.78
8740.66
7476.25
8238.77
9171.31
10019.73
32838.93
43841 .35
51756.70
30764.85
17570.34
17294.53
16487.60
14248.45
13854.01
16874.80
19639.35
22663.99
D-44
-------
o
g
o
en
ro
Q
•Observed Flow Duration (1/1/1993 to 12/31/2002 )
Modeled Flow Duration (1/1/1993 to 12/31/2002 )
1000000
1 00000
10000
1000
10%
20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90%
100%
Figure 26. Flow exceedence: Model DSN 9001 vs. USGS 02358000 Apalachicola River At
Chattahoochee, FL- calibration period (HSPF).
o
o
ro
T3
I
.0
E
"o
^3
LL
T3
.N
(0
•Observed Flow Volume (1/1/1993 to 12/31/2002 )
Modeled Flow Volume (1/1/1993 to 12/31/2002 )
120%
100%
80%
60%
40%
20%
Jan-93
Jul-94
Jan-96
Jul-97
Jan-99
Jul-00
Jan-02
Figure 27. Flow accumulation: Model DSN 9001 vs. USGS 02358000 Apalachicola River At
Chattahoochee, FL- calibration period (HSPF).
D-45
-------
Table 12. Summary statistics: Model DSN 9001 vs. USGS 02358000 Apalachicola River At
Chattahoochee, FL- calibration period (HSPF).
HSPF Simulated Flow
REACH OUTFLOW FROM DSN 9001
10-Year Analysis F^riod: 1/1/1993 - 12/31/2002
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9)
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Srjring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
NasjT^SuteliffeJDpjsfficj^^
Baseline adjusted coefficient (Garrick), E':
Monthly NSE ~
17.06
5.55
4.09
3.07
3.46
6.82
3.72
4.27
0.72
Error Statistics
7.35
12.13
8.50
9.50
USGS 02358000 APALACHICOLA RIVER AT CHATTAHOOCHEE FLA
Hydrologic Unit Code: 3130011
Latitude: 30.701 0251
Longitude: -84.8590871
Drainage Area (sq-rri): 17200
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
14.07 » | 30
4.69
4.78
15.09
12.02
0.769
0.575
0.922
30
30
20
50
Model accuracy increases
as E or E' approaches 1 .0
15.89
5.11
3.65
2.80
3.03
6.51
3.55
3.71
0.64
Clear [
D-46
-------
The calibration and validation statistical measurements, at all USGS gages used in the ACF basin for the 20
Watershed project, are shown in Tables 13 and 14 respectively.
Table 13. Summary statistics (percent error) for all stations - calibration period 1993-2002
(HSPF)
Station
Error in
total
volume:
Error in
50%
lowest
flows:
Error in
10%
highest
flows:
Seasonal
volume
error -
Summer:
Seasonal
volume
error- Fall:
Seasonal
volume
error -
Winter:
Seasonal
volume
error -
Spring:
Error in
storm
volumes:
Error in
summer
storm
volumes:
Daily
Nash-
Sutcliffe
Coefficient
of
Efficiency,
E:
Monthly
Nash-
Sutcliffe
Coefficient:
of
02349605
5.50
-1.49
6.32
10.14
5.17
4.48
4.21
1.25
-2.14
0.707
0.934
02331600
0.14
3.09
4.62
2.28
6.04
4.19
-10.90
-0.25
-28.59
0.640
0.862
02336300
17.16
2.56
6.29
1.30
19.26
27.56
13.02
8.76
3.05
0.536
0.477
02353500
3.93
17.11
9.69
1.33
-8.97
9.57
9.42
6.61
-34.98
0.339
0.652
02336000
24.16
59.65
-7.60
8.20
42.13
16.81
35.91
-56.00
-62.85
0.539
0.683
02339500
15.40
74.07
-0.36
9.33
19.50
9.17
27.67
-70.75
-87.05
0.591
0.821
02343801
16.79
49.13
8.63
11.54
26.84
10.78
24.10
-37.96
-60.30
0.717
0.858
02353000
8.33
4.70
14.97
16.06
7.59
7.54
4.65
4.85
12.41
0.607
0.928
02358000
7.35
12.13
8.50
9.50
14.07
4.69
4.78
15.09
12.02
0.769
0.922
D-47
-------
Station
Efficiency
02349605
02331600
02336300
02353500
02336000
02339500
02343801
02353000
02358000
Table 14. Summary statistics (percent error) for all stations - validation period 1983-1992
(HSPF)
Station
Error in
total
volume:
Error in
50%
lowest
flows:
Error in
10%
highest
flows:
Seasonal
volume
error -
Summer:
Seasonal
volume
error- Fall:
Seasonal
volume
error -
Winter:
Seasonal
volume
error -
Spring:
Error in
storm
volumes:
Error in
summer
storm
volumes:
Daily
Nash-
Sutcliffe
Coefficient
of
Efficiency,
E:
Monthly
Nash-
Sutcliffe
Coefficient:
of
02349605
5.79
-3.16
3.69
17.40
1.92
3.33
6.93
-2.74
9.14
0.651
0.899
02331600
-8.32
-4.98
-6.88
-9.94
-4.13
-2.73
-17.08
-22.55
-50.35
0.696
0.865
02336300
13.73
4.87
3.05
5.82
19.78
19.85
7.49
6.12
7.78
0.553
0.654
02353500
1.19
-1.92
6.48
23.48
-14.67
-3.02
7.15
-0.18
-22.89
0.385
0.652
02336000
12.01
47.20
-14.34
-2.55
8.67
22.43
19.52
-61.77
-71.06
0.479
0.845
02339500
10.86
80.89
-5.02
6.67
-1.60
16.08
21.34
-74.01
-87.14
0.566
0.797
02343801
13.28
37.90
7.57
13.98
8.04
14.33
15.92
-56.18
-77.18
0.698
0.858
02353000
6.91
-1.14
11.27
17.92
7.94
4.62
3.64
-10.57
-23.32
0.682
0.890
02358000
2.21
-9.05
8.49
3.55
1.33
4.57
-1.81
20.87
-5.49
0.707
0.914
D-48
-------
Station
Efficiency
02349605
02331600
02336300
02353500
02336000
02339500
02343801
02353000
02358000
D-49
-------
An additional check was done on the reservoirs of the Chattahoochee River mainstem. Observed and modeled
reservoir elevations were compared by using a histogram approach. This check ensured that the storage contained
within a reservoir was accounted for even though the flow calibration downstream of the reservoir didn't simulate
well. These comparisons were performed for the time period from January 1993 to December 2002. All lake
elevation simulations closely compared to the observed elevations except for Lake Lanier (Figures 28 through
31).
Lake Lanier-Simulated and Measured Elevation
1000
I 500
Stage (ft)
Figure 28. Histogram of simulated and measured elevation for Lake Lanier from 1/1/1993 to
12/31/2002.
D-50
-------
West Point - Simulated and Observed Elevation
600
500
400
Q
300
200
100
PI n n
-------
Walter F. George - Simulated and Observed Elevation
1800
1600
1400
1200
1000
ra
Q
800
600
400
200
CD
05 05
Stage (ft)
Figure 30. Histogram of simulated and measured elevation for Lake Walter F. George from 1/1/1993 to
12/31/2002.
D-52
-------
Lake Seminole - Simulated and Measured Elevation
1 Ann
1 onn
onn -
1
0
n
PI
— .
—
O IO O LO O
n
m o o o o o o
D Simulated
• Measured
Stage (ft)
Figure 31. Histogram of simulated and measured elevation for Lake Seminole from 1/1/1993 to
12/31/2002.
Water Quality Calibration and Validation
The 20 Watershed models are designed to provide water quality simulation for total suspended solids (TSS), total
nitrogen, and total phosphorus. TSS is simulated with the standard HSPF approach (USEPA 2006). In contrast to
TSS, total nitrogen and total phosphorus are simulated in this application in a simplistic fashion, as HSPF general
quality constituents (GQUALs) subject to an exponential decay rate during transport.
The water quality calibration focuses on the replication of monthly loads, as specified in the project QAPP. Given
the approach to water quality simulation in the 20 Watershed model, a close match to individual concentration
observations cannot be expected. Comparison to monthly loads presents challenges, as monthly loads are not
observed. Instead, monthly loads must be estimated from scattered concentration grab samples and continuous
flow records. As a result, the monthly load calibration is inevitably based on the comparison of two uncertain
numbers. Nonetheless, calibration is able to achieve a reasonable agreement. Further, the load comparisons were
supported by detailed examinations of the relationships of flows to loads and concentrations and the distribution
of concentration prediction errors versus flow, time, and season, as well as standard time series plots.
For application on a nationwide basis, the 20 Watershed protocols assume that TSS and total phosphorus loads
will likely exhibit a strong positive correlation to flow (and associated erosive processes), while total nitrogen
loads, which often have a dominant groundwater component, will not. Accordingly, TSS and total phosphorus
D-53
-------
loads were estimated from observations using a flow-stratified log-log regression approach, while total nitrogen
loads were estimated using a flow-stratified averaging estimator, consistent with the findings of Preston et al.
(1989).
Similarly to hydrology, initial calibration and validation of water quality was done on the Upper Flint River,
comparing model results to data from USGS 02349605. The initial calibration used calendar years 1999-2002 for
calibration and calendar years 1991-1998 for validation as there were no data available prior to 1992 for this gage.
As stated above, initial water quality parameters were obtained from an LSPC model for a TMDL done for the
Lake Allatoona watershed in North Georgia. The Lake Allatoona watershed LSPC model was parameterized with
values from literature, information collected in the field, and from previous modeling work done in the state of
Georgia. With the exception of shrub lands, both the TMDL model and 20 Watershed model had similar land
uses. Shrub lands had parameters assigned from forest lands since these two land uses should behave similarly to
each other.
Time series of simulated and estimated TSS loads at the Upper Flint gage for both periods are shown in Figure 32
and statistics for the two periods are provided separately in Table 15. Results of the TSS calibration are generally
acceptable. Visually, the model is roughly simulating the trends contained in the observed data but the loading
estimates are on the high side. The statistics performed on the comparison between the simulated results and
observed data also indicate that TSS loading is slightly high. The key statistic in the table (consistent with the
QAPP) is the relative percent error, which shows the error in the prediction of monthly load normalized to the
estimated load. The table also shows the relative average absolute error, which is the average of the relative
magnitude of errors in individual monthly load predictions. This number is inflated by outlier months in which the
simulated and estimated loads differ by large amounts (which may be as easily due to uncertainty in the estimated
load due to limited data) and the third statistic, the relative median absolute error, is likely more relevant and
shows acceptable agreement.
TSS
1,000,000
o
«
I
-Regression Loads
-Simulated Loads
Figure 32. Fit for monthly load of TSS USGS 02349605 Flint River at Ga 26, near Montezuma, GA
(HSPF).
D-54
-------
Table 15. Model fit statistics (observed minus predicted) for monthly TSS loads using
stratified regression
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1999-2002)
-117%
129%
38.1%
Validation period
(1991-1998)
-78%
110%
44.5%
A variety of other diagnostics were also pursued to ensure agreement between the model and observations. These
are available in full in the calibration spreadsheets, but a few examples are provided below. First, load-flow power
plots were compared for individual days (Figures 33 and 34). These confirm that the relationship between flow
and load is consistent across the entire range of observed flows, for both the calibration and validation periods.
TSS Load, tons/day
FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA 1999-2002
1 nnnnn
I UUUUU
10000 -
1000 -
100 -
10
1
0.1 -
0.01 -
Onni
* »
" : j^
l^K^^^
JiSsir *'
AJpf*
i) :
: *
4
*
1 10 100 1000 10000 100000
Flow, cfs
« Simulated A Observed ^^"Fbwer (Simulated) ^^~ Power (Observed)
Figure 33. Power plot for observed and simulated TSS at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - calibration period (HSPF).
D-55
-------
FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA 1986-1998
-1
"5>
8
1000000
100000
10000
1000
100
10
1
0.1
10 100 1000 10000 100000 1000000
Flow, cfs
» Simulated A Observed
Power (Simulated)
Power (Observed)
Figure 34. Power plot for observed and simulated TSS at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - validation period (HSPF).
Standard time series plots (Figure 35) show that observed and simulated concentrations achieve good agreement,
although individual observations may deviate. Plots of concentration error versus flow and versus month (not
shown) were used to guard against hydrologic and temporal bias. Finally, statistics on concentration (Table 16)
show that acceptable median errors are achieved for the calibration period.
D-56
-------
FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA
_. I .
A
Observed
1000
100
CO
1999
2000
2001
Year
2002
Figure 35. Time series plot of TSS concentration at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - calibration period (HSPF).
Table 16. Relative errors (observed minus predicted), TSS concentration at USGS 02349605
Flint River at Ga 26, near Montezuma, GA (HSPF).
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1999-2002)
49
3.30%
19.78%
Validation period
(1991-1998)
24
-97.85%
-33.27%
The general quality constituent parameters in the Lake Allatoona TMDL model for total phosphorus were
essentially directly transferred to the 20 Watershed model. As stated earlier, shrub land was parameterized
similarly to forested lands. The model simulates total phosphorus from the uplands as having both sediment-
associated and buildup-washoff components. The sediment-associated component of the surface load reflects
mineral phosphorus, while the buildup-washoff component addresses organic phosphorus.
Initial parameter assignments for phosphorus for the Upper Flint calibration focus area performed well. Minor
adjustments were made to the interflow and groundwater component. The same percent adjustment was made to
all land uses in order to keep the land use associated loading rates, developed in the TMDL model, intact. All
streams in the calibration focus area were supplied with the same first order decay rate. This decay rate was
obtained from the TMDL model and is consistent with other modeling work conducted throughout the state of
Georgia. Adjustment was not made to this parameter while calibration was performed on the calibration focus
area.
Monthly loading series for total phosphorus are shown in Figure 36 and load statistics are summarized in Table
17. In general, the observed and simulated total phosphorus loads attain an acceptable match for both the
calibration and validation periods. There are a few locations where the simulation is not trending and the
D-57
-------
simulated loads are higher than the observed loads. These errors are most likely attributed to the error in the TSS
simulation during the same time period.
Total P
1000
100 --
o
E
In
I
-Regression Loads
-Simulated Loads
10 -----
Figure 36. Fit for monthly load of total phosphorus at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA (HSPF).
Table 17. Model fit statistics (observed minus predicted) for monthly total phosphorus loads
using stratified regression
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1999-2002)
-59%
69%
35.3%
Validation period
(1986-1998)
-23%
35%
18.5%
As with TSS, additional diagnostics for total phosphorus included flow-load power plots (Figures 37 and 38),
time series plots (Figure 39) and analysis of concentration errors (Table 18). All show acceptable agreement.
D-58
-------
FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA 1999-2002
•\ nn
I UU
10 -
>
ra
•o
1 1
3
§ 0.1
Q.
0.01 -
Onni
»
««»
•>••••**•*
«**i» * *£**
-------
FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA
A A A AAA A AAA
1999
2000
2001
Year
2002
Figure 39. Time series plot of total phosphorus concentration at USGS 02349605 Flint River at Ga 26,
near Montezuma, GA- calibration period (HSPF).
Table 18. Relative errors (observed minus predicted), total phosphorus concentration at
USGS 02349605 Flint River at Ga 26, near Montezuma, GA (HSPF)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1999-2002)
48
-31 .2%
-26.5%
Validation period
(1986-1998)
80
-2.2%
-12.5%
As discussed above in the Calibration Data and Locations section of this report, the number of measured total
nitrogen observations was very limited for the ACF. The approach used to estimate the observed total nitrogen
should give reasonable values for total nitrogen since two of the three components making up total nitrogen were
measured. Similarly to total phosphorus, total nitrogen parameters in the Lake Allatoona TMDL model were
easily transferred to the 20 Watershed model. Also, similarly to total phosphorus, initial interflow and
groundwater total nitrogen concentrations were adjusted for all land uses together in order to keep the land use
associated loading rates, developed in the TMDL model, intact.
Results for total nitrogen are summarized in Figures 40 through 43 and Tables 19 and 20, following the same
format as total phosphorus. The results are acceptable, and generally better than those for total phosphorus. This is
because nitrogen is not sediment-associated, therefore, problems with sediment are not reflected in the calibration
for total nitrogen.
D-60
-------
3,000
Total N
-Averaging Loads
-Simulated Loads
Figure 40. Fit for monthly load of total nitrogen at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA (HSPF).
Table 19. Model fit statistics (observed minus predicted) for monthly total nitrogen loads
using averaging estimator (HSPF)
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1999-2002)
-30%
54%
26.1%
Validation period
(1986-1998)
-22%
42%
18.2
D-61
-------
FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA 1999-2002
•\ nnn
I UUU
> 100-
ra
•o
«
J. 10-
•o
ra
o
* 1
01
•
„_._
» » »
.. .. ».«.•«
* ,».>»V>
* »*s
°t*tf ¥>
ji^1
-#r-
10 100 1000 10000 100000
Flow, cfs
• Simulated A Observed ^^"Fbwer (Simulated) ^^~ Power (Observed)
Figure 41. Power plot for observed and simulated total nitrogen at USGS 02349605 Flint River at Ga
26, near Montezuma, GA - calibration period (HSPF).
FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA 1986-1998
1000
100
ra
•o
«
c
o
•o
ra
o
10 100 1000 10000 100000 1000000
Flow, cfs
• Simulated A Observed ^^™Power (Simulated) ^^™Power (Observed)
Figure 42. Power plot for observed and simulated total nitrogen at USGS 02349605 Flint River at Ga
26, near Montezuma, GA - validation period (HSPF).
D-62
-------
FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA
1999
2000
2001
Year
2002
Figure 43. Time series plot of total nitrogen concentration at USGS 02349605 Flint River at Ga 26,
near Montezuma, GA- calibration period (HSPF).
Table 20. Relative errors (observed minus predicted), total nitrogen concentration at USGS
02349605 Flint River at Ga 26, near Montezuma, GA (HSPF)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1999-2002)
48
25.5%
19.3%
Validation period
(1986-1998)
76
27.2%
14.7%
Water Quality Results for Larger Watershed
Similar to hydrology, the Upper Flint water quality parameterization was not directly transferable to other areas
of the watershed. The Upper Flint parameters were utilized as starting parameters at the other calibration
locations. Once those locations reasonably agreed with the observed data, they were transferred to other parts of
the watershed, as with the hydrology calibration. The decay rates assigned to the streams in the calibration focus
area were also assigned to all streams in the ACF basin.
Upon initial water quality parameterization mapping, the water quality simulation below the reservoirs was
checked. At this point the simulation suggested that each of the reservoirs consumed all of the nitrogen and
phosphorus entering the reservoir and the water leaving the reservoir was free of all total nitrogen and total
phosphorus. This was taken to mean that due to longer residence time within reservoir reaches, the first order
decay rate applied to the reservoirs was too high. Total nitrogen and total phosphorus first order decay rates were
lowered until the simulation below each of the reservoirs better matched the observed water quality. This resulted
D-63
-------
in decay rates that were an order of magnitude lower than those applied to the other stream reaches. HSPF
assumes that the water within a reach is both vertically and horizontally mixed; therefore it does not take into
account nutrient transformations and cycling occurring within a reservoir. The lowered decay rate was utilized as
a parameter to capture nutrient dynamics within the reservoirs, therefore they decay rate was the only parameter
adjusted in order to get a reasonable representation of water quality below the reservoirs.
Summary statistics for the water quality calibration and validation at all stations in the watershed are provided in
Tables 21 and 22. The results of the water quality calibration and validation are much better at some gages than
others. The gages that are simulating poorly are probably doing so because of a lack of data to reasonably
construct observed monthly loadings. Another source of error that can balloon error statistics is poor hydrology
simulation at some of the monitoring locations. These two errors coupled together can severely impact the error
statistics presented.
Table 21. Summary statistics for water quality (observed minus predicted) for all stations -
calibration period 1999-2002 (HSPF)
Station
Relative Percent
Error TSS Load
TSS
Concentration
Median Percent
Error
Relative Percent
Error TP Load
TP
Concentration
Median Percent
Error
Relative Percent
Error TN Load
TN
Concentration
Median Percent
Error
02349605
-117%
19.78%
-59%
-26.5%
-30%
19.3%
02331600
-4%
16.73%
-1%
12.8%
-25%
4.2%
02336300
74%
10.16%
24%
-13.8%
-38%
16.8%
02353500
-438%
-38.90%
-205%
-224.5%
-38%
46.7%
02336000
-215%
1.72%
-77%
-40.1%
-35%
35.8%
02339500
-141%
-120.07%
-272%
-759.0%
2%
46.4%
02343801
34%
28.57%
-202%
-1107.9%
3%
54.8%
02353000
-63%
-64.66%
-82%
-89.2%
32%
53.1%
D-64
-------
Table 22. Summary statistics for water quality (observed minus predicted) for all stations -
validation period 1986-1998 (HSPF)
Station
Relative Percent
Error TSS Load
TSS
Concentration
Median Percent
Error
Relative Percent
Error TP Load
TP
Concentration
Median Percent
Error
Relative Percent
Error TN Load
TN
Concentration
Median Percent
Error
02349605
-78%
-33.3%
-23%
-12.5%
-22%
14.7%
02331600
90%
7.2%
53%
43.6%
-8%
1.3%
02336300
89%
3.5%
47%
5.9%
-21%
20.7%
02353500
-570%
-103.8%
-94%
-17.0%
-20%
32.7%
02336000
18%
6.7%
-16%
-9.1%
10%
20.0%
02339500
-7%
-36.9%
-64%
-85.2%
25%
-28.2%
02343801
84%
8.0%
54%
-499.6%
7%
31.2%
02353000
-77%
-1.0%
-22%
-26.1%
16%
-39.9%
D-65
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
to
As mentioned in the above section for the HSPF model, no changes were made to the meteorological or land use
base data for the SWAT model. However, an error in the connectivity of reaches was found in the predefined
reach file, which was rectified. The impoundments of Lake Blackshear and Lake Seminole have large tributaries
contributing to them. The pre-defined reach file had reaches of the upstream tributaries draining into the next
downstream reach rather than into the reach within the subwatershed containing the lakes. This made the
contributing drainage areas for these lakes incorrect. The connectivity was modified so that the tributaries pour
into the reaches within the subwatersheds containing the lakes. Incorrect connectivity of reaches would pose a
problem because of the incorrect drainage area contributing flow to the impoundments. In the SWAT model,
reservoirs are modeled to be simulated at the outlet of the subwatershed in which they are located and receive
flow from all the upstream drainage area.
Jim Woodruff Dam, Muckafoonee Creek Dam, West Point, and Buford (Table 6) reservoirs were represented in
the ACF basin 20 Watershed SWAT model. Pertinent reservoir information including surface area and storage at
principal (normal) and emergency spillway levels for the reservoirs modeled were obtained from the National
Inventory of Dams (NID) database (USAGE 1982). The SWAT model provides four options to simulate reservoir
outflow: 1) measured daily outflow, 2) measured monthly outflow, 3) average annual release rate for uncontrolled
reservoir, and 4) controlled outflow with target release. Keeping the goals of the 20 Watershed climate change
impact evaluation application, it was assumed that the best representation of the reservoirs was to simulate them
without supplying time series of outflow records. Therefore, a target release approach was used in the GCRP-
SWAT model. The average release rate was estimated using the outflow data available at the (USAGE 2010).
The number of days to reach target storage was assumed to be 90 days for all lakes except Lake Worth, which was
assumed to be 10 days.
Croplands occupy about eight percent of the total watershed area. It was found that irrigation occurred on about
5.5 percent of the total watershed area with 4.18 percent being irrigated by groundwater and 1.35 percent being
irrigated by surface water (Hook 2009). To simulate irrigation in the SWAT model, the auto-irrigation feature was
used in the management set-up on those HRUs that represented cotton and peanut crops.
The SWAT model setup for the ACF basin was set up fresh, with no prior-existing SWAT model for the
watershed. The model calibration period was set to calendar years 1993-2002.
Consistent with the HSPF modeling efforts, the specific site chosen for initial calibration was the Flint River at
GA 26, near Montezuma, GA (USGS 02349605) (Table 5). Most of the calibration efforts were geared toward
getting a closer match between simulated and observed flows at the outlet of the calibration focus area. Initially,
the parameters set for this area were applied across the watershed and the model performance was verified at other
D-66
-------
stations. Model performance was not the same as it was for the calibration focus area, mostly due to the
dominance of different land uses in different parts of the watershed. In response to the variations in spatial
characteristics of the subwatersheds, a systematic adjustment of parameters individually, by land use type has
been adopted and the same adjustment is applied throughout the watershed. Observed data at other gaging stations
with the dominance of a different land use type was used to adjust the corresponding parameters. For example, at
the gaging station that drains predominantly urban land, the area was used to set the parameters for the urban land
areas.
It is acknowledged that a hydrologic/water quality model can be precisely calibrated, given the degree of freedom,
resources, time, and data. Keeping in view the interests of this project, which are to study the land use change and
climate change impacts on flow and water quality, a site specific calibration was deliberately not attempted. To
some extent, the limitation of this approach is that the local differences in soil, weather, management, and
hydrology is not thoroughly accounted for. This approach will provide an idea of the model performance when it
is not spatially-tightly calibrated and what to expect when transferring the parameters to other ungaged
watersheds or to watersheds where detailed modeling is not practical due to limited resources.
Land Use/Soil/Slope Definition
A 5/10/5 percent threshold was used for land use/soil/slope in the SWAT model while defining the HRUs. The
cropland HRUs were split into cotton, peanuts, and corn in proportions of 48, 30, and 22 percent, respectively.
Further these classes and the urban (including current and future urban class types) classes were exempt from
applying the thresholds.
The calibration focus area represents 21 subwatersheds that, together, consist of 1,342 HRUs. The parameters
were adjusted within the practical range to obtain a reasonable fit between the simulated and measured flows in
terms of Nash-Sutcliffe modeling efficiency and the high flow and low flow components as well as the seasonal
flows. The general land use characteristics of the watershed were represented well in the calibration focus area.
Two other locations: one predominantly forested (Chartahoochee River near Cornelia, GA; USGS 02331600) and
the other, predominantly urban (Peachtree Creek at Atlanta, GA; USGS 02336300), were chosen to set the
parameters for forest and urban areas, respectively. These parameters were then applied across the entire
watershed. There is essentially one set of parameters for a land use type for the entire watershed.
During calibration, parameters were carefully adjusted such that different components of streamflow contribution
were adequately simulated. For instance, the observed and simulated baseflow and surface runoff contributions to
streamflow, as well as seasonal flows, matched well. Reasonable estimations of actual ET and crop yields were
also given consideration.
Calibration adjustments focused on the following parameters:
• Curve numbers (varied systematically by land use)
• ESCO (soil evaporation compensation factor)
• SURLAG (surface runoff lag coefficient)
• Groundwater "revap" rates
• Baseflow factor
• GW_DELAY (groundwater delay time)
• GWQMN (threshold depth of water in the shallow aquifer required for return flow to occur)
• RevapMN (threshold depth of water in the shallow aquifer required for "revap" or percolation to the deep
aquifer to occur
• CANMAX (maximum canopy storage)
• Manning's "n" value for overland flow, main channels, and tributary channels
• Sol_AWC (available water capacity of the soil layer, mm water/mm of soil)
D-67
-------
Initial calibrations were performed for the Flint River at GA 26, near Montezuma, GA and are summarized in
Figures 44 through 50 and Tables 23 and 24. As evidenced through the time series plots and the Nash-Sutcliffe
modeling efficiency, the model performed well in simulating the timing and magnitude of streamflow for various
seasons.
I Avg Monthly Rainfall (in)
-Avg Observed Flow (1/1/1993 to 12/31/2002 )
Avg Modeled Flow (Same Period)
i?nnnn
mnnnn
onnnn
finnnn
4nnnn
onnnn
Ljjt
"|
Jt
Hf'F
i
J
Jan-93 Jul-94
r
i,
m
,!
i
Jan-96
Ifll If
ill I
PT|
'
I
iJ
Jt ,
Jul-97
Date
Jj^
W
']
P|||"l '
_i
IP
_ o
4
m
19
Jan-99 Jul-00 Jan-02
Figure 44. Mean daily flow at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - calibration
period (SWAT).
30000
20000
10000
J-93
Avg Monthly Rainfall (in)
-Avg Observed Flow (1/1/1993 to 12/31/2002 )
Avg Modeled Flow (Same Period)
J-02
•E
o
Figure 45. Mean monthly flow at USGS 02349605 Flint River at Ga 26, near Montezuma, GA
calibration period (SWAT).
D-68
-------
i
I
30000
T3
JD
CD
T3
O
• Avg Flow (1/1/1993 to 12/31/2002)
Line of Equal Value
Best-Fit Line
20000
CD
D)
5
CD
<
10000
10000 20000 30000
Average Observed Flow (cfs)
T3
O
CD
O
C
JS
ro
m
(D
"5
Avg Observed Flow (1/1/1993 to 12/31/2002)
Avg Modeled Flow (1/1/1993 to 12/31/2002 )
Line of Equal Value
J-93 J-94 J-96
J-97 J-99
Month
J-00 J-02
Figure 46. Monthly flow regression and temporal variation at USGS 02349605 Flint River at Ga 26,
near Montezuma, GA- calibration period (SWAT).
Avg Flow (1/1/1993 to 12/31/2002)
• Line of Equal Value
-Best-Fit Line
8000
£
| 6000
Avg Monthly Rainfall (in)
-Avg Observed Flow (1/1/1993 to 12/31/2002)
-Avg Modeled Flow (Same Period)
8000
I
I
6000
4000
2000
2000 4000 6000
Average Observed Flow (cfs)
8000
2 3 4 5 6 7 8 9 10 11 12
Month
Figure 47. Seasonal regression and temporal aggregate at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - calibration period (SWAT).
D-69
-------
Average Monthly Rainfall (in)
-Median Observed Flow (1/1/1993 to 12/31/2002)
12000 -
I
I
• Observed (25th, 75th)
Modeled (Median, 25th, 75th)
11
12
Figure 48. Seasonal medians and ranges at USGS 02349605 Flint River at Ga 26, near Montezuma, GA
- calibration period (SWAT).
Table 23. Seasonal summary at USGS 02349605 Flint River at Ga 26, near Montezuma, GA
calibration period (SWAT)
MONTH
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
3955.13
5767.27
6976.42
3970.93
2277.06
1783.49
3617.54
1440.71
1260.28
1816.83
2297.06
3137.85
3665.00
3915.00
5150.00
3065.00
1810.00
1670.00
1290.00
1035.00
964.00
1105.00
1660.00
2240.00
2110.00
2572.50
2932.50
2210.00
1400.00
1180.00
977.25
752.00
770.50
775.75
1220.00
1490.00
5155.00
6887.50
8377.50
4297.50
2470.00
2170.00
1765.00
1420.00
1252.50
1855.00
2685.00
3497.50
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
4206.77
6093.10
7590.84
4303.82
2405.19
1937.85
3856.12
1715.12
1538.55
1874.05
2484.96
3082.68
3393.41
3996.73
5958.04
3329.58
2322.73
1659.04
1342.07
1139.52
957.90
1162.76
1619.56
1810.72
1700.42
2395.72
2700.58
2078.74
1358.75
909.57
644.18
673.31
710.14
673.08
854.25
1116.26
5504.55
7738.14
10476.33
5450.56
3121.85
2654.73
2287.50
1722.66
1670.33
2586.26
3456.35
361 1 .32
D-70
-------
•Observed Flow Duration (1/1/1993 to 12/31/2002 )
•Modeled Flow Duration (1/1/1993 to 12/31/2002 )
1000000
i
o
0
D)
ro
Q
100000
10000
1000 =
100
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 49. Flow exceedence at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - calibration
period (SWAT).
o
o
T3
I
0
0
_g
o
o
T3
0
N
•Observed Flow Volume (1/1/1993 to 12/31/2002 )
Modeled Flow Volume (1/1/1993 to 12/31/2002 )
120%
100%
80% -
60%
40%
20%
Jan-93
Jul-94
Jan-96
Jul-97
Jan-99
Jul-00
Jan-02
Figure 50. Flow accumulation at USGS 02349605 Flint River at Ga 26, near Montezuma, GA
calibration period (SWAT).
D-71
-------
Table 24. Summary statistics at USGS 02349605 Flint River at Ga 26, near Montezuma, GA
calibration period (SWAT)
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 38
10-Year Analysis F^riod: 1/1/1993 - 12/31/2002
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe CoefficjejTtjjfJEfficjejTC^JE^
Baseline adjusted coefficient (Garrick), E':
Monthly NSE 1
15.88
6.39
2.49
2.79
2.91
6.85
3.33
4.22
0.90
Error Statistics
7.28
-9.39
11.34
12.46
Observed Flow Gage
USGS 02349605 FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA
Hydrologic Unit Code: 3130006
Latitude: 32.29305556
Longitude: -84.04361 11
Drainage Area (sq-rri): 2920
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
2.56 » | 30
7.19
7.64
-12.03
-12.28
0.624
0.442
0.876 1
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
14.80
5.74
2.75
2.48
2.83
6.39
3.10
4.80
1.02
Clear [
Hydrology Validation
Consistent with HSPF modeling efforts, validation for the Upper Flint calibration focus area was performed at the
same location but for calendar years 1983-1992. Results are presented in Figures 51 through 57 and Tables 25 and
26. Although, the Nash-Sutcliffe modeling efficiency is not as good as it was for the calibration period, the model
performance was adequate for the validation period. None of the metrics fall out of the range set for the 20
Watershed study.
D-72
-------
Avg Monthly Rainfall (in)
-Avg Observed Flow (1/1/1983to 12/31/1992)
Avg Modeled Flow (Same Period)
70000
Jul-84
Jan-86
Jul-87 Jan-89
Date
Jul-90
Jan-92
c
'ro
ro
Q
Figure 51. Mean daily flow at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - validation
period (SWAT).
I
15000
10000
5000
J-83
J-84
Avg Monthly Rainfall (in)
-Avg Observed Flow (1/1/1983 to 12/31/1992 )
Avg Modeled Flow (Same Period)
J-92
Figure 52. Mean monthly flow at USGS 02349605 Flint River at Ga 26, near Montezuma, GA -
validation period (SWAT).
D-73
-------
Avg Flow (1/1/1983 to 12/31/1992)
• Line of Equal Value
Best-Fit Line
t
_o
U_
•a
_CD
CD
•a
o
CD
O)
15000
10000
5000
5000
10000
15000
100% i
f 90% ^
Avg Observed Flow (1/1/1983 to 12/31/1992 )
Avg Modeled Flow (1/1/1983 to 12/31/1992 )
- Line of Equal Value
Average Observed Flow (cfs)
J-83 J-84 J-86 J-87 J-89 J-90 J-92
Month
Figure 53. Monthly flow regression and temporal variation at USGS 02349605 Flint River at Ga 26,
near Montezuma, GA - validation period (SWAT).
8000
• Avg Flow (1 /1 /1983 to 12/31 /1992)
Line of Equal Value
Best-Fit Line
2000 4000 6000
Average Observed Flow (cfs)
8000
I
8000
6000
4000
2000
Avg Monthly Rainfall (in)
-Avg Observed Flow (1/1/1983 to 12/31/1992)
•Avg Modeled Flow (Same Period)
Illllllllll
2345678
Month
9 10 11 12
Figure 54. Seasonal regression and temporal aggregate at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - validation period (SWAT).
D-74
-------
Average Monthly Rainfall (in)
-Median Observed Flow (1/1/1983 to 12/31/1992)
8000
• Observed (25th, 75th)
Modeled (Median, 25th, 75th)
I
I
11
12
to
Figure 55. Seasonal medians and ranges at USGS 02349605 Flint River at Ga 26, near Montezuma, GA
- validation period (SWAT).
Table 25. Seasonal summary at USGS 02349605 Flint River at Ga 26, near Montezuma, GA
validation period (SWAT)
MONTH
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
4168.00
5242.97
5976.65
4328.17
2451 .47
1753.24
1908.70
1632.54
1283.93
1425.06
2275.14
3667.95
3115.00
4150.00
4310.00
3275.00
2135.00
1 300.00
1 330.00
1150.00
1120.00
1 020.00
1 330.00
2275.00
2152.50
2850.00
2830.00
2277.50
1420.00
986.75
927.50
816.50
881 .75
869.50
1120.00
1602.50
5350.00
6430.00
6542.50
4880.00
2937.50
2072.50
2257.50
1830.00
1500.00
1390.00
2277.50
4277.50
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
4143.78
5285.84
6060.01
4531.12
2658.75
2095.76
2160.97
1732.26
1534.60
1509.13
2284.09
3322.66
2886.66
4532.61
4515.09
3509.51
2418.76
1629.70
1141.19
1482.31
1359.25
945.55
1154.11
1848.91
1525.30
2618.46
2739.77
2517.97
1 824.58
1 060.96
757.51
655.52
812.62
654.00
608.47
833.94
5836.80
7460.70
7304.52
5265.76
3102.27
2490.67
2715.43
2563.83
2005.97
1823.54
2252.94
4660.44
D-75
-------
•Observed Flow Duration (1/1/1983 to 12/31/1992)
Modeled Flow Duration (1/1/1983 to 12/31/1992)
t
o
O)
CO
Q
100000
10000
1000
100
10%
20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90% 100%
Figure 56. Flow exceedence at USGS 02349605 Flint River at Ga 26, near Montezuma, GA - validation
period (SWAT).
o
o
CO
T3
.Q
O
o
_o
LJ_
T3
N
"CD
•Observed Flow Volume (1/1/1983 to 12/31/1992)
Modeled Flow Volume (1/1/1983 to 12/31/1992)
120%
100%
80% -
60% -
40% -
20%
Jan-83
Jul-84
Jan-86
Jul-87
Jan-89
Jul-90
Jan-92
Figure 57. Flow accumulation at USGS 02349605 Flint River at Ga 26, near Montezuma, GA -
validation period (SWAT)
D-76
-------
Table 26. Summary statistics at USGS 02349605 Flint River at Ga 26, near Montezuma, GA
validation period (SWAT)
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 37
10-Year Analysis F^riod: 1/1/1983 - 12/31/1992
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient _of_EfficiencyJ_Ej___
Baseline adjusted coefficient (Garrick), E':
Monthly NSE 1
14.42
5.01
2.57
2.12
2.78
5.93
3.58
3.67
0.46
Error Statistics
3.33
-7.57
2.89
12.43
Observed Flow Gage
USGS 02349605 FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA
Hydrologic Unit Code: 3130006
Latitude: 32.29305556
Longitude: -84.04361 1 1
Drainage Area (sq-rri): 2920
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Obsei-ved Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
-3.46 » | 30
0.66
8.82
-16.93
-10.04
0.559
0.444
0.833
30
30
20
50
Model accuracy increases
as E or E' approaches 1 .0
13.95
4.87
2.78
1.89
2.88
5.89
3.29
4.42
0.51
Clear [
Hydrology Results for Larger Watershed
Calibration and validation results at all gages are summarized in Tables 29 and 30. As discussed above, a detailed
spatial calibration was not conducted for the GCRP-SWAT model of the ACF basin. The parameterization is
identical across the entire watershed, although, measured flow and water quality data at other stations (other than
the calibration focus area) where land use dominance occurred were used to set the corresponding parameters. A
better model fit could perhaps be achieved if the model was more tightly calibrated but this was not attempted
deliberately keeping in view the intended bigger scope of the project.
In general, the model performance was good, as noticed from the Nash-Sutcliffe modeling efficiency statistics,
except for forest and urban dominated subwatersheds. The model over-predicted flow in winter months and
under-predicted flow in spring months. Also, the Ichawaynochaway Creek subwatershed (station 02353500)
simulation statistics are relatively lower than those at the other stations. The model over-predicted high flows and
under-predicted low flows. The simulated peak flows correlated well with the observed high rainfall events;
however, watershed response, as noticed from the measured streamflows, didn't result in such high flows.
D-77
-------
All the reservoir parameters were set such that inflow equals outflow and this approach worked well for all
reservoirs except Lake Lanier. Similar to the HSPF results, after the confluence of Chattahoochee and Flint rivers
and downstream of Lake Seminole, the simulation and the model fit greatly improved as shown in Figures 58
through 64 and Tables 27 and 28.
^H Avg Monthly Rainfall (in)
Avg Observed Flow (1/1/1993 to 12/31/2002)
Avg Modeled Flow (Same Period)
250000
200000
% 150000
£ 100000
50000
Jan-93
Jul-94
Jan-96
Jul-97
Jan-99
Jul-00
Jan-02
Date
Figure 58. Mean daily flow at USGS 02358000 Apalachicola River at Chattahoochee, FL- calibration
period (SWAT)
100000
80000
J-93
Avg Monthly Rainfall (in)
-Avg Observed Flow (1/1/1993 to 12/31/2002 )
•Avg Modeled Flow (Same Period)
J-94
J-02
f
ro
01
o
Figure 59. Mean monthly flow at USGS 02358000 Apalachicola River at Chattahoochee, FL-
calibration period (SWAT)
D-78
-------
• Avg Flow (1/1/1993 to 12/31/2002 )
• - - - • Line of Equal Value
Best-Fit Line
100000
O)
TO
20000
O
0)
o
ro
m
_
0)
to
100%
90% -
80%
70% -
60%
50%
Avg Observed Flow (1/1/1993 to 12/31/2002 )
I Avg Modeled Flow (1/1/1993 to 12/31/2002 )
-Line of Equal Value
40% -
30% -
20% -
10% -
0%
0 20000 40000 60000 80000 100000
Average Observed Flow (cfs)
J-93 J-94 J-96
J-97 J-99
Month
J-00 J-02
Figure 60. Monthly flow regression and temporal variation at USGS 02358000 Apalachicola River at
Chattahoochee, FL- calibration period (SWAT)
Avg Monthly Rainfall (in)
-Avg Observed Flow (1/1/1993 to 12/31/2002)
Avg Modeled Flow (Same Period)
50000
40000
Best-Fit Line
c;nnnn -,
&
-ii/innnn
O
LL
-n ^nnnn
"oi
T3
° onnnn
0
D)
2 1 nnnn
^j 1UUUU
n -
y = 0.8975X +
p2 = n 9fl
»
itf*
.,--
3900.1
R2
,x
X
f '
10000 20000 30000 40000 50000
Average Observed Flow (cfs)
10 11 12
Figure 61. Seasonal regression and temporal aggregate at USGS 02358000 Apalachicola River at
Chattahoochee, FL- calibration period (SWAT)
D-79
-------
Average Monthly Rainfall (in)
-Median Observed Flow (1/1/1993 to 12/31/2002)
60000
50000
40000
30000
20000 -
10000
• Observed (25th, 75th)
Modeled (Median, 25th, 75th)
11
12
c
'co
>,
o
Figure 62. Seasonal medians and ranges at USGS 02358000 Apalachicola River at Chattahoochee,
FL- calibration period (SWAT)
Table 27. Seasonal summary at USGS 02358000 Apalachicola River at Chattahoochee, FL-
calibration period (SWAT)
MONTH
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
24730.16
34453.44
41056.97
25249.20
15972.58
12939.23
18873.90
12411.87
10850.80
12458.10
13896.10
19332.77
19950.00
29050.00
36850.00
19800.00
15250.00
12600.00
11950.00
11050.00
8235.00
10750.00
13500.00
14550.00
13500.00
18275.00
19300.00
16000.00
9570.00
8237.50
7665.00
7102.50
6530.00
6112.50
6557.50
9150.00
30475.00
43900.00
53300.00
30250.00
19675.00
17600.00
15575.00
14300.00
12900.00
13900.00
18250.00
23800.00
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
26487.83
33462.26
41547.96
27986.54
18337.31
15481.95
21090.83
15037.00
14122.01
15677.74
16163.41
18812.57
20770.32
29041 .02
36338.79
24610.79
18704.41
16359.52
15028.16
12370.73
10885.75
11788.04
12783.91
13546.71
12429.88
15105.85
17536.38
15952.52
11760.67
9413.12
10404.58
9075.87
7701 .25
8509.07
9989.64
9380.46
33656.64
44319.91
56838.96
37760.21
22816.81
19775.33
18080.23
15658.52
17196.48
17775.64
20016.35
22589.03
D-80
-------
o
D)
ro
Q
•Observed Flow Duration (1/1/1993 to 12/31/2002 )
Modeled Flow Duration (1/1/1993 to 12/31/2002 )
1000000
100000
10000
1000
100
10
1
0.1
0% 10% 20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90%
100%
Figure 63. Flow exceedence at USGS 02358000 Apalachicola River at Chattahoochee, FL- calibration
period (SWAT)
o
o
ro
T3
I
.a
E
"o
^3
LL
T3
.N
(0
•Observed Flow Volume (1/1/1993 to 12/31/2002 )
Modeled Flow Volume (1/1/1993 to 12/31/2002 )
120%
100%
80%
60%
40%
20%
Jan-93
Jul-94
Jan-96
Jul-97
Jan-99
Jul-00
Jan-02
Figure 64. Flow accumulation at USGS 02358000 Apalachicola River at Chattahoochee, FL-
calibration period (SWAT)
D-81
-------
Table 28. Summary statistics at USGS 02358000 Apalachicola River at Chattahoochee, FL-
calibration period (SWAT)
REACH OUTFLOW FROM OUTLET 13
10-Year Analysis F^riod: 1/1/1993 - 12/31/2002
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Srjring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
_Nas^3ut£liffeJDpj5^
Baseline adjusted coefficient (Garrick), E':
Monthly NSE H
17.35
5.19
4.21
3.34
3.36
6.60
4.05
3.13
0.62
Error Statistics
9.16
15.49
1.57
19.16
Observed Flow Gage
USGS 02358000 APALACHICOL/
Hydrologic Unit Code: 3130011
Latitude: 30.701 0251
Longitude: -84.8590871
Drainage Area (sq-rri): 17200
i RIVER AT CHATTAHOOCHEE FLA
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
10.82 » | 30
1.39
14.12
-15.66
-3.56
0.793
0.543
0.919 \
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
15.89
5.11
3.65
2.80
3.03
6.51
3.55
3.71
0.64
Clear [
D-82
-------
Table 29. Summary statistics (percent error) for all stations - calibration period 1993-2002
(SWAT)
Station
Error in total
volume:
Error in 50%
lowest flows:
Error in 10%
highest flows:
Seasonal
volume error
- Summer:
Seasonal
volume error
- Fall:
Seasonal
volume error
- Winter:
Seasonal
volume error
- Spring:
Error in storm
volumes:
Error in
summer
storm
volumes:
Daily Nash-
Sutcliffe
Coefficient of
Efficiency, E:
Monthly
Nash-
Sutcliffe
Coefficient of
Efficiency, E:
02349605
7.28
-9.39
11.34
12.46
2.56
7.19
7.64
-12.03
-12.28
0.624
0.876
02331600
1.74
-44.46
28.59
9.11
22.61
16.05
-27.05
71.60
107.67
0.358
0.560
02336300
-10.17
-22.69
-27.08
26.93
-15.44
10.95
-24.11
-38.66
-49.29
0.334
0.336
02336000
17.52
34.07
-10.28
-4.30
25.99
13.79
37.03
-69.57
-77.02
0.489
0.631
02353500
5.89
-20.35
37.75
0.78
-2.69
25.04
-21.27
25.40
-10.26
0.335
0.532
02339500
6.26
33.88
-7.29
-8.34
0.51
5.34
25.48
-75.17
-82.70
0.542
0.837
02343801
16.53
62.17
-10.00
28.70
19.75
-2.93
41.14
-52.52
-58.88
0.642
0.837
02353000
10.45
-2.48
16.02
17.53
1.02
9.50
15.31
-48.51
-42.31
0.697
0.830
02358000
9.16
15.49
1.57
19.16
10.82
1.39
14.12
-15.66
-3.56
0.793
0.919
D-83
-------
Table 30. Summary statistics (percent error) for all stations - validation period 1983-1992
(SWAT)
Station
Error in total
volume:
Error in 50%
lowest flows:
Error in 10%
highest flows:
Seasonal
volume error
- Summer:
Seasonal
volume error
- Fall:
Seasonal
volume error
- Winter:
Seasonal
volume error
- Spring:
Error in storm
volumes:
Error in
summer
storm
volumes:
Daily Nash-
Sutcliffe
Coefficient of
Efficiency, E:
Monthly
Nash-
Sutcliffe
Coefficient of
Efficiency, E:
02349605
3.33
-7.57
2.89
12.43
-3.46
0.66
8.82
-16.93
-10.04
0.559
0.833
02331600
-4.33
-49.69
26.93
-13.83
9.43
11.99
-27.97
56.71
66.81
0.347
0.578
02336300
-12.32
-23.46
-29.96
-18.32
-12.56
-3.15
-18.51
-40.06
-43.98
0.420
0.587
02336000
5.62
18.44
-16.50
-15.29
-3.92
23.35
18.15
-73.11
-81.91
0.369
0.484
02353500
1.03
-38.73
35.73
14.60
-15.20
9.61
-11.43
26.32
14.15
0.222
0.543
02339500
0.92
36.21
-12.47
-11.16
-15.65
12.01
15.39
-77.02
-83.37
0.481
0.758
02343801
11.47
50.10
-10.38
25.32
1.66
1.04
27.43
-65.86
-76.19
0.599
0.829
02353000
6.48
-7.51
5.33
21.21
-0.29
3.04
8.68
-52.34
-52.71
0.685
0.785
02358000
2.17
-7.83
-2.31
9.39
-3.96
1.04
3.84
-8.64
-6.41
0.770
0.901
Water Quality Calibration and Validation
Initial calibration and validation of water quality was done on the Flint River near Montezuma (USGS02349605),
using calendar years 1999-2002 for calibration and calendar years 1991-1998 for validation. As with hydrology,
calibration was performed on the later period as this better reflects the land use included in the model. The start of
the validation period is constrained by data availability.
Calibration adjustments for sediment focused on the following parameters:
• PRF (Peak rate adjustment factor for sediment routing in the main channel)
• SPCON and SPEXP (Linear and Exponent parameters for estimating maximum amount of sediment that
can be reentrained during channel sediment routing)
• RSDCO (Residue decomposition coefficient)
• USLE-P (USLE equation support practice factor
D-84
-------
Simulated and estimated sediment loads at the Montezuma station for both periods are shown in Figures 65
through 68 and statistics for the two periods are provided separately in Tables 31 and 32. The key statistic in
Table 31 is the relative percent error, which shows the error in the prediction of monthly load normalized to the
estimated load. Table 31 also shows the relative average absolute error, which is the average of the relative
magnitude of errors in individual monthly load predictions. This number is inflated by outlier months in which the
simulated and estimated loads differ by large amounts (that may be as easily due to uncertainty in the estimated
load due to limited data as to problems with the model) and the third statistic, the relative median absolute error,
is likely more relevant and shows good agreement.
1,000,000
TSS
-Regression Loads
-Simulated Loads
Figure 65. Fit for monthly load of TSS at USGS 02349605 Flint River at Ga 26, near Montezuma, GA -
calibration period (SWAT).
Table 31. Model fit statistics (observed minus predicted) for monthly TSS loads using
stratified regression at USGS 02349605 Flint River at Ga 26, near Montezuma, GA
(SWAT)
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1999-2002)
-9%
66%
39.1%
Validation period
(1991-1998)
17%
42%
25.8%
D-85
-------
FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA 1999-2002
o
•o
ra
o
CO
V)
10
100 1000
Flow, cfs
10000
100000
Simulated A Observed
Fbw er (Simulated)
Power (Observed)
Figure 66. Power plot for observed and simulated TSS at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - calibration period (SWAT).
FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA 1986-1998
^
"5>
•o
ra
o
CO
V)
10
100 1000
Flow, cfs
10000
100000
• Simulated A Observed
Row er (Simulated)
Power (Observed)
Figure 67. Power plot for observed and simulated TSS at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA - validation period (SWAT).
D-86
-------
A low baseflow recession factor used in the calibrated model setup allowed streamflow to reach very low values,
which in turn simulated numerous days with extremely low sediment values as seen in Figure 68.
FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA
• Simulated A Observed
1000
100
O)
eo
CO
1999
2000
2001
Year
2002
Figure 68. Time series plot of TSS concentration at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA (SWAT).
Table 32. Relative errors (observed minus predicted), TSS concentration at USGS 02349605
Flint River at Ga 26, near Montezuma, GA (SWAT)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1999-2002)
49
63.01%
64.14%
Validation period
(1991-1998)
24
60.61%
52.79%
Calibration adjustments for total phosphorus and total nitrogen focused on the following parameters:
• PPERCO (phosphorus percolation coefficient)
• NPERCO (nitrogen percolation coefficient)
• PHOSKD (phosphorus soil partitioning coefficient)
• HLIFE_NGW (half life of nitrate in the shallow aquifer)
• SOL_CBN1 (organic carbon in the first soil layer)
• QUAL2E parameters such as algal, organic nitrogen, and organic phosphorus settling rate in the reach,
benthic source arte for dissolved phosphorus and NH4-N in the reach, fraction of algal biomass that is
nitrogen and phosphorus, Michaelis-Menton half-saturation constant for nitrogen and phosphorus
D-87
-------
In general, the match between observed and measured total phosphorus and total nitrogen was acceptable. Total
phosphorus and total nitrogen calibration results are presented in Figures 69 through 76 and Tables 33 through 36.
Total P
1000
100 --
o
«
I
-Regression Loads
-Simulated Loads
10 ---
Figure 69. Fit for monthly load of total phosphorus at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA (SWAT).
Table 33. Model fit statistics (observed minus predicted) for monthly total phosphorus loads
using stratified regression at USGS 02349605 Flint River at Ga 26, near Montezuma,
GA (SWAT)
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1999-2002)
-50%
72%
31.8%
Validation period
(1991-1998)
-30%
49%
18.9%
D-88
-------
5
"5>
•o
ra
o
0.01
0.001
FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA 1999-2002
10
100 1000
Flow, cfs
10000 100000
Simulated A Observed Power (Simulated) ^^™Power (Observed)
Figure 70. Power plot for observed and simulated total phosphorus at USGS 02349605 Flint River at
Ga 26, near Montezuma, GA - calibration period (SWAT).
FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA 1986-1998
1000
0.01
10
100 1000
Flow, cfs
10000
100000
Simulated A Observed ^^™Power (Simulated) ^^™Power (Observed)
Figure 71. Power plot for observed and simulated total phosphorus at USGS 02349605 Flint River at
Ga 26, near Montezuma, GA - validation period (SWAT).
D-89
-------
FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA
1999
2000
2001
Year
2002
Figure 72. Time series plot of total phosphorus concentration at USGS 02349605 Flint River at Ga 26,
near Montezuma, GA (SWAT).
Table 34. Relative errors (observed minus predicted), total phosphorus concentration, USGS
02349605 Flint River at Ga 26, near Montezuma, GA (SWAT)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1999-2002)
48
-32.67%
-41.28%
Validation period
(1991-1998)
80
-8.3%
-11.99%
D-90
-------
Total N
-Averaging Loads
-Simulated Loads
O T- CN
999
c c c
03 03 03
Figure 73. Fit for monthly load of total nitrogen at USGS 02349605 Flint River at Ga 26, near
Montezuma, GA (SWAT).
Table 35. Model fit statistics (observed minus predicted) for monthly total nitrogen loads
using averaging estimator at USGS 02349605 Flint River at Ga 26, near Montezuma,
GA (SWAT)
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1999-2002)
-18%
31%
15.7%
Validation period
(1991-1998)
9%
30%
19.9%
D-91
-------
FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA 1999-2002
100
10
100 1000
Flow, cfs
10000
100000
• Simulated A Observed ^^~Power (Simulated) ^^"Fbwer (Observed)
Figure 74. Power plot for observed and simulated total nitrogen at USGS 02349605 Flint River at Ga
26, near Montezuma, GA - calibration period (SWAT).
FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA 1986-1998
1000
100
5
"5>
•o
ra
o
10
100 1000
Flow, cfs
10000
100000
Simulated A Observed ^^™Power (Simulated) ^^"Fbwer (Observed)
Figure 75. Power plot for observed and simulated total nitrogen at USGS 02349605 Flint River at Ga
26, near Montezuma, GA - validation period (SWAT).
D-92
-------
1999
FLINT RIVER AT GA 26, NEAR MONTEZUMA, GA
2000
2001
Year
2002
Figure 76. Time series plot of total nitrogen concentration at USGS 02349605 Flint River at Ga 26,
near Montezuma, GA (SWAT).
Table 36. Relative errors (observed minus predicted), total nitrogen concentration, USGS
02349605 Flint River at Ga 26, near Montezuma, GA (SWAT)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1999-2002)
48
-14.52%
-10.44%
Validation period
(1991-1998)
76
2.26%
8.07%
Water Quality Results for Larger Watershed
As with hydrology, the Flint River watershed parameters for water quality were directly transferred to other
portions of the watershed. In general, simulated sediment was low and simulated total phosphorus and total
nitrogen was high at most of the stations. Ortho phosphorus and mineral nitrogen made up most of the total
phosphorus and total nitrogen as organic components corresponded to the sediment fraction. Summary statistics
for the water quality calibration and validation at other stations in the watershed are provided in Tables 37 and 38,
respectively.
D-93
-------
Table 37. Summary statistics (observed minus predicted) for water quality for all stations -
calibration period 1999-2002 (SWAT)
Station
Relative
Percent Error
TSS Load
TSS
Concentration
Median
Percent Error
Relative
Percent Error
TP Load
TP
Concentration
Median
Percent Error
Relative
Percent Error
TN Load
TN
Concentration
Median
Percent Error
02349605
-9
64.1
-50
-41.3
-18
-10.4
02331600
37
-26.0
25
0.0
13
54.3
02336300
91
-20.4
40
-21.6
-59
15.8
02336000
-3
3.2
-29
-56.4
-117
-46.1
02353500
33
41.9
-305
-124.5
0
63.9
02339500
71
16.9
-317
-1136.7
-606
-461.8
02343801
83
32.6
-85
-814.9
-310
-335.7
02353000
66
33.7
-45
-24.8
26
25.8
D-94
-------
Table 38. Summary statistics (observed minus predicted) for water quality for all stations -
validation period 1986-1998 (SWAT)
Station
Relative
Percent
Error TSS
Load
TSS
Concentrati
on Median
Percent
Error
Relative
Percent
Error TP
Load
TP
Concentrati
on Median
Percent
Error
Relative
Percent
Error TN
Load
TN
Concentrati
on Median
Percent
Error
02349605
17
53.8
-30
-12.0
9
8.1
02331600
91
0.9
62
46.6
44
59.7
02336300
93
1.7
18
11.2
-75
37.4
02336000
33
14.4
14
2.8
-52
-50.3
02353500
42
89.9
-177
-15.1
14
70.6
02339500
87
54.9
-59
-134.0
-280
-467.1
02343801
95
31.1
72
-291.1
-144
-229.4
02353000
65
33.8
-10
-12.1
37
20.2
D-95
-------
References
Alabama River Alliance. 2007. Water Wars Background: http://www.alabamarivers.org/current_work/water-wars
(Accessed January 2, 2010).
Cherry, R.N. 1961. Chemical Quality of Water of Georgia Stream. 1957-58, U.S. Geological Survey Bulletin No.
69.
Couch, C.A. 1993. Environmanetal Seetting of the Apalachicola-Chattahoochee_Flint River Basin. Proceedings
of the 1993 Georgia Water Resources Conference. April 20-21, 1993, University of Georgia.
Hook, J.E., 2009. Agricultural Irrigation Water Demand (1 June 2009):
http: //www .ne spal.org/SIRP/waterinfo/State/common/AgWaterDemand .htm.
Preston, S.D., V.J. Bierman, Jr., and S.E. Silliman. 1989. An evaluation of methods for the estimation of tributary
mass loads. Water Resources Research. 25(6): 1379-1389.
USAGE (Unites States Army Corps of Engineers). 2010. Apahalachicola-Chattahoochee-Flint River (1/23/2009):
http://water.sam.usace.army.mil/acfframe.htm (Accessed August 10, 2010).
U. S. Army Corps of Engineers (USAGE). 1982. National inventory of dams database in card format, available
from National Technical Information Service, Springfield, VA 22162, #ADA 118670.
USEPA (United States Environmental Protection Agency). 2008. Using the BASINS Meteorological Database
(Version 2006). BASINS Technical Note 10. Office of Water, U.S. Environmental Protection Agency.
Washington, DC. http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf
(Accessed June, 2009).
USEPA (United States Environmental Protection Agency). 2007. Creating Hydraulic Function Tables
(FTABLES) for Reservoirs in BASINS. BASINS Technical Note 1. Office of Water, U.S. Environmental
Protection Agency. Washington, DC.
http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel.pdf (Accessed June,
2009).
USEPA (United States Environmental Protection Agency). 2006. Sediment Parameter and Calibration Guidance
for HSPF. BASINS Technical Note 8. Office of Water, U.S. Environmental Protection Agency. Washington, DC.
http://water.epa.gov/scitech/datait/models/basins/upload/2006_02_02_BASINS_tecnote8.pdf (Accessed June,
2009).
USEPA (United States Environmental Protection Agency). 2000. Estimating Hydrology and Hydraulic
Parameters for HSPF. BASINS Technical Note 6. EPA-823-ROO-012. Office of Water, U.S. Environmental
Protection Agency. Washington, DC.
http://water.epa.gov/scitech/datait/models/basins/upload/2000_08_14_BASINS_tecnote6.pdf (Accessed June,
2009).
USGS (United States Geological Survey). 2008. Description of the ACF River Basin Study Area (July 17, 2008):
http://ga.water.usgs.gov/nawqa/basinall.html, Jan. 2, 2010.
D-96
-------
Appendix E
Model Configuration, Calibration and
Validation
Basin: Arizona: Salt, Verde, and San
Pedro Rivers (Ariz)
E-l
-------
Contents
Watershed Background E-8
Water Body Characteristics E-8
Soil Characteristics E-13
Land Use Representation E-13
Topography E-18
Point Sources E-20
Meteorological Data E-22
Watershed Segmentation E-26
Calibration Data and Locations E-29
Other Relevant Features E-29
HSPF Modeling E-31
Changes Made to Base Data Provided E-31
Assumptions E-31
Hydrology Calibration E-31
Hydrology Validation E-37
Hydrology Results for Larger Watershed E-42
Water Quality Calibration and Validation E-49
Water Quality Results for Larger Watershed E-59
SWAT Modeling E-61
Changes Made to Base Data Provided E-61
Assumptions E-61
Hydrology Calibration E-62
Hydrology Validation E-67
Hydrology Results for Larger Watershed E-72
Water Quality Calibration and Validation E-79
Water Quality Results for Larger Watershed E-89
References E-91
E-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes E-16
Table 2. Land use distribution for the Arizona basins (2001 NLCD) (mi2) E-17
Table 3. Major point source discharges in the Arizona basins E-20
Table 4. Precipitation stations for the Arizona models E-22
Table 5. Calibration and validation locations in the Arizona basins E-29
Table 6. Seasonal summary at USGS 09504000 Verde River near Clarkdale, AZ - calibration period
(HSPF) E-35
Table 7. Summary statistics at USGS 09504000 Verde River near Clarkdale, AZ - calibration period
(HSPF) E-37
Table 8. Seasonal summary at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(HSPF) E-40
Table 9. Summary statistics at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(HSPF) E-42
Table 10. Summary statistics (percent error) for all stations - calibration period WY 1992-2002 (HSPF) ...E-43
Table 11. Seasonal summary at USGS 09498500 Salt River near Roosevelt, AZ - calibration period
(HSPF) E-46
Table 12. Summary statistics at USGS 09498500 Salt River near Roosevelt, AZ - calibration period
(HSPF) E-48
Table 13. Summary statistics for all stations - validation period WY 1982-1992 (HSPF) E-49
Table 14. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 095 04000 Verde River near Clarkdale, AZ (HSPF) E-51
Table 15. Relative errors (observed minus predicted) for TSS concentration at USGS 09504000 Verde
River near Clarkdale, AZ (HSPF) E-53
Table 16. Model fit statistics (observed minus predicted) for monthly total phosphorus loads using
stratified regression E-54
Table 17. Relative errors (observed minus predicted) for total phosphorus concentration at USGS
09504000 Verde River near Clarkdale, AZ (HSPF) E-56
Table 18. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator (HSPF) E-57
Table 19. Relative errors (observed minus predicted) for total nitrogen concentration at USGS
09504000 Verde River near Clarkdale, AZ (HSPF) E-59
Table 20. Summary statistics (observed minus predicted) for water quality for all stations - calibration
period 1993-2002 (HSPF) E-60
Table 21. Summary statistics (observed minus predicted) for water quality for all stations - validation
period 1986-1992 (HSPF) E-60
Table 22. Seasonal summary at USGS 09504000 Verde River near Clarkdale, AZ - calibration period
(SWAT) E-65
Table 23. Summary statistics at USGS 09504000 Verde River near Clarkdale, AZ - calibration period
(SWAT) E-66
Table 24. Seasonal summary at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(SWAT) E-70
Table 25. Summary statistics at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(SWAT) E-72
Table 26. Summary statistics (percent error) at all stations - calibration period WY 1992-2002 (SWAT) ...E-73
Table 27. Seasonal summary at USGS 09498500 Salt River near Roosevelt, AZ - calibration period
(SWAT) E-76
Table 28. Summary statistics at USGS 09498500 Salt River near Roosevelt, AZ - calibration period
(SWAT) E-78
Table 29. Summary statistics at all stations - validation period WY 1982-1992 (SWAT) E-79
Table 30. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 09504000 Verde River near Clarkdale, AZ (SWAT) E-81
E-3
-------
Table 31. Relative errors (observed minus predicted) for TSS concentration at USGS 09504000 Verde
River near Clarkdale, AZ (SWAT) E-83
Table 32. Model fit statistics (observed minus predicted) for monthly total phosphorus loads using
tratified regression (SWAT) E-84
Table 33. Relative errors (observed minus predicted) for total phosphorus concentration at USGS
09504000 Verde River near Clarkdale, AZ (SWAT) E-86
Table 34. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator (SWAT) E-87
Table 35. Relative errors (observed minus predicted) for total nitrogen concentration at USGS
09504000 Verde River near Clarkdale, AZ (SWAT) E-89
Table 36. Summary statistics (observed minus predicted) for water quality at all stations - calibration
period 1993-2002 (SWAT) E-90
Table 37. Summary statistics (observed minus predicted) for water quality at all stations - validation
period 1986-1992 (SWAT) E-90
Figures
Figure 1. The Arizona basins -Verde and Salt River sections E-ll
Figure 2. The Arizona basins - San Pedro River section E-12
Figure 3. Land use in the Arizona basin - Verde and Salt River section E-14
Figure 4. Land use in the Arizona basin - San Pedro River section E-15
Figure 5. Topography of the Arizona Basins E-19
Figure 6. Major point sources in the Arizona basins E-21
Figure 7. Weather stations for the Arizona basins model -Verde and Salt River section E-24
Figure 8. Weather stations for the Arizona basins model - San Pedro River section E-25
Figure 9. Model segmentation USGS stations utilized for the Arizona basins - Verde and Salt River
section E-27
Figure 10. Model segmentation USGS stations utilized for the Arizona basins - San Pedro River section E-28
Figure 11. Mean daily flow at USGS 09504000 Verde River near Clarkdale, AZ - calibration period
(HSPF) E-32
Figure 12. Mean monthly flow at USGS 09504000 Verde River near Clarkdale, AZ - calibration period
(HSPF) E-33
Figure 13. Monthly flow regression and temporal variation at USGS 09504000 Verde River near Clarkdale,
AZ - calibration period (HSPF) E-33
Figure 14. Seasonal regression and temporal aggregate at USGS 09504000 Verde River near Clarkdale,
AZ - calibration period (HSPF) E-34
Figure 15. Seasonal medians and ranges at USGS 09504000 Verde River near Clarkdale, AZ - calibration
period (HSPF) E-34
Figure 16. Flow exceedance at USGS 09504000 Verde River near Clarkdale, AZ - caalibration period
(HSPF) E-35
Figure 17. Flow accumulation at USGS 09504000 Verde River near Clarkdale, AZ - calibration period
(HSPF) E-36
Figure 18. Mean daily flow at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(HSPF) E-38
Figure 19. Mean monthly flow at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(HSPF) E-38
Figure 20. Monthly flow regression and temporal variation at USGS 09504000 Verde River near Clarkdale,
AZ - validation period (HSPF) E-39
Figure 21. Seasonal regression and temporal aggregate at USGS 09504000 Verde River near Clarkdale,
AZ - validation period (HSPF) E-39
E-4
-------
Figure 22. Seasonal medians and ranges at USGS 09504000 Verde River near Clarkdale, AZ - validation
period (HSPF) E-40
Figure 23. Flow exceedance at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(HSPF) E-41
Figure 24. Flow accumulation at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(HSPF) E-41
Figure 25. Mean daily flow at USGS 09498500 Salt River near Roosevelt, AZ - calibration period (HSPF)..E-44
Figure 26. Mean monthly flow at USGS 09498500 Salt River near Roosevelt, AZ - calibration period
(HSPF) E-44
Figure 27. Monthly flow regression and temporal variation at USGS 09498500 Salt River near Roosevelt,
AZ - calibration period (HSPF) E-45
Figure 28. Seasonal regression and temporal aggregate at USGS 09498500 Salt River near Roosevelt, AZ -
calibration period (HSPF) E-45
Figure 29. Seasonal medians and ranges at USGS 09498500 Salt River near Roosevelt, AZ - calibration
period (HSPF) E-46
Figure 30. Flow duration at USGS 09498500 Salt River near Roosevelt, AZ - calibration period (HSPF) E-47
Figure 31. Flow accumulation at USGS 09498500 Salt River near Roosevelt, AZ - calibration period
(HSPF) E-47
Figure 32. Fit for monthly load of TSS at USGS 09504000 Verde River near Clarkdale, AZ (HSPF) E-50
Figure 33. Power plot for observed and simulated TSS at USGS 09504000 Verde River near Clarkdale, AZ -
calibration period (HSPF) E-51
Figure 34. Power plot for observed and simulated TSS at USGS 09504000 Verde River near Clarkdale, AZ -
validation period (HSPF) E-52
Figure 35. Time series plot of TSS concentration at USGS 09504000 Verde River near Clarkdale, AZ
(HSPF) E-53
Figure 36. Fit for monthly load of total phosphorus at USGS 09504000 Verde River near Clarkdale, AZ
(HSPF) E-54
Figure 37. Power plot for observed and simulated total phosphorus at USGS 09504000 Verde River near
Clarkdale, AZ - calibration period (HSPF) E-55
Figure 38. Power plot for observed and simulated total phosphorus at USGS 09504000 Verde River near
Clarkdale, AZ - validation period (HSPF) E-55
Figure 39. Time series plot of total phosphorus concentration at USGS 09504000 Verde River near
Clarkdale, AZ (HSPF) E-56
Figure 40. Fit for monthly load of total nitrogen at USGS 09504000 Verde River near Clarkdale, AZ
(HSPF) E-57
Figure 41. Power plot for observed and simulated total nitrogen at USGS 09504000 Verde River near
Clarkdale, AZ - calibration period (HSPF) E-58
Figure 42. Power plot for observed and simulated total nitrogen at USGS 09504000 Verde River near
Clarkdale, AZ - validation period (HSPF) E-58
Figure 43. Time series plot of total nitrogen concentration at USGS 09504000 Verde River near Clarkdale,
AZ(HSPF) E-59
Figure 44. Mean daily flow at USGS 09504000 Verde River near Clarkdale, AZ - calibration period
(SWAT) E-62
Figure 45. Mean monthly flow at USGS 09504000 Verde River near Clarkdale, AZ - calibration period
(SWAT) E-63
Figure 46. Monthly flow regression and temporal variation at USGS 09504000 Verde River near Clarkdale,
AZ - calibration period (SWAT) E-63
Figure 47. Seasonal regression and temporal aggregate at USGS 09504000 Verde River near Clarkdale,
AZ - calibration period (SWAT) E-64
Figure 48. Seasonal medians and ranges at Verde River near Clarkdale, AZ - calibration period (SWAT) E-64
Figure 49. Flow exceedance at USGS 09504000 Verde River near Clarkdale, AZ - calibration period
(SWAT) E-65
E-5
-------
Figure 50. Flow accumulation at USGS 09504000 Verde River near Clarkdale, AZ - calibration period
(SWAT) E-66
Figure 51. Mean daily flow at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(SWAT) E-68
Figure 52. Mean monthly flow at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(SWAT) E-68
Figure 53. Monthly flow regression and temporal variation at USGS 09504000 Verde River near Clarkdale,
AZ - validation period (SWAT) E-69
Figure 54. Seasonal regression and temporal aggregate at USGS 09504000 Verde River near Clarkdale, AZ -
validation period (SWAT) E-69
Figure 55. Seasonal medians and ranges at USGS 09504000 Verde River near Clarkdale, AZ - validation
period (SWAT) E-70
Figure 56. Flow exceedance at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(SWAT) E-71
Figure 57. Flow accumulation at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(SWAT) E-71
Figure 58. Mean daily flow at USGS 09498500 Salt River near Roosevelt, AZ - calibration period
(SWAT) E-74
Figure 59. Mean monthly flow at USGS 09498500 Salt River near Roosevelt, AZ - calibration period
(SWAT) E-74
Figure 60. Monthly flow regression and temporal variation at USGS 09498500 Salt River near Roosevelt,
AZ - calibration period (SWAT) E-75
Figure 61. Seasonal regression and temporal aggregate at USGS 09498500 Salt River near Roosevelt, AZ -
calibration period (SWAT) E-75
Figure 62. Seasonal medians and ranges at USGS 09498500 Salt River near Roosevelt, AZ - calibration
period (SWAT) E-76
Figure 63. Flow exceedence at USGS 09498500 Salt River near Roosevelt, AZ - calibration period
(SWAT) E-77
Figure 64. Flow accumulation at USGS 09498500 Salt River near Roosevelt, AZ - calibration period
(SWAT) E-77
Figure 65. Fit for monthly load of TSS at USGS 09504000 Verde River near Clarkdale, AZ (SWAT) E-81
Figure 66. Power plot for observed and simulated TSS at USGS 09504000 Verde River near Clarkdale,
AZ - calibration period (SWAT) E-82
Figure 67. Power plot for observed and simulated TSS at USGS 09504000 Verde River near Clarkdale,
Z - validation period (SWAT) E-82
Figure 68. Time series plot of TSS concentration at USGS 09504000 Verde River near Clarkdale, AZ
(SWAT) E-83
Figure 69. Fit for monthly load of total phosphorus at USGS 09504000 Verde River near Clarkdale, AZ
(SWAT) E-84
Figure 70. Power Plot for Observed and Simulated total phosphorus at USGS 09504000 Verde River near
Clarkdale, AZ - calibration period (SWAT) E-85
Figure 71. Power plot for observed and simulated total phosphorus at USGS 09504000 Verde River near
Clarkdale, AZ - validation period (SWAT) E-85
Figure 72. Time series plot of total phosphorus concentration at USGS 09504000 Verde River near
Clarkdale, AZ (SWAT) E-86
Figure 73. Fit for monthly load of total nitrogen at USGS 09504000 Verde River near Clarkdale, AZ
(SWAT) E-87
Figure 74. Power plot for observed and simulated total nitrogen at USGS 09504000 Verde River near
Clarkdale, AZ - calibration period (SWAT) E-88
Figure 75. Power plot for observed and simulated total nitrogen at USGS 09504000 Verde River near
Clarkdale, AZ - validation period (SWAT) E-88
E-6
-------
Figure 76 Time series plot of total nitrogen concentration at USGS 09504000 Verde River near Clarkdale,
AZ(SWAT) E-89
E-7
-------
The Arizona (Salt, Verde, and San Pedro) basins are located in central and southern Arizona in EPA Region 9
(Cordy et al. 2000). The watershed includes large parts of two hydrologic provinces—the Central Highlands in
the north and the Basin and Range Lowlands in the south. Five major river systems drain the area: the Gila, Salt,
Verde, Santa Cruz and San Pedro Rivers. The selected model area includes perennial portions of the Salt and
Verde River basins (in HUC 1506) that lie upstream of major impoundments, along with the San Pedro River
(HUC 1505), for a total of 10 HUCSs with an area of 14,910 mi2 (Figure 1 and Figure 2).
Land cover is primarily desert scrub and rangeland at low elevations with sparse forest at higher elevations
(USGS, 2004; Cordy et al., 2000). The two major population centers of Arizona, Phoenix and Tucson, are located
just downstream of the model area, while portions of Flagstaff, Prescott, and several smaller towns are within the
Verde River watershed. Population growth is resulting in increasing demands on the limited water resources of
the area. The climate is arid to semiarid and is characterized by variability from place to place as well as large
differences in precipitation from one year to the next. Precipitation can be three times greater in wet years than in
dry years.
The Verde and Salt River watersheds are in the Central Highlands hydrologic province, characterized by
mountainous terrain with shallow, narrow intermountain basins. Forests and rangeland cover most of the area.
The largest town in the province is Prescott and other small rural towns dot the region. Most of the perennial
streams in the study area are in the Central Highlands. These streams derive their flow from precipitation in the
mountains and from rainfall and snowmelt along the northeastern border of the basins. Many of the major streams
with headwaters in the Central Highlands are perennial in their upper reaches but are captured for water supply for
metropolitan Phoenix, power generation, and flood control before they reach the Basin and Range Lowlands.
The San Pedro watershed is in the Basin and Range Lowlands hydrologic province. The Basin and Range
Lowlands are characterized by ephemeral streams, the largest water demands, and reliance on groundwater. Deep,
broad alluvial basins separated by mountain ranges of small areal extent characterize this hydrologic province.
There is very little natural streamflow because of an average annual rainfall of less than 10 to 15 inches except at
the highest elevations. With the exception of some small, higher elevation streams and sections of the San Pedro
River, most perennial streams in the Basin and Range Lowlands are dependent on treated wastewater effluent for
their year-round flow. Water use in the Basin and Range Lowlands represents 96 percent of all water use in the
Arizona basins. Agriculture is the largest water user. Because of the general lack of surface water resources in the
Basin and Range Lowlands, groundwater is relied upon heavily to meet agricultural and municipal demands.
The lower portions of the rivers in the Arizona basins have been extensively engineered for water supply purposes
(e.g., the Salt River Project) and also contain many reaches that flow only intermittently. Larger reservoirs are
problematic for scenario simulations as future demands and reservoir management are not fully known, while
intermittent streams are difficult to calibrate and can present problems for model performance. Therefore, the
portions of the watershed chosen for simulation are upstream of major reservoirs and focus on perennial streams.
The resulting three distinct study areas are the Verde, Salt, and San Pedro rivers (Figure 1 and Figure 2).
Water Body Characteristics
Verde River
The first area of study is the Verde River watershed upstream of Horseshoe Reservoir (Figure 1). The Verde River
watershed comprises approximately 6,577 square miles (mi2), while the area upstream of the northern end of
Horseshoe Reservoir contains approximately 5,563 mi2. The watershed trends south-southeast from Fraziers Well,
E-8
-------
immediately south of the Colorado River watershed and the Grand Canyon National Park to its confluence with
the Salt River on the east side of Phoenix. The study area ranges in elevation from over 11,000 feet (ft) where it
drains a portion of Humphreys Peak north of Flagstaff to around 2,100 ft at the confluence of the Verde River
with Horseshoe Reservoir.
The Verde Valley, which descends into the Central Highland province, is bounded by the Mogollon Rim to the
north and northeast and by the Black Hills to the southwest (Owen-Joyce and Bell 1983). The headwaters of the
Verde River are considered to be just below Sullivan Lake, an impoundment of Big Chino Wash. Upstream of
this point lies a large drainage area that is dominated by intermittent flow (HUC 15060201). Within the Upper
Verde watershed (HUC 15060202) from Sullivan Lake to Camp Verde, the Verde River flows through rugged
country and drains high mountains to the north and east. Perennial flow in the Verde River is usually considered
to start at the confluence with Granite Creek, just below Sullivan Lake. Granite Creek and its two tributaries
originate in the mountainous area outside of Prescott. All three of these tributaries are dammed to provide water to
the city of Prescott and the Chino Valley Irrigation District. Flow in Granite Creek is ephemeral at the point of
confluence with the Verde River; however, about 25 percent of the baseflow in the Verde River at this point is
believed to derive from groundwater transport out of the Granite Creek drainage (ADWR 2000).
The baseflow in the upper reaches of the Verde River is supported by groundwater discharges between Granite
Creek and Paulden (Owen-Joyce and Bell 1983). From Paulden to Sycamore Creek the river gains additional
groundwater discharges, primarily at Mormon Pocket. Sycamore Creek is an important tributary of the Verde
River, draining the area west of Flagstaff, and has a spring-fed baseflow. The net result of these groundwater
sources is a nearly constant baseflow of around 75 to 80 cfs at Clarkdale.
Groundwater throughout the Big Chino subwatershed occurs under both confined and unconfmed conditions.
Groundwater levels range from above surface due to confined conditions to over 200 feet below surface, with a
depth to water in most wells of less than 80 ft (Schwab 1995). The major source of recharge for the Big Chino
subwatershed is infiltration of runoff from the mountain fronts and flow within the major washes. Only a small
percentage of the annual precipitation in the subwatershed reaches the groundwater table because the majority
occurs in high intensity summer storm events and is lost as surface runoff, evaporation and transpiration by
vegetation (Schwab 1995).
ADWR (2000) examined water budgets for 1996-97 for the Big Chino subwatershed plus the uppermost part of
the Verde River to the USGS gage at Paulden and concluded that there was no net change in the groundwater
storage. Inflows were estimated to be 26,760 acre-feet from natural recharge plus 8,010 acre-feet from incidental
anthropogenic recharge. Of the total discharges, 19,050 acre-feet (55 percent) occurred as flow in the Verde River
near Paulden and the remainder as groundwater pumpage.
An additional important factor in the hydrology of the Verde River watershed, particularly upstream of Paulden,
is the construction of numerous stock pond impoundments used to capture surface runoff to support cattle
ranching. These impoundments may act as recharge basins, but impede the flow of runoff that would otherwise
have occurred. A survey of small impoundments upstream of Camp Verde was conducted in 1996. Approximately
2,635 impoundments ranging in size from 0.1 acres to approximately 350 acres in surface area were identified
(ADWR 2000). No estimate of recharge has been calculated for these impoundments and no determination of the
impact from restricting and/or impounding the natural runoff has ever been studied.
Salt River
The Salt River (including Tonto Creek) lies immediately to the east of the Verde River watershed and shares
many similar characteristics. The model simulates these streams down to Roosevelt Reservoir (Figure 1). Like the
Verde River watershed, the Salt River and Tonto Creek watersheds have high relief and are bounded by the
Mogollon Rim. However, unlike the Verde watershed, these watersheds have less in the way of teleconnections to
deep groundwater. Perennial springs are important in the upper reaches of the Salt; however, most of the water
E-9
-------
discharged by these springs appears to derive from local sources. The Salt River watershed also has much less
human influence than the Verde, with only 1.5 percent of the land area in private ownership. The bulk of the
watershed is under tribal or US Forest Service ownership.
San Pedro River
The San Pedro River, a tributary of the Gila River, flows northward from the Arizona-Mexico border (Figure 2).
The watershed consists of a large alluvial valley flanked by mountain ranges. The river is perennial in the
southern (upstream) reaches, but only intermittent in the northern (downstream) reaches. As with the Verde River
and Salt River watersheds, precipitation and temperature vary strongly with elevation, with most of the
precipitation occurring at the higher elevations. The perennial portions of the river support important desert
riparian forest habitat, and most of this section is contained within the San Pedro Riparian National Conservation
Area.
E-10
-------
Hydrography
Water (Nat. Atlas Dataset)
US Census Populated Places
Municipalities (pop 2 50,000)
County Boundaries
^ Watershed witfi HUCSs
Salt and Verde
River Basins
Big
Chino-William
Valley
(15060201
Upper Verde
(15060202)
Lower Verd
Tonto
(15060105)
Carrizo
(15060104)
Upper Salt
(15060103)
White
(15060102)
sevelt
ervoir
V
ack
(15060101)
GCRP Model Areas - Salt and Verde River Basins
Base Map
Figure 1. The Arizona basins - Verde and Salt River sections.
E-ll
-------
Lower San Pedro
(15050203)
Upper San Pedro
(15050202)
Hydrography
Interstate
^B Water (Nat. Atlas Dataseti
US Census Populated Places
^H Municipalities (pop 2 50.000)
County Boundaries
1 I Watershed with HUCBs
GCRP Model Areas - San Pedro River Basin
Base Map
NAD 1983 Albers_metets
20
Kilometers
Figure 2. The Arizona basins - San Pedro River section.
E-12
-------
Soil Characteristics
The hydrology of the Arizona Basin is strongly influenced by the soils and underlying geology of the watershed.
These in turn reflect the complex geologic history of Arizona, which includes periods of marine inundation,
volcanism, and uplift.
One of the most important characteristics of soils for watershed modeling is their hydrologic soil group (HSG).
The 20 Watershed study utilized STATSGO soil survey HSG information during model set-up. Soils are
classified into four hydrologic groups (SCS 1986), separated by runoff potential, as follows:
A Low runoff potential and high infiltration rates even when thoroughly wetted. Chiefly deep, well
to excessively drained sands or gravels. High rate of water transmission (> 0.75 cm/hr).
B Moderate infiltration rates when thoroughly wetted. Chiefly moderately deep to deep, moderately
well to well drained soils with moderately fine to moderately coarse textures. Moderate rate of
water transmission (0.40—0.75 cm/hr).
C Low infiltration rates when thoroughly wetted. Chiefly soils with a layer that impedes downward
movement of water, or soils with moderately fine to fine texture. Low rate of water transmission
(0.15—0.40 cm/hr).
D High runoff potential. Very low infiltration rates when thoroughly wetted. Chiefly clay soils with
a high swelling potential, soils with a permanent high water table, soils with a claypan or clay
layer at or near the surface, or shallow soils over nearly impervious material. Very low rate of
water transmission (0—0.15 cm/hr).
The soils in the Verde River watershed are predominantly hydrologic group B soils while soils in the San Pedro
River watershed are predominantly hydrologic group C soils. The Salt River watershed contains almost equal
amounts of B, C, and D soils with a slight dominance of B soils.
Land Use Representation
Land use in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage (Figure 3 and
Figure 4) and is predominantly scrub/shrub chaparral blending into Sonoran paloverde at lower elevations and
pinyon-juniper evergreen forest at higher elevations. Only a few small municipalities are located in the study
watersheds and much of the land is in federal ownership.
E-13
-------
Hydrography
Interstate
Water (Nat. Atlas Dataset)
I I County Boundaries
Watershed
2001 NLCD Land Use
^B °Pen water
^ Developed, open space
| Developed, low intensity
| Developed, medium intensity
| Developed, high intensity
^ Barren land
| Deciduous forest
| Evergreen forest
^ Mixed forest
I | Scrub/shrub
] Grassland/herbaceous
Pasture/hay
^| Cultivated crops
^ Woody wetlands
n Emergent herbaceous wetlands
Verc/e
R/Ver
GCRP Model Areas - Salt and Verde River Basins
Land Use Map
Figure 3. Land use in the Arizona basin - Verde and Salt River section.
E-14
-------
Legend
Hydrography
^^= Interstate
^B Water (Nat. Atlas Dataset)
3 County Boundaries
n Watershed
2001 NLCD Land Use
^^| Open water
^ Developed, open space
| Developed, low intensity
| Developed, medium intensity
^H Developed, high intensity
^ Barren land
| Deciduous forest
| Evergreen forest
^ Mixed forest
I | Scrub/shrub
I | Grassland/herbaceous
^ Pasture/hay
| Cultivated crops
^ Woody wetlands
^ Emergent herbaceous wetlands
San Pedro
River
MEXICO
GCRP Model Areas - San Pedro River Basin
Land Use Map
NAD_1983_Albers_meters
Map produced 2-8-2010- B. Tucker
40
• Miles
TETRATECH
Figure 4. Land use in the Arizona basin - San Pedro River section.
E-15
-------
NLCD land cover classes were aggregated according to the scheme shown in Table 1 then overlain with the soils
HSG grid. Minor land uses with less than 5 percent coverage within a subwatershed were reassigned to more
dominant classes. Pervious and impervious lands are specified separately for HSPF, so only one developed
pervious class is used, along with an impervious class. HSPF simulates impervious land areas separately from
pervious land. Impervious area distributions were also determined from the NLCD Urban Impervious data
coverage. Specifically, percent impervious area was calculated over the whole basin for each of the four
developed land use classes. These percentages were then used to separate out impervious land. NLCD
impervious area data products are known to underestimate total imperviousness in rural areas. However, the
model properly requires connected impervious area, not total impervious area, and the NLCD tabulation is
assumed to provide a reasonable approximation of connected impervious area. Different developed land classes
are specified separately in SWAT. In HSPF the WATER, BARREN, DEVPERV, and WETLAND classes are not
subdivided by HSG; SWAT uses the built-in HRU overlay mechanism in the ArcSWAT interface.
Table 1. Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area
usually accounted for as
reach area
Deciduous
Evergreen
Mixed
Emergent & woody
wetlands
Aquatic bed wetlands (not
emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY or GRASS
AGRR
WETF, WETL,
WETN
WATR
HSPF (after processing)
WATER
BARREN, Assume HSG D
DEVPERV;
IMPERV
BARREN (D)
FOREST (A,B,C,D)
SHRUB (A,B,C,D)
GRASS (A,B,C,D), BARREN (D)
GRASS (A,B,C,D)
AGRI (A,B,C,D)
WETLAND, Assume HSG D
WATER
The distribution of land use in the watershed is summarized in Table 2. Note that the small areas in crop and hay
production along the Verde mainstem and elsewhere do not meet the 5 percent threshold requirement in SWAT
and are thus not explicitly included in the model; instead, the developed pervious land use implicitly includes
those areas in crop production for SWAT.
E-16
-------
Table 2. Land use distribution for the Arizona basins (2001 NLCD) (mi2)
HUC8
watershed
Upper San
Pedro
15050202
Lower San
Pedro
15050203
Black
15060101
White
15060102
Upper Salt
15060103
Carrizo
15060104
Tonto
15060105
Big Chino-
Williamson
Valley
15060201
Upper
Verde
15060202
Lower
Verde
15060203
Total
Open
Water
0.1
1.4
1.8
1.6
16.7
0.0
0.0
0.1
1.2
0.3
23.2
Developed9
Open
space
19.2
8.3
1.8
4.5
10.4
2.7
4.6
11.8
48.0
9.8
121.2
Low
Density
4.9
2.2
0.2
1.2
5.1
0.2
1.2
3.5
25.1
4.0
47.7
Medium
Density
1.2
0.3
0.0
0.2
1.4
0.0
0.1
0.2
5.3
0.4
9.0
High
Density
0.2
0.1
0.0
0.0
0.3
0.0
0.0
0.0
0.6
0.1
1.3
Barren
Land
0.4
6.6
0.6
0.5
14.0
1.4
0.6
3.8
15.3
1.0
44.3
Forest
64.2
179.1
1,026.3
537.2
950.1
586.4
443.7
615.8
1,256.8
582.0
6,241.7
Shru bland
1,129.2
1,791.7
217.9
89.9
1,251.9
118.5
494.9
1,514.8
1,143.2
601.3
8,353.2
Pasture/Hay
1.5
1.5
0.0
0.1
0.7
0.0
0.3
1.4
2.5
1.9
10.0
Cultivated
9.9
7.5
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
17.4
Wetland
4.5
12.5
2.0
3.2
4.7
0.4
1.9
0.9
6.1
4.6
40.7
Total
1,235.2
2,011.2
1,250.7
638.4
2,255.2
709.8
947.1
2,152.4
2,504.1
1,205.4
14,909.6
The percent imperviousness applied to each of the developed land uses is as follows: open space (7.37%), low density (29.66%), medium density (53.71%), and high
density (73.85%).
E-17
-------
The HSPF model is set up on a hydrologic response unit (HRU) basis. For HSPF, HRUs were formed from an
intersection of land use and hydrologic soil group, then further subdivided by precipitation gage and slope. SWAT
HRUs were formed from an intersection of land use and SSURGO major soils.
Topography
The Salt, Verde, and San Pedro River watersheds are characterized by high relief (Figure 5) and precipitation and
temperature vary greatly with elevation. The largest precipitation amounts and lowest temperatures occur at the
high elevations along the Mogollon Rim on the north and east sides of the Salt and Verde River watersheds.
E-18
-------
GCRP Model Areas - Verde and Salt Rivers
Elevation Map
o 25 so
100
• Kilometers
0 12.5 25
50
• Miles
TETRATECH
GCRP Model Areas - San Pedro Basin
Elevation Map
Legend
Interstate
^H Water (Nat. Atlas Dataset)
| Municipalities (pop > 50,000)
County Boundaries
I | Watershed
Elevation
Meters
High : 3848
Low: 585
Figure 5. Topography of the Arizona Basins.
E-19
-------
Only the two major dischargers with a design flow greater than 1 MOD are included in the simulation (Table 3
and Figure 6). These dischargers are Page Springs Fish Hatchery and the Final Creek Wastewater Treatment Plant
(WWTP) in Globe, Arizona. Because of the arid climate and low population in the study watersheds, much of the
wastewater that is generated is either used for irrigating golf courses or discharged to ephemeral washes that lack
a direct surface connection to the river system.
Table 3. Major point source discharges in the Arizona basins
NPDES ID
AZ0021245
AZ0020249
Name
Page Springs Fish Hatchery (AZ Game and
Fish Department)
Final Creek WWTP (City of Globe, AZ)
Design Flow (MGD)
20.35
1.20
Observed Flow (MGD)
(1991-2006 average)
21.92
12.54
The discharges from Page Springs Fish Hatchery to Oak Creek are largely composed of natural groundwater.
Some of this groundwater arises within the local subwatershed, and is thus already accounted for in the model. To
prevent double-counting of this water, the reported discharges were reduced significantly to provide an
approximate match to observed base flows in Oak Creek.
Several other smaller discharges reported in the study area were determined to be used primarily for irrigation or
discharge to dry washes, do not cause live stream discharges and so are not explicitly included in the model. The
San Jose WWTP major discharge at Bisbee, Arizona is in part used for irrigation, but also discharges to
Greenbush Draw, tributary to the San Pedro. However, it enters the San Pedro upstream of the modeled area.
E-20
-------
Legend
Point Sources
Hydrography
Interstate
Water (Nat. Atlas Dataset)
US Census Populated Places
Municipalities (pop > 50,000)
] County Boundaries
Watershed with HUCSs
Verde
River
PAGE SPRINGS / //
FISH HATCHERY
Rgpseyelt
Reservoir
PINAL CR. WWTP
GCRP Model Areas - Salt and Verde River Basins
Major Point Sources
Figure 6. Major point sources in the Arizona basins.
E-21
-------
Meteorological Data
The required meteorological data series for the 20 Watershed study are precipitation, air temperature, and
potential evapotranspiration. The 20 Watershed model does not include water temperature or algal simulation and
uses a degree-day method for snowmelt. These meteorological data are drawn from the BASINS4 Meteorological
Database (USEPA 2008), which provides a consistent, quality-assured set of nationwide data with gaps filled and
records disaggregated. Scenario application will require simulation over 30 years, so the available stations are
those with a common 30-year period of record (or one that can be filled from an approximately co-located station)
that covers the year 2001. A total of 29 precipitation stations were identified for use in the Arizona basins model
with a common period of record of 10/1/1972-9/30/2002 (Table 4 and Figure 7 and Figure 8). Temperature
records are sparser; where these are absent, temperature is taken from nearby stations with an elevation correction.
For each weather station, Penman-Monteith reference evapotranspiration was calculated for use in HSPF using
observed precipitation and temperature coupled with SWAT weather generator estimates of solar radiation, wind
movement, cloud cover, and relative humidity.
For the 20 Watershed model applications, SWAT uses daily meteorological data, while HSPF requires hourly
data. It is important to note that a majority of the meteorological stations available for the Arizona basins are
Cooperative Summary of the Day stations that do not report sub-daily data. The BASINS4 dataset already has
versions of the daily data that have been disaggregated to an hourly time step using template stations. For each
daily station, this disaggregation was undertaken in reference to a single disaggregation template. Occasionally,
this automated procedure provides undesirable results, particularly when the total rainfall for the day is very
different between the subject station and the disaggregation template.
Table 4. Precipitation stations for the Arizona models
COOP ID
AZ020159
AZ020487
AZ020670
AZ020683
AZ020808
AZ021231
AZ021330
AZ021614
AZ021654
AZ021870
AZ022140
AZ023010
AZ023828
AZ024453
AZ025512
AZ026323
AZ026601
AZ026653
AZ026796
AZ026840
Name
Alpine
Ash Fork 3
Beaver Creek
Benson 6 SE
Black River Pumps
Canelo 1 NW
Cascabel
Childs
Chino Valley
Cochise 4 SSE
Coronado NM Hdqtrs
Flagstaff AP
Happy Jack RS
Jerome
Miami
Payson
Pinetop 2E
Pleasant Valley RS
Prescott
Punkin Center
Latitude
33.8493
35.199
34.6418
31.8803
33.4783
31.559
32.3208
34.3495
34.757
32.059
31.3457
35.1442
34.7433
34.7523
33.4045
34.2315
34.1243
34.099
34.5706
33.8557
Longitude
-109.146
-112.488
-111.783
-110.24
-109.751
-110.529
-110.413
-111.698
-112.456
-109.89
-110.254
-111.666
-111.413
-112.111
-110.87
-111.339
-109.921
-110.944
-112.432
-111.306
Temperature
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Elevation
(ft)
8049
5074
3523
1125
6065
1527
959
2650
4749
1274
1598
7003
7478
4950
3559
4907
7200
5048
5202
2326
E-22
-------
COOP ID
AZ027281
AZ027530
AZ027708
AZ027716
AZ028619
AZ028650
AZ029158
AZ029271
AZ029359
Name
Roosevelt 1 WNW
San Manuel
Sedona
Seligman
Tombstone
Tonto Creek Fish Hatchery 2
Walnut Creek
Whiteriver 1 SW
Williams
Latitude
33.6731
32.6014
34.8957
35.3323
31.7057
34.3839
34.9282
33.8169
35.2407
Longitude
-111.15
-110.633
-111.764
-112.879
-110.056
-111.097
-112.809
-109.983
-112.19
Temperature
X
X
X
X
X
X
X
X
Elevation
(ft)
2204
1055
4218
5248
1405
6389
5087
5120
6747
Orographic effects on precipitation and temperature are important throughout the region. This is addressed
through use of the elevation bands option and the imposition of precipitation and temperature lapse rates in the
SWAT model. All SWAT model subwatersheds are assigned at least one elevation band, and multiple elevation
bands are used when the interquartile range of elevations within a subwatershed exceeds 375 m. For HSPF,
whenever the precipitation station was located outside or near the edge of a model segment, a multiplier was
applied to the data based on the ratio of the estimated median annual rainfall from isohyetal information and the
long term annual average for the station. The evaporation data appeared to be estimates of pan evaporation, and
ranged from 70 to 100 inches per year. They were therefore adjusted by a factor of 0.7 to reduce them to potential
evapotranspiration. Some of the multipliers were adjusted slightly during the hydrology calibration.
E-23
-------
A Weather Stations
— Hydrography
Interstate
Water (Nat. Atlas Dataset)
US Census Populated Places
Municipalities (pop > 50,000)
^] County Boundaries
Watershed with HUCSs
Verde
River
AZ020487
AZ029359
AZ029-158
^
AZ021654
AZ026796
AZ024453
AZ020670
7
AZ028650
AZ026653
/ AZ026323
GCRP Model Areas - Salt and Verde River Basin
V\feather Stations
NADJ 983_Albere_metere
Map produced 02-11-2010- B. Tucker
Figure 7. Weather stations for the Arizona basins model - Verde and Salt River section.
E-24
-------
A Weather Stations
— Hydrography
Interstate
Water (Nat. Atlas Dataset)
US Census Populated Places
Municipalities (pop > 50,000)
^] County Boundaries
Watershed with HUCSs
San Pedro
River
GCRP Model Areas - San Pedro River Basin
Weather Stations
NADJ 983_Albere_metere
Map produced 02-11-2010- B. Tucker
Figure 8. Weather stations for the Arizona basins model - San Pedro River section.
E-25
-------
Watershed Segmentation
The Arizona basins were divided into 81 subwatersheds for the purposes of modeling (Figure 9 and Figure 10) -
30 in the Verde, 28 in the Salt, and 23 in the San Pedro river models. Initial calibration was conducted on the
Verde River at Clarkdale. However, the parameters derived at this station were not fully transferable to other
portions of the watershed, and additional calibration was conducted at multiple gage locations.
The Verde and Salt River models encompass entire watersheds upstream of major reservoirs - thus upstream
boundary conditions are not required. However, for the Verde River watershed, boundary conditions are needed to
account for the large influx of deep groundwater (much of it ultimately derived from infiltration many miles away
in the Chino watershed) that enters the river in the reach near Paulden, Arizona.
The San Pedro River watershed extends into Mexico; however, the geospatial and meteorological data used to
build the 20 Watershed models do not cover Mexico. Therefore, the San Pedro is simulated with an upstream
boundary condition at the USGS gage on the San Pedro River at Charleston, Arizona (09471000). This is the most
upstream gage with near complete records for the simulation period; the gage at Palominas (09470500), although
closer to the Mexican border, has long periods of missing records.
Major reservoirs are generally avoided in the model setup; however, it is also necessary to account for storage in
smaller reservoirs and stock ponds. For SWAT, these are specified using the Ponds option, based on information
in ADWR (2009) and, for the Verde watershed, Tetra Tech (2001). Significant pond storage is considered in
subwatersheds 11, 13, 15, 16, 17, 18, 19, 23, 25, 26, and 27 for the Verde River watershed. In the Salt River
watershed there are reservoirs with nominal storage capacity near 25,000 acre feet in subwatersheds 20 and 21,
although the normal capacity is only a fraction of this total. As these are headwater subwatersheds, these
reservoirs are also treated as ponds. No reservoirs or ponds are simulated in the San Pedro watershed.
It should be noted that Sullivan Lake, at the head of the perennial portion of the Verde River watershed
(subwatershed 14), intercepts flows out of the Chino watershed and has a significant impact on the progression of
flood waves downstream. This lake is not directly represented in the model due to lack of information on storage
characteristics.
E-26
-------
Legend
USGS Gages
Hydrography
Interstate
Water (Nat. Atlas Dataset)
US Census Populated Places
Municipalities (pop > 50,000)
^| County Boundaries
Model Subbasins
] Initial Calibration Watershed
Minnesota River Basin
USGS 09503700
USGS 09504420
USGS 09504500
USGS 09505800
USGS 09504000 '
USGS 09507980
USGS 09506000
USGS 09508500
USGS 09498500
GCRP Model Areas - Salt and Verde River Basins
Model Segmentation
NAD_1983_Albers_meters
Map produced 2-11-2010- B. Tucker
Figure 9. Model segmentation USGS stations utilized for the Arizona basins - Verde and Salt River
section.
Note: SWAT subwatersheds numbering is shown; the HSPF model for this watershed uses the same subwatershed boundaries with an
alternative internal numbering scheme.
E-27
-------
Legend
USGS Gages
Hydrography
Interstate
Water (Nat. Atlas Dataset)
US Census Populated Places
Municipalities (pop > 50,000)
USGS 09473000
^| County Boundaries
Model Subbasins
] Initial Calibration Watershed
Minnesota River Basin
USGS 09471000
GCRP Model Areas - San Pedro River Basin
Model Segmentation
NADJ 983_Albers_meters
Map produced 2-11-2010 - B. Tucker
Figure 10. Model segmentation USGS stations utilized for the Arizona basins - San Pedro River
section.
Note: SWAT subwatersheds numbering is shown; the HSPF model for this watershed uses the same subwatershed boundaries with an
alternative internal numbering scheme.
E-28
-------
Calibration Data and Locations
The site selected for initial calibration was the Verde River near Clarkdale, AZ (USGS gage 09504000); however,
calibration and validation were pursued at multiple locations (Table 5, Figure 9 and Figure 10).
Table 5. Calibration and validation locations in the Arizona basins
Station name
San Pedro River near Redington, AZ
Aravaipa Creek near Mammoth, AZ
Salt River near Roosevelt, AZ
Verde River near Paulden, AZ
Verde River near Clarkdale, AZ
Oak Creek near Cornville, AZ
West Clear Creek near Camp Verde, AZ
Verde River near Camp Verde, AZ
East Verde River near Childs, AZ
Verde River below Tangle Creek
USGS ID
09472000
09473000
09498500
09503700
09504000
09504500
09505800
09506000
09507980
09508500
Drainage area
(mi2)
2,927
537
4,306
2,507
3,503
355
241
5,009
331
5,858
Hydrology
calibration
X
X
X
X
X
X
X
X
X
X
Water quality
calibration
X
X
X
X
The model hydrology calibration period was set to Water Years 1993-2002, with some variation according to
gage variability. The end date was constrained by the common period of the set of 20 Watershed meteorological
stations available for the watershed, and a 10 year calibration period was desired. Calibration was done on the
later data, because of concerns that there may have been changes in land use and management over time.
Hydrologic validation was then performed on Water Years 1983-1992. Water quality calibration used calendar
years 1993-2002, while validation used 1986-1992, as limited data were available prior to 1986.
Other Relevant Features
Along the mainstem of the Verde River between Clarkdale and Camp Verde and on several tributaries there are
substantial water diversions to support riparian agricultural production, primarily hay. ADWR (2000) identifies 24
diversion structures on the Verde River proper from near Clarkdale to below Camp Verde, 32 diversions along
Oak Creek, and 12 diversions along Wet Beaver Creek, as well as several in other locations, and estimates that the
total agricultural diversion amount between Perkinsville and Horseshoe Reservoir (most of it occurring in the
Verde Valley, Oak Creek, and Wet Beaver Creek) amounts to 31,668 acre-feet per year. Very little of the water
diverted for irrigation returns as surface flow (Owen-Joyce and Bell 1983); however, a substantial portion may
return as subsurface flow. The water applied from these diversions is represented as irrigation applications in the
model. During development of the previous SWAT model for the Verde it was found that a direct linkage of
irrigation applications to river withdrawals did not provide satisfactory results and indeed tended to cause model
instability. Therefore, the withdrawals and irrigation are uncoupled in the model: irrigation is represented as
nominally occurring from an external source, while withdrawals from the river are specified separately as a
consumptive use that occurs during the April-September growing season. Consumptive use withdrawals are
applied to Verde model subwatersheds 6, 8, 9, 10, and 30. The status of agricultural diversions in the Salt and San
Pedro watersheds is not fully known; however, growing season diversions from the river are assigned to improve
flow closure, being assigned to subwatershed 8 in the San Pedro watershed and subwatershed 5 in the Salt River
watershed.
E-29
-------
A separate representation is used for the Prescott Valley area of the Verde River watershed model (subwatershed
16). Here, agriculture is supported by water stored in two small reservoirs (Granite and Willow). For this
subwatershed, irrigation is represented as linked to and derived from water stored in these reservoirs (represented
as ponds in the SWAT model).
Special notes are required regarding the East Verde River (Verde model subwatershed 4). The town of Payson,
Arizona obtains its municipal supply from groundwater, which is pumped from the alluvium of the East Verde.
The groundwater supply appears to be directly connected to surface water, and causes the East Verde to go dry at
times. However, there is also a source of imported water in the East Verde, as water is brought from across the
Mogollon Rim divide and discharged into the East Verde to augment Payson supplies. Detailed documentation
was not obtained. For the purposes of the 20 Watershed model it is assumed that the imported water is essentially
all consumed by Payson. Therefore, the stream is simulated as a losing reach, but the imported water is not
explicitly simulated. Payson's wastewater discharges leave the watershed, and are primarily used for golf course
irrigation in the Tonto Creek watershed (subwatershed 26 in the Salt River model).
E-30
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a si
No changes were made to the meteorological or land use base data. Similar to the SWAT modeling, the Globe,
Arizona point source total nitrogen and total phosphorus concentrations were reduced, since the recommended
concentrations did not permit a reasonable calibration of total nitrogen and total phosphorus at the Roosevelt
station.
An important feature of the Arizona basins is the complex interaction of surface and groundwater. As noted
earlier, neither SWAT nor HSPF is capable of providing a detailed, process-based simulation of groundwater
flow. It is therefore assumed that interactions with groundwater can be handled with the following simplifying
assumptions:
• The local (within subwatershed) accumulation and discharge of shallow groundwater is adequately
addressed by HSPF's active groundwater formulation.
• Discharges to streams from deep groundwater are represented as constant point sources, with discharge
rates set based on flow information on major springs identified in the Arizona Water Atlas (ADWR
2009). This means that the model cannot account for seasonal variability in deep groundwater discharge,
nor can it evaluate how such discharges may evolve in response to climate change.
• Losses to groundwater from stream reaches in alluvial basins are simulated based on channel
conductivity. This makes such losses a function of flow and depth in the affected reaches. The HSPF
model formulation, which is incorporated in the stream reach FTABLEs as a volume-based loss term does
not take into account changes in local groundwater head; instead, the loss occurs continuously.
In the Verde River watershed, irrigation withdrawals and applications were modeled (similarly to the SWAT
model) in the Verde Valley between Clarkdale and Camp Verde based on information provided in ADWR (2009).
A total of 31,668 ac-ft/yr is withdrawn during April-September from selected reaches based on the relative
amounts of grass and developed area in the reach watersheds. The HSPF irrigation module was used to apply this
water to the grass and developed PERLND's using a constant application of 0.11 inches/day during the April-
September period.
The starting parameters for the Arizona HSPF model were developed from an HSPF model of the San Francisco
Bay area watersheds, particularly watersheds in eastern Alameda County. After the starting parameters were
inserted into the model input files, average annual potential evapotranspiration values were computed and
compared to published values. Through this process it was determined the input potential evapotranspiration time
series should be reduced by multipliers, since the computation of these time series produced more PET on an
average annual basis than the published values indicate. The default multipliers used for PET were 0.70; however,
some of the multipliers were adjusted slightly during the hydrology calibration. Calibration adjustments focused
on the following parameters:
• LZSN (lower zone nominal storage): LZSN was generally reduced from the initial values to shift flows to
the wet period and reduce them in the summer. It was also used to increase total runoff.
E-31
-------
• INFILT (index to mean soil infiltration rate): Infiltration was generally decreased from the high initial
values to increase storm peaks, reduce low flows, and increase surface runoff.
• DEEPFR (fraction of groundwater inflow that will enter deep groundwater): small values of DEEPFR
were used to attempt to reduce low flows and to reduce total flow volume. In the Salt River, the initial
low values were not adjusted. In the Verde River at Paulden, DEEPFR was increased to a high value to
represent the recharge losses in the Chino Basin; some of this groundwater returns to the river below the
Paulden gage.
• BASETP (ET by riparian vegetation): Generally BASETP was increased over the initial values in order to
provide some ET by riparian vegetation and improved the simulation of low flows.
• LZETP (lower zone E-T parameter): LZETP was generally increased to reduce flow, particularly the low
flows, and to reduce total volumes.
• AGWRC (Groundwater recession rate): AGWRC was typically reduced from the initial values to help
reproduce the brief, sudden storms that are experienced in the Arizona basin.
Obtaining a high quality fit to hydrology in the Arizona basin is difficult with HSPF due to the importance of
groundwater, which is simplistically represented in the model. As in the SWAT model, the specification of
groundwater discharges as constant values and the simulation of reach losses by channel conductivity without
feedback from local groundwater elevations both introduce uncertainty.
Initial calibrations were performed for the two Verde River gages at Paulden and Clarkdale. The calibration
period was set to the 10 water years from 10/01/1992 to 09/30/2002. The results at Clarkdale are summarized in
Figures 11 through 17 and Tables 6 and 7. The fit at Clarkdale is fairly good, although the summer storm volumes
are over-simulated. Predictions at Clarkdale are largely determined by model fit upstream at Paulden, where flows
about 95 percent of the time consist of approximately constant base flow. Spring peaks occasionally push through
from the Chino subwatershed. Accuracy in simulating these peaks is primarily affected by lack of an accurate
representation of the hydraulic behavior of Sullivan Lake, and somewhat caused by the necessity of specifying
constant values for channel conductivity to account for transmission losses, when in fact these loss rates are likely
much reduced during the spring wet period. Parameter modifications to improve the peak spring flows out of the
Chino subwatershed result in significant over-prediction of summer storm events.
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002 )
•Avg Modeled Flow (Same Period)
35000
Apr-94
Oct-95
Apr-97
Date
Oct-98
Apr-00
Oct-01
6
8
10
12
14
Figure 11. Mean daily flow at USGS 09504000 Verde River near Clarkdale, AZ - calibration period
(HSPF).
E-32
-------
•6
6000
4000
2000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002 )
•Avg Modeled Flow (Same Period)
0-92
A-94
0-01
Figure 12. Mean monthly flow at USGS 09504000 Verde River near Clarkdale, AZ - calibration period
(HSPF).
_ 6000
I
I
LJ- 4000
2000
O)
TO
• Avg Flow (10/1 /1992 to 9/30/2002 )
• - - - • Line of Equal Value
Best-Fit Line
2000 4000
Average Observed Flow (cfs)
6000
O
£=
_
ro
m
100%
90%
80%
70% -
60%
50%
40% -
30%
20%
10% -
0%
Avg Observed Flow (10/1/1992 to 9/30/2002 )
Avg Modeled Flow (10/1/1992 to 9/30/2002 )
-Line of Equal Value
.JL.
I Ij,
O-92 A-94 O-95
A-97 O-98
Month
A-00 O-01
Figure 13. Monthly flow regression and temporal variation at USGS 09504000 Verde River near
Clarkdale, AZ - calibration period (HSPF).
E-33
-------
Avg Flow (10/1/1992 to 9/30/2002)
• Line of Equal Value
Best-Fit Line
800
¥
I600
T3
0)
^400
o
0)
200
y=1.|)498x-3.8J436
° = 0.9747
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002)
Avg Modeled Flow (Same Period)
800
600 4
400 ]
200 400 600
Average Observed Flow (cfs)
800
200 -4— V
10 11 12 1 2 3 4 5 6 7
Month
Figure 14. Seasonal regression and temporal aggregate at USGS 09504000 Verde River near
Clarkdale, AZ - calibration period (HSPF).
• Observed (25th, 75th)
-Median Observed Flow (10/1/1992 to 9/30/2002)
Average Monthly Rainfall (in)
Modeled (Median, 25th, 75th)
o
600
400
200
10 11 12 1
Figure 15. Seasonal medians and ranges at USGS 09504000 Verde River near Clarkdale, AZ -
calibration period (HSPF).
E-34
-------
Table 6. Seasonal summary at USGS 09504000 Verde River near Clarkdale, AZ - calibration
period (HSPF)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
87.85
84.09
101.70
370.90
529.79
290.55
129.11
79.98
76.29
78.20
83.85
100.71
80.00
84.00
83.50
87.00
87.00
92.00
83.00
79.00
75.00
77.00
78.00
81.00
77.00
78.00
80.00
81.00
80.00
80.00
77.00
72.00
69.00
71.00
75.00
73.00
84.00
89.00
90.00
93.00
101.50
232.00
93.00
87.00
81.00
82.00
84.00
85.00
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
96.91
99.66
108.76
324.23
600.69
282.20
126.31
87.10
70.29
77.73
88.43
104.90
82.51
85.28
81.17
90.51
101.20
120.53
100.56
73.67
68.05
69.49
79.82
83.32
71.73
72.52
75.07
74.56
78.14
74.05
72.53
62.50
59.97
64.23
69.25
70.17
99.47
120.36
108.70
122.50
136.39
227.47
157.49
108.75
76.39
78.65
97.29
112.11
•Observed Flow Duration (10/1/1992 to 9/30/2002 )
Modeled Flow Duration (10/1/1992 to 9/30/2002 )
o
100000
10000
1000
'ro
Q
100 = =3
10%
20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90% 100%
Figure 16. Flow exceedance at USGS 09504000 Verde River near Clarkdale, AZ - caalibration period
(HSPF).
E-35
-------
•Observed Flow Volume (10/1/1992 to 9/30/2002 )
Modeled FlowVolume (10/1/1992 to 9/30/2002 )
120%
o
o
-------
Table 7. Summary statistics at USGS 09504000 Verde River near Clarkdale, AZ - calibration
period (HSPF)
HSPF Simulated Flow
REACH OUTFLOW FROM DSN 101
10-Year Analysis F^riod: 10/1/1992 - 9/30/2002
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
_Jotak3fj3irrujlatejdjT^^
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9)
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3^:
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
^EjTOnnJ^yTlih^SLIlP^J-^
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE ~
0.66
0.35
0.14
0.09
0.10
0.38
0.09
0.25
0.02
Error Statistics
2.43
-7.64
-1.58
3.15
Observed Flow Gage
USGS 09504000 VERDE RIVER NEAR CLARKDALE, AZ
Hydrologic Unit Code: 15060202
Latitude: 34.8522416
Longitude: -112.065994
Drainage Area (sq-rri): 3503
Total Observed In-stream Flow:
Total of Observed higjiesMO%Jows:_^
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
11.52 »| 30
0.83
-0.50
-15.79
44.47
0.481
0.485
0.803
30
30
20
50
Model accuracy increases
as E or E' approaches 1 .0
0.64
0.36
0.15
0.09
0.09
0.38
0.09
0.29
0.01
Clear [
Hydrology Validation
Like the SWAT modeling, validation for the Verde River near Clarkdale was performed for the period 10/1/1982
through 9/30/1992. Results are presented in Figures 18 through 24 and Tables 8 and 9. The HSPF validation
results are fair, but are generally worse than during the calibration period. In particular, the storm peak volumes
are under-predicted, likely due to the effort to reduce summer storm peaks in the calibration.
E-37
-------
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1982 to 9/30/1992 )
•Avg Modeled Flow (Same Period)
i
I
onnn
vnnn
finnn
A nnn
onnn
1 nnn
1 1
1
i
yi
I,
11111
1
If'i'l
1
,1
1 1
Jn
ll
njFTt
, pi.
• " |i[r -
1
1
t
j
2
4
fi
8
10
12
c
'ro
Oct-82 Apr-84 Oct-85 Apr-87 Oct-88 Apr-90 Oct-91
Date
Figure 18. Mean daily flow at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(HSPF).
1500
i
I
1000 -
500 - -
O-82
A-84
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1982 to 9/30/1992
•Avg Modeled Flow (Same Period)
O-85
A-87
Month
O-88
A-90
O-91
)
10
9
8
7
6
5
4
3
2
1
0
'ro
a:
Figure 19. Mean monthly flow at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(HSPF).
E-38
-------
Avg Flow (10/1/1982 to 9/30/1992 )
• Line of Equal Value
Best-Fit Line
1500
I
o
LJ- 1000 -
O)
TO
y = 0.5957X + 72.209
R2 = 0.666
500 - — —» -.'-
0 500 1000 1500
Average Observed Flow (cfs)
100%
Avg Observed Flow (10/1/1982 to 9/30/1992 )
Avg Modeled Flow (10/1/1982 to 9/30/1992 )
-Line of Equal Value
en
O
8
ro
m
o>
to
A-84 O-85
A-87 O-88
Month
A-90 O-91
Figure 20. Monthly flow regression and temporal variation at USGS 09504000 Verde River near
Clarkdale, AZ - validation period (HSPF).
Avg Flow (10/1/1982 to 9/30/1992)
•Line of Equal Value
•Best-Fit Line
500 n
100 200 300 400
Average Observed Flow (cfs)
500
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1982 to 9/30/1992)
Avg Modeled Flow (Same Period)
ONDJFMAMJJAS
10 11 12 1 2 3 4 5 6 7
Month
Figure 21. Seasonal regression and temporal aggregate at USGS 09504000 Verde River near
Clarkdale, AZ - validation period (HSPF).
E-39
-------
• Observed (25th, 75th)
-Median Observed Flow (10/1/1982 to 9/30/1992)
Average Monthly Rainfall (in)
Modeled (Median, 25th, 75th)
600
500
400
t
o
10 11 12 1
CD
or
Figure 22. Seasonal medians and ranges at USGS 09504000 Verde River near Clarkdale, AZ -
validation period (HSPF).
Table 8. Seasonal summary at USGS 09504000 Verde River near Clarkdale, AZ - validation
period (HSPF)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
101.69
129.11
160.55
97.85
232.11
422.76
206.73
86.31
80.68
89.01
104.45
146.81
85.00
87.00
89.00
89.00
99.00
175.50
88.00
83.00
79.00
81.00
83.00
83.00
79.00
84.00
85.00
86.00
85.00
87.00
83.00
80.00
76.75
77.00
79.00
79.00
88.00
89.00
98.00
94.75
251.00
510.50
113.00
88.00
85.00
86.75
93.00
89.00
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
150.26
148.72
165.82
159.89
242.85
293.60
191.47
119.35
86.81
105.71
152.50
157.50
89.46
105.69
151.17
143.90
156.86
219.66
149.35
99.69
75.84
87.23
102.41
95.00
80.07
81.49
88.08
116.52
99.85
118.37
98.93
76.30
68.37
76.53
80.53
78.80
122.85
160.11
189.05
188.96
273.47
364.94
249.64
136.06
89.21
109.19
154.85
132.27
E-40
-------
•Observed Flow Duration (10/1/1982 to 9/30/1992 )
Modeled Flow Duration (10/1/1982 to 9/30/1992 )
10000
10%
20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90% 100%
Figure 23. Flow exceedance at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(HSPF).
o
O
CO
JD
o
o
I
"(D
E
•Observed Flow Volume (10/1/1982 to 9/30/1992 )
Modeled Flow Volume (10/1/1982 to 9/30/1992 )
120%
100%
80% —
60% —
40% —
20% —
Oct-82
Apr-84
Oct-85
Apr-87
Oct-88
Apr-90
Oct-91
Figure 24. Flow accumulation at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(HSPF).
E-41
-------
Table 9. Summary statistics at USGS 09504000 Verde River near Clarkdale, AZ - validation
period (HSPF)
HSPF Simulated Flow
REACH OUTFLOW FROM DSN 101
10-Year Analysis F^riod: 10/1/1982 - 9/30/1992
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9)
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3^:
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE ~
0.64
0.21
0.16
0.14
0.15
0.22
0.13
0.17
0.05
Error Statistics
6.31
3.34
-27.25
22.38
Observed Flow Gage
USGS 09504000 VERDE RIVER NEAR CLARKDALE, AZ
Hydrologic Unit Code: 15060202
Latitude: 34.8522416
Longitude: -112.065994
Drainage Area (sq-rri): 3503
Total Observed In-stream Flow:
Total of Observed highesMO%Jows:_^
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
18.80 » | 30
-7.82
6.64
-27.53
39.20
0.451
0.325
0.655
30
30
20
50
Model accuracy increases
as E or E' approaches 1 .0
0.60
0.29
0.16
0.11
0.13
0.24
0.12
0.23
0.03
Clear [
Hydrology Results for Larger Watershed
As described above, parameters determined through calibration to the Verde River near Paulden and Clarkdale
gages were not fully transferable to other gages in the watershed. Hydrology calibration was performed at a total
of 10 gages in the Arizona basins - 6 in the Verde River watershed, 1 in the Salt River watershed, and 2 in the
San Pedro River watershed. Only the gage at Roosevelt provides a long period of record for the Salt River
watershed. The mainstem gage for the San Pedro River does not provide a rigorous calibration test because its
flow is largely determined by the upstream boundary condition. Therefore, calibration was also performed on
perennial Aravaipa Creek. The San Pedro gages ceased operation in 1995; therefore calibration was pursued over
an earlier time period without a separate validation test.
Calibration results at all gages are summarized in Table 10 and are generally of similar quality to the fit obtained
on the Verde River near Clarkdale. The generally close match between observed and predicted flow at the Salt
River gage is shown in Figures 25 through 31 and Tables 11 and 12. Results of the validation exercise are
summarized in Table 13. In general, the quality of fit during the validation period is similar to that in the
calibration period, with some reductions in fit for some of the seasonal volume error terms, and also some
improvements. The model is judged to be useful for scenario evaluation.
E-42
-------
Table lO.Summary statistics (percent error) for all stations - calibration period WY 1992-2002
(HSPF)
Station
Error in
total
volume:
Error in
50%
lowest
flows:
Error in
10%
highest
flows:
Seasonal
volume
error -
Summer:
Seasonal
volume
error- Fall:
Seasonal
volume
error -
Winter:
Seasonal
volume
error -
Spring:
Error in
storm
volumes:
Error in
summer
storm
volumes:
Daily
Nash-
Sutcliffe
Coefficient,
E:
Monthly
Nash-
Sutcliffe
Coefficient,
E:
09472000
San
Pedro nr
Redington
(1972-95)*
8.73
NA*
-5.26
22.67
8.64
-3.27
25.98
-4.82
11.07
0.574
0.908
09473000
Aravaipa
Crk nr
Mammoth
(1972-95)
-2.49
-3.74
1.89
-0.61
-7.68
2.64
-11.01
-13.40
-3.99
0.553
0.425
09498500
Salt River
nr
Roosevelt
4.48
2.24
7.56
20.18
11.78
2.79
-1.82
30.67
21.62
0.529
0.930
09503700
Verde
River nr
Paulden
9.41
8.37
5.57
20.17
15.98
5.12
9.84
-12.00
111.20
0.624
0.835
09504000
Verde
River nr
Clarkdale
2.43
-7.64
-1.58
3.15
11.52
0.83
-0.50
-15.79
44.47
0.481
0.803
09504500
Oak
Creek nr
Cornville
2.64
-12.56
0.51
-5.58
24.57
7.67
-33.23
-18.12
-8.02
0.078
0.443
09505800
W Clear
Cr nr
Camp
Verde
7.50
-7.97
2.60
-3.67
13.45
1.00
37.02
-33.88
-44.53
0.451
0.803
09506000
Verde
River nr
Camp
Verde
-2.41
-34.63
6.79
-34.15
-5.83
6.53
-18.47
-10.81
-42.63
0.661
0.803
09507980
E Verde
River nr
Childs
1.00
64.01
-1.53
-4.49
2.62
1.47
-0.25
-11.92
-57.76
0.689
0.946
09508500
Verde R
below
Tangle
Cr
-5.03
-17.25
-2.42
-19.87
0.44
-1.97
-14.97
-16.42
-36.53
0.703
0.921
*Note that median flow for the San Pedro River nr Redington is 0.
E-43
-------
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002 )
•Avg Modeled Flow (Same Period)
100000
90000
80000
70000
60000
50000
Q: 40000
30000
20000
10000
1
1
ip '
,
IL
—
,1
m
I1
—
Hi
in .
i
i
—
i'|i'"|'
, j
—
•1
' i i
T '
,
|
"ii
T
i
—
.
o
Oct-92
603
*£
-8 rS
^s
10 g
- 12
14
Apr-94 Oct-95 Apr-97 Oct-98 Apr-00 Oct-01
Date
Figure 25. Mean daily flow at USGS 09498500 Salt River near Roosevelt, AZ - calibration period
(HSPF).
i
o
15000
10000 - —
5000 - -
0-92
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002 )
•Avg Modeled Flow (Same Period)
A-94
0-01
Figure 26. Mean monthly flow at USGS 09498500 Salt River near Roosevelt, AZ - calibration period
(HSPF).
E-44
-------
15000
I
o
10000
5000
O)
TO
Avg Flow (10/1/1992 to 9/30/2002 )
•Line of Equal Value
Best-Fit Line
y = 0.9806X + 54.209
R2 = 0.9333
0 5000 10000 15000
Average Observed Flow (cfs)
100%
Avg Observed Flow (10/1/1992 to 9/30/2002 )
Avg Modeled Flow (10/1/1992 to 9/30/2002 )
-Line of Equal Value
en
JD
O
ro
m
_
0)
to
A-94 O-95
A-97 O-98
Month
A-00 O-01
Figure 27. Monthly flow regression and temporal variation at USGS 09498500 Salt River near
Roosevelt, AZ - calibration period (HSPF).
Avg Flow (10/1/1992 to 9/30/2002)
• Line of Equal Value
Best-Fit Line
2500
y = CJ.9922X f 44.478
-20001- -^B^=Loapa5_
-0 1500 -
1000 -
0
D)
ro
500 -
500 1000 1500 2000
Average Observed Flow (cfs)
2500
Avg Monthly Rainfall (in)
—•-Avg Observed Flow (10/1/1992 to 9/30/2002)
Avg Modeled Flow (Same Period)
2500 T—-—-—-—-—-—-—-—-—-—-—-—r 0
2000 -
•g 1500
r? 1000
500 - —a
3.5
10 11 12 1 23456789
Month
Figure 28. Seasonal regression and temporal aggregate at USGS 09498500 Salt River near Roosevelt,
AZ - calibration period (HSPF).
E-45
-------
• Observed (25th, 75th)
-Median Observed Flow (10/1/1992 to 9/30/2002)
Average Monthly Rainfall (in)
Modeled (Median, 25th, 75th)
3000
2500
2000
t
o
Figure 29. Seasonal medians and ranges at USGS 09498500 Salt River near Roosevelt, AZ -
calibration period (HSPF).
Table 11.Seasonal summary at USGS 09498500 Salt River near Roosevelt, AZ - calibration
period (HSPF)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
256.42
450.48
507.61
1831.48
1511.59
1875.51
1429.55
768.86
277.08
226.12
377.92
368.88
200.50
243.00
267.50
267.00
455.50
1080.00
953.50
535.00
187.50
185.50
267.50
249.50
168.25
195.00
210.00
204.00
206.00
211.50
224.00
150.25
121.00
129.25
205.50
174.00
264.00
352.75
356.25
413.25
1135.00
2682.50
1722.50
1047.50
335.25
269.50
389.50
402.25
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
376.63
490.17
490.28
1674.43
1731.21
1974.34
1465.01
718.83
247.91
272.76
452.19
444.35
271.58
278.03
291.03
251.55
340.99
725.57
750.52
397.70
194.37
193.92
229.13
316.73
214.51
229.85
228.54
204.76
182.31
200.23
218.88
136.42
104.06
123.47
158.00
198.86
372.78
404.40
484.15
491.75
1074.86
2429.07
1974.37
761 .58
307.05
274.39
366.45
415.12
E-46
-------
(0
Q
•Observed Flow Duration (10/1/1992 to 9/30/2002 )
Modeled Flow Duration (10/1/1992 to 9/30/2002 )
100000
10000
1000
100 - =
10%
90%
100%
20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
Figure 30. Flow duration at USGS 09498500 Salt River near Roosevelt, AZ - calibration period (HSPF).
o
o
ro
T3
(D
-------
Table 12.Summary statistics at USGS 09498500 Salt River near Roosevelt, AZ - calibration
period (HSPF)
HSPF Simulated Flow
REACH OUTFLOW FROM DSN 103
10-Year Analysis F^riod: 10/1/1992 - 9/30/2002
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
_JoJak3fj3imuljjt^
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9)
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3^:
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
^EjTOnnJ^yTlih^SLIlP^J-^
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE 1
2.70
1.67
0.29
0.31
0.36
1.40
0.64
1.22
0.11
Error Statistics
4.48
2.24
7.56
20.18
Observed Flow Gage
USGS 09498500 SALT RIVER NEAR ROOSEVELT, AZ
Hydrologic Unit Code: 15060103
Latitude: 33.61 94949
Longitude: -110.9215037
Drainage Area (sq-rri): 4306
Total Observed In-stream Flow:
Total of Observed higjiesMO%Jows:_^
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
11.78 »| 30
2.79
-1.82
30.67
21.62
0.529
0.539
30
30
20
50
Model accuracy increases
as E or E' approaches 1 .0
2.59
1.56
0.29
0.26
0.32
1.36
0.65
0.94
0.09
Clear [
0.930
E-48
-------
Table 13.Summary statistics for all stations - validation period WY 1982-1992 (HSPF)
Station
Error in total
volume:
Error in 50%
lowest flows:
Error in 10%
highest flows:
Seasonal
volume error -
Summer:
Seasonal
volume error -
Fall:
Seasonal
volume error -
Winter:
Seasonal
volume error -
Spring:
Error in storm
volumes:
Error in
summer storm
volumes:
Daily Nash-
Sutcliffe
Coefficient of
Efficiency, E:
Monthly Nash-
Sutcliffe
Coefficient of
Efficiency, E
09498500 Salt
River nr
Roosevelt
-7.09
-9.03
6.64
16.38
4.66
-26.48
1.49
16.98
40.28
0.354
0.786
09503700
Verde River
nr Paulden
8.35
5.07
-11.57
-4.29
13.22
16.64
8.64
-14.00
-20.94
0.443
0.614
09504000
Verde River
nrClarkdale
6.31
3.34
-27.25
22.38
18.80
-7.82
6.64
-27.53
39.20
0.451
0.320
09504500
Oak Creek nr
Cornville
15.32
3.54
-9.47
61.67
15.76
6.84
12.03
-17.61
86.24
0.545
0.755
09505800 W
ClearCr nr
Camp Verde
-9.69
-9.99
-17.07
-26.78
-37.41
7.66
-17.93
-51.19
-74.51
0.232
0.655
09507980
E Verde
River nr
Childs
31.18
8.49
45.18
-17.15
28.39
65.55
-26.11
34.57
-43.69
-0.119
0.335
09508500
Verde R
below Tangle
Cr
-6.93
-21.11
-13.91
24.90
-10.34
-5.86
-25.48
-16.03
78.31
0.510
0.809
Water Quality Calibration and Validation
The 20 Watershed models are designed to provide water quality simulation for total suspended solids (TSS), total
nitrogen, and total phosphorus. The water quality calibration focuses on the replication of monthly loads, as
specified in the project QAPP. Given the simplified approach to water quality simulation in the 20 Watershed
model a close match to individual concentration observations cannot be expected. However, comparison to
monthly loads presents challenges, as monthly loads are not observed. Instead, monthly loads must be estimated
from scattered concentration grab samples and continuous flow records. Such estimation presents some
uncertainty because it depends on the degree and form in which concentration and flow are correlated with one
another. Further, the bulk of the load of sediment and sediment-associated phosphorus is likely to move through
the system in a limited number of high flow events, which usually have not been monitored. As a result, the
monthly load calibration is inevitably based on the comparison of two uncertain numbers. Nonetheless,
calibration is able to achieve a fair agreement. The load comparisons were supported by detailed examinations of
E-49
-------
the relationships of flows to loads and concentrations and the distribution of concentration prediction errors versus
flow, time, and season, as well as standard time series plots.
For application on a nationwide basis, the 20 Watershed protocols assume that TSS and total phosphorus loads
will likely exhibit a strong positive correlation to flow (and associated erosive processes), while total nitrogen
loads, which often have a dominant groundwater component, will not. Accordingly, TSS and total phosphorus
loads were estimated from observations using a flow-stratified log-log regression approach, while total nitrogen
loads were estimated using a flow-stratified averaging estimator, consistent with the findings of Preston et al.
(1989).
Water quality calibration and validation was done on the Verde River near Clarkdale, using 1993-2002 for
calibration and 1986-1992 for validation. As with hydrology, calibration was performed on the later period as this
better reflects the land use included in the model. The start of the validation period is constrained by data
availability.
TSS calibration was performed by adjusting the coefficients in the soil detachment (KRER) and soil washoff
(KSER) equations along with changes to the seasonal vegetation COVER. Furthermore, it was necessary to model
scour of the soil matrix (i.e., gully erosion) in addition to losses of detached sediment. The washoff of detached
sediment did not provide sufficient sediment losses to calibrate the model without severely degrading the channel
bed.
Time series of simulated and estimated TSS loads at the Clarkdale station for both periods are shown in Figure 32
and statistics for the two periods are provided separately in Table 14. Visually, the model is roughly simulating
the trends contained in the observed data. The key statistic in Table 14 (consistent with the QAPP) is the relative
percent error, which shows the error in the prediction of monthly load normalized to the estimated load. Table 14
also shows the relative average absolute error, which is the average of the relative magnitude of errors in
individual monthly load predictions. This number is inflated by outlier months in which the simulated and
estimated loads differ by large amounts (which may be as easily due to uncertainty in the estimated load due to
limited data as to problems with the model) and the third statistic, the relative median absolute error, is likely
more relevant and shows good agreement.
TSS
1,000,000
- Regression Loads
-Simulated Loads
OO
O->-rMCO-^-ir)(DI^
O)O)O)O)O)O)O)O)
g 8 5 8
Figure 32. Fit for monthly load of TSS at USGS 09504000 Verde River near Clarkdale, AZ (HSPF).
E-50
-------
Table 14. Model fit statistics (observed minus predicted) for monthly sediment loads using
stratified regression at USGS 09504000 Verde River near Clarkdale, AZ (HSPF)
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1993-2002)
31%
40%
1.1%
Validation period
(1986-1992)
-41%
123%
8.5%
Several other diagnostics were also examined to evaluate agreement between the model and observations. These
are available in full in the calibration spreadsheets, but a few examples are provided below. First, load-flow power
plots were compared for individual days (Figures 33 and 34). These show that the relationship between flow and
load is reasonably consistent across the entire range of observed flows, for both the calibration and validation
periods.
Verde River -Clarkdale
1993-2002
-I nnnnnn
I UUUUUU
100000
>
| 10000
I/)
I 1000
§ 100
| 10
1
0-1
. 1 n
1
••
^^^
• • >v-.-*-t^i^^
4^^'-^>
jjKj&**_
IT
10 100 1000 10000 100000
Flow, cfs
• Simulated A Observed ^^™ Power (Simulated) Power (Observed)
Figure 33. Power plot for observed and simulated TSS at USGS 09504000 Verde River near Clarkdale,
AZ - calibration period (HSPF).
E-51
-------
100000
ra
•o
ra
o
(0
10000
1000
100
10
1 -f
0.1
1
Verde River - Clarkdale
1986-1992
10
100
Flow, cfs
1000
10000
Simulated A Observed
Power (Simulated)
Power (Observed)
Figure 34. Power plot for observed and simulated TSS at USGS 09504000 Verde River near Clarkdale,
AZ - validation period (HSPF).
A standard time series plot (Figure 35) shows that observed and simulated concentrations achieve at best a fair
agreement, and the model may deviate substantially from individual observations. However, the concentration
statistics (Table 15) show that reasonably low median errors are achieved.
E-52
-------
Verde River - Clarkdale
1993-2002
• Simulated A Observed
10000
O)
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Year
Figure 35. Time series plot of TSS concentration at USGS 09504000 Verde River near Clarkdale, AZ
(HSPF).
Table 15.Relative errors (observed minus predicted) for TSS concentration at USGS 09504000
Verde River near Clarkdale, AZ (HSPF).
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1993-2002)
47
-104%
-19%
Validation period
(1986-1992)
62
7.3%
-1.0%
For simulation of total phosphorus, calibration was performed primarily through adjustment of the potency factors
and the subsurface concentrations. Total nitrogen calibration was accomplished primarily by adjusting the
subsurface concentrations and secondarily by the accumulation-washoff parameters. Monthly loading time series
for total phosphorus are shown in Figure 36 and the load statistics are summarized in Table 16. The model
reproduces the general trend in monthly loads, but is significantly lower than the peak loads predicted by the
regression method, resulting in high relative percent errors for both the calibration and validation periods. It
should be noted that the available data are limited, particularly for high flow events. Thus, the estimates of
"observed" load are also subject to considerable uncertainty.
E-53
-------
10000
Total P
9 9 9
- Regression Loads
-Simulated Loads
Figure 36. Fit for monthly load of total phosphorus at USGS 09504000 Verde River near Clarkdale, AZ
(HSPF).
Table 16. Model fit statistics (observed minus predicted) for monthly total phosphorus loads
using stratified regression
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1993-2002)
87%
87%
0.6%
Validation period
(1986-1992)
66%
78%
5.1%
Additional diagnostics for total phosphorus included flow-load power plots (Figures 37 and 38), concentration
time series plots (Figure 39) and analysis of concentration errors (Table 17). While these show approximate
agreement, the model often overpredicts total phosphorus concentrations under lower flow conditions.
E-54
-------
Verde River - Clarkdale
1993-2002
w
•o
•o
a
o
100
10
1
0.1
0.01
0.001
10
100 1000
Flow, cfs
10000
100000
• Simulated A Observed
Row er (Simulated)
Power (Observed)
Figure 37. Power plot for observed and simulated total phosphorus at USGS 09504000 Verde River
near Clarkdale, AZ - calibration period (HSPF).
Verde River - Clarkdale
1986-1992
10
w
•o
w
c
s
•a
a
o
0.1
0.01
0.001
10
100
Flow, cfs
1000
10000
Simulated A Observed
Fbw er (Simulated)
Fbw er (Observed)
Figure 38. Power plot for observed and simulated total phosphorus at USGS 09504000 Verde River
near Clarkdale, AZ - validation period (HSPF).
E-55
-------
Verde River - Clarkdale
1993-2002
• Simulated A Observed
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Year
Figure 39. Time series plot of total phosphorus concentration at USGS 09504000 Verde River near
Clarkdale, AZ (HSPF).
Table 17. Relative errors (observed minus predicted) for total phosphorus concentration at
USGS 09504000 Verde River near Clarkdale, AZ (HSPF)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1993-2002)
57
16%
-23%
Validation period
(1986-1992)
75
54%
-8.5%
Fewer data are available for total nitrogen because many sampling events omitted one or more nitrogen species.
This increases the uncertainty of the comparison. Results for total nitrogen are summarized in Figures 40 through
43 and Tables 18 and 19 following the same format as total phosphorus. The loading results are fair, and are
generally better than those obtained for total phosphorus; however, there is significant uncertainty in the
prediction of individual total nitrogen observations. Total nitrogen concentrations at base flow are generally over-
predicted.
E-56
-------
180
c
ro
Total N
-Averaging Loads
-Simulated Loads
c
ro
c
ro
9
c
ro
•fincor^oocno-j-CN
o>o>o>o>o>o>ooo
c
ro
c
ro
c
ro
c
ro
c
ro
c
ro
Figure 40. Fit for monthly load of total nitrogen at USGS 09504000 Verde River near Clarkdale, AZ
(HSPF).
Table 18. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator (HSPF)
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1993-2002)
1 .6%
45%
17%
Validation period
(1986-1992)
-2.7%
37%
18%
E-57
-------
Verde River - Clarkdale
1993-2002
TN Load, tons/day
100
O
1
0.1
0.01
10
100 1000
Flow, cfs
10000
100000
• Simulated A Observed ^^™ Power (Simulated) ^^"Fbwer (Observed)
Figure 41. Power plot for observed and simulated total nitrogen at USGS 09504000 Verde River near
Clarkdale, AZ - calibration period (HSPF).
Verde River - Clarkdale
1986-1992
10
re
5
In
c
o
•4-i
-6
re
o
1
0.1
0.01
10
100
Flow, cfs
1000
10000
Simulated A Observed
Fbw er (Simulated)
Row er (Observed)
Figure 42. Power plot for observed and simulated total nitrogen at USGS 09504000 Verde River near
Clarkdale, AZ - validation period (HSPF).
E-58
-------
2.5
D)
2
1.5
Verde River - Clarkdale
1993-2002
•Simulated A Observed
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Year
Figure 43. Time series plot of total nitrogen concentration at USGS 09504000 Verde River near
Clarkdale, AZ (HSPF).
Table 19. Relative errors (observed minus predicted) for total nitrogen concentration at USGS
09504000 Verde River near Clarkdale, AZ (HSPF)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1993-2002)
46
-63%
-70%
Validation period
(1986-1992)
75
22%
1.9%
Water Quality Results for Larger Watershed
Summary statistics for the water quality calibration and validation at two other stations in the watershed (i.e., Salt
River near Roosevelt and Verde River below Tangle Creek) are provided in Tables 20 and 21 along with the
Clarkdale statistics. Water quality was not calibrated at the Camp Verde station on the Verde River because of the
lack of observed data at that location. And no water quality calibration was done in the San Pedro River. In most
cases, total nitrogen loads are better predicted than total phosphorus and TSS loads. Simulated TSS and total
phosphorus loads in the Verde River are lower than those estimated from observations, but this may reflect, in
part, the uncertainty in the regression-based load estimates as water quality observations during high flows are
sparse. Water quality results in the Salt River are generally better than the Verde River.
E-59
-------
Table 20. Summary statistics (observed minus predicted) for water quality for all stations
calibration period 1993-2002 (HSPF)
Station
Relative Percent Error
TSS Load
TSS Concentration
Median Percent Error
Relative Percent Error
TP Load
TP Concentration
Median Percent Error
Relative Percent Error
TN Load
TN Concentration
Median Percent Error
09498500
Salt River nr
Roosevelt
-10
0.80
-29
-0.53
-8.2
-3.5
09504000
Verde River nr
Clarkdale
31
-19
87
-23
1.6
-70
09508500
Verde River below
Tangle Cr
81
-5.6
75
-17
-2.7
-64
Table 21. Summary statistics (observed minus predicted) for water quality for all stations
validation period 1986-1992 (HSPF)
Station
Relative Percent Error
TSS Load
TSS Concentration
Median Percent Error
Relative Percent Error
TP Load
TP Concentration
Median Percent Error
Relative Percent Error
TN Load
TN Concentration
Median Percent Error
09498500
Salt River nr
Roosevelt
4.8
1.8
52
8.9
10
-21
09504000
Verde River nr
Clarkdale
-41
-1.0
78
-8.5
-2.7
1.9
09508500
Verde River below
Tangle Cr
2.1
-28
46
-12
10
-15
E-60
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
A SWAT model already exists for the Verde River portion of the watershed (Tetra Tech 2001). This model was
calibrated for hydrology and nutrients. The existence of this earlier model provides a useful basis for parameter
initialization, and is one of the reasons that the Arizona basins were selected as a pilot site. However, there are
also significant differences to the 20 Watershed model. Because of these differences in approach the two models
are substantively different, and not all model parameters are transferable. Nonetheless, the earlier model does
provide important insights and parameter starting values that are incorporated into the 20 Watershed model.
A key aspect of the Arizona basins models is the intimate linkage of surface and groundwater hydrology.
Perennial flow in various river segments is supported by groundwater discharge that, in the case of the Verde
watershed, may arise from distant teleconnections. Many of the river reaches in alluvial valleys also lose flow to
groundwater, at least on a seasonal basis. Both SWAT and HSPF models include simplified mass-balance
accounting of groundwater at a local (subwatershed) scale, but neither model contains a detailed simulation of
surface and groundwater interactions. In SWAT, the presence of deep groundwater discharges derived from
sources outside a model subwatershed can be addressed through specification of these inflows as point sources.
This is adequate for calibration, but there is no provision in the model for direct consideration of how these
sources may change in the face of climate change. SWAT also provides for simulation of losing river reaches
through specification of a rate of bed conductivity. This approach does not account for interaction with the
seasonal water table. In many cases, alluvial river reaches may gain from groundwater during the wet season
when water tables are high and lose to the alluvial aquifer during dry seasons. SWAT is, however, constrained to
simulation of these interactions solely as a bed conductivity rate. This means that the behavior of seasonally
losing reaches can only be roughly approximated in the model.
to
No changes were made to the meteorological or land use base data. For the Globe, Arizona point source it is clear
that phosphorus concentrations downstream during baseflow conditions are much less than would be expected
from the estimated phosphorus load from this WWTP. This may be due to a rapid loss of phosphorus in the near
field immediately downstream of the discharge, likely due to settling of particulate matter. Discounting total
phosphorus load in the effluent to 1/8 of the nominal value provided resolved this problem during calibration.
A key feature of the Arizona basins is the complex interaction of surface and groundwater. As noted above,
neither SWAT nor HSPF is capable of providing a detailed, process-based simulation of groundwater flow. It is
therefore assumed that interactions with groundwater can be handled with the following simplifying assumptions:
• The local (within a subwatershed) accumulation and discharge of shallow groundwater is adequately
addressed by SWAT's linear storage reservoir formulation.
• Discharges to stream from deep groundwater are represented as constant point sources, with discharge
rates set based on flow information on major springs identified in the Arizona Water Atlas (ADWR
2009). This means that the model cannot account for seasonal variability in deep groundwater discharge,
nor can it evaluate how such discharges may evolve in response to climate change.
• Losses to groundwater from stream reaches in alluvial basins are simulated based on channel
conductivity. This makes such losses a function of flow, wetted perimeter, and travel time in the affected
reaches. The SWAT model formulation does not take into account changes in local groundwater head;
instead, the loss occurs continuously. In addition, SWAT partitions channel losses to deep groundwater
and bank storage using a fixed ratio.
E-61
-------
Hydrology Calibration
Obtaining a high quality fit to hydrology in the Arizona basins is difficult with SWAT due to the importance of
groundwater interaction terms, which are simplistically addressed in the model. The specification of groundwater
discharges as constant values and the simulation of reach losses by channel conductivity without feedback from
local groundwater elevations both introduce uncertainty. In addition, the model formulation requires specification
of several factors at a global level, not allowing for spatial variability, including the fraction of transmission losses
assigned to deep groundwater (TRNSRCH), the intensity of direct evaporation from the channel (EVRCH), and
the parameters controlling snow melt.
Calibration adjustments focused on the following parameters:
• Curve numbers (varied systematically by land use)
• ESCO (soil evaporation compensation factor)
• Reach conductivity
• Bank storage and recession rates
• Groundwater "revap" rates
Initial calibrations were performed for the Verde River at Clarkdale and are summarized in Figures 44 through 50
Tables 22 and 23. The fit is fair at best, although the total volume errors are small. However, predictions at
Clarkdale are largely determined by model fit upstream at Paulden, where flows about 95 percent of the time
consist of approximately constant baseflow. Spring peaks occasionally push through from the Chino watershed.
Accuracy in simulating these peaks appears to be affected by 1) lack of an accurate representation of the hydraulic
behavior of Sullivan Lake, and 2) the necessity of specifying constant values for channel conductivity to account
for transmission losses, when in fact these loss rates are likely much reduced during the spring wet period.
Modifications to increase spring flows out of the Chino watershed rapidly results in severe over-prediction of
response to summer storm events. The Nash-Sutcliffe coefficient is near zero because flows tend to remain
approximately constant during dry periods.
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002 )
•Avg Modeled Flow (Same Period)
35000
30000
_ 2500°
;§, 20000
1 15000
10000
Tyyy
1 1 1 1 1 ] 1
14
Oct-95
Apr-97 Oct-98
Date
Apr-00
Oct-01
Figure 44. Mean daily flow at USGS 09504000 Verde River near Clarkdale, AZ - calibration period
(SWAT).
E-62
-------
i
o
4000
3000 -
2000 -
1000 - -
0-92
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002 )
•Avg Modeled Flow (Same Period)
A-94
0-01
Figure 45. Mean monthly flow at USGS 09504000 Verde River near Clarkdale, AZ - calibration period
(SWAT).
Avg Flow (10/1/1992 to 9/30/2002 )
• Line of Equal Value
Best-Fit Line
4000
1000 2000 3000 4000
Average Observed Flow (cfs)
£
_
ro
m
Avg Observed Flow (10/1/1992 to 9/30/2002 )
Avg Modeled Flow (10/1/1992 to 9/30/2002 )
-Line of Equal Value
O-92 A-94 O-95
A-97 O-98
Month
A-00 O-01
Figure 46. Monthly flow regression and temporal variation at USGS 09504000 Verde River near
Clarkdale, AZ - calibration period (SWAT).
E-63
-------
• Avg Flow (10/1 /1992 to 9/30/2002)
Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002)
Avg Modeled Flow (Same Period)
600
i
I
600
400
200
200 400
Average Observed Flow (cfs)
600
10 11 12 1 23456789
Month
Figure 47. Seasonal regression and temporal aggregate at USGS 09504000 Verde River near
Clarkdale, AZ - calibration period (SWAT).
i Observed (25th, 75th)
•Median Observed Flow (10/1/1992 to 9/30/2002)
Average Monthly Rainfall (in)
Modeled (Median, 25th, 75th)
•e
1
800
600
400
200
10 11 12
Figure 48. Seasonal medians and ranges at at USGS 09504000 Verde River near Clarkdale, AZ -
calibration period (SWAT).
E-64
-------
Table 22.Seasonal summary at USGS 09504000 Verde River near Clarkdale, AZ - calibration
period (SWAT)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
87.85
84.09
101.70
370.90
529.79
290.55
129.11
79.98
76.29
78.20
83.85
100.71
80.00
84.00
83.50
87.00
87.00
92.00
83.00
79.00
75.00
77.00
78.00
81.00
77.00
78.00
80.00
81.00
80.00
80.00
77.00
72.00
69.00
71.00
75.00
73.00
84.00
89.00
90.00
93.00
101.50
232.00
93.00
87.00
81.00
82.00
84.00
85.00
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
152.63
107.60
138.79
235.56
340.35
181.43
119.69
90.37
86.15
117.83
154.15
228.15
91.66
92.48
89.91
104.44
99.91
103.13
94.57
72.78
72.08
84.24
113.30
107.68
82.39
82.01
80.83
78.08
76.04
73.67
72.58
64.29
60.86
66.98
86.81
79.88
113.64
112.22
114.00
147.50
154.28
161.90
127.94
102.97
89.61
114.69
163.60
147.35
•Observed Flow Duration (10/1/1992 to 9/30/2002 )
•Modeled Flow Duration (10/1/1992 to 9/30/2002 )
100000
o
-------
•Observed Flow Volume (10/1/1992 to 9/30/2002 )
•Modeled Flow Volume (10/1/1992 to 9/30/2002 )
120%
o
o
T3
I
-------
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 10
10-Year Analysis F^riod: 10/1/1992 - 9/30/2002
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3^:
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error injH)%Jik|hest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error- Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
0.63
0.29
0.14
0.16
0.13
0.24
0.10
0.28
0.08
Error Statistics
-2.46
-1.74
-19.34
89.89
Observed Flow Gage
USGS 09504000 VERDE RIVER NEARCLARKDALE, AZ
Hydrologic Unit Code: 15060202
Latitude: 34.852241 6
Longitude: -11 2.065994
Drainage Area (sq-rri): 3503
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
46.00 » | 30
-36.45
3.88
-3.32
593.89
0.030
0.236
0.685
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
0.64
0.36
0.15
0.09
0.09
0.38
0.09
0.29
0.01
Clear [
Hydrology Validation
Validation for the Verde River near Clarkdale was performed for the period 10/1/1982 through 9/30/1992. Results
are presented in Figures 51 through 57 and Table 24 and 25. The validation results are generally similar to those
of the calibration period, and indeed are better on some statistics, indicating that the model is not over-fit to the
specific conditions of the calibration period.
It is important to recognize that the validation uses the 2001 land use as a static representation. While the
watershed has remained largely in National Forest, Indian Reservations and unoccupied rangeland, important
temporal changes have occurred as a result of intermittent wildfires. Areas of recent burns typically have
decreased evapotranspiration and increased direct runoff. These, however, are not represented in the model. In
addition, the PET estimates for the 20 Watershed model use SWAT weather generator statistics for solar
radiation, cloud cover, wind, and relative humidity - essentially assuming that the central tendency of these
factors has not changed over time.
E-67
-------
Avg Monthly Rainfall (in)
Avg Observed Flow (10/1/1982 to 9/30/1992 )
Avg Modeled Flow (Same Period)
18000
16000
_ 14000
.(/) -i onnn
7" 10000
Q: 8000
6000
4000
2000
I'll » I
1-
r, j" •
jttltr
1
"
tF,
"
±f.
} ] i
j j
1 ~[ 1
ii tirH
Oct-82 Apr-84 Oct-85 Apr-87 Oct-88
m "•) |1
j-L-
rr
if
li
" r "
,t
-2
-4 1
603
•E
-8 |
-10 f
- 12
Apr-90 Oct-91
Date
Figure 51. Mean daily flow at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(SWAT).
1500
i
I
1000 -
500 - -
O-82
A-84
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1982 to 9/30/1992
•Avg Modeled Flow (Same Period)
O-85
A-87
Month
O-88
A-90
O-91
)
10
9
8
7
6
5
4
3
2
1
0
I
'S
a:
Figure 52. Mean monthly flow at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(SWAT).
E-68
-------
Avg Flow (10/1/1982 to 9/30/1992 )
• Line of Equal Value
Best-Fit Line
1500
I
JD
LJ- 1000--
O)
TO
y = 0.4995x + 86.113
R2 = 0.368f
500 - - * -.'--
0 500 1000 1500
Average Observed Flow (cfs)
100%
Avg Observed Flow (10/1/1982 to 9/30/1992 )
Avg Modeled Flow (10/1/1982 to 9/30/1992 )
-Line of Equal Value
en
O
8
ro
m
o>
to
A-84 O-85
A-87 O-88
Month
A-90 O-91
Figure 53. Monthly flow regression and temporal variation at USGS 09504000 Verde River near
Clarkdale, AZ - validation period (SWAT).
Best-Fit Line
&
Q
400
o
rr
-n ^nn
-------
• Observed (25th, 75th)
-Median Observed Flow (10/1/1982 to 9/30/1992)
Average Monthly Rainfall (in)
Modeled (Median, 25th, 75th)
600
500
o
10 11 12 1
Figure 55. Seasonal medians and ranges at USGS 09504000 Verde River near Clarkdale, AZ -
validation period (SWAT).
Table 24. Seasonal summary at USGS 09504000 Verde River near Clarkdale, AZ - validation
period (SWAT)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
101.69
129.11
160.55
97.85
232.11
422.76
206.73
86.31
80.68
89.01
104.45
146.81
85.00
87.00
89.00
89.00
99.00
175.50
88.00
83.00
79.00
81.00
83.00
83.00
79.00
84.00
85.00
86.00
85.00
87.00
83.00
80.00
76.75
77.00
79.00
79.00
88.00
89.00
98.00
94.75
251 .00
510.50
113.00
88.00
85.00
86.75
93.00
89.00
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
133.41
177.41
189.36
148.63
201.41
209.87
131.86
106.03
91.66
146.37
194.00
230.93
95.42
102.19
113.19
115.42
112.81
125.61
101.34
89.26
80.87
102.31
123.61
100.94
80.47
85.14
93.62
103.86
98.46
94.45
88.26
80.14
72.39
82.18
91.04
84.97
115.23
132.23
145.09
146.28
161.08
179.31
122.62
104.86
92.45
143.73
185.62
121.65
E-70
-------
o
0
D)
0
ro
Q
•Observed Flow Duration (10/1/1982 to 9/30/1992 )
•Modeled Flow Duration (10/1/1982 to 9/30/1992 )
100000
10000
1000
100
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 56. Flow exceedance at USGS 09504000 Verde River near Clarkdale, AZ - validation period
(SWAT).
o
o
in
ro
T3
0
0
-------
Table 25. Summary statistics at USGS 09504000 Verde River near Clarkdale, AZ - validation
period (SWAT)
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 10
10-Year Analysis F^riod: 10/1/1982 - 9/30/1992
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe CoefficjejTtjDfJEfficiency^E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
0.63
0.26
0.16
0.19
0.16
0.18
0.11
0.26
0.10
Error Statistics
5.68
5.17
-9.70
68.04
Observed Flow Gage
USGS 09504000 VERDE RIVER NEAR CLARKDALE, AZ
Hydrologic Unit Code: 15060202
Latitude: 34.852241 6
Longitude: -11 2.065994
Drainage Area (sq-rri): 3503
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
27.70 » | 30
-25.96
-11.56
12.44
197.67
-0.996
0.121
0.320
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
0.60
0.29
0.16
0.11
0.13
0.24
0.12
0.23
0.03
Clear [
Hydrology Results for Larger Watershed
As described above, parameters determined through calibration to the Verde River near Clarkdale gage were not
fully transferable to other gages in the watershed. Therefore, calibration was pursued at a total of 10 gages in the
Arizona basins - six in the Verde River watershed, one in the Salt River watershed, and two in the San Pedro
River watersheds. Only the gage at Roosevelt provides a long period of record for the Salt River watershed. The
mainstem gage for the San Pedro River does not provide a rigorous calibration test because its flow is largely
determined by the upstream boundary condition. Therefore, calibration was also pursued on perennial Aravaipa
Creek. The two San Pedro gages ceased operation in 1995; therefore calibration was pursued over an earlier time
period without a separate validation test.
Calibration results at all gages are summarized in Table 26 and are generally of similar quality to the fit obtained
on the Verde River near Clarkdale. The fit for the East Verde River is believed to be relatively poor due to the
influences of Payson water withdrawals. The generally close match between observed and predicted flow at the
Salt River gage is shown in Figures 58 through 64 and Tables 27 and 28. Results of the validation exercise are
summarized in Table 29. In general, the quality of fit during the validation period is similar to that in the
calibration period. Thus, the model is judged to be useful for scenario evaluation.
E-72
-------
Table 26. Summary statistics (percent error) at all stations - calibration period WY 1992-2002
(SWAT)
Station
Error in
total
volume:
Error in
50%
lowest
flows:
Error in
10%
highest
flows:
Seasonal
volume
error -
Summer:
Seasonal
volume
error -
Fall:
Seasonal
volume
error -
Winter:
Seasonal
volume
error -
Spring:
Error in
storm
volumes:
Error in
summer
storm
volumes:
Daily
Nash-
Sutcliffe
Coefficient
of
Efficiency,
E:
Monthly
Nash-
Sutcliffe
Coefficient
of
Efficiency,
E:
09472000
San
Pedro nr
Redington
(1972-95)*
-6.74
NA*
-7.97
-4.82
8.09
-20.25
-7.04
1.49
-3.89
-0.362
0.690
09473000
Aravaipa
Crknr
Mammoth
(1972-95)
3.46
-1.25
2.62
53.40
-3.31
-10.77
7.82
-27.95
73.05
0.629
0.717
09498500
Salt River
nr
Roosevelt
9.43
-7.16
4.52
58.85
103.52
6.95
-51.64
16.40
103.32
0.222
0.783
09503700
Verde
River nr
Paulden
9.15
-1.14
8.65
110.89
56.82
-24.75
-1.39
19.48
875.65
-0.864
0.777
09504000
Verde
River nr
Clarkdale
-2.46
-1.74
-19.34
89.89
46.00
-36.45
3.88
-3.32
593.89
0.030
0.320
09504500
Oak
Creek nr
Cornville
-2.63
-17.90
-10.31
96.08
27.01
-22.05
-4.10
-10.98
205.71
0.454
0.728
09505800
W Clear
Crnr
Camp
Verde
9.45
-10.22
7.29
49.24
26.17
2.87
2.23
21.24
96.30
-0.021
0.804
09506000
Verde
River nr
Camp
Verde
7.68
8.49
-5.00
100.98
44.07
-13.37
3.38
9.94
153.62
0.225
0.880
09507980
E Verde
River nr
Childs
-6.21
56.51
-6.56
156.62
58.54
-35.02
0.08
17.92
338.25
0.311
0.736
09508500
Verde R
below
Tangle
Cr
1.68
-17.85
-7.27
96.88
47.60
-19.30
-7.60
16.89
262.34
0.250
0.860
! Note that median flow for the San Pedro River nr Redington is 0.
E-73
-------
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002 )
•Avg Modeled Flow (Same Period)
t
o
100000
90000
80000
70000
60000
50000
40000
30000
20000
10000
0
IJT' '
I
fi
|l
u
u
jfpt
,l
1 1 1
['1 1 I1
I1
ill
1
,
J
1
1
i
.1 tf J
1*
1
i
• I
t
JLL -i
n"l T
| J
t f
—
j
illl
- 2
8
10 «
12
14
Oct-92 Apr-94 Oct-95 Apr-97 Oct-98
Date
Apr-00
Oct-01
Figure 58. Mean daily flow at USGS 09498500 Salt River near Roosevelt, AZ - calibration period
(SWAT).
i
o
15000
10000 -
5000 - -
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002 )
Avg Modeled Flow (Same Period)
O-92
A-94
O-01
Figure 59. Mean monthly flow at USGS 09498500 Salt River near Roosevelt, AZ - calibration period
(SWAT).
E-74
-------
Avg Flow (10/1/1992 to 9/30/2002 )
•Line of Equal Value
Best-Fit Line
15000
I
o
y = 0.8091 x +234.63
R2 = 0.786
10000
5000
O)
TO
100%
Avg Observed Flow (10/1/1992 to 9/30/2002 )
Avg Modeled Flow (10/1/1992 to 9/30/2002 )
-Line of Equal Value
en
JD
O
ro
m
_
0)
to
0 5000 10000 15000
Average Observed Flow (cfs)
O-92 A-94 O-95
A-97 O-98
Month
A-00 O-01
Figure 60. Monthly flow regression and temporal variation at USGS 09498500 Salt River near
Roosevelt, AZ - calibration period (SWAT).
Avg Flow (10/1/1992 to 9/30/2002)
• Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002)
Avg Modeled Flow (Same Period)
2500
I
'2000
-a 1500 -
1000 -
0
D)
ro
500 -
2500 -i
500 1000 1500 2000
Average Observed Flow (cfs)
2500
10 11 12 1 23456789
Month
Figure 61. Seasonal regression and temporal aggregate at USGS 09498500 Salt River near Roosevelt,
AZ - calibration period (SWAT).
E-75
-------
• Observed (25th, 75th)
-Median Observed Flow (10/1/1992 to 9/30/2002)
Average Monthly Rainfall (in)
Modeled (Median, 25th, 75th)
3000
2500
2000
1500
1000
500
10
11
^ J-U
12
1
10 11
12
234
Month
0
1
2 ^
3 5
4 £
c
o
5 S
Figure 62. Seasonal medians and ranges at USGS 09498500 Salt River near Roosevelt, AZ -
calibration period (SWAT).
Table 27.Seasonal summary at USGS 09498500 Salt River near Roosevelt, AZ - calibration
period (SWAT)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
256.42
450.48
507.61
1831.48
1511.59
1875.51
1429.55
768.86
277.08
226.12
377.92
368.88
200.50
243.00
267.50
267.00
455.50
1080.00
953.50
535.00
187.50
185.50
267.50
249.50
168.25
195.00
210.00
204.00
206.00
211.50
224.00
150.25
121.00
129.25
205.50
174.00
264.00
352.75
356.25
413.25
1135.00
2682.50
1722.50
1047.50
335.25
269.50
389.50
402.25
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
608.17
791 .68
1067.82
2253.82
1755.74
1584.27
683.25
324.01
191.56
275.63
628.63
643.08
344.34
218.51
403.31
489.18
642.09
794.67
331.92
236.85
199.95
180.78
216.21
293.31
198.53
204.55
196.21
188.76
204.73
261.51
207.45
184.00
141.41
113.41
140.01
144.25
657.22
804.13
1094.30
1909.69
2372.74
2058.55
998.65
380.12
225.64
207.08
844.51
699.35
E-76
-------
I
ro
Q
•Observed Flow Duration (10/1/1992 to 9/30/2002 )
Modeled Flow Duration (10/1/1992 to 9/30/2002 )
100000
10000
1000
100 - =
10%
20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90%
100%
Figure 63. Flow exceedence at USGS 09498500 Salt River near Roosevelt, AZ - calibration period
(SWAT).
o
o
ro
T3
(D
-------
Table 28. Summary statistics at USGS 09498500 Salt River near Roosevelt, AZ - calibration
period (SWAT)
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 3
10-Year Analysis F^riod: 10/1/1992 - 9/30/2002
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
_Jotak3fj3irrujlatejdjT^^
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9)
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3^:
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring;
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
2.83
1.63
0.27
0.41
0.65
1.46
0.31
1.09
0.18
Error Statistics
9.43
-7.16
4.52
58.85
Observed Flow Gage
USGS 09498500 SALT RIVER NEAR ROOSEVELT, AZ
Hydrologic Unit Code: 15060103
Latitude: 33.61 94949
Longitude: -110.9215037
Drainage Area (sq-rri): 4306
Total Observed In-stream Flow:
Total of Observed higjiesMO%Jows:_^
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
103.52 » | 30
6.95
-51.64
16.40
103.32
0.222
0.314
0.783
30
30
20
50
Model accuracy increases
as E or E' approaches 1 .0
Run (n-1)
2.65
3.27
-3.14
73.05
2.59
1.56
0.29
0.26
0.32
1.36
0.65
0.94
0.09
Run (n-2)
5.65
13.54
-2.75
86.02
77.83 Clear [ 91.08
0.17
-57.34
5.26
110.10
0.363
0.335
0.25
-57.27
5.60
112.10
0.360
0.317
E-78
-------
Table 29. Summary statistics at all stations - validation period WY 1982-1992 (SWAT)
Station
Error in total
volume:
Error in 50%
lowest flows:
Error in 10%
highest flows:
Seasonal
volume error -
Summer:
Seasonal
volume error -
Fall:
Seasonal
volume error -
Winter:
Seasonal
volume error -
Spring:
Error in storm
volumes:
Error in
summer storm
volumes:
Daily Nash-
Sutcliffe
Coefficient of
Efficiency, E:
Monthly Nash-
Sutcliffe
Coefficient of
Efficiency, E:
09498500 Salt
River nr
Roosevelt
-1.76
-18.12
-2.86
58.01
54.85
-0.60
-60.95
6.42
99.23
-0.072
0.446
09503700
Verde River
nr Paulden
38.73
-2.09
78.10
50.18
67.27
31.00
-1.16
108.63
107.87
-2.028
-0.034
09504000
Verde River
nr Clarkdale
5.68
5.17
-9.70
68.04
27.70
-25.96
-11.56
12.44
197.67
-0.996
0.320
09504500
Oak Creek nr
Cornville
15.80
13.08
-3.77
197.59
24.82
-16.44
-8.90
18.06
370.81
-0.061
0.395
09505800 W
ClearCr nr
Camp Verde
-0.57
-28.71
-0.02
18.18
-10.18
11.25
-33.19
34.57
-9.73
-1.666
-0.161
09507980 E
Verde River
nrChilds
23.55
1.12
53.33
141.05
45.31
-6.66
-18.58
91.61
310.39
-1.193
0.736
09508500
Verde R
below Tangle
Cr
7.96
11.91
-2.37
157.67
17.43
-17.53
-30.60
19.33
332.99
-0.191
0.441
Water Quality Calibration and Validation
The 20 Watershed models are designed to provide water quality simulation for total suspended solids (TSS), total
nitrogen, and total phosphorus. The water quality calibration focuses on the replication of monthly loads, as
specified in the project QAPP. Given the simplified approach to water quality simulation in the 20 Watershed
model a close match to individual concentration observations cannot be expected. However, comparison to
monthly loads presents challenges, as monthly loads are not observed. Instead, monthly loads must be estimated
from scattered concentration grab samples and continuous flow records. Such estimation is fraught with
uncertainty because it depends on the degree and form in which concentration and flow are correlated with one
another. Further, the bulk of the load of sediment and sediment-associated phosphorus is likely to move through
the system in a limited number of high flow events that typically are not monitored. As a result, the monthly load
calibration is inevitably based on the comparison of two uncertain numbers. Nonetheless, calibration is able to
achieve a fair agreement. The load comparisons were supported by detailed examinations of the relationships of
E-79
-------
flows to loads and concentrations and the distribution of concentration prediction errors versus flow, time, and
season, as well as standard time series plots.
For application on a nationwide basis, the 20 Watershed protocols assume that sediment and phosphorus loads
will likely exhibit a strong positive correlation to flow (and associated erosive processes), while total nitrogen
loads, which often have a dominant groundwater component, will not. Accordingly, sediment and phosphorus
loads were estimated from observations using a flow-stratified log-log regression approach, while total nitrogen
loads were estimated using a flow-stratified averaging estimator, consistent with the findings of Preston et al.
(1989).
Initial calibration and validation of water quality was done on the Verde River near Clarkdale, using 1993-2002
for calibration and 1986-1992 for validation. As with hydrology, water quality calibration was performed on the
later period as this belter reflects the land use included in the model. The start of the validation period was
constrained by data availability.
Sediment concentrations in larger, higher-order alluvial streams are largely determined by channel scour,
deposition, and transport capacity rather than by upland sediment load. The SWAT representation of these
channel processes is rather simplistic. First, the maximum transport capacity concentration (Cmx) is determined as
C^ = SPCON • VpkSPEXP, where Vpk is the peak velocity, estimated by a simple ratio to the average rate of flow,
and SPCON and SPEXP are user-defined parameters. When the predicted sediment concentration in the reach
exceeds Cmx the excess is assumed to settle out. If the predicted sediment concentration in the reach is less than
C^, additional sediment may be scoured from the channel to make up the difference, depending on the channel
erodibility factor, Kch (cm/hr/Pa). There is no provision for the different transport characteristics of different
sediment size fractions. Further, the SPCON and SPEXP parameters are specified at the global level and do not
vary by reach. As a result, the ability of SWAT to match individual TSS observations is rather limited.
By judicious adjustment of the SPCON, SPEXP, and Kch parameters combined with the SWAT default MUSLE
representation of upland sediment yield a reasonable representation of sediment load can be obtained. Time series
of simulated and estimated sediment loads at the Clarkdale station for both periods are shown in Figure 65 and
statistics for the two periods are provided separately in Table 30. The key statistic in the table is the relative
percent error, which shows the error in the prediction of monthly load normalized to the estimated load. The table
also shows the relative average absolute error, which is the average of the relative magnitude of errors in
individual monthly load predictions. This number is inflated by outlier months in which the simulated and
estimated loads differ by large amounts (which may be as easily due to uncertainty in the estimated load due to
limited data as to problems with the model) and the third statistic, the relative median absolute error, is likely
more relevant and shows good agreement.
E-80
-------
TSS
1,000,000
-Regression Loads
-Simulated Loads
Figure 65. Fit for monthly load of TSS at USGS 09504000 Verde River near Clarkdale, AZ (SWAT).
Table 30. Model fit statistics (observed minus predicted) for monthly sediment loads using
stratified regression at USGS 09504000 Verde River near Clarkdale, AZ (SWAT)
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1993-2002)
16.9%
64%
0.5%
Validation period
(1986-1992)
-42.6%
122%
0.8%
A variety of other diagnostics were also examined to evaluate agreement between the model and observations.
These are available in full in the calibration spreadsheets, but a few examples are provided below. First, load-flow
power plots were compared for individual days (Figures 66 and 67). These show that the relationship between
flow and load is reasonably consistent across the entire range of observed flows, for both the calibration and
validation periods.
E-81
-------
Verde River - Clarkdale
1993-2002
•o
«
c
o
•o
ra
o
(0
(0
10
100 1000
Flow, cfs
10000
100000
• Simulated A Observed
Row er (Simulated)
Fbwer (Observed)
Figure 66. Power plot for observed and simulated TSS at USGS 09504000 Verde River near Clarkdale,
AZ - calibration period (SWAT).
Verde River - Clarkdale
1986-1992
1
•a
ra
o
w
100000
10000
1000
1 00
10000
Simulated A Observed
Fbwer (Simulated)
Power (Observed)
Figure 67. Power plot for observed and simulated TSS at USGS 09504000 Verde River near Clarkdale,
AZ - validation period (SWAT).
E-82
-------
Standard time series plots (Figure 68) show that observed and simulated concentrations achieve at best a fair
agreement, and the model may deviate substantially from individual observations. However, statistics on
concentration (0) show that low median errors are achieved (Table 31).
Verde River - Clarkdale
1993-2002
• Simulated A Observed
10000
1000
100
(0
(0
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Year
Figure 68. Time series plot of TSS concentration at USGS 09504000 Verde River near Clarkdale, AZ
(SWAT).
Table 31. Relative errors (observed minus predicted) for TSS concentration at USGS 09504000
Verde River near Clarkdale, AZ (SWAT)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1993-2002)
47
-22%
3.3%
Validation period
(1986-1992)
62
32%
9.3%
For simulation of total phosphorus and total nitrogen, calibration was advanced primarily through adjustment of
the PPERCO and NPERCO coefficients. Monthly loading series for total phosphorus are shown in Figure 69 and
load statistics are summarized in Table 32. The model reproduces the general trend in monthly loads, but is
significantly lower than the peak loads predicted by the regression method, resulting in high relative percent
errors for both the calibration and validation periods. It should be noted that the available data are limited,
particularly for high flow events. Thus, the estimates of "observed" load are also subject to considerable
uncertainty.
E-83
-------
Total P
10000
Regression Loads
Simulated Loads
_ 8 8 _
CM CM CM CM
Figure 69. Fit for monthly load of total phosphorus at USGS 09504000 Verde River near Clarkdale, AZ
(SWAT).
Table 32. Model fit statistics (observed minus predicted) for monthly total phosphorus loads
using stratified regression (SWAT)
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1993-2002)
83.5%
93%
0.3%
Validation period
(1986-1992)
31.4%
83%
13.1%
As with TSS, additional diagnostics for total phosphorus included flow-load power plots (Figures 70 and 71),
time series plots (Figure 72) and analysis of concentration errors (Table 33). While these show approximate
agreement, the model often overpredicts total phosphorus concentrations under lower flow conditions, although
not on a consistent basis.
E-84
-------
Verde River - Clarkdale
1993-2002
100
ra
•o
«
c
o
•o
ra
o
0.001
0.0001
10
100 1000
Flow, cfs
10000 100000
• Simulated A Observed ^^™Power (Simulated) ^^™Power (Observed)
Figure 70. Power Plot for Observed and Simulated total phosphorus at USGS 09504000 Verde River
near Clarkdale, AZ - calibration period (SWAT).
Verde River - Clarkdale
1986-1992
100
10000
• Simulated A Observed
Power (Simulated)
Power (Observed)
Figure 71. Power plot for observed and simulated total phosphorus at USGS 09504000 Verde River
near Clarkdale, AZ - validation period (SWAT).
E-85
-------
Verde River - Clarkdale
1993-2002
• Simulated A Observed
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Year
Figure 72. Time series plot of total phosphorus concentration at USGS 09504000 Verde River near
Clarkdale, AZ (SWAT).
Table 33. Relative errors (observed minus predicted) for total phosphorus concentration at
USGS 09504000 Verde River near Clarkdale, AZ (SWAT)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1993-2002)
57
-11%
-33%
Validation period
(1986-1992)
75
31%
-17%
For total nitrogen, fewer data are available because many sampling events omitted one or more nitrogen species.
This increases the uncertainty of the comparison. Results for total nitrogen are summarized in Figures 73 through
76 and Tables 34 and 35, following the same format as total phosphorus. The loading results are acceptable, and
generally better than those obtained for total phosphorus. However, there is significant uncertainty in the
prediction of individual nitrogen observations.
E-86
-------
Total N
120
Averaging Loads
Simulated Loads
COCOCOCOCOCO
O5O5O5O5O5O5O5O5O5O5O5O5O5O5O5O5O5O5O5
0000
O O O O
CM CM CM CM
Figure 73. Fit for monthly load of total nitrogen at USGS 09504000 Verde River near Clarkdale, AZ
(SWAT).
Table 34. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator (SWAT)
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1993-2002)
-14.4%
84%
13%
Validation period
(1986-1992)
-15.9%
67%
48%
E-87
-------
Verde River - Clarkdale
1993-2002
100
ra
•o
«
c
o
•o
ra
o
0.1
0.01
10
100 1000
Flow, cfs
10000
100000
• Simulated A Observed ^^™Power (Simulated) ^^™Power (Observed)
Figure 74. Power plot for observed and simulated total nitrogen at USGS 09504000 Verde River near
Clarkdale, AZ - calibration period (SWAT).
Verde River - Clarkdale
1986-1992
0.01
10000
• Simulated A Observed ^^™Power (Simulated) ^^"Fbwer (Observed)
Figure 75. Power plot for observed and simulated total nitrogen at USGS 09504000 Verde River near
Clarkdale, AZ - validation period (SWAT).
E-88
-------
2.5 -i
Verde River - Clarkdale
1993-2002
• Simulated A Observed
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Year
Figure 76. Time series plot of total nitrogen concentration at USGS 09504000 Verde River near
Clarkdale, AZ (SWAT).
Table 35. Relative errors (observed minus predicted) for total nitrogen concentration at USGS
09504000 Verde River near Clarkdale, AZ (SWAT)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1993-2002)
46
-83%
-90%
Validation period
(1986-1992)
75
1.3%
-18%
Water Quality Results for Larger Watershed
Summary statistics for the water quality calibration and validation at other stations in the watershed are provided
in Tables 36 and 37, respectively. In most cases, nitrogen loads are better predicted than phosphorus and TSS
loads. In a majority of cases simulated TSS and total phosphorus loads are lower than those estimated from
observations, but this may reflect in part the uncertainty in the regression-based load estimates as water quality
observations during high flows are sparse.
E-89
-------
Table 36. Summary statistics (observed minus predicted) for water quality at all stations
calibration period 1993-2002 (SWAT)
Station
Relative Percent Error TSS
Load
TSS Concentration Median
Percent Error
Relative Percent Error TP
Load
TP Concentration Median
Percent Error
Relative Percent Error TN
Load
TN Concentration Median
Percent Error
09498500 Salt River
nr Roosevelt
41.1
-15
61.0
-2
9.5
-4
09504000 Verde River
nrClarkdale
16.9
3.4
83.5
-33
-14.4
-90
09508500 Verde River
below Tangle Cr
17.7
2.0
33.8
-19.9
17.0
-60
Table 37. Summary statistics (observed minus predicted) for water quality at all stations
validation period 1986-1992 (SWAT)
Station
Relative Percent Error TSS
Load
TSS Concentration Median
Percent Error
Relative Percent Error TP
Load
TP Concentration Median
Percent Error
Relative Percent Error TN
Load
TN Concentration Median
Percent Error
09498500 Salt River
nr Roosevelt
-0.6
-16
54.2
-18
18.0
-36
09504000 Verde River
nrClarkdale
-42.6
9.3
31.4
-17
-15.9
-18
09508500 Verde River
below Tangle Cr
-55.1
0.35
-41.4
-5.8
-20.3
-13
E-90
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
ADWR (Arizona Department of Water Resources). 2000. Verde River Watershed Study. Arizona Department of
Water Resources. Phoenix, AZ. https://portal.azoah.com/oedf/documents/08A-AWS001-
DWR/Omnia/200004%20DWR%20Verde%20River%20Watershed%20Report.pdf
ADWR (Arizona Department of Water Resources). 2009. Arizona Water Atlas. Arizona Department of Water
Resources. Phoenix, AZ. Published online at
http://www.azwater.gov/AzDWR/StatewidePlanningAVaterAtlas/default.htm.
Anderson, J. R., E.E. Hardy, J.T. Roach, and R.E. Witmer. 1976. A Land Use and Land Cover Classification
System for Use with Remote Sensor Data. Professional Paper 964, U.S. Geological Survey. Reston, VA.
Cordy, G.E., D.J. Gellenbeck, J.B. Gebler, D.W. Anning, A.L. Goes, R.J. Edmonds, J. A.H. Rees, and H.W.
Sanger. 2000. Water Quality in the Arizona Basins, Arizona, 1995-98. U.S. Geological Survey Circular 1213.
Reston, VA.
Owen-Joyce, S.J., and C. K. Bell. 1983. Appraisal of Water Resources in the Upper Verde River Area: Yapai and
Coconino Counties, Arizona. Arizona Department of Water Resources Bulletin 2. Prepared by U.S. Geological
Survey. Phoenix, Arizona.
Preston, S.D., V.J. Bierman, Jr., and S.E. Silliman. 1989. An evaluation of methods for the estimation of tributary
mass loads. Water Resources Research, 25(6): 1379-1389.
Schwab, K.J. 1995. Maps Showing Groundwater Conditions in the Big Chino Sub-Basin of the Verde River
Basin, Coconino and Yavapai Counties, Arizona—1992. Arizona Department of Water Resources. Phoenix,
Arizona.
SCS (Soil Conservation Service). 1986. Urban Hydrology for Small Watersheds. Technical Release 55 (2nd
edition). U.S. Department of Agriculture. Washington, DC.
Tetra Tech. 2001. Verde River Assimilative Capacity Study, Modeling and Assimilative Capacity Report.
Prepared for Arizona Department of Environmental Quality. Tetra Tech, Inc. Research Triangle Park, NC.
USEPA (United States Environmental Protection Agency). 2008. Using the BASINS Meteorological Database
(Version 2006). BASINS Technical Note 10. Office of Water, U.S. Environmental Protection Agency.
Washington, DC. http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf
(Accessed June, 2009).
Wallace, B.L. and R. L. Laney. 1976. Maps Showing Ground-water Conditions in Lower Big Chino Valley and
Williamson Valley Areas, Yavapai and Coconino Counties, Arizona, 1975 - 1976. Water Resources
Investigations 76 - 78. U.S. Geological Survey. Washington, DC.
E-91
-------
Appendix F
Model Configuration, Calibration and
Validation
Basin: Susquehanna River
F-l
-------
Contents
Watershed Background F-7
Water Body Characteristics F-7
Soil Characteristics F-10
Land Use Representation F-10
Point Sources F-14
Meteorological Data F-20
Watershed Segmentation F-24
Calibration Data and Locations F-26
HSPF Modeling F-27
Changes Made to Base Data Provided F-28
Assumptions F-28
Hydrology Calibration F-28
Hydrology Validation F-34
Hydrology Results for Larger Watershed F-39
Water Quality Calibration and Validation F-46
Water Quality Results for Larger Watershed F-55
SWAT Modeling F-57
Changes Made to Base Data Provided F-57
Assumptions F-57
Hydrology Calibration F-57
Hydrology Validation F-63
Hydrology Results for Larger Watershed F-68
Water Quality Calibration and Validation F-75
Water Quality Results for Larger Watershed F-84
References F-86
F-2
-------
Tables
Table 1. Aggregation ofNLCD land cover classes F-ll
Table 2. Land use distribution for the Susquehanna River watershed (2001 NLCD, mi2) F-13
Table 3. Major point source discharges in the Susquehanna River watershed F-14
Table 4. Precipitation stations for the Susquehanna River watershed model F-20
Table 5. Calibration and validation locations in the Susquehanna River watershed F-26
Table 6. Seasonal summary at USGS 01562000 Raystown Branch Juniata River at Saxton, PA -
calibration period (HSPF) F-32
Figure 12. Flow accumulation at USGS 01562000 Raystown Branch Juniata River at Saxton,
PA - calibration period (HSPF) F-33
Table 7. Summary statistics at USGS 01562000 Raystown Branch Juniata River at Saxton, PA -
calibration period (HSPF) F-34
Table 8. Seasonal summary at USGS 01562000 Raystown Branch Juniata River at Saxton, PA -
validation period (HSPF) F-37
Table 9. Summary statistics at USGS 01562000 Raystown Branch Juniata River at Saxton, PA -
validation period (HSPF) F-39
Table 10. Seasonal summary at USGS 01576000 Susquehanna River at Marietta, PA - calibration
period (HSPF) F-42
Table 11. Summary statistics at USGS 01576000 Susquehanna River at Marietta, PA - calibration period
(HSPF) F-44
Table 12. Summary statistics (percent error) for all stations - calibration period (HSPF) F-45
Table 13. Summary statistics (percent error) for all stations - validation period (HSPF) F-46
Table 14. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression - USGS 01562000 Raystown Branch Juniata River at Saxton, PA (HSPF) F-47
Table 15. Relative errors,(observed minus simulated) for TSS concentrations at USGS 01562000
Raystown Branch Juniata River at Saxton, PA (HSPF) F-49
Table 16. Model fit statistics (observed minus predicted) for monthly total phosphorus loads using
stratified regression - USGS 01562000 Raystown Branch Juniata River at Saxton, PA (HSPF).. F-50
Table 17. Relative errors (observed minus simulated) for total phosphorus concentrations at USGS
01562000 Raystown Branch Juniata River at Saxton, PA (HSPF) F-52
Table 18. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using averaging
estimator - USGS 01562000 Raystown Branch Juniata River at Saxton, PA (HSPF) F-53
Table 19. Relative errors (observed minus simulated) for total nitrogen concentration at USGS 01562000
Raystown Branch Juniata River at Saxton, PA (HSPF) F-55
Table 20. Summary statistics for water quality for all stations - calibration period (HSPF) F-5 6
Table 21. Summary statistics for water quality for all stations - validation period (HSPF) F-5 6
Table 22. Seasonal summary at USGS 01562000 Raystown Branch Juniata River at Saxton, PA -
calibration period (SWAT) F-61
Table 23. Summary statistics at USGS 01562000 Raystown Branch Juniata River at Saxton, PA -
calibration period (SWAT) F-63
Table 24. Seasonal summary at USGS 01562000 Raystown Branch Juniata River at Saxton, PA -
validation period (SWAT) F-66
Table 25. Summary statistics at USGS 01562000 Raystown Branch Juniata River at Saxton, PA -
validation period (SWAT) F-68
Table 26. Summary statistics (percent error) for all stations - calibration period (SWAT) F-69
Table 27. Seasonal summary at USGS 01576000 Susquehanna River at Marietta, PA - calibration period
(SWAT) F-72
Table 28. Summary statistics at USGS 01576000 Susquehanna River at Marietta, PA - calibration period
(SWAT) F-74
Table 29. Summary statistics (percent error) for all stations - validation period (SWAT) F-75
Table 30. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression - USGS 01562000 Raystown Branch Juniata River at Saxton, PA (SWAT) F-76
~
-------
Table 31. Relative errors (observed minus simulated) for TS S concentrations at USGS 015 62000 Raystown
Branch Juniata River at Saxton, PA (SWAT) F-78
Table 32. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression - USGS 01562000 Raystown Branch Juniata River at Saxton, PA (SWAT) F-79
Table 33. Relative errors (observed minus predicted), total phosphorus concentration, USGS 01562000
Raystown Branch Juniata River at Saxton, PA (SWAT) F-81
Table 34. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using averaging
estimator - USGS 01562000 Raystown Branch Juniata River at Saxton, PA (SWAT) F-82
Table 35. Relative errors (observed minus predicted), total nitrogen concentration, USGS 01562000
Raystown Branch Juniata River at Saxton, PA (SWAT) F-84
Table 36. Summary statistics for water quality for all stations - calibration period (SWAT) F-85
Table 37. Summary statistics for water quality for all stations - validation period (SWAT) F-85
Figures
Figure 1. Location of the Susquehanna River watershed F-9
Figure 2. Land use in the Susquehanna River watershed F-12
Figure 3. Major point sources in the Susquehanna River watershed F-19
Figure 4. Weather stations forthe Susquehanna River watershed model F-23
Figure 5. Model segmentation and USGS stations utilized for the Susquehanna River watershed F-25
Figure 6. Mean daily flow at USGS 01562000 Raystown Branch Juniata River at Saxton, PA - calibration
period (HSPF) F-30
Figure 7. Mean monthly flow at USGS 01562000 Raystown Branch Juniata River at Saxton, PA -
calibration period (HSPF) F-30
Figure 8. Monthly flow regression and temporal variation at USGS 01562000 Raystown Branch Juniata River at
Saxton, PA - calibration period (HSPF) F-31
Figure 9. Seasonal regression and temporal aggregate at USGS 01562000 Raystown Branch Juniata River at
Saxton, PA - calibration period (HSPF) F-31
Figure 10. Seasonal medians and ranges at USGS 01562000 Raystown Branch Juniata River at Saxton, PA -
calibration period (HSPF) F-32
Figure 11. Flow exceedence at USGS 01562000 Raystown Branch Juniata River at Saxton, PA - calibration
period (HSPF) F-33
Figure 13. Mean daily flow at USGS 01562000 Raystown Branch Juniata River at Saxton, PA - validation
period (HSPF) F-35
Figure 14. Mean monthly flow at USGS 01562000 Raystown Branch Juniata River at Saxton, PA - validation
period (HSPF) F-35
Figure 15. Monthly flow regression and temporal variation at USGS 01562000 Raystown Branch Juniata
River at Saxton, PA - validation period (HSPF) F-36
Figure 16. Seasonal regression and temporal aggregate at USGS 01562000 Raystown Branch Juniata River at
Saxton, PA - validation period (HSPF) F-36
Figure 17. Seasonal medians and ranges at USGS 01562000 Raystown Branch Juniata River at Saxton, PA -
validation period (HSPF) F-37
Figure 18. Flow exceedence at USGS 01562000 Raystown Branch Juniata River at Saxton, PA - validation
period (HSPF) F-38
Figure 19. Flow accumulation at USGS 01562000 Raystown Branch Juniata River at Saxton, PA - validation
period (HSPF) F-38
Figure 20. Mean daily flow at USGS 01576000 Susquehanna River at Marietta, PA - calibration period
(HSPF) F-40
Figure 21. Mean monthly flow at USGS 01576000 Susquehanna River at Marietta, PA - calibration period
(HSPF) F-40
F-4
-------
Figure 22. Monthly flow regression and temporal variation at USGS 01576000 Susquehanna River at Marietta,
PA - calibration period (HSPF) F-41
Figure 23. Seasonal regression and temporal aggregate at USGS 01576000 Susquehanna River at Marietta,
PA - calibration period (HSPF) F-41
Figure 24. Seasonal medians and ranges at USGS 01576000 Susquehanna River at Marietta, PA - calibration
period (HSPF) F-42
Figure 25. Flow exceedence at USGS 01576000 Susquehanna River at Marietta, PA - calibration period
(HSPF) F-43
Figure 26. Flow accumulation at USGS 01576000 Susquehanna River at Marietta, PA - calibration period
(HSPF) F-43
Figure 27. Fit for monthly load of TSS at Raystown Branch Juniata River at Saxton, PA (HSPF) F-47
Figure 28. Power plot for observed and simulated TSS at Raystown Branch Juniata River at Saxton, PA -
calibration period (HSPF) F-48
Figure 29. Power plot for observed and simulated TSS at Raystown Branch Juniata River at Saxton, PA -
validation period (HSPF) F-48
Figure 30. Time series plot of TSS concentration at Raystown Branch Juniata River at Saxton, PA - calibration
period (HSPF) F-49
Figure 31. Fit for monthly load of total phosphorus at Raystown Branch Juniata River at Saxton, PA
(HSPF) F-50
Figure 32. Power plot for observed and simulated total phosphorus at Raystown Branch Juniata River at
Saxton, PA - calibration period (HSPF) F-51
Figure 33. Power plot for observed and simulated total phosphorus at Raystown Branch Juniata River at
Saxton, PA - validation period (HSPF) F-51
Figure 34. Time series plot of total phosphorus concentration, at Raystown Branch Juniata River at Saxton,
PA (HSPF) F-52
Figure 35. Fit for monthly load of total nitrogen at Raystown Branch Juniata River at Saxton, PA (HSPF) F-53
Figure 36. Power plot for observed and simulated total nitrogen at Raystown Branch Juniata River at Saxton,
PA - calibration period (HSPF) F-54
Figure 37. Power plot for observed and simulated total nitrogen at Raystown Branch Juniata River at Saxton,
PA - validation period (HSPF) F-54
Figure 38. Time series plot of total nitrogen concentration, at Raystown Branch Juniata River at Saxton, PA
(HSPF) F-55
Figure 39. Mean daily flow at USGS 01562000 Raystown Branch Juniata River at Saxton, PA - calibration
period (SWAT) F-59
Figure 40. Mean monthly flow at USGS 01562000 Raystown Branch Juniata River at Saxton, PA - calibration
period (SWAT) F-59
Figure 41. Monthly flow regression and temporal variation at USGS 01562000 Raystown Branch Juniata River
at Saxton, PA - calibration period (SWAT) F-60
Figure 42. Seasonal regression and temporal aggregate at USGS 01562000 Raystown Branch Juniata River at
Saxton, PA - calibration period (SWAT) F-60
Figure 43. Seasonal medians and ranges at USGS 01562000 Raystown Branch Juniata River at Saxton, PA -
calibration period (SWAT) F-61
Figure 44. Flow exceedence at USGS 01562000 Raystown Branch Juniata River at Saxton, PA - calibration
period (SWAT) F-62
Figure 45. Flow accumulation at USGS 01562000 Raystown Branch Juniata River at Saxton, PA - calibration
period (SWAT) F-62
Figure 46. Mean daily flow at USGS 01562000 Raystown Branch Juniata River at Saxton, PA - validation
period (SWAT) F-64
Figure 47. Mean monthly flow at USGS 01562000 Raystown Branch Juniata River at Saxton, PA - validation
period (SWAT) F-64
Figure 48. Monthly flow regression and temporal variation at USGS 01562000 Raystown Branch Juniata
River at Saxton, PA - validation period (SWAT) F-65
~
-------
Figure 49. Seasonal regression and temporal aggregate at USGS 01562000 Raystown Branch Juniata River at
Saxton, PA - validation period (SWAT) F-65
Figure 50. Seasonal medians and ranges at USGS 01562000 Raystown Branch Juniata River at Saxton, PA -
validation period (SWAT) F-66
Figure 51. Flow exceedence at USGS 01562000 Raystown Branch Juniata River at Saxton, PA - validation
period (SWAT) F-67
Figure 52. Flow accumulation at USGS 01562000 Raystown Branch Juniata River at Saxton, PA - validation
period (SWAT) F-67
Figure 53. Mean daily flow at USGS 01576000 Susquehanna River At Marietta, PA - calibration period
(SWAT) F-69
Figure 54. Mean monthly flow at USGS 01576000 Susquehanna River at Marietta, PA - calibration period
(SWAT) F-70
Figure 55. Monthly flow regression and temporal variation at USGS 01576000 Susquehanna River at Marietta,
PA - calibration period (SWAT) F-70
Figure 56. Seasonal regression and temporal aggregate at USGS 01576000 Susquehanna River at Marietta,
PA - calibration period (SWAT) F-71
Figure 57. Seasonal medians and ranges at USGS 01576000 Susquehanna River at Marietta, PA - calibration
period (SWAT) F-71
Figure 58. Flow exceedence at USGS 01576000 Susquehanna River at Marietta, PA - calibration period
(SWAT) F-72
Figure 59. Flow accumulation at USGS 01576000 Susquehanna River at Marietta, PA - calibration period
(SWAT) F-73
Figure 60. Fit for monthly load of TSS at USGS 01562000 Raystown Branch Juniata River at Saxton, PA
(SWAT) F-76
Figure 61. Power plot for observed and simulated TSS at USGS 01562000 Raystown Branch Juniata River at
Saxton, PA - calibration period (SWAT) F-77
Figure 62. Power plot for observed and simulated TSS at USGS 01562000 Raystown Branch Juniata River at
Saxton, PA - validation period (SWAT) F-77
Figure 63. Correlation between observed and predicted TSS concentration at USGS 01562000 Raystown
Branch Juniata River at Saxton, PA (SWAT) F-78
Figure 64. Fit for monthly load of total phosphorous at USGS 01562000 Raystown Branch Juniata River at
Saxton, PA (SWAT) F-79
Figure 65. Power plot for observed and simulated total phosphorus at USGS 01562000 Raystown Branch
Juniata River at Saxton, PA - calibration period (SWAT) F-80
Figure 66. Power plot for observed and simulated total phosphorus at USGS 01562000 Raystown Branch
Juniata River at Saxton, PA - validation period (SWAT) F-80
Figure 67. Time series plot of total phosphorus concentration at USGS 01562000 Raystown Branch Juniata
River at Saxton, PA (SWAT) F-81
Figure 68. Fit for monthly load of total nitrogen at USGS 01562000 Raystown Branch Juniata River at
Saxton, PA (SWAT) F-82
Figure 69. Power plot for observed and simulated total nitrogen at USGS 01562000 Raystown Branch Juniata
River at Saxton, PA - calibration period (SWAT) F-83
Figure 70. Power plot for observed and simulated total nitrogen at USGS 01562000 Raystown Branch Juniata
River at Saxton, PA - validation period (SWAT) F-83
Figure 71. Time series plot of total nitrogen concentration at USGS 01562000 Raystown Branch Juniata
River at Saxton, PA (SWAT) F-84
F-6
-------
Water Body Characteristics
The Susquehanna River drains about 27,500 mi2 in the states of New York, Pennsylvania, and Maryland and
includes a total of 19 HUCSs in HUC 2050 (Figure 1). The watershed makes up 43 percent of the Chesapeake
Bay's drainage area and consists of six major subwatersheds (Chemung, Upper Susquehanna, West Branch
Susquehanna, Middle Susquehanna, Juniata, and Lower Susquehanna). The Susquehanna River flows about 444
miles from its headwaters at Otsego Lake in Cooperstown, New York to Havre de Grace, Maryland, where the
river flows into the Chesapeake Bay. The river is the largest tributary to the Chesapeake Bay, providing 50
percent of its freshwater flows (SRBC 2008).
The Susquehanna River watershed includes three physiographic provinces: the Appalachian Plateau, the Valley
and Ridge, and the Piedmont Provinces. The Appalachian Plateau Province is characterized by high, flat-topped
hills and deep valleys cut by the Susquehanna River and its tributaries. The Valley and Ridge physiographic
province contains steep mountains and ridges separated by valleys. The Piedmont physiographic province consists
of uplands and lowlands. The Piedmont physiographic province generally has terrain that is gently rolling to hilly.
Sixty-nine percent of the watershed is forested. However, the well-drained areas with rolling hills and valleys in
the southern part of the watershed contain most of the population and some of the most productive agricultural
land in the US. Groundwater maintains the base flow of perennial streams during periods of little or no
precipitation and constitutes an average of 50 percent of the flow of most streams at other times.
The U.S. Army Corps of Engineers (USAGE) operates and maintains 13 dams and reservoirs that are located in
all six major subwatersheds. USAGE also regulates the operation of a state of Pennsylvania reservoir (George B.
Stevenson) in the West Branch Susquehanna subwatershed for the purpose of flood damage reduction. These 14
reservoirs provide most of the floodwater storage in the watershed. The Natural Resources Conservation Service
and the state of Pennsylvania have also constructed reservoirs in the watershed that reduce flood damages;
however, these reservoirs are typically smaller in scale than the USAGE reservoirs.
In addition to the many flood storage dams and reservoirs, there are 20 major electric power generating plants
located in the Susquehanna River watershed that use water resources in their operation. Many of these
hydroelectric dams are located in the lower Susquehanna watershed. Just below Harrisburg, Pennsylvania the
Susquehanna River flows through a series of gorges dammed by hydroelectric power facilities. There are also 13
approved water diversions from the Susquehanna River watershed.
The Susquehanna River basin has a continental type of climate. The average annual temperature in the basin
ranges from about 44 degrees in the northern part of the basin to about 53 degrees in the southern part. Average
annual precipitation is about 40 inches over the entire basin and ranges from 33 inches in the northern part of the
basin to 46 inches in the southern part. Virtually all the major streams experience their highest flows in March,
April, and May, when melting snows combine with spring rains. These three months account for about one-half of
the yearly runoff. Flows are lowest in these streams during the summer and early fall months, with most streams
falling to their lowest levels in September. The Susquehanna River basin is one of the country's most flood prone
areas. Generally, floods occur each year somewhere in the basin, and major floods can occur in all seasons of the
year, and a major flood occurs on average every 13 years.
Groundwater flow maintains the base flow of perennial streams during periods of little or no precipitation and
constitutes an average of 50 percent of the flow of most streams at other times. The use of groundwater resources
~
-------
in the basin is extensive. Groundwater plays a critical role in supplying drinking water and maintaining economic
viability. Outside of the major population centers, drinking water supplies are heavily dependent on groundwater
wells. Approximately 20 percent of the basin population is served by public water suppliers that use groundwater
as a source.
F-8
-------
Lake Ontario
•Upper Susquehanna
(2050101)
Chenango
(2050102)
o-Wappasem
(2050103)
Ghemung
2050105}
Upper Susquehanna-Tunkhannoc
(2050106)
me Pennsylvania/^
(2050205)
iddle West Branch I
Susquehanna
(2050203)*y ';
\ Uppey
Susquehanna-L&ckawanna
(2050,ltT7)
Sinnemahomng
(2050*202) " Lower West Branc
Susquehanna
" (2050206)
Bald Eagle
(2050204)
Lower Susquehanna
(2050301)
Pehrisvlvania
ower Juniata
(2050304)
Philadelphia
yrSusquehanna-Swatara
'(2050305)
Lower Susquehanna
(2050306)
Raystown
(2050303)
Hydrography
Water (Nat. Atlas Dasaseti
US Census Populated Places
Municipalities (pop a 50,000)
County Boundaries
Watershed with HUC8s
„.
Baltimore
GCRP Model Areas - Susquehanna River Basin
Base Map
Figure 1. Location of the Susquehanna River watershed.
F-9
-------
Soil Characteristics
One of the most important characteristics of soils for watershed modeling is their hydrologic soil group (HSG).
The 20 Watershed study utilized STATSGO soil survey HSG information during model set-up. Soils are
classified into four hydrologic groups (SCS 1986), separated by runoff potential, as follows:
Group A Soils Have low runoff potential and high infiltration rates even when thoroughly wetted. They
consist chiefly of deep, well to excessively drained sands or gravels and have a high rate
of water transmission.
Group B Soils Have moderate infiltration rates when wet and consist chiefly of soils that are moderately
deep to deep, moderately well to well drained, and moderately fine to moderately course
textures.
Group C Soils Have low infiltration rates when thoroughly wetted and consist chiefly of soils having a
layer that impedes downward movement of water with moderately fine to fine structure.
Group D Soils Have high runoff potential, very low infiltration rates and consist chiefly of clay soils
with high swelling potential, soils with a permanent water table, soils with a claypan or
clay layer at or near the surface and shallow soils over nearly impervious material.
The Susquehanna River watershed contains all four HSGs in the watershed. However, soils in the watershed, as
described in STATSGO soil surveys, fall primarily into HSGs B (moderately high infiltration capacity) and C
(low infiltration capacity). Hydrologic group C soils dominate the northern portion of the watershed while a
mixture of B and C soils dominate the southern portion of the watershed.
Land Use Representation
Land use/cover in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage and is
predominantly forested with some agricultural and developed land (Figure 2). Agriculture and pasture are more
predominant in the downstream, eastern portions of the watershed. Urban development is found throughout the
watershed; however, the major concentration is in the eastern portions of the watershed.
NLCD land cover classes were aggregated according to the scheme shown in Table 1 for representation in the 20
Watershed model, then overlain with the soils HSG grid. Pervious and impervious lands are specified separately
for HSPF, so only one developed pervious class is used, along with an impervious class. HSPF simulates
impervious land areas separately from pervious land. Impervious area distributions were also determined from the
NLCD Urban Impervious data coverage. Specifically, percent impervious area was calculated over the whole
basin for each of the four developed land use classes. These percentages were then used to separate out
impervious land. NLCD impervious area data products are known to underestimate total imperviousness in rural
areas. However, the model properly requires connected impervious area, not total impervious area, and the NLCD
tabulation is assumed to provide a reasonable approximation of connected impervious area. Different developed
land classes are specified separately in SWAT. In HSPF the WATER, BARREN, DEVPERV, and WETLAND
classes are not subdivided by HSG; SWAT uses the built-in HRU overlay mechanism in the ArcSWAT interface.
The distribution of land use in the watershed is summarized in Table 2.
F-10
-------
Table 1. Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area
usually accounted for as
reach area
Deciduous
Evergreen
Mixed
Emergent & woody
wetlands
Aquatic bed wetlands (not
emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
HSPF (after processing)
WATER
BARREN, Assume HSG D
DEVPERV;
IMPERV
BARREN (D)
FOREST (A,B,C,D)
SHRUB (A,B,C,D)
GRASS (A,B,C,D), BARREN (D)
GRASS(A,B,C,D)
AGRI (A,B,C,D)
WETLAND, Assume HSG D
WATER
F-ll
-------
Legend
— Hydrography
^^= Interstate
^H Water (Nat. Atlas Dataset)
^] County Boundaries
n Watershed
2001 NLCD Land Use
I Open water
^ Developed, open space
| Developed, low intensity
H Developed, medium intensity
I Developed, high intensity
^ Barren land
H Deciduous forest
| Evergreen forest
| | Mixed forest
] Scrub/shrub
^ Grassland/herbaceous
^ PastureAiay
^ Cultivated crops
| Woody wetlands
^ Emergent herbaceous wetlands
GCRP Model Areas - Susquenanna River Basin
Figure 2. Land use in the Susquehanna River watershed.
F-12
-------
Table 2. Land use distribution for the Susquehanna River watershed (2001 NLCD, mi2)
HUC8
watershed
02050101
02050102
02050103
02050104
02050105
02050106
02050107
02050201
02050202
02050203
02050204
02050205
02050206
02050301
02050302
02050303
02050304
02050305
02050306
Total
Open
water
29.7
13.2
7.8
7.3
9.2
24.3
32.4
10.6
1.4
5.6
4.4
2.4
14.1
32.8
3.3
17.3
12.4
33.5
51.5
313.2
Developed3
Open
space
89.0
56.8
45.3
47.4
55.7
88.2
117.5
90.3
12.1
12.8
45.9
16.9
80.6
87.1
54.4
51.5
72.7
139.6
109.8
1,273.6
Low
density
18.1
14.3
18.3
7.1
18.1
9.4
60.5
14.7
1.7
2.3
19.0
2.4
23.6
24.7
26.5
12.6
18.0
113.1
99.7
504.1
Medium
density
5.7
4.8
6.9
2.5
5.9
2.9
42.9
2.5
0.5
0.9
4.5
0.7
8.5
7.8
8.6
2.9
4.1
35.6
40.4
188.6
High
density
1.5
1.1
1.7
0.5
1.2
0.7
12.1
0.5
0.1
0.1
1.5
0.2
2.1
2.4
3.5
0.8
1.1
15.1
16.0
62.5
Barren
land
0.5
1.7
0.5
2.7
1.2
1.9
11.8
30.1
2.5
2.1
1.4
2.1
2.3
9.3
1.8
1.4
0.3
3.5
23.0
100.0
Forest
1,345.3
909.5
659.5
833.2
704.5
1,177.4
1,083.8
1,229.8
921.2
728.9
543.1
816.2
1,170.3
860.0
688.8
655.6
1,019.2
858.3
603.5
16,808.1
Shrubland
73.0
75.2
25.0
39.8
45.4
29.0
23.7
32.9
56.5
18.6
0.0
42.0
33.1
0.0
0.0
0.0
0.0
0.0
0.0
494.4
Pasture/Hay
459.3
319.1
200.8
299.3
196.1
340.7
230.2
147.0
27.9
21.1
64.8
68.0
282.9
219.9
101.9
140.8
185.0
391.9
1,015.3
4,712.1
Cultivated
163.0
146.5
56.5
137.3
152.5
314.0
126.1
36.2
2.8
7.7
70.0
25.9
184.5
201.7
102.2
78.8
137.2
273.5
489.1
2,705.6
Wetland
104.6
65.0
23.7
6.0
22.1
18.1
22.9
1.9
7.3
3.1
0.1
4.3
7.8
3.0
0.1
0.0
0.7
12.3
39.4
342.2
Total
2,289.7
1,607.3
1,046.1
1,383.2
1,211.8
2,006.7
1,764.0
1,596.7
1,033.8
803.2
754.6
981.0
1,809.9
1,448.7
991.1
961.7
1,450.7
1,876.5
2,487.5
27,504.3
aThe percent imperviousness applied to each of the developed land uses is as follows: open space (6.90%), low density (31.26%), medium density (60.90%), and high
density (85.41%).
F-13
-------
Point Sources
There are numerous point source discharges in the watershed. For the purposes of 20 Watershed modeling, only
the 147 major dischargers with a design flow greater than 1 MGD are included in the simulation (Table 3 and
Figure 3). The major dischargers account for the majority of the facilities, so the effect of the omitted sources
distributed throughout the watershed will be relatively small, except during extreme low flow conditions. The
major dischargers are represented at long-term average flows, without accounting for changes over time or
seasonal variations.
Data from 1991-2006 were compiled from the PCS database and the median total nitrogen, total phosphorus, and
TSS values were estimated. The facilities that were missing a total nitrogen, total phosphorus, and TSS
concentration value were filled with atypical pollutant concentration value from literature (Tetra Tech 1990)
based on the SIC classification. The median concentrations for the nutrient species were estimated based on the
values reported in the Chesapeake Bay Phase 5 Model documentation (USEPA 2010).
Table 3. Major point source discharges in the Susquehanna River watershed
NPDES ID
MD0002518
NY0003824
NY0003859
NY0003867
NY0003875
NY0004057
NY0004081
NY0004138
NY0004146
NY0004243
NY0020672
NY0021423
NY0021431
NY0022357
NY0022730
NY0023591
NY0023647
NY0023906
NY0024414
NY0025712
Name
EXELON POWER GENERATION
AEROSPACE OPERATIONS
AES HICKLING, LLC
AES-JENNISON, LLC
AES WESTOVER
SYSTEMS INTEGRATION - OWEGO
MOTOR COMPONENTS, LLC
OSG NORWICH PHARMACEUTICALS
WOODS CORNER PLANT
KERRY BIO-SCIENCE
HAMILTON (V)WPCP
NORWICH (C) WWTP
BATH (V) WWTP
ALFRED (V) WWTP
OWEGO (T) SD#1
COOPERSTOWN (V) STP
HORNELL(C)WPCP
ERWIN (T) STP
BINGHAMTON-JOHNSON (C) JNT STP
PAINTED POST (V) STP
Design flow
(MGD)
47.74
1.35
77.34
65.34
101.90
1.61
0.93
0.80
0.30
2.30
0.85
2.20
1.00
0.98
0.50
0.52
4.00
0.80
20.00
0.50
Observed flow
(MGD)
(1991 -2006 average)
0.00
0.22
1.10
0.03
0.44
0.62
3.00
0.08
0.07
0.39
0.53
2.14
0.74
0.60
0.67
0.65
2.94
0.64
22.20
0.25
F-14
-------
NPDES ID
NY0025721
NY0025798
NY0027561
NY0027669
NY0029262
NY0029271
NY0031151
NY0035742
NY0036986
PA0007498
PA0007919
PA0008231
PA0008265
PA0008281
PA0008303
PA0008419
PA0008443
PA0008451
PA0008508
PA0008575
PA0008869
PA0008885
PA0008923
PA0009024
PA0009164
PA0009202
PA0009229
PA0009253
PA0009270
PA0009733
PA0009920
PAOO 10031
Name
CORNING (C)STP
OWEGO WPCP #2
LE ROY R SUMMERSON WWTF
ENDICOTT (V) WPCP
OWEGO (V) STP
SIDNEY (V) WWTP
ONEONTA (C) WWTP
CHEMUNG CO ELMIRA SD STP
CHEMUNG CO SD#1 STP
WISE FOODS INC
CASCADES TISSUE GROUP - PA INC
GOLD MILLS INC
APPLETON PAPERS INC - SPRING M
PPL BRUNNER ISLAND LLC
ISG STEELTON LLC
MERCK & CO INC
PPL MONTOUR LLC
SUNBURY GENERATION LLC
BURLE BUSINESS PARK LP
WIREROPE WORKS INC
PH GLATFELTER CO
PROCTER & GAMBLE PRODUCTS CO
CORNING ASAHI VIDEO PROD CO
OSRAM SYLVANIA PRODUCTS INC
STANDARD STEEL LLC
CERRO METAL PRODUCTS CO
NORFOLK SOUTHERN RAILWAY CO -
UNITED DEFENSE LP
DEL MONTE CORP
EXELON GENERATION CO LLC - PEA
AMERGEN ENERGY CO LLC - THREE
RELIANT ENERGY MID-ATLANTIC PO
Design flow
(MGD)
2.13
2.00
10.00
10.00
1.00
1.70
4.00
12.00
9.50
0.59
1.25
2.00
4.84
621.00
27.60
12.20
0.46
3.38
0.32
0.07
13.70
7.60
1.97
1.22
1.45
0.23
0.50
0.03
0.67
0.05
81.02
0.01
Observed flow
(MGD)
(1991 -2006 average)
1.56
1.14
6.96
7.48
0.57
0.62
2.47
7.49
7.94
0.24
1.56
17.94
4.45
6.46
28.01
8.70
8.44
6.50
0.11
0.05
28.27
7.68
1.27
0.93
15.56
2.36
0.24
0.03
0.41
0.12
20.07
0.70
F-15
-------
NPDES ID
PAOO 10430
PAOO 13862
PA0020273
PA0020320
PA0020486
PA0020567
PA0020664
PA0020826
PA0020885
PA0020893
PA0020923
PA0021067
PA0021571
PA0021687
PA0021814
PA0021890
PA0022209
PA0022535
PA0023108
PA0023248
PA0023531
PA0023558
PA0023744
PA0024040
PA0024287
PA0024325
PA0024406
PA0024431
PA0024759
PA0024902
PA0025933
PA0026077
Name
HANOVER FOODS CORP
CORIXA CORP
MILTON REGIONAL SEW AUTH
LITITZSEWAUTH
BELLEFONTE BORO
NORTHUMBERLAND SEW AUTH
MIDDLETOWN BORO AUTH
DOVER TWP
MECHANICSBURG BORO
MANHEIM BORO
NEW OXFORD MUN AUTH
MOUNT JOY BORO AUTH
MARYSVILLE BORO COUNCIL
WELLSBORO MUN AUTH
MANSFIELD BORO MUN AUTH
NEW HOLLAND BORO
BEDFORD BORO MUN AUTH
MILLERSBURG AREA AUTH
ELIZABETHTOWN BORO
BERWICK AREA JNT SEW AUTH
DANVILLE BORO
ASHLAND BORO
NORTHEASTERN YORK CO SEW AUTH
HIGHSPIREBORO
PALMYRA BORO STP
MUNCY BORO MUN AUTH
MOUNT CARMEL MUN AUTH
DILLSBURG AREA AUTH
CURWENSVILLE MUN AUTH
UPPER ALLEN TWP BRD OF COMMRS
LOCK HAVEN CITY
CARLISLE BORO
Design flow
(MGD)
0.23
0.32
3.42
3.85
3.22
1.13
2.20
8.00
2.08
1.14
1.79
1.53
1.25
2.00
1.00
1.34
1.50
1.00
3.00
3.70
3.62
1.30
1.70
2.00
1.42
1.40
1.50
1.53
0.75
1.10
3.75
4.63
Observed flow
(MGD)
(1991 -2006 average)
0.18
0.18
78.54
2.62
2.21
11.68
1.32
3.71
1.05
6.29
1.11
0.78
0.80
1.07
0.53
0.96
0.79
0.47
2.10
1.38
2.13
12.19
0.73
1.11
0.83
0.76
1.01
0.68
0.41
5.81
2.29
3.34
F-16
-------
NPDES ID
PA0026107
PA0026123
PA0026191
PA0026239
PA0026263
PA0026280
PA0026310
PA0026361
PA0026441
PA0026484
PA0026492
PA0026557
PA0026620
PA0026654
PA0026727
PA0026735
PA0026743
PA0026808
PA0026875
PA0026921
PA0027014
PA0027022
PA0027049
PA0027057
PA0027065
PA0027090
PA0027171
PA0027189
PA0027197
PA0027316
PA0027324
PA0027405
Name
WYOMING VALLEY SAN AUTH
COLUMBIA MUN AUTH
HUNTINGTON BORO
UNIVERSITY AREA JOINT AUTH -
YORK CITY SEW AUTH
LEWISTOWN BORO
CLEARFIELD MUN AUTH
LOWER LACKAWANNA VLY SAN AUTH
LEMOYNE BORO
DERRY TWP MUN AUTH - CLEARWATE
SCRANTON CITY SEW AUTH
SUNBURY CITY MUN AUTH
MILLERSVILLE BORO
NEW CUMBERLAND BORO
TYRONE BORO
SWATARA TWP AUTH
LANCASTER CITY
SPRINGETTSBURY TWP
HANOVER BOROUGH
GREATER HAZELTON JNT SEW AUTH
ALTOONA CITY AUTH - EAST
ALTOONA CITY AUTH - WEST
WILLIAMSPORT SAN AUTH-WEST
WILLIAMSPORT SAN AUTH-CENTRAL
LACKAWANNA RIVER BASIN SEW AUT
LACKAWANNA RIVER BASIN SAN AUT
BLOOMSBURG MUN AUTH
LOWER ALLEN TWP AUTH
HARRISBURG AUTHORITY
LEBANON CITY
SHAMOKIN-COAL TWP JNT SEW AUTH
EPHRATA BORO AUTH - WWTP #1
Design flow
(MGD)
32.00
2.00
5.90
6.00
26.00
2.82
4.50
6.00
2.09
5.00
20.00
4.20
1.85
1.25
9.00
6.30
29.73
15.00
5.50
8.90
8.00
9.00
3.92
10.50
6.00
7.00
4.29
6.25
37.70
8.00
7.00
3.80
Observed flow
(MGD)
(1991 -2006 average)
24.07
0.80
2.50
4.62
10.66
1.70
3.46
3.36
1.72
3.48
13.03
3.40
0.65
3.98
5.97
3.71
19.72
10.04
4.16
7.20
6.77
8.03
2.89
7.38
2.78
5.28
2.58
4.38
25.06
5.31
3.74
2.56
F-17
-------
NPDES ID
PA0027553
PA0028142
PA0028461
PA0028576
PA0028665
PA0028681
PA0028746
PA0030643
PA0032883
PA0034576
PA0037150
PA0037966
PA0038415
PA0042269
PA0043257
PA0043273
PA0043681
PA0044661
PA0045985
PA0047325
PA0062219
PA0070041
PA0070386
PA0080314
PA0083011
PA0087181
PA01 10582
PA01 10965
PA0111759
PA0208779
PA0209228
PA0228818
Name
PINE CREEK MUN AUTH
PA NATIONAL GUARD - FORT INDIA
MIFFLINBURG BORO
CLARKS SUMMIT - S ABINGTON JSA
JERSEY SHORE BORO
KELLY TWP MUN AUTH
HAMPDEN TWP
SHIPPENSBURG BORO
DUNCANSVILLE BORO
TOWANDA MUN AUTH
PENN TWP BOARD OF COMMISSIONER
MOSHANNON VALLEY JT SEW AUTH
EAST PENNSBORO TWP
LANCASTER AREA SEW AUTH
NEW FREEDOM BORO AUTH
HOLLIDAYSBURG SEW AUTH
VALLEY JOINT SEW AUTH
LEWISBURG AREA JOINT SEW AUTH
MOUNTAINTOP AREA JNT SAN AUTH
PPL SUSQUEHANNA LLC
FRACKVILLE AREA MUN AUTH
MAHANOY CITY SEW AUTH
SHENANDOAH MUN SEW AUTH
HAMPDEN TWP - ROTH LANE
NEWBERRY TWP MUN AUTH
EPHRATA BORO AUTH - WWTP #2
EASTERN SNYDER CO REG AUTH
MID-CENTRE COUNTY AUTH
CARGILL MEAT SOLUTIONS CORP
CLEARFIELD LEATHER INC DBA WIC
LYCOMING CO WATER & SEWER AUTH
FIRST QUALITY TISSUE LLC
Design flow
(MGD)
1.30
1.00
1.40
2.50
1.05
3.75
1.76
3.30
1.22
1.16
4.20
1.73
3.70
15.00
7.20
6.00
2.25
2.42
4.16
0.08
1.40
1.38
2.00
4.65
1.30
2.30
2.80
1.00
0.80
0.12
1.50
3.95
Observed flow
(MGD)
(1991 -2006 average)
39.24
1.84
0.77
2.38
0.69
14.25
1.30
1.79
0.73
0.75
1.75
1.53
3.03
7.52
2.17
3.46
1.04
1.30
2.64
0.11
1.05
0.57
1.35
2.11
0.53
1.24
1.62
10.99
0.56
0.13
0.61
1.63
F-18
-------
Point Sources
Hydrography
Interstate
Water (Nat. Atlas Dataset)
US Census Populated Places
Municipalities (pop > 50,000)
] County Boundaries
Watershed with HUCSs
Pennsylvania
GCRP Model Areas - Susquehanna River Basin
Major Point Sources
Figure 3. Major point sources in the Susquehanna River watershed.
F-19
-------
Meteorological Data
The required meteorological data series for the 20 Watershed model simulations are precipitation, air temperature,
and potential evapotranspiration (PET). The 20 Watershed model does not include water temperature or algal
simulation and uses a degree-day method for snowmelt, so additional meteorological variables such as solar
radiation are needed only for the calculation of PET. These meteorological data are drawn from the BASINS4
Meteorological Database (USEPA 2008), which provides a consistent, quality-assured set of nationwide data with
gaps filled and records disaggregated. Scenario application will require simulation over 30 years, so the available
stations are those with a common 30-year period of record (or one that can be filled from an approximately co-
located station) that covers the year 2001. A total of 62 precipitation stations were identified for use in the
Susquehanna River model with a common period of record of 2/1/1972-3/31/2004 for the entire watershed (Table
4 and Figure 4). The majority of the stations are available through at least 9/30/2005. Temperature records are
sparser; where these are absent temperature is taken from nearby stations with an elevation correction. For each
weather station, Penman-Monteith reference evapotranspiration was calculated for use in HSPF using observed
precipitation and temperature coupled with SWAT weather generator estimates of solar radiation, wind
movement, cloud cover, and relative humidity.
SWAT uses daily meteorological data for the 20 Watershed model applications, while HSPF requires hourly data.
It is important to note that a majority of the meteorological stations available for the Susquehanna River
watershed are Cooperative Summary of the Day stations that do not report sub-daily data. The BASINS4 dataset
already has versions of the daily data that have been disaggregated to an hourly time step using template stations.
Table 4. Precipitation stations for the Susquehanna River watershed model
COOP ID
NY300085
NY300270
NY300448
NY300687
NY301168
NY301173
NY301424
NY301752
NY302454
NY302610
NY303979
NY303983
NY304772
NY305512
NY306085
NY307195
NY308498
NY309442
Name
ALFRED
ARNOT FOREST
BATH
BINGHAMTON WSO AP
CANDOR 2 SE
CANISTEO 1 SW
CHEPACHET
COOPERSTOWN
EAST SIDNEY
ELMIRA
HORNBY
HORNELL ALMOND DAM
LINDLEY2N
MORRISVILLE 6 SW
NORWICH
ROCKDALE
THURSTON
WHITNEY POINT DAM
Latitude
42.2614
42.2670
42.3500
42.2078
42.1947
42.2667
42.9097
42.7150
42.3333
42.1000
42.2330
42.3500
42.0500
42.8333
42.5011
42.3833
42.2000
42.3500
Longitude
-77.7850
-76.6330
-77.3500
-75.9814
-76.3133
-77.6167
-75.1108
-74.9283
-75.2333
-76.8000
-77.0500
-77.7000
-77.1333
-75.7333
-75.5194
-75.4000
-77.3330
-75.9670
Temperature
Yes
No
Yes
Yes
No
No
No
Yes
No
Yes
No
No
No
Yes
Yes
No
No
No
Elevation (ft)
1,770
1,200
1,120
1,600
920
1,155
1,320
1,200
1,155
844
1,795
1,325
1,040
1,300
1,020
1,030
1,620
1,040
F-20
-------
COOP ID
PA360140
PA360147
PA360457
PA360482
PA360656
PA360725
PA360763
PA361087
PA361480
PA361833
PA361961
PA362013
PA362245
PA362629
PA362721
PA363130
PA364047
PA364763
PA364778
PA364853
PA364896
PA364992
PA365344
PA365915
PA366289
PA366916
PA367409
PA367727
PA367730
PA367931
PA368057
PA368073
PA368449
Name
ALTOONA 3 W
ALVIN R BUSH DAM
BEAR GAP
BEAVERTOWN 1 NE
BIGLERVILLE
BLAIN 5SW
BLOSERVILLE 1 N
BUFFALO MILLS
CLARENCE
COVINGTON 2 WSW
CURWENSVILLE LAKE
DANVILLE
DRIFTWOOD
EMPORIUM
EVERETT
GALETON
HONEY BROOK 2 SSE
LANCASTER 2 NE FILT PLANT
LANDISVILLE 2 NW
LAURELTON CENTER
LEBANON 2 W
LEWISTOWN
MAHANOY CITY 2 N
MONTROSE
NEW PARK
PHILIPSBURG8E
RENOVO
RUSHVILLE
SABINSVILLE 3 SE
SELINSGROVE2S
SHICKSHINNY3N
SHIPPENSBURG
STATE COLLEGE
Latitude
40.4950
41.3670
40.8236
40.7667
39.9356
40.3000
40.2636
39.9461
41.0456
41.7331
40.9500
40.9483
41.3419
41.5067
40.0136
41.7356
40.0789
40.0500
40.1167
40.9017
40.3333
40.5869
40.8344
41.8667
39.7350
40.9167
41.3297
41.7833
41.8422
40.7831
41.2000
40.0500
40.7933
Longitude
-78.4667
-77.9330
-76.4983
-77.1500
-77.2578
-77.5833
-77.3639
-78.6458
-77.9453
-77.1167
-78.5330
-76.6036
-78.1403
-78.2275
-78.3653
-77.6519
-75.8975
-76.2742
-76.4333
-77.2139
-76.4667
-77.5697
-76.1353
-75.8500
-76.5061
-78.0667
-77.7381
-76.1167
-77.4747
-76.8611
-76.1500
-77.5167
-77.8672
Temperature
Yes
Yes
No
No
Yes
No
Yes
No
No
No
No
No
No
Yes
Yes
No
No
No
Yes
Yes
Yes
Yes
No
Yes
No
No
Yes
No
No
Yes
No
Yes
Yes
Elevation (ft)
1,320
930
900
540
720
820
700
1,310
1,390
1,745
1,165
475
820
1,040
1,000
1,345
665
270
360
800
450
460
1,710
1,420
800
1,945
660
870
1,999
420
780
680
1,170
F-21
-------
COOP ID
PA368491
PA368692
PA368905
PA368959
PA369408
PA369705
PA369714
PA369728
PA369823
PA369933
PA369950
Name
STILLWATER
SUSQUEHANNA
TOWANDA 1 S
TROY 1 NE
WELLSBORO 4 SW
WILKES BARRE SCRANTON
WSOAP
WILLIAMSBURG
WILLIAMSPORT RGNL AP
WOLFSBURG
YORK 3 SSW PUMP STN
YORK HAVEN
Latitude
41.6830
41.9483
41.7506
41.7833
41.7003
41.3389
40.4667
41.2433
40.0417
39.9167
40.1167
Longitude
-75.4830
-75.6050
-76.4428
-76.7833
-77.3894
-75.7267
-78.2000
-76.9217
-78.5278
-76.7500
-76.7167
Temperature
No
No
Yes
No
Yes
Yes
No
Yes
No
Yes
No
Elevation (ft)
1,650
910
750
1,110
1,818
930
840
520
1,185
390
310
F-22
-------
Legend
^ Weather Stations
= Interstate
Hydrography
•1 Water (Nat. Atlas Dataset)
US Census Populated Places
HI Municipalities (pop > 50,000)
^] County Boundaries
l~~l Watershed with HUCSs
'tnajiiio' •/*
\ t f & y^Y5P2*54
NY309442' ~k^~^
~~~f\f^J-f\
NY300687 \ New York
N"°±i^^ip^ >4;iyiA368491
^.i.«%^ x^ ^ y>fA368959A^ /T y &?^i
r^^^4^U\PA367'730\ %J^>^PA368905^/^nV/J*
W363130V1>^-V io.il.,,X^ "f^^V/ IS [ ;
Figure 4. Weatherstations for the Susquehanna River watershed model.
F-23
-------
Watershed Segmentation
At about 27,000 square miles, the Susquehanna River basin is one of the largest modeling areas considered for the
20 Watershed project. It encompasses a complete drainage area, with no need for upstream boundary conditions.
There is also an existing detailed HSPF model of the basin (the Chesapeake Bay Model or CBM; USEPA 2010).
Watershed segmentation for the Susquehanna River basin is based on the segmentation used in the CBM,
resulting in 278 subbasins for modeling (Figure 5). The initial calibration watershed (Raystown Branch Juniata
River) is highlighted.
The model subbasins approximate the HUC-10 scale, but are subdivided as needed to account for the connection
of tributaries and location of flow gages. The subbasins range in size from 1.04 to 516 mi2.
F-24
-------
Legend
A USGS Gages
Hydrography
Interstate
Water (Nat. Atlas Dataset)
US Census Populated Places
Municipalities (pop > 50,000)
] County Boundaries
I | Initial Calibration Watershed
Model Subbasins
Pennsylvania
USGS 1562000
GCRP Model Areas - Susquenanna River Basin
Model Segmentation
Figure 5. Model segmentation and USGS stations utilized for the Susquehanna River watershed.
F-25
-------
Calibration Data and Locations
The Susquehanna River was selected as a pilot site because of extensive previous experience with the CBM.
Given the existence of this calibrated model, the approach to the Susquehanna pilot study was to start from
parameters derived from the existing model, evaluate the parameterization through detailed application to an
initial calibration focus area, then evaluate the quality of the fit through comparison to data at two monitoring
points representing the larger watershed. Spatial calibration at multiple locations was not pursued. It should be
noted, however, that the 20 Watershed approach is based on a different land use coverage and uses different
weather data and, in particular, a different estimator of PET (Penman-Monteith PET using local temperature and
weather generator insolation and auxiliary variables for 20 Watershed versus Hamon PET in the Phase 5 CBM
and Penman pan evaporation at first-order weather stations in early versions of the CBM). These differences may
result in systematic differences in model parameters.
Three sites with known high quality flow gaging and water quality data were selected for both hydrology and
water quality calibration (Table 5). The first of these (Raystown Branch Juniata River) was selected for initial
calibration, with subsequent adjustments based on comparison to data at two stations on the mainstem
Susquehanna.
Table 5. Calibration and validation locations in the Susquehanna River watershed
Station Name
Raystown Branch Juniata River at Saxton, PA
Susquehanna River at Danville, PA
Susquehanna River at Marietta, PA
USGS ID
01562000
01540500
01576000
Drainage Area
(mi2)
756
11,220
25,990
Hydrology
Calibration
X
X
X
Water Quality
Calibration
X
X
X
The model hydrology calibration period for Raystown Branch Juniata River was set to Water Years 1995-2005,
while the mainstem stations use Water Years 1993-2003 to fall within the common period of record of the
weather data. Calibration was done on the later data due to concerns that there have been significant changes in
land use and agricultural management practices since the 1980s. Hydrologic validation was then performed on the
10 water years priorto the calibration period. The bulk of available water quality data are from the early 1990s.
Therefore, water quality calibration used calendar year periods beginning in 1991, with earlier data reserved for
validation.
F-26
-------
Initial hydrologic parameterization for the Susquehanna River calibration focus area came from the Chesapeake
Bay Program's (CBP) model for the Susquehanna River watershed (CBM). The CBM has undergone a series of
revisions over many years, with the most current version of the model being known as Phase 5 (USEPA 2010).
The Phase 5 parameters were obtained from the CBP and reviewed. Through this process it was identified that the
CBM set for Phase 5 was much more complex than that used for a typical HSPF model, with parameters and land
use categories provided in a multi-file database, with values changing over the time span of the simulation. In the
CBP Phase 5 model, HSPF input sequences are developed through an elaborate scripting scheme, with multiple
HSPF input sequences developed for more than 20 land use categories (including various agricultural
management practices) and multiple time spans to determine flows and loads on a unit area basis. These unit area
flows and loads are then combined through scripts to create input for an in-stream simulation.
The complexity of the Phase 5 model far exceeded the constraints of this study, so rather than using the Phase 5
parameters, it was decided to parameterize Susquehanna River watershed model using the previous version of the
CBM, Phase 4. The Phase 4 parameters are readily available, as they are incorporated into the BASINS
companion program HSPFParm, and these parameters are available in a format directly akin to that needed by
HSPF. The six pervious (forest, high till cropland, low till cropland, pasture, urban, and hay) and two impervious
(animal/feedlot and urban) land use categories of the Phase 4 model are more analogous to the categories of the
Susquehanna model for this study.
Even using the simpler Phase 4 model from the CBP, there was no one-to-one correlation between land use
categories of the CBM and this study. The Phase 4 model of the Susquehanna River watershed consisted of 12
land segments in 4 UCI files, and these parameters needed to be applied to the 21 land uses in this project.
Moreover, for this project the land use categories explicitly represented hydrologic soil groups, while the CBM
parameters did not. A method was developed for creating the Susquehanna model parameters for this study from
these Phase 4 parameters, where approximate average values were applied from the CBP Upper Susquehanna
simulations for model section 020501, from the CBP Western Branch Susquehanna simulation for model section
020502, and from the Lower Susquehanna simulation for model section 020503. Since there was no explicit
representation of hydrologic soil groups in the CBM, parameter values were assigned the same values across each
of the Susquehanna model soil classifications. For example, the CBM forest parameters for the Lower
Susquehanna were applied to the land use categories Forest_A through Forest_D in the model parameters for
020503.
The USGS gage on the Raystown Branch of the Juniata River at Saxton, PA (USGS 02050303) was used as the
primary calibration location, while the gages on the Susquehanna River at Danville, PA (USGS 01540500), on the
West Branch Susquehanna River at Lewisburg, PA (USGS 01553500), and on the Susquehanna River at Marietta,
PA (USGS 01576000) were used as additional calibration checks. Calibrated parameters from the Raystown
Branch gage were applied to the Lower Susquehanna portion of the study area (020503), while calibration
adjustments for the Danville and Lewisburg gages were applied to the Upper and Western sections of the study
area respectively (020501 and 020502). The Susquehanna River at Marietta, PA (USGS 01576000), the most
downstream of the gages, was used to verify that the calibration parameters applied at the two upstream gages
were applicable to the entire watershed.
Once the hydrology calibration was complete for the entire Susquehanna River watershed, the focus turned to
sediment and water quality representation. Extracting parameters from the CBP Susquehanna model for sediment
and water quality was even less straightforward than that for hydrology parameters, since the CBM used the more
complicated NITR and PHOS modules of the HSPF pervious land simulation operations. Initial parameterization
for sediment and water quality simulation was taken from the loadings used to set up the Willamette model in this
20 Watershed study, adjusted based on the parameters from the CBP Phase 4 model where available.
-------
Changes Made to Base Data Provided
No changes were made to the meteorological, point source, or land use base data.
Assumptions
Reservoirs
While there are many dams in the study area, their influence was not explicitly included in the study. The largest
of the reservoirs are on the Susquehanna River near the outlet of the study area, well below the USGS gage on the
Susquehanna River at Marietta. This USGS gage on the Susquehanna River at Marietta was selected for use in
calibration because it is the most downstream main stem gage that is still upstream of the influence of the major
reservoirs. While one would assume the major main stem reservoirs influence the flow and water quality exiting
the Susquehanna River at the outlet, for this model the impacts of these reservoirs are assumed to be implicitly
represented through the tabular representation of reach hydrologic response (FTables).
The primary intention of the 20 Watershed simulations is to examine relative changes in response of large
watersheds. Information is not available to specify future time series of operations or boundary conditions for
these reservoirs. Therefore, representation of the reservoirs through the stage-discharge relationships expressed in
the FTables provides the most useful basis for evaluating relative changes in response.
Withdrawals and Point Sources
A variety of water withdrawals occur in the Susquehanna, but these have a relatively small effect on the overall
water balance. In addition, future changes in water withdrawals are not known. Therefore, withdrawals were not
included in the 20 Watershed model application. In contrast to withdrawals, point sources must be included for
model water quality calibration because they represent a significant fraction of nutrient loads in the system.
Existing major point source flows and loads are represented in the model, but will be held at current levels for
simulation of future conditions to better isolate the potential direct impacts of climate and land use change.
Snow Simulation
The Susquehanna HPSF model includes snow simulation using the degree-day method for snowmelt. The initial
values extracted from the Chesapeake Bay Program Model are assumed to be appropriate and the initial
parameterization was not adjusted.
Hydrology Calibration
As explained above, the starting parameters for the Susquehanna HSPF model came from a Chesapeake Bay
Program model of the Susquehanna River watershed. Once the starting parameters were inserted into the model
input files, average annual precipitation and potential evapotranspiration values were computed and compared to
published values. Through this process it was determined the input potential evapotranspiration time series should
be reduced by multipliers, since the computation of these time series produced more PET on an average annual
basis than the published values indicate. The multipliers used for PET were 0.75 for the Lower Susquehanna and
Western Branch, and 0.8 for the Upper Susquehanna. Calibration adjustments focused on the following
parameters:
• BASETP (ET by riparian vegetation): The model was significantly oversimulating the 50 percent low
flows at the primary calibration location. Slightly increasing the BASETP value provided some ET by
riparian vegetation and thus improved the simulation of low flows.
-------
• DEEPFR (fraction of groundwater inflow which will enter deep groundwater): Adding a modest amount
of DEEPFR above the primary calibration location improved the overall water balance and improved the
simulation of low flows.
• INFILT (index to mean soil infiltration rate): In the upper portions of the study area the peak flows were
simulating too low, while the low flows were simulating too high. Decreasing INFILT for the upper
portions of the watershed shifted flows to a faster response, increasing the peaks and reducing the low
flows.
• AGWRC (Groundwater recession rate): Adjusted slightly in order to replicate groundwater recession in
the observed data.
Initial calibration was performed at the USGS gage on the Raystown Branch of the Juniata River at Saxton, PA
(USGS 02050303), and is summarized in Figures 6 through 12 and Tables 6 and 7. The model fit is of high
quality overall, but simulates low on the lowest 10 percent of flows. This could be due to something not
accounted for in the model, such as reservoir operations or other discharges. Given that these low flows are not
critical to the purposes of this study, the issue is being noted as an area with potential further refinement. None of
the metrics fall beyond the range of those set for the 20 Watershed study. The model calibration period was set to
the 10 water years from 10/01/1995 to 09/30/2005.
F-29
-------
I Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1995 to 9/30/2005 )
-Avg Modeled Flow (Same Period)
35000
o
(0
«
Oct-95 Apr-97 Oct-98 Apr-00 Oct-01
Date
Apr-03
Oct-04
Figure 6. Mean daily flow at USGS 01562000 Raystown Branch Juniata River at Saxton, PA -
calibration period (HSPF).
^•Avg Monthly Rainfall (in)
-»-Avg Observed Flow (10/1/1995 to 9/30/2005
—Avg Modeled Flow (Same Period)
6000
4000 -
LJ-
2000 -
Figure 7. Mean monthly flow at USGS 01562000 Raystown Branch Juniata River at Saxton, PA
calibration period (HSPF).
F-30
-------
6000
•s
1
4000 -
_
-s
o
O)
o>
£
01
2000 -
Avg Flow (10/1/1995 to 9/30/2005
Line of Equal Value
Best-Fit Line
0.937x + 65.513
R2 = 0.8993
2000 4000
Average Observed Flow (cfs)
6000
T3
O
.a
g
O)
o
ro
uD
_
O>
Avg Obsen/ed Flow (10/1/1995 to 9/30/2005 )
^Avg Modeled Flow (10/1/1995 to 9/30/2005 )
Line of Equal Value
100% -,
90% -
80% -
70% -
60% -
50%
40% -
30% -
20% -
10% -
0%
L J 1 L , J
0-95 A-97 0-98
A-00 0-01
Month
A-03 0-04
Figure 8. Monthly flow regression and temporal variation at USGS 01562000 Raystown Branch Juniata
River at Saxton, PA - calibration period (HSPF).
Avg Flow (10/1/1995 to 9/30/2005)
-Line of Equal Value
-Best-Fit Line
2500
500 1000 1500 2000 2500
Average Observed Flow (cfs)
Avg Monthly Rainfall (in)
-Avg Obsen/ed Flow (10/1/1995 to 9/30/2005)
-Avg Modeled Flow (Same Period)
2500 -i
10 11 12 1 2 3 4 5 6
Month
Figure 9. Seasonal regression and temporal aggregate at USGS 01562000 Raystown Branch Juniata
River at Saxton, PA - calibration period (HSPF).
F-31
-------
Average Monthly Rainfall (in)
-Median Observed Flow (10/1/1995 to 9/30/2005)
I Observed (25th, 75th)
I Modeled (Median, 25th, 75th)
3000 T
2500 -
2000 -
1500 -
1000 -
500 - --
r 0
- 1
- 2
- 3
- 5
10
11
12
3 4
Month
Figure 10. Seasonal medians and ranges at USGS 01562000 Raystown Branch Juniata River at
Saxton, PA - calibration period (HSPF).
Table 6. Seasonal summary at USGS 01562000 Raystown Branch Juniata River at Saxton,
PA - calibration period (HSPF)
MON I H
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
438. 69 1 187. 00 1 119.00 530.75
1024.44 647. 50 1 149.75 1170.00
1071.221 682. 00l 280.751 1350.00
1255. 25 1 475. 00| 260.00 1180.00
1290.35 900. 00 1 470.00 1775.00
2177.77! 1690. OOl 1052.50! 2742.50
1771.52] 1180.00| 843.75) 2102.50
1310.00 739. 50 1 414.00 1777.50
895. 83 1 448. OOl 287.50! 879.00
326.40] 207. 00 1 146.00 324.00
278.62 157.50] 115.25 339.75
907.63 124. 50 1 104.00 360.25
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
599.79 301. 11 1 128.74] 788.27
1101.45 658. 92 1 138. 80 1 1324.27
1029. 71 1 830.45! 374.78! 1250.32
1394.02] 892.68] 351.51] 1730.37
1330.42 1074.05] 739. 83 ] 1743.98
1963.581 1594.05! 1008.22! 2348.98
1490.45) 989. 11 1 640.38! 1736.15
1010.40 563.51 303. 89| 1180.35
939.20I 441.19! 220.33! 910.95
407.92 216. 08 1 114. 91 1 368.66
345.53 174.23 74.24] 430.05
1120.57 133.58] 62.75| 649.56
F-32
-------
•Obsen/ed Flow Duration (10/1/1995 to 9/30/2005 )
•Modeled Flow Duration (10/1/1995 to 9/30/2005 )
100000
10000
1000
10
10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 11. Flow exceedence at USGS 01562000 Raystown Branch Juniata River at Saxton, PA -
calibration period (HSPF).
o
o
T3
0)
0)
0)
E
0)
.N
"re
E
o
-Observed Flow Volume (10/1/1995 to 9/30/2005 )
-Modeled Flow Volume (10/1/1995 to 9/30/2005 )
120%
100% -
80% -
60% -
40% -
20% -
Oct-95
Apr-97
Oct-9£
Apr-00
Oct-01
Apr-03
Oct-04
Figure 12. Flow accumulation at USGS 01562000 Raystown Branch Juniata River at Saxton, PA
- calibration period (HSPF).
F-33
-------
Table 7. Summary statistics at USGS 01562000 Raystown Branch Juniata River at Saxton,
PA - calibration period (HSPF)
HSPF Simulated Flow
REACH OUTFLOW FROM DSN 101
10-Year Analysis Period: 10/1/1995 - 9/30/2005
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3^:
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% hig_hest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
19.01
8.15
2.19
2.80
4.11
6.97
5.13
7.78
1.62
Error Statistics
-0.16
6.72
-3.92
23.90
Observed Flow Gage
USGS 01562000 Raystown Branch Juniata River at Saxton, PA
Hydrologic Unit Code: 2050303
Latitude: 40.21591249
Longitude: -78.2652901
Drainage Area (sq-rri): 756
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
7.76 » | 30
-0.84
-13.61
-0.63
39.10
0.698
0.552
0.898
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
19.04
8.48
2.05
2.26
3.81
7.03
5.93
7.83
1.16
Clear [
Hydrology Validation
Validation for the Susquehanna River calibration focus was performed at the same gage (USGS 01562000
Raystown Branch Juniata River at Saxton, PA) but for the water years 10/01/1985 to 09/30/1995. Results are
presented in Figures 13 through 19 and Tables 8 and 9. Similar to the calibration years, the validation years'
model fit is of high quality, although the validation simulates the summer storm volumes somewhat high while
undersimulating the overall storm volume. This may in part reflect differences in land use and management
practices relative to the 200 NLCD. The remaining metrics fall within the acceptable range set for the 20
Watershed study.
F-34
-------
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1985 to 9/30/1995
-Avg Modeled Flow (Same Period)
30000
o
Oct-94
Figure 13. Mean daily flow at USGS 01562000 Raystown Branch Juniata River at Saxton, PA
validation period (HSPF).
Avg Monthly Rainfall (in)
-»-Avg Observed Flow (10/1/1985 to 9/30/1995 )
—Avg Modeled Flow (Same Period)
8000
•S
6000 -
4000
2000 -
0-85 A-87 0-88 A-90 O-91
Month
A-93
0-94
Figure 14. Mean monthly flow at USGS 01562000 Raystown Branch Juniata River at Saxton, PA -
validation period (HSPF).
F-35
-------
• Avg Flow (10/1/1985 to 9/30/1995 )
Line of Equal Value
— Best-Fit Line
8000
T3
O
+
CO
JS
ro
00
100% -,
90% -
80% -
70% -
60% -
50%
Avg Observed Flow (10/1/1985 to 9/30/1995 )
Avg Modeled Flow (10/1/1985 to 9/30/1995 )
-Line of Equal Value
40% -
30% -
20% -
10% -
0%
0 2000 4000 6000
Average Observed Flow (cfs)
8000
0-85 A-87 0-88
A-90 0-91
Month
A-93 0-94
Figure 15. Monthly flow regression and temporal variation at USGS 01562000 Raystown Branch
Juniata River at Saxton, PA - validation period (HSPF).
• Avg Flow (10/1/1985 to 9/30/1995)
Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1985 to 9/30/1995)
-Avg Modeled Flow (Same Period)
2500
= 0,7072x+192.32
..R2 = 0.8986
2500 -i
500 1000 1500 2000 2500
Average Observed Flow (cfs)
10 11 12 1 23456789
Month
Figure 16. Seasonal regression and temporal aggregate at USGS 01562000 Raystown Branch Juniata
River at Saxton, PA - validation period (HSPF).
F-36
-------
i Observed (25th, 75th)
• Median Observed Flow (10/1/1985 to 9/30/1995)
Average Monthly Rainfall (in)
Modeled (Median, 25th, 75th)
2500
10 11
Figure 17. Seasonal medians and ranges at USGS 01562000 Raystown Branch Juniata River at
Saxton, PA - validation period (HSPF).
Table 8. Seasonal summary at USGS 01562000 Raystown Branch Juniata River at Saxton,
PA - validation period (HSPF)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
351.63
128.00
104. 00 1 200.00
771.90! 411.00! 211.75! 649.00
936.58
603.50
1063.43 562.00
302.50! 1157.50
322.50! 1167.50
1214.57 754.00 410.00! 1495.00
2016.51
1240.00
670.75! 2140.00
1772. 93| 965. 00| 613. 50| 1935.00
1102.77
676.00
479.25! 1122.50
559.15! 275.50 201.00! 479.50
545.15
252.88
180.00
136.00
138.00! 418.00
108.00! 216.75
195.59! 160.00! 107.75! 233.50
MODELED FLOW (CFS)
MEAN MEDIAN
429.02
25TH 75TH
157. 78 1 101. 27 1 307.10
798.12 454.17! 239.35! 790.07
993.77
1044.74
1256.37
1591.71
1379.12
638.99
779.69
787.94
1003.08
1053.09
399.19! 1215.70
549.18! 1243.93
702.93! 1494.87
578.34! 1910.94
653. 23| 398.61 1345.43
384.81
225.71! 698.65
619.87 217.91! 135.17! 385.61
621.12
328.81
160.93
115.81
89.83! 396.62
54.58J 259.23
231.64 169.21! 67.62! 328.28
F-37
-------
-Obsen/ed Flow Duration (10/1/1985 to 9/30/1995
Modeled Flow Duration (10/1/1985 to 9/30/1995 )
100000
10000
£
0)
E
T3
O)
ro
o
-Obsenyed Flow Volume (10/1/1985 to 9/30/1995
-Modeled Flow Volume (10/1/1985 to 9/30/1995 )
120%
Oct-85 Apr-87
Oct-88 Apr-90 Oct-91 Apr-93 Oct-94
Figure 19. Flow accumulation at USGS 01562000 Raystown Branch Juniata River at Saxton, PA
validation period (HSPF).
F-38
-------
Table 9. Summary statistics at USGS 01562000 Raystown Branch Juniata River at Saxton,
PA - validation period (HSPF)
REACH OUTFLOW FROM DSN 101
10-Year Analysis F^riod: 10/1/1985 - 9/30/1995
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total ojj3inTulatejdjTic|h^
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of EfficjencyJ^___
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
14.83
6.77
1.74
1.79
3.35
5.76
3.93
5.52
0.85
Error Statistics
-8.00
-4.61
-12.82
18.92
USGS 01562000 Raystown Branch Juniata River at Saxton, PA
Hydrologic Unit Code: 2050303
Latitude: 40.21 591 249
Longitude: -78.2652901
Drainage Area (sq-rri): 756
Total Observed In-stream Flow:
_J]oJal^f^teervedJTig_hest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow Volume (1-3):
Observed Spring Flow Volume (4-6):
Total Obsei-ved Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
7.86 » | 30
-9.69
-23.40
-14.91
56.16
0.553
0.512
0.868
30
30
20
50
Model accuracy increases
as E or E' approaches 1 .0
16.12
7.77
1.82
1.51
3.10
6.38
5.12
6.49
0.54
Clear [
Hydrology Results for Larger Watershed
Since the calibration location above represents only a small portion of the drainage area for this project, results
near the outlet of the entire watershed were examined at the Susquehanna River at Marietta, PA USGS gage
(01576000). The results at this gage look fairly good as well. The simulated output is just a little high on the low
flows and a little high on the summer volumes. A few of the metrics were exceeded, but most of the metrics fall
within the acceptable range set for the 20 Watershed study including a daily Nash-Sutcliffe of 0.77 at the Marietta
gage. The calibration results at the Susquehanna River at Marietta, PA USGS gage (01576000) are presented in
Figures 20 through 26 and Tables 10 and 11. The calibration and validation statistical measurements at all USGS
gages used in the Susquehanna River watershed for the 20 Watershed project are shown in Tables 12 and 13,
respectively.
F-39
-------
^•Avg Monthly Rainfall (in)
Avg Observed Flow (10/1/1993 to 9/30/2003 )
Avg Modeled Flow (Same Period)
600000
t
o
Apr-95
Oct-96
Apr-98 Oct-99
Date
Apr-01
Oct-02
Figure 20. Mean daily flow at USGS 01576000 Susquehanna River at Marietta, PA - calibration period
(HSPF).
200000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1993 to 9/30/2003
•Avg Modeled Flow (Same Period)
0-02
Figure 21. Mean monthly flow at USGS 01576000 Susquehanna River at Marietta, PA - calibration
period (HSPF).
F-40
-------
Avg Flow (10/1/1993 to 9/30/2003 )
• Line of Equal Value
Best-Fit Line
200000
150000
100000
o
0) 50000
y = 0.816x +7854.6
0 50000 100000 150000 200000
Average Observed Flow(cfs)
100% -i
Avg Observed Flow (10/1/1993 to 9/30/2003 )
Avg Modeled Flow (10/1/1993 to 9/30/2003 )
-Line of Equal Value
O
£=
_
ro
m
A-95 O-96
A-98 O-99
Month
A-01 O-02
Figure 22. Monthly flow regression and temporal variation at USGS 01576000 Susquehanna River at
Marietta, PA - calibration period (HSPF).
Avg Flow (10/1/1993 to 9/30/2003)
• Line of Equal Value
Best-Fit Line
100000
80000
60000
40000
-------
o
i Observed (25th, 75th)
• Median Observed Flow (10/1/1993 to 9/30/2003)
Average Monthly Rainfall (in)
Modeled (Median, 25th, 75th)
100000
90000
80000
70000
60000
50000 -
£ 40000
30000
20000
10000
0
10 11
Figure 24. Seasonal medians and ranges at USGS 01576000 Susquehanna River at Marietta, PA -
calibration period (HSPF).
Table 10. Seasonal summary at USGS 01576000 Susquehanna River at Marietta, PA
calibration period (HSPF)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
18258.39
29811.13
43879.65
46873.61
45647.16
78282.90
74228.00
46236.13
32080.10
16327.52
15446.55
16977.37
10650.00
19300.00
31150.00
20700.00
35800.00
64900.00
63000.00
33750.00
24450.00
13800.00
7330.00
7645.00
7305.00
9002.50
19700.00
14425.00
25050.00
44725.00
43400.00
23100.00
16575.00
8122.50
5292.50
4757.50
19225.00
37325.00
54050.00
45175.00
50975.00
93150.00
90750.00
54575.00
38925.00
22075.00
15175.00
21650.00
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
26039.03
33731 .91
42761 .05
50733.93
56828.16
74939.27
53288.33
36901.17
32128.54
19391.00
20184.23
26001 .69
16034.54
24238.21
32769.07
32088.13
46881 .28
64475.93
45390.82
25714.54
24520.57
14448.96
9362.29
11403.96
11020.63
10179.07
21414.73
23521.55
35098.74
47051.59
31740.68
15935.99
14349.20
8844.17
6647.42
6133.33
26062.98
40953.88
54042.56
55480.61
66429.31
90047.35
66146.48
46244.86
42832.37
24338.26
20176.83
36050.57
F-42
-------
t
I
D)
(0
(0
Q
•Observed Flow Duration (10/1/1993 to 9/30/2003 )
Modeled Flow Duration (10/1/1993 to 9/30/2003 )
1000000
100000
10000
1000
10%
20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90%
100%
Figure 25. Flow exceedence at USGS 01576000 Susquehanna River at Marietta, PA - calibration
period (HSPF).
o
o
ro
T3
o
I
-a
-------
Table 11. Summary statistics at USGS 01576000 Susquehanna River at Marietta, PA-
calibration period (HSPF)
HSPF Simulated Flow
REACH OUTFLOW FROM DSN 101
10-Year Analysis Period: 10/1/1993 - 9/30/2003
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3^:
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
20.55
6.79
3.76
2.87
4.50
7.87
5.30
7.51
1.19
Error Statistics
1.79
14.99
-7.21
34.30
Observed Flow Gage
USGS 01576000 Susquehanna River at Marietta, PA
Hydrologic Unit Code: 2050306
Latitude: 40.054541 3
Longitude: -76.5307992
Drainage Area (sq-rri): 25990
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
11.49 »| 30
6.41
-19.82
2.72
56.64
0.771
0.582
0.861
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
20.19
7.31
3.27
2.14
4.04
7.39
6.62
7.32
0.76
Clear [
F-44
-------
Table 12. Summary statistics (percent error) for all stations - calibration period (HSPF)
Station
Error in total volume:
Error in 50% lowest
flows:
Error in 10% highest
flows:
Seasonal volume error
- Summer:
Seasonal volume error
-Fall:
Seasonal volume error
-Winter:
Seasonal volume error
- Spring:
Error in storm
volumes:
Error in summer storm
volumes:
Daily Nash-Sutcliffe
Coefficient of
Efficiency, E:
Monthly Nash-Sutcliffe
Coefficient of
Efficiency, E
01562000
Raystown Branch
Juniata River at
Saxton, PA
(1995-2005)
-0.16
6.72
-3.92
23.9
7.76
-0.84
-13.61
-0.63
39.10
0.698
0.898
01540500
Susquehanna River at
Danville, PA
(1993-2003)
6.67
25.34
-4.71
45.64
14.79
13.54
-15.66
7.37
75.91
0.786
0.837
01576000
Susquehanna River at
Marietta, PA
(1993-2003)
1.79
14.99
-7.21
34.30
11.49
6.41
-19.82
2.72
56.64
0.771
0.861
F-45
-------
Table 13. Summary statistics (percent error) for all stations - validation period (HSPF)
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error -
Summer:
Seasonal volume error -
Fall:
Seasonal volume error -
Winter:
Seasonal volume error -
Spring:
Error in storm volumes:
Error in summer storm
volumes:
Nash-Sutcliffe Coefficient
of Efficiency, E:
Baseline adjusted
coefficient (Garrick), E':
01562000
Ray stow n Branch
Juniata River at Saxton,
PA
(1985-1995)
-8.00
-4.61
-12.82
18.92
7.86
-9.69
-23.40
-14.91
56.16
0.553
0.868
01540500
Susquehanna River at
Danville, PA
(1983-1993)
2.12
30.70
-13.14
54.06
13.13
8.76
-23.37
0.31
85.91
0.714
0.782
01576000
Susquehanna River at
Marietta, PA
(1983-1993)
-1.64
18.99
-13.32
35.02
15.62
2.93
-27.27
-1.26
64.86
0.665
0.777
Water Quality Calibration and Validation
The 20 Watershed models are designed to provide water quality simulation for total suspended solids (TSS), total
nitrogen, and total phosphorus. TSS is simulated with the standard HSPF approach (USEPA 2006). In contrast to
TSS, total nitrogen and total phosphorus are simulated in this application in a simplistic fashion, as HSPF general
quality constituents (GQUALs) subject to an exponential decay rate during transport.
The water quality calibration focuses on the replication of monthly loads, as specified in the project QAPP. Given
the approach to water quality simulation in the 20 Watershed model, a close match to individual concentration
observations cannot be expected. Comparison to monthly loads presents challenges, as monthly loads are not
observed. Instead, monthly loads must be estimated from scattered concentration grab samples and continuous
flow records. As a result, the monthly load calibration is inevitably based on the comparison of two uncertain
numbers. Nonetheless, calibration is able to achieve a reasonable agreement. Further, the load comparisons were
supported by detailed examinations of the relationships of flows to loads and concentrations and the distribution
of concentration prediction errors versus flow, time, and season, as well as standard time series plots.
For application on a nationwide basis, the 20 Watershed protocols assume that TSS and total phosphorus loads
will likely exhibit a strong positive correlation to flow (and associated erosive processes), while total nitrogen
loads, which often have a dominant groundwater component, will not. Accordingly, TSS and total phosphorus
loads were estimated from observations using a flow-stratified log-log regression approach, while total nitrogen
F-46
-------
loads were estimated using a flow-stratified averaging estimator, consistent with the findings of Preston et al.
(1989).
Similar to hydrology, initial calibration of water quality was done on the Raystown Branch Juniata River at
Saxton, PA, comparing model results to data from USGS 01562000. The calibration used the time period 1991-
2000 and 1990 was used for validation.
Results of the TSS calibration are generally acceptable. Visually, the model is roughly simulating the trends
contained in the observed data. A variety of other diagnostics were also pursued to ensure agreement between the
model and observations. These are available in full in the calibration spreadsheets, but a few examples are
provided below. Figure 27 presents the monthly load of TSS. Load-flow power plots were compared for
individual days (Figures 28 and 29). This confirms that the relationship between flow and load is consistent across
the entire range of observed flows, for both the calibration and validation periods. Tables 14 and 15 provide
model statistics and relative errors for the TSS calibration and validation periods.
TSS
1,000,000
-Regression Loads
-Simulated Loads
CpCpCpCpCpCpCpCpCpCpCpCpCpCpCpCpCpCpCpCpO O
CO —. CD —. CD —. CD —. CD —.CD —. CD —. CD —. CD —. CD —.CD —.
-—i J —\ J —~i J —~i J —~i J —~i J —\ J —~i J —~i J —~i J —~i J
Figure 27. Fit for monthly load of TSS at Raystown Branch Juniata River at Saxton, PA (HSPF).
Table 14. Model fit statistics (observed minus predicted) for monthly sediment loads using
stratified regression - USGS 01562000 Raystown Branch Juniata River at Saxton,
PA (HSPF)
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1991-2000)
-78.2%
146%
20.3%
Validation period
(1990)
-89.7%
124%
58.3%
F-47
-------
Raystown Branch Juniata River at Saxton, PA 1991-2000
"35
c
o
•o
ra
o
V)
10
100 1000
Flow, cfs
10000
100000
• Simulated A Observed ^^™Power (Simulated) ^^™Power (Observed)
Figure 28. Power plot for observed and simulated TSS at Raystown Branch Juniata River at Saxton,
PA - calibration period (HSPF).
Raystown Branch Juniata River at Saxton, PA 1990-1990
10000
10000
• Simulated A Observed ^^™Power (Simulated) ^^™Power (Observed)
Figure 29. Power plot for observed and simulated TSS at Raystown Branch Juniata River at Saxton,
PA - validation period (HSPF).
F-48
-------
Standard time series plots (Figure 30) show that observed and simulated concentrations achieve good agreement,
although individual observations may deviate. Plots of concentration error versus flow and versus month (not
shown) were used to guard against hydrologic and temporal bias.
Raystown Branch Juniata River at Saxton, PA
• Simulated A Observed
10000
1000
O)
eo
CO
100
1993 1994 1995 1996 1997
Year
1998
1999
Figure 30. Time series plot of TSS concentration at Raystown Branch Juniata River at Saxton, PA
calibration period (HSPF).
Table 15. Relative errors,(observed minus simulated) for TSS concentrations at USGS
01562000 Raystown Branch Juniata River at Saxton, PA (HSPF)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1991-2000)
106
-138.4%
-29.73%
Validation period
(1990)
15
-30.82%
24.63%
The total phosphorus load calibration performed well for the Susquehanna River watershed calibration focus area.
Adjustments were made to the accumulation rate and storage limits for the impervious surfaces. In general, the
observed and simulated total phosphorus loads attained an acceptable match for the simulation period (Figure 31).
As with TSS, additional diagnostics for total phosphorus included flow-load power plots (Figures 32 and 33) and
time series plots (Figure 34). All show acceptable agreement. Tables 16 and 17 provide model statistics and
relative errors for the total phosphorus calibration and validation periods. In contrast to load, phosphorus
concentrations are generally over-estimated (observed minus simulated concentration less than zero). This is due
to an over-estimation of observed phosphorus concentrations at low flows that may be due to the simplistic
representation of point source discharges in the model.
F-49
-------
o
E
In
I
Total P
100
-Regression Loads
-Simulated Loads
Figure 31. Fit for monthly load of total phosphorus at Raystown Branch Juniata River at Saxton, PA
(HSPF).
Table 16. Model fit statistics (observed minus predicted) for monthly total phosphorus loads
using stratified regression - USGS 01562000 Raystown Branch Juniata River at
Saxton, PA (HSPF)
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1991-2000)
26.0%
49%
22.7%
Validation period
(1990)
21.5%
45%
35.4%
F-50
-------
Raystown Branch Juniata River at Saxton, PA 1991-2000
10
100 1000
Flow, cfs
10000
100000
• Simulated A Observed ^^~Power (Simulated) ^^™ Power (Observed)
Figure 32. Power plot for observed and simulated total phosphorus at Raystown Branch Juniata
River at Saxton, PA - calibration period (HSPF).
Raystown Branch Juniata River at Saxton, PA 1990-1990
10
>
ra
•o
«
c
o
•o
re
o
0.1
0.01
0.001
10
100
Flow, cfs
1000
10000
» Simulated A Observed ^^~ Power (Simulated) ^^~ Power (Observed)
Figure 33. Power plot for observed and simulated total phosphorus at Raystown Branch Juniata
River at Saxton, PA - validation period (HSPF).
F-51
-------
Raystown Branch Juniata River at Saxton, PA
1994
1995
1996 1997
Year
1998
1999
Figure 34. Time series plot of total phosphorus concentration, at Raystown Branch Juniata River at
Saxton, PA (HSPF).
Table 17. Relative errors (observed minus simulated) for total phosphorus concentrations at
USGS 01562000 Raystown Branch Juniata River at Saxton, PA (HSPF)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1990-1990)
122
-121.72%
-50.63%
Validation period
(1990)
18
-0.88%
-12.24%
Results for total nitrogen are summarized in Figures 35 through 38. The results are acceptable, and generally
better than those for total phosphorus. This is due to total nitrogen not being sediment associated, therefore,
problems with sediment are not reflected in the calibration for total nitrogen. Tables 18 and 19 provide model
statistics and relative errors for the total nitrogen calibration and validation periods.
F-52
-------
Total N
1,400
-Averaging Loads
-Simulated Loads
Figure 35. Fit for monthly load of total nitrogen at Raystown Branch Juniata River at Saxton, PA
(HSPF).
Table 18. Model fit statistics (observed minus predicted) for monthly total nitrogen loads
using averaging estimator- USGS 01562000 Raystown Branch Juniata River at
Saxton, PA (HSPF)
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1991-2000)
7.0%
34%
16.8%
Validation period
(1990)
17.2%
29%
26.2%
F-53
-------
Raystown Branch Juniata River at Saxton, PA 1991-2000
1000
0.01
10
100 1000
Flow, cfs
10000 100000
• Simulated A Observed ^^™Power (Simulated) ^^™Power (Observed)
Figure 36. Power plot for observed and simulated total nitrogen at Raystown Branch Juniata River at
Saxton, PA - calibration period (HSPF).
Raystown Branch Juniata River at Saxton, PA 1990-1990
100
10000
Simulated A Observed ^^™Power (Simulated) ^^™Power (Observed)
Figure 37. Power plot for observed and simulated total nitrogen at Raystown Branch Juniata River at
Saxton, PA - validation period (HSPF).
F-54
-------
Raystown Branch Juniata River at Saxton, PA
• Simulated A Observed
O)
20
18
16
14
12
10
8
6
4
2
1986 1987 1988 1989 1990
Year
1991
1992
Figure 38. Time series plot of total nitrogen concentration, at Raystown Branch Juniata River at
Saxton, PA (HSPF).
Table 19. Relative errors (observed minus simulated) for total nitrogen concentration at
USGS 01562000 Raystown Branch Juniata River at Saxton, PA (HSPF)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1991-2000)
13
-22.89%
-22.60%
Validation period
(1990)
6
11.10%
7.42%
Water Quality Results for Larger Watershed
As with hydrology, the Raystown Branch Juniata River (USGS 01562000) watershed parameters for water quality
were directly transferred to other portions of the watershed. Summary statistics for the water quality calibration
and validation at other stations in the watershed are provided in Tables 20 and 21, respectively.
F-55
-------
Table 20. Summary statistics for water quality for all stations - calibration period (HSPF)
Station
Relative Percent Error TSS Load
TSS Concentration Median Error
Relative Percent Error TP Load
TP Concentration Median Error
Relative Percent Error TN Load
TN Concentration Median Error
01576000
Susquehanna
River at Marietta,
PA
(1991-1995)
26.5%
-24.8%
44.0%
-1.0%
-14.4%
-34.8%
01540500
Susquehanna River at
Danville, PA
(1991-1994)
27.5%
-6.3%
50.0%
24.6%
6.2%
-4.8%
01562000
Raystown Branch
Juniata River at Saxton,
PA
(1991-2000)
-78.2%
-29.7%
26.0%
-50.6%
7.0%
-22.6%
Table 21. Summary statistics for water quality for all stations - validation period (HSPF)
Station
Relative Percent Error TSS Load
TSS Concentration Median Error
Relative Percent Error TP Load
TP Concentration Median Error
Relative Percent Error TN Load
TN Concentration Median Error
01576000
Susquehanna
River at Marietta,
PA
(1980-90)
-0.6%
-27.3%
38.8%
5.1%
-7.1%
-17.2%
01540500
Susquehanna River at
Danville, PA
(1986-90)
-11.1%
-1 1 .2%
40.5%
20.7%
10.0%
5.9%
01562000
Raystown Branch
Juniata River at Saxton,
PA
(1990)
-89.7%
24.6%
21.5%
-12.2%
17.2%
7.4%
F-56
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a si
The SWAT model for the Susquehanna River watershed was set up with the ArcSWAT Version 2.3.3 interface
using the subwatersheds and stream network layers obtained from CBP, and other geospatial coverages described
above for the HSPF model. The precipitation and temperature data were preprocessed from BASINS Weather
Data Management (WDM) files to obtain the daily values.
The SWAT modeling process started with hydrology calibration, followed by calibration of sediment, and then
calibration of nitrogen and phosphorous. The USGS gage on the Raystown Branch of the Juniata River at Saxton,
PA (USGS 02050303) was used as the initial calibration location. The parameters were then transferred to the
entire Susquehanna River watershed and results were evaluated at the gages on the Susquehanna River at
Danville, PA (USGS 01540500) and on the Susquehanna River at Marietta, PA (USGS 01576000). While
hydrology parameters were readily transferable, water quality parameters, especially those related to sediment
needed some adjustment at the larger watershed scale.
No changes were made to the meteorological or land use base data for the SWAT model.
Though there are a number of reservoirs present in Susquehanna River watershed, they are located below the
calibration and validation locations. Hence the information regarding these reservoirs was not specifically
addressed in the SWAT model, consistent with the approach used for HSPF.
The point source data were specified for all the major active point sources in the Susquehanna study area. The
point source flows and concentrations for each facility in the watershed were assumed to be constant throughout
the simulation period. The data from the time period from 1991-2006 were compiled from the PCS database and
the median values were estimated. The facilities that were missing a total nitrogen, total phosphorus, and TSS
concentration value were filled with a typical pollutant concentration value from literature (Typical Pollutant
Concentration for NCPDI Discharge Categories -Improving Point Source Loading Data for Reporting National
Water Quality Indicators) prepared for Jim Home, EPA/OWM (Tetra Tech 1990) based on the SIC
classification. All POTWs were assumed to have secondary treatment. The median concentrations for the nutrient
species were estimated based on the values reported in Cheaspake Bay Phase 5 Model report for species
relationship for point sources and used in the model.
Similar to HSPF, hydrology calibration was performed at the USGS gage on the Raystown Branch of the Juniata
River at Saxton, PA (USGS 01562000). Though some adjustments are made at the major watershed level, a
spatial calibration approach was not adopted for Susquehanna River watershed SWAT modeling. The calibration
efforts were geared toward getting a closer match between simulated and observed flows at the outlet of the
calibration focus area and to limit the error statistics within the acceptable ranges listed in the QAPP.
F-57
-------
Land Use/Soil/Slope definition
A 5/10/5 percent threshold was used for land use/soil/slope in the SWAT model while defining the HRUs. Urban
(including current and future urban class types) classes were exempt from applying the thresholds.
Elevation Bands
The topographical analysis of Susquehanna River watersheds showed a significant range of elevations within
some individual modeling subwatersheds. This is likely to result in orographic variability in precipitation. Eight
elevation bands were used to account for the orographic effects on temperature and precipitation.
Calibration Parameters
The initial values of the parameters were set by ArcGIS based on various geospatial datasets and the defaults set
in the SWAT database. During the calibration process, adjustments were focused on the following parameters:
• ICN (Daily curve number calculation method) - In order to make the CN less dependent on the soil
moisture content and more dependent on the antecedent conditions, the ICN was set to 1 (CN as function
orET)
• FFCB (Initial soil water storage expressed as a fraction of field capacity water content)
• CN_FROZ (Frozen curve number active)
• Tlaps (Temperature laps rate)
• Flaps (Preciptation laps rate)
• ALAI_MIN (Minimum leaf area index for plant during dormant period)
• CurYr_Mat (Current age of trees)
• LAI_Ini (Initial leaf area index)
• EPCO (Plant uptake compensation factor)
• CN (Curve Number)
• TIMP (Snow pack temperature lag factor)
• ESCO (Soil evaporation compensation factor)
• CANMX (Maximum canopy storage)
• Alpha_bf (Baseflow alpha factor)
• GW_Delay (Groundwater delay)
• GWQMN (Threshold depth of water in the shallow aquifer for return flow to occur)
• REVAPMN (Threshold depth of water in the shallow aquifer for revap to occur)
• GW_REVAP (Groundwater revap coefficient)
• SURLAG (Surface runoff lag time)
• CH_K1 (Effective hydraulic conductivity in tributary channel alluvium)
• CH_K2 (Effective hydraulic conductivity in main channel alluvium)
• SFTMP (Snowfall temperature)
• SMTMP (Snow melt base temperature)
• SMFMX (Maximum melt rate for snow during year)
• SMFMN (Minimum melt rate for snow during year)
• CNCOEFF (Plant ET curve number coefficient)
• CH_N2 (Manning's n value for the main channel)
• CH_N1 (Manning's "n" value for the tributary channels)
• HRU_Slope (Average slope steepness)
• Slsubbsn (Average slope length)
F-58
-------
Initial calibrations were performed for the Raystown Branch of the Juniata River, comparing model results to data
from USGS 01562000 (Raystown Branch Juniata River At Saxton, PA), and are summarized in Figures 39
through 45 and Tables 22 and 23. The model fit is of good quality, but summer volumes are over estimated. The
model calibration period was set to the 10 water years from 10/01/1995 to 09/30/2005.
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1995 to 9/30/2005 )
•Avg Modeled Flow (Same Period)
35000
30000
_ 2500°
o, 20000
1 15000
LJ_
10000
5000 4-
IWT
- 2
Oct-04
Figure 39. Mean daily flow at USGS 01562000 Raystown Branch Juniata River at Saxton, PA
calibration period (SWAT).
5000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1995 to 9/30/2005 )
•Avg Modeled Flow (Same Period)
0-95
A-97
0-04
Figure 40. Mean monthly flow at USGS 01562000 Raystown Branch Juniata River at Saxton, PA
calibration period (SWAT).
F-59
-------
Avg Flow (10/1/1995 to 9/30/2005 )
• Line of Equal Value
Best-Fit Line
5000
100% -,
"g 90% -
5 80% -
en 70% -
O^ 60% -
g 50% -
j| 40% -
,S 30% -
jy 20% -
;> 10% -
0%
Avg Observed Flow (10/1/1995 to 9/30/2005 )
Avg Modeled Flow (10/1/1995 to 9/30/2005 )
-Line of Equal Value
IJ i
mil
0 1000 2000 3000 4000 5000
Average Observed Flow (cfs)
O-95 A-97 O-98 A-00 O-01
Month
A-03 O-04
Figure 41. Monthly flow regression and temporal variation at USGS 01562000 Raystown Branch
Juniata River at Saxton, PA - calibration period (SWAT).
• Avg Flow (10/1 /1995 to 9/30/2005)
• - - - • Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1995 to 9/30/2005)
•Avg Modeled Flow (Same Period)
2500
I
o
2000
-o 1500
J5
-------
Average Monthly Rainfall (in)
- Median Observed Flow (10/1/1995 to 9/30/2005)
[Observed (25th, 75th)
Modeled (Median, 25th, 75th)
3000 i
2500
2000
1500 -
1000 - —
500 - —
10 11
Figure 43. Seasonal medians and ranges at USGS 01562000 Raystown Branch Juniata River at
Saxton, PA - calibration period (SWAT).
Table 22. Seasonal summary at USGS 01562000 Raystown Branch Juniata River at Saxton,
PA - calibration period (SWAT)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
438.69
1024.44
1071.22
1255.25
1290.35
2177.77
1771.52
1310.00
895.83
326.40
278.62
907.63
187.00
647.50
682.00
475.00
900.00
1690.00
1180.00
739.50
448.00
207.00
157.50
124.50
119.00
149.75
280.75
260.00
470.00
1052.50
843.75
414.00
287.50
146.00
115.25
104.00
530.75
1170.00
1350.00
1180.00
1775.00
2742.50
2102.50
1777.50
879.00
324.00
339.75
360.25
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
692.82
1034.60
969.20
998.54
1120.65
1804.98
1425.47
1035.99
999.51
440.54
404.37
1129.77
371.51
754.67
765.27
505.18
539.96
1195.40
986.87
630.37
535.90
222.43
249.52
239.93
162.66
141.71
263.19
171.86
290.45
691.99
613.33
283.51
258.19
144.38
100.82
73.81
959.94
1346.72
1190.02
1224.18
935.31
2568.17
1666.15
1281.83
1082.22
426.51
519.83
1015.30
F-61
-------
t
I
D)
(0
•Observed Flow Duration (10/1/1995 to 9/30/2005 )
•Modeled Flow Duration (10/1/1995 to 9/30/2005 )
100000
10000
1000 -
100 -T
10%
20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90%
100%
Figure 44. Flow exceedence at USGS 01562000 Raystown Branch Juniata River at Saxton, PA
calibration period (SWAT).
o
o
ro
T3
o
I
-a
-------
Table 23. Summary statistics at USGS 01562000 Raystown Branch Juniata River at Saxton,
PA - calibration period (SWAT)
REACH OUTFLOW FROM OUTLET 26
10-Year Analysis F^riod: 10/1/1995 - 9/30/2005
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9)
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3):
Simulated Srjring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error -Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash^3ut£liffeJDpj5fficje^^
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
18.01
7.87
1.94
2.96
4.06
5.83
5.16
6.86
1.30
Error Statistics
-5.41
-5.67
-7.15
30.66
Observed Flow Gage
USGS 01562000 Raystown Bran
Hydrologic Unit Code: 2050303
Latitude: 40.21591249
Longitude: -78.2652901
Drainage Area (sq-rri): 756
ch Juniata River at Saxton, PA
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow Volume (1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
6.47 » | 30
-17.01
-13.07
-12.43
11.80
0.294
0.395
0.669
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
19.04
8.48
2.05
2.26
3.81
7.03
5.93
7.83
1.16
Clear [
Hydrology Validation
Hydrology validation for the Susquehanna River watershed model was performed at the same gage location but
for the period 10/1/1985 through 9/30/1995. Results are presented in Figure 49 through 52 and Tables 24 and 25.
The validation achieves a reasonable coefficient of model fit efficiency, but many of statistics show that simulated
values were underestimated compared to the observed values.
In general, the sign of the errors in the validation period are similar to those in the calibration period, but the
discrepancies are larger. Additional factors that may have contributed to the difference in the flows between the
calibration and validation period are:
• Drainage area of the observed USGS gage is about 6% higher than that of the calibration watershed.
• Increase in urban impervious surface areas between the 1980s and present.
F-63
-------
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1985 to 9/30/1995 )
•Avg Modeled Flow (Same Period)
30000
Apr-87 Oct-88 Apr-90 Oct-91
Date
Apr-93 Oct-94
Figure 46. Mean daily flow at USGS 01562000 Raystown Branch Juniata River at Saxton, PA
validation period (SWAT).
8000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1985 to 9/30/1995 )
•Avg Modeled Flow (Same Period)
O-85
A-87
O-94
Figure 47. Mean monthly flow at USGS 01562000 Raystown Branch Juniata River at Saxton, PA
validation period (SWAT).
F-64
-------
8000
Avg Flow (10/1/1985 to 9/30/1995 )
• Line of Equal Value
Best-Fit Line
2000 4000 6000 8000
Average Observed Flow (cfs)
100% -,
+
&
0)
O
ro
m
Avg Observed Flow (10/1/1985 to 9/30/1995 )
Avg Modeled Flow (10/1/1985 to 9/30/1995 )
-Line of Equal Value
A-87 O-88
A-90 O-91
Month
A-93 O-94
Figure 48. Monthly flow regression and temporal variation at USGS 01562000 Raystown Branch
Juniata River at Saxton, PA - validation period (SWAT).
• Avg Flow (10/1 /1985 to 9/30/1995)
• - - - • Line of Equal Value
Best-Fit Line
2500
I
o
2000
-o 1500
-------
• Observed (25th, 75th) Average Monthly Rainfall (in) -Median Observed Flow (10/1/1985 to 9/30/1995) Modeled (Median, 25th, 75th)
2500
10 11 12 1
8 9
Figure 50. Seasonal medians and ranges at USGS 01562000 Raystown Branch Juniata River at
Saxton, PA - validation period (SWAT).
Table 24. Seasonal summary at USGS 01562000 Raystown Branch Juniata River at Saxton,
PA - validation period (SWAT)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
351 .63
771 .90
936.58
1063.43
1214.57
2016.51
1772.93
1102.77
559.15
545.15
252.88
195.59
128.00
41 1 .00
603.50
562.00
754.00
1240.00
965.00
676.00
275.50
180.00
136.00
160.00
104.00
211.75
302.50
322.50
410.00
670.75
613.50
479.25
201.00
138.00
108.00
107.75
200.00
649.00
1157.50
1167.50
1495.00
2140.00
1935.00
1122.50
479.50
418.00
216.75
233.50
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
479.66
769.29
790.68
731 .64
615.95
1497.03
1417.56
652.39
552.94
739.08
441 .53
310.53
229.97
500.41
506.59
553.91
454.68
838.37
554.79
355.80
232.64
192.13
223.70
246.34
83.68
209.87
271.91
238.23
238.94
522.30
298.14
230.15
136.75
108.06
89.30
56.63
417.42
857.53
1027.22
916.06
827.25
1559.50
1296.67
793.79
476.75
694.82
404.09
481 .69
F-66
-------
t
I
D)
(0
(0
Q
•Observed Flow Duration (10/1/1985 to 9/30/1995 )
•Modeled Flow Duration (10/1/1985 to 9/30/1995 )
100000
10000
1000
100 -E
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Percent of Time that Flow is Equaled or Exceeded
Figure 51. Flow exceedence at USGS 01562000 Raystown Branch Juniata River at Saxton, PA
validation period (SWAT).
•Observed Flow Volume (10/1/1985 to 9/30/1995 )
Modeled Flow Volume (10/1/1985 to 9/30/1995 )
o
o
ro
T3
_a
O
_
o
o
ro
120%
100%
80%
60%
40%
20%
Oct-85
Apr-87
Oct-88 Apr-90
Oct-91
Apr-93 Oct-94
Figure 52. Flow accumulation at USGS 01562000 Raystown Branch Juniata River at Saxton, PA
validation period (SWAT).
F-67
-------
Table 25. Summary statistics at USGS 01562000 Raystown Branch Juniata River at Saxton,
PA - validation period (SWAT)
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 26
10-Year Analysis F^riod: 10/1/1985 - 9/30/1995
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9)
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error -Summer:
Seasonal volume error - Fall:
Seasonal volume error- Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
13.49
6.35
1.59
2.26
3.07
4.25
3.90
4.60
0.77
Error Statistics
-16.30
-12.87
-18.30
50.02
Observed Flow Gage
USGS 01562000 Raystown Branch Juniata River at Saxton, PA
Hydrologic Unit Code: 2050303
Latitude: 40.21591249
Longitude: -78.2652901
Drainage Area (sq-rri): 756
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
-1.00 » | 30
-33.35
-23.82
-29.17
42.83
0.415
0.383
0.664
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
16.12
7.77
1.82
1.51
3.10
6.38
5.12
6.49
0.54
Clear [
Hydrology Results for Larger Watershed
The parameters determined for the Raystown Branch of the Juniata River gage were transferred to the remainder
of the watershed without detailed spatial calibration. Tests of calibration and validation were pursued at a total of
three gages throughout the watershed, all of them at the outlet of 8-digit HUCs. Calibration results were generally
acceptable at all gages, as summarized in Table 26. The match between observed and predicted flow
corresponding to the largest watershed of the three gages (USGS 01576000, Susquehanna River at Marietta) is
shown in Figures 53 through 59 and Tables 27 and 28. Validation results were also generally in the acceptable
range for all the gages, as summarized in Table 29. It appears, however, that there are some systematic biases in
the model, including under-prediction of the 10 percent highest flows and winter flows, coupled with over-
prediction of summer flows
F-68
-------
Table 26. Summary statistics (percent error) for all stations - calibration period (SWAT)
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient of
Efficiency, E:
Monthly Nash-Sutcliffe Coefficient of
Efficiency,
01576000
Susquehanna River
at Marietta, PA
(1993-2003)
-9.74
2.03
-19.80
15.58
-2.46
-31.54
1.99
-22.08
9.52
0.451
0.669
01540500
Susquehanna River
at Danville, PA
(1993-2003)
-4.51
5.90
-11.24
11.96
-4.24
-38.92
26.18
-29.54
-12.84
0.327
0.573
01562000 Raystown
Branch Juniata River
at Saxton, PA
(1995-2005)
-5.41
-5.67
-7.15
30.66
6.47
-17.01
-13.07
-12.43
11.80
0.294
0.669
I
I
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1993 to 9/30/2003 )
•Avg Modeled Flow (Same Period)
600000
500000
400000
300000
200000
100000
T
TTNiT1
T 1—1___^ r—_|___T ,_
Oct-93 Apr-95
Oct-96 Apr-98 Oct-99
Date
Apr-01
Oct-02
Figure 53. Mean daily flow at USGS 01576000 Susquehanna River At Marietta, PA - calibration period
(SWAT).
F-69
-------
o
200000
150000 --
100000 -----
50000 -
O-93
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1993 to 9/30/2003
•Avg Modeled Flow (Same Period)
A-95
O-02
Figure 54. Mean monthly flow at USGS 01576000 Susquehanna River at Marietta, PA - calibration
period (SWAT).
Avg Flow (10/1/1993 to 9/30/2003 )
• Line of Equal Value
Best-Fit Line
200000
|
150000 -
T3
O
D)
100000 -
50000
y = 0.7524X + 5781.3
R2 = 0.7671
0 50000 100000 150000 200000
Average Observed Flow(cfs)
£
O
_
ro
m
o>
-t-j
TO
100% -,
90%
80%
70% -
60%
50%
Avg Observed Flow (10/1/1993 to 9/30/2003 )
Avg Modeled Flow (10/1/1993 to 9/30/2003 )
-Line of Equal Value
O-93 A-95 O-96
A-98 O-99
Month
A-01 O-02
Figure 55. Monthly flow regression and temporal variation at USGS 01576000 Susquehanna River at
Marietta, PA - calibration period (SWAT).
F-70
-------
Avg Flow (10/1/1993 to 9/30/2003)
• Line of Equal Value
Best-Fit Line
100000
•6
20000 40000 60000 80000 10000
0
Average Observed Flow (cfs)
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1993 to 9/30/2003)
Avg Modeled Flow (Same Period)
o
10 11 12 1 2 3 4 5 6 7 8 9
Month
Figure 56. Seasonal regression and temporal aggregate at USGS 01576000 Susquehanna River at
Marietta, PA - calibration period (SWAT).
• Observed (25th, 75th)
•Median Observed Flow (10/1/1993 to 9/30/2003)
Average Monthly Rainfall (in)
Modeled (Median, 25th, 75th)
120000
o
10 11 12 1
Figure 57. Seasonal medians and ranges at USGS 01576000 Susquehanna River at Marietta, PA-
calibration period (SWAT).
F-71
-------
Table 27. Seasonal summary at USGS 01576000 Susquehanna River at Marietta, PA - calibration
period (SWAT)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
18258.39
29811.13
43879.65
46873.61
45647.16
78282.90
74228.00
46236.13
32080.10
16327.52
15446.55
16977.37
10650.00
19300.00
31150.00
20700.00
35800.00
64900.00
63000.00
33750.00
24450.00
13800.00
7330.00
7645.00
7305.00
9002.50
19700.00
14425.00
25050.00
44725.00
43400.00
23100.00
16575.00
8122.50
5292.50
4757.50
19225.00
37325.00
54050.00
45175.00
50975.00
93150.00
90750.00
54575.00
38925.00
22075.00
15175.00
21650.00
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
22700.72
32093.73
34985.85
27590.78
29342.81
59826.28
74819.70
47504.67
33247.00
16522.99
17697.63
22211.19
16857.46
22972.19
28272.92
19412.47
19732.07
54243.33
68934.23
39005.05
26980.41
14343.05
9907.53
12176.50
11114.41
12402.51
19425.72
14285.67
13347.18
29365.91
41733.11
21727.35
14564.65
8986.70
6522.62
7015.26
28029.25
41600.68
39499.45
32425.04
30222.29
81082.47
95561 .49
63884.23
43437.04
20941 .60
17877.17
30420.94
•Observed Flow Duration (10/1/1993 to 9/30/2003 )
Modeled Flow Duration (10/1/1993 to 9/30/2003 )
1000000
o
D)
ro
(D
ro
Q
100000
10000
1000
10%
20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90%
100%
Figure 58. Flow exceedence at USGS 01576000 Susquehanna River at Marietta, PA - calibration
period (SWAT).
F-72
-------
o
o
ro
T3
o
^3
LL
T3
N
15
•Observed Flow Volume (10/1/1993 to 9/30/2003 )
Modeled Flow Volume (10/1/1993 to 9/30/2003 )
120%
100% -
80%
60%
40% - -
20% —
Apr-95
Oct-96
Apr-98
Oct-99
Apr-01
Oct-02
Figure 59. Flow accumulation at USGS 01576000 Susquehanna River at Marietta, PA - calibration
period (SWAT).
F-73
-------
Table 28. Summary statistics at USGS 01576000 Susquehanna River at Marietta, PA - calibration
period (SWAT)
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 75
10-Year Analysis F^riod: 10/1/1993 - 9/30/2003
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spjinj Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error -Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
18.22
5.86
3.34
2.47
3.94
5.06
6.75
5.70
0.84
Error Statistics
-9.74
* 2.03
15.58
Observed Flow Gage
USGS 01576000 Susquehanna River at Marietta, PA
Hydrologic Unit Code: 2050306
Latitude: 40.0545413
Longitude: -76.5307992
Drainage Area (sq-rri): 25990
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
-2.46 » | 30
-31.54
1.99
-22.08
9.52
0.451
0.443
0.641
30
30
20
50
Model accuracy increases
as E or E' approaches 1 .0
20.19
7.31
3.27
2.14
4.04
7.39
6.62
7.32
0.76
Clear [
F-74
-------
Table 29. Summary statistics (percent error) for all stations - validation period (SWAT)
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error -
Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of
Efficiency, E:
Baseline adjusted coefficient
(Garrick), E':
01576000 Susquehanna
River at Marietta, PA
(1983-1993)
-15.17
1.37
-24.62
22.63
-8.86
-34.95
-12.26
-28.54
29.42
0.485
0.657
01540500
Susquehanna River
at Danville, PA
(1983-1993)
-10.17
1.84
-17.11
17.20
-11.29
-40.55
9.55
-36.20
-15.18
0.372
0.573
01562000 Raystown
Branch Juniata River
at Saxton, PA
(1985-1995)
-16.30
-12.87
-18.30
50.02
-1.00
-33.35
-23.82
-29.17
42.83
0.415
0.664
Water Quality Calibration and Validation
Initial calibration and validation of water quality was done on data from the Raystown Branch Juniata River at
Saxton (USGS 01562000), using 1991-2000 for calibration and 1990 for validation. As with hydrology,
calibration was performed on the later period as this better reflects the land use included in the model. The start of
the validation period is constrained by data availability.
Calibration adjustments for sediment focused on the following parameters:
• SPCON (Linear parameters for estimating maximum amount of sediment that can be re-entrained during
channel sediment routing)
• SPEXP (Exponent parameter for calculating sediment re-entrained in channel sediment routing)
• CH_COV (Channel cover factor)
• CH_EROD (Channel credibility factor)
Various plots that compare TSS simulated by SWAT against the observed data are shown in Figures 60 through
63. The comparison statistics are provided in Tables 30 and 31. The fit to monthly sediment loads is generally
better than that obtained with the HSPF model. However, the correlation between observed and predicted
concentrations is weak - in part because many of the observed data are reported as less than a detection limit of 2
mg/L (plotted at 1 mg/L in Figure 63).
F-75
-------
TSS
100,000
CpCpCpCpCpCpCpCpCpCpCpCpCpCpCpCpCpCpCpCpOO
-Regression Loads
-Simulated Loads
Figure 60. Fit for monthly load of TSS at USGS 01562000 Raystown Branch Juniata River at Saxton,
PA (SWAT).
Table 30. Model fit statistics (observed minus predicted) for monthly sediment loads using
stratified regression - USGS 01562000 Raystown Branch Juniata River at Saxton,
PA (SWAT)
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1991-2000)
-10.1%
80%
11.1%
Validation period
(1990)
-33.6%
67%
41.1%
F-76
-------
Raystown Branch Juniata River at Saxton, PA 1991-2002
•o
«
c
o
•o
ra
(0
(0
0
0.0001
0.00001
100000
Flow, cfs
Simulated A Observed'
Fbw er (Simulated)
Power (Observed)
Figure 61. Power plot for observed and simulated TSS at USGS 01562000 Raystown Branch Juniata
River at Saxton, PA - calibration period (SWAT).
Raystown Branch Juniata River at Saxton, PA 1986-1990
10000
» Simulated A Observed
Power (Simulated)
Power (Observed)
Figure 62. Power plot for observed and simulated TSS at USGS 01562000 Raystown Branch Juniata
River at Saxton, PA - validation period (SWAT).
F-77
-------
Raystown Branch Juniata River at Saxton, PA 1990-2002
Paired data ^-^ Equal fit
1000
T 100
0.01
0.01
0.1 1 10
Observed TSS (mg/L)
100
1000
Figure 63. Correlation between observed and predicted TSS concentration at USGS 01562000
Raystown Branch Juniata River at Saxton, PA (SWAT).
Table 31. Relative errors (observed minus simulated) for TSS concentrations at USGS
01562000 Raystown Branch Juniata River at Saxton, PA (SWAT)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1991-2000)
106
-26.49%
12.04%
Validation period
(1990)
14
73.88%
73.84%
Calibration adjustments for the simulation of total phosphorus and total nitrogen focused on the following
parameters:
• PHOSKD (phosphorus soil partitioning coefficient)
• NPERCO (nitrogen percolation coefficient)
• PPERCO (phosphorus percolation coefficient)
• SOL_NO3 (initial nitrate concentration in soil layers)
• SOL_ORGN (initial organic nitrogen concentration in soil layers)
• SOL_SOLP (initial soluble phosphorus concentration in soil layers)
• SOL_ORGP (initial organic phosphorus concentration in soil layers)
Various plots that compare total phosphorous simulated by SWAT against the observed data are shown in Figures
64 through 67. The comparison statistics are provided in Tables 32 and 33. Similarly, the results corresponding to
total nitrogen are shown in Figures 68 though 71 and Tables 34 and 35. The model representation of total load is
generally acceptable, although better for phosphorus than for nitrogen. As with the HSPF application, phosphorus
F-78
-------
loads and concentrations tend to be overestimated at lower flows, likely as a result of the simplified representation
of point sources.
Total P
100
o
E
In
I
-Regression Loads
-Simulated Loads
Figure 64. Fit for monthly load of total phosphorous at USGS 01562000 Raystown Branch Juniata
River at Saxton, PA (SWAT).
Table 32. Model fit statistics (observed minus predicted) for monthly phosphorus loads using
stratified regression - USGS 01562000 Raystown Branch Juniata River at Saxton,
PA (SWAT)
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1991-2000)
-0.5%
73%
48.6%
Validation period
(1990)
9.2%
54%
44.6%
F-79
-------
Raystown Branch Juniata River at Saxton, PA 1991-2002
100
10
100 1000
Flow, cfs
10000
100000
• Simulated A Observed ^^~Power (Simulated) ^^"Fbwer (Observed)
Figure 65. Power plot for observed and simulated total phosphorus at USGS 01562000 Raystown
Branch Juniata River at Saxton, PA - calibration period (SWAT)
Raystown Branch Juniata River at Saxton, PA 1986-1990
•o
ra
o
0.01
0.001
10000
» Simulated A Observed
Power (Simulated)
Power (Observed)
Figure 66. Power plot for observed and simulated total phosphorus at USGS 01562000 Raystown
Branch Juniata River at Saxton, PA - validation period (SWAT)
F-80
-------
Raystown Branch Juniata River at Saxton, PA
1993 1994 1995 1996 1997
Year
1998
1999
Figure 67. Time series plot of total phosphorus concentration at USGS 01562000 Raystown Branch
Juniata River at Saxton, PA (SWAT).
Table 33. Relative errors (observed minus predicted), total phosphorus concentration, USGS
01562000 Raystown Branch Juniata River at Saxton, PA (SWAT)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1991-2000)
122
-377.89%
-71.74%
Validation period
(1990)
18
-58.65%
-9.69%
F-81
-------
Total N
1,400
-Averaging Loads
-Simulated Loads
Figure 68. Fit for monthly load of total nitrogen at USGS 01562000 Raystown Branch Juniata River at
Saxton, PA (SWAT)
Table 34. Model fit statistics (observed minus predicted) for monthly total nitrogen loads
using averaging estimator- USGS 01562000 Raystown Branch Juniata River at
Saxton, PA (SWAT)
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1991-2000)
28.6%
45%
19.7%
Validation period
(1990)
43.9%
53%
58.3%
F-82
-------
Raystown Branch Juniata River at Saxton, PA 1991-2002
1000
10
100 1000
Flow, cfs
10000
100000
• Simulated A Observed ^^™Power (Simulated) ^^™Power (Observed)
Figure 69. Power plot for observed and simulated total nitrogen at USGS 01562000 Raystown Branch
Juniata River at Saxton, PA - calibration period (SWAT).
100 -
| 10
-------
30
25
20
15
10
Raystown Branch Juniata River at Saxton, PA
• Simulated A Observed
1986 1987 1988 1989 1990 1991 1992
Year
Figure 71. Time series plot of total nitrogen concentration at USGS 01562000 Raystown Branch
Juniata River at Saxton, PA (SWAT).
Table 35. Relative errors (observed minus predicted), total nitrogen concentration, USGS
01562000 Raystown Branch Juniata River at Saxton, PA (SWAT)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1991-2002)
13
-13.82%
21.58%
Validation period
(1990)
6
32.07%
36.92%
Water Quality Results for Larger Watershed
Similar to hydrology calibration, water quality results were compared at other gages. Note that in contrast to the
HSPF model, water quality for the SWAT model for Susquehanna River at Marietta, PA was calibrated to the
stratified regression monthly load estimates for the entire 1991-2005 period, although the observed data stop with
1995. Summary statistics for the water quality calibration and validation at other stations in the watershed are
provided in Tables 36 and 37.
F-84
-------
Table 36. Summary statistics for water quality for all stations - calibration period (SWAT)
Station
Relative Percent Error TSS
Load
TSS Concentration Median
Error
Relative Percent Error TP
Load
TP Concentration Median
Error
Relative Percent Error TN
Load
TN Concentration Median
Error
01576000
Susquehanna River at
Marietta, PA
(1991-1995)
25.2%
-34.0%
-1 1 .4%
-23.2%
-14.0%
-39.2%
01540500 Susquehanna
River at Danville, PA
(1991-1994)
28.4%
-19.9%
22.6%
-2.4%
-1.6%
-7.8%
01562000 Raystown
Branch Juniata River at
Saxton, PA (1991 -2000)
-10.1%
12.04%
-0.5%
71.74%
28.6%
21.6%
Table 37. Summary statistics for water quality for all stations - validation period (SWAT)
Station
Relative Percent Error TSS
Load
TSS Concentration Median
Error
Relative Percent Error TP
Load
TP Concentration Median
Error
Relative Percent Error TN
Load
TN Concentration Median
Error
01576000
Susquehanna River at
Marietta, PA
(1980-1990)
15.2%
-22.8%
0.9%
-21.2%
-0.1%
-16.6%
01540500 Susquehanna
River at Danville, PA
(1986-1990)
17.1%
-3.4%
10.9%
-21.9%
15.7%
16.0%
01562000 Raystown
Branch Juniata River at
Saxton, PA (1990)
-33.6%
-73.8%
9.2%
9.7%
43.9%
36.9%
F-85
-------
Preston, S.D., V.J. Bierman, Jr., and S.E. Silliman. 1989. An evaluation of methods for the estimation of tributary
mass loads. Water Resources Research. 25(6): 1379-1389.
SRBC (Susquehanna River Basin Commission). 2008. Comprehensive Plan for the Water Resources of the
Susquehanna River Basin. Harrisburg, PA.
http://www.srbc.net/planning/ComprehensivePlanwithoutAppendices62012.pdf (Accessed June, 2012)
Tetra Tech. 1990. Typical Pollutant Concentration for NCPDI Discharge Categories -Improving Point Source
Loading Data for Reporting National Water Quality Indicators. Prepared for Jim Home, EPA/OWM.
USEPA (United States Environmental Protection Agency). 2010. Chesapeake Bay Phase 5 Community
Watershed Model. (In preparation). U.S. Environmental Protection Agency, Chesapeake Bay Program Office,
Annapolis, MD. http://www.chesapeakebay.net/modeljhase5.aspx7menuitenrf6169.
F-86
-------
Appendix G
Model Configuration, Calibration and
Validation
Basin: Minnesota River (Minn)
G-l
-------
Contents
Watershed Background G-7
Water Body Characteristics G-7
Soil Characteristics G-9
Land Use Representation G-9
Point Sources G-13
Meteorological Data G-15
Watershed Segmentation G-18
Calibration Data and Locations G-20
HSPF Modeling G-21
Changes Made to Base Data Provided G-21
Assumptions G-22
Hydrology Calibration G-23
Hydrology Validation G-28
Hydrology Results for Larger Watershed G-34
Water Quality Calibration and Validation G-40
Water Quality Results for Larger Watershed G-51
SWAT Modeling G-53
Changes Made to Base Data Provided G-53
Assumptions G-53
Hydrology Calibration G-53
Hydrology Validation G-59
Hydrology Results for Larger Watershed G-64
Water Quality Calibration and Validation G-71
Water Quality Results for Larger Watershed G-80
References G-83
G-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes G-ll
Table 2. Land use distribution for the Minnesota River watershed (2001 NLCD) (mi2) G-12
Table 3. Major point source discharges in the Minnesota River watershed G-13
Table 4. Precipitation stations for the Minnesota River watershed model G-15
Table 5. Calibration and validation locations in the Minnesota River basin G-20
Table 6. Seasonal summary at USGS 05317000 Cottonwood River near New Ulm, MN - calibration
period (HSPF) G-26
Table 7. Summary statistics at USGS 05317000 Cottonwood River near New Ulm, MN - calibration
period (HSPF) G-28
Table 8. Seasonal summary at USGS 05317000 Cottonwood River near New Ulm, MN - validation
period (HSPF) G-31
Table 9. Summary statistics at USGS 05317000 Cottonwood River near New Ulm, MN - validation
period (HSPF) G-34
Table 10. Summary statistics (percent error): all stations - calibration period (HSPF) G-3 5
Table 11. Seasonal summary at USGS 05330000 Minnesota River near Jordan, MN - calibration period
(HSPF) G-38
Table 12. Summary statistics: all stations - validation period (HSPF) G-40
Table 13. Model fit statistics (observed minus predicted) for monthly TSS loads using stratified regression
(HSPF) G-42
Table 14. Relative errors (observed minus predicted), TSS concentration at USGS 05317000 Cottonwood
River (HSPF) G-44
Table 15. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression (HSPF) G-45
Table 16. Relative errors (observed minus predicted), total phosphorus concentration at USGS 05317000
Cottonwood River (HSPF) G-48
Table 17. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using averaging
estimator (HSPF) G-49
Table 18. Relative errors (observed minus predicted), total nitrogen concentration at USGS 05317000
Cottonwood River (HSPF) G-51
Table 19. Summary statistics for water quality: all stations - calibration period 1993-2002 (HSPF) G-51
Table 20. Summary statistics for water quality: all stations - validation period 1986-1992 (HSPF) G-52
Table 21. Seasonal summary at USGS 05317000 Cottonwood River near New Ulm, MN - calibration
period (SWAT) G-57
Table 22. Summary statistics at USGS 05317000 Cottonwood River near New Ulm, MN - calibration
period (SWAT) G-59
Table 23. Seasonal summary at USGS 05317000 Cottonwood River near New Ulm, MN - validation
period (SWAT) G-62
Table 24. Summary statistics at USGS 05317000 Cottonwood River near New Ulm, MN - validation
period (SWAT) G-64
Table 25. Summary statistics (percent error): all stations - calibration period (SWAT) G-65
Table 26. Seasonal summary at USGS 05330000 Minnesota River near Jordan, MN - calibration period
(SWAT) G-68
Table 27. Summary statistics at USGS 05330000 Minnesota River near Jordan, MN - calibration period
(SWAT) G-70
Table 28. Summary statistics: all stations - validation period (SWAT) G-71
Table 29. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 05317000 Cottonwood River (SWAT) G-72
Table 30. Relative errors (observed minus predicted), TSS concentration at USGS 05317000
Cottonwood River (SWAT) G-74
Table 31. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 05317000 Cottonwood River (SWAT) G-75
-------
Table 32. Relative errors (observed minus predicted), total phosphorus concentration at USGS
05 317000 Cottonwood River (SWAT) G-77
Table 33. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 05317000 Cottonwood River (SWAT) G-78
Table 34. Relative errors (observed minus predicted), total nitrogen concentration at USGS 05317000
Cottonwood River (SWAT) G-80
Table 35. Summary statistics for water quality at all stations - calibration period 1993-2002 (SWAT).. G-82
Table 36. Summary statistics for water quality at all stations - validation period 1986-1992 (SWAT)... G-82
Figures
Figure 1. Location of the Minnesota River watershed G-8
Figure 2. Land use in the Minnesota River watershed G-10
Figure 3. Major point sources in the Minnesota River watershed G-15
Figure 4. Weatherstations for the Minnesota River watershed model G-17
Figure 5. Model segmentation and USGS stations utilized for the Minnesota River watershed G-19
Figure 6. Mean daily flow at USGS 05317000 Cottonwood River near New Ulm, MN - calibration period
(HSPF) G-24
Figure 7. Mean monthly flow at USGS 05317000 Cottonwood River near New Ulm, MN - calibration period
(HSPF) G-24
Figure 8. Mean monthly flow regression and temporal variation at USGS 05317000 Cottonwood River near
New Ulm, MN - calibration period (HSPF) G-25
Figure 9. Seasonal regression and temporal aggregate at USGS 05317000 Cottonwood River near New Ulm,
MN - calibration period (HSPF) G-25
Figure 10. Seasonal medians and ranges at USGS 05317000 Cottonwood River near New Ulm, MN -
calibration period (HSPF) G-26
Figure 11. Flow exceedance at USGS 05317000 Cottonwood River near New Ulm, MN - calibration period
(HSPF) G-27
Figure 12. Flow accumulation at USGS 05317000 Cottonwood River near New Ulm, MN - calibration period
(HSPF) G-27
Figure 13. Mean daily flow at USGS 05317000 Cottonwood River near New Ulm, MN - validation period
(HSPF) G-29
Figure 14. Mean monthly flow at USGS 05317000 Cottonwood River near New Ulm, MN - validation
period (HSPF) G-30
Figure 15. Monthly flow regression and temporal variation at USGS 05317000 Cottonwood River near
New Ulm, MN - validation period (HSPF) G-30
Figure 16. Seasonal regression and temporal aggregate at USGS 05317000 Cottonwood River near New
Ulm, MN - validation period (HSPF) G-30
Figure 17. Seasonal medians and ranges at USGS 05317000 Cottonwood River near New Ulm, MN -
validation period (HSPF) G-31
Figure 18. Flow exceedance at USGS 05317000 Cottonwood River near New Ulm, MN - validation period
(HSPF) G-32
Figure 19. Flow accumulation at USGS 05317000 Cottonwood River near New Ulm, MN - validation period
(HSPF) G-33
Figure 20. Mean daily flow simulation at USGS 05330000 Minnesota River near Jordan, MN - calibration
period (HSPF) G-36
Figure 21. Mean monthly flow simulation at USGS 05330000 Minnesota River near Jordan, MN - calibration
period (HSPF) G-36
Figure 22. Monthly flow regression and temporal variation at USGS 05330000 Minnesota River near Jordan,
MN - calibration period (HSPF) G-37
CM
-------
Figure 23. Seasonal regression and temporal aggregate at USGS 05330000 Minnesota River near Jordan,
MN - calibration period (HSPF) G-37
Figure 24. Seasonal medians and ranges at USGS 05330000 Minnesota River near Jordan, MN - calibration
period (HSPF) G-38
Figure 25. Flow exceedence at USGS 05330000 Minnesota River near Jordan, MN - calibration period
(HSPF) G-39
Figure 26. Flow accumulation at USGS 05330000 Minnesota River near Jordan, MN - calibration period
(HSPF) G-39
Figure 27. Fit for monthly load of TSS at USGS 05317000 Cottonwood River (HSPF) G-42
Figure 28. Power plot for observed and simulated TSS at USGS 05317000 Cottonwood River - calibration
period (HSPF) G-43
Figure 29. Power plot for observed and simulated TSS at USGS 05317000 Cottonwood River - validation period
(HSPF) G-43
Figure 30. Time series plot of TSS concentration at USGS 05317000 Cottonwood River (HSPF) G-44
Figure 31. Fit for monthly load of total phosphorus at USGS 05317000 Cottonwood River (HSPF) G-45
Figure 32. Power plot for observed and simulated total phosphorus at USGS 05317000 Cottonwood River -
calibration period (HSPF) G-46
Figure 33. Power plot for observed and simulated total phosphorus at USGS 05317000 Cottonwood River -
validation period (HSPF) G-47
Figure 34. Time series plot of total phosphorus concentration at USGS 05317000 Cottonwood River
(HSPF) G-47
Figure 35. Fit for monthly load of total nitrogen at USGS 05317000 Cottonwood River (HSPF) G-48
Figure 36. Power plot for observed and simulated total nitrogen at USGS 05317000 Cottonwood River -
calibration period (HSPF) G-49
Figure 37. Power plot for observed and simulated total nitrogen at USGS 05317000 Cottonwood River -
validation period (HSPF) G-50
Figure 38. Time series plot of total nitrogen concentration at USGS 05317000 Cottonwood River (HSPF).... G-50
Figure 39. Mean daily flow at USGS 05317000 Cottonwood River near New Ulm, MN - calibration period
(SWAT) G-55
Figure 40. Mean monthly flow at USGS 05317000 Cottonwood River near New Ulm, MN - calibration period
(SWAT) G-55
Figure 41. Monthly flow regression and temporal variation at USGS 05317000 Cottonwood River near New
Ulm, MN - calibration period (SWAT) G-56
Figure 42. Seasonal regression and temporal aggregate at USGS 05317000 Cottonwood River near New Ulm,
MN - calibration period (SWAT) G-56
Figure 43. Seasonal medians and ranges at USGS 05317000 Cottonwood River near New Ulm, MN -
calibration period (SWAT) G-57
Figure 44. Flow exceedance at USGS 05317000 Cottonwood River near New Ulm, MN - calibration period
(SWAT) G-58
Figure 45. Flow accumulation at USGS 05317000 Cottonwood River near New Ulm, MN - calibration period
(SWAT) G-58
Figure 46. Mean daily flow at USGS 05317000 Cottonwood River near New Ulm, MN - validation period
(SWAT) G-60
Figure 47. Mean monthly flow at USGS 05317000 Cottonwood River near New Ulm, MN - validation period
(SWAT) G-60
Figure 48. Monthly flow regression and temporal variation at USGS 05317000 Cottonwood River near New
Ulm, MN - validation period (SWAT) G-61
Figure 49. Seasonal regression and temporal aggregate at USGS 05317000 Cottonwood River near New Ulm,
MN - validation period (SWAT) G-61
Figure 50. Seasonal medians and ranges at USGS 05317000 Cottonwood River near New Ulm, MN -
validation period (SWAT) G-62
Figure 51. Flow exceedance at USGS 05317000 Cottonwood River near New Ulm, MN - validation period
(SWAT) G-63
-------
Figure 52. Flow accumulation at USGS 05317000 Cottonwood River near New Ulm, MN - validation period
(SWAT) G-63
Figure 53. Monthly flow simulation: USGS 05330000 Minnesota River near Jordan, MN - calibration period
(SWAT) G-66
Figure 54. Monthly flow regression and temporal variation at USGS 05330000 Minnesota River near Jordan,
MN - calibration period (SWAT) G-66
Figure 55. Seasonal regression and temporal aggregate at USGS 05330000 Minnesota River near Jordan, MN -
calibration period (SWAT) G-67
Figure 56. Seasonal medians and ranges at USGS 05330000 Minnesota River near Jordan, MN - calibration
period (SWAT) G-67
Figure 57. Flow exceedence at USGS 05330000 Minnesota River near Jordan, MN - calibration period
(SWAT) G-68
Figure 58. Flow accumulation at USGS 05330000 Minnesota River near Jordan, MN - calibration period
(SWAT) G-69
Figure 59. Fit for monthly load of TSS at USGS 05317000 Cottonwood River (SWAT) G-72
Figure 60. Power plot for observed and simulated TSS at USGS 05317000 Cottonwood River - calibration
period (SWAT) G-73
Figure 61. Power plot for observed and simulated TSS at USGS 05317000 Cottonwood River - validation
period (SWAT) G-73
Figure 62. Time series plot of TSS concentration at USGS 05317000 Cottonwood River (SWAT) G-74
Figure 63. Fit for monthly load of total phosphorus at USGS 05317000 Cottonwood River (SWAT) G-75
Figure 64. Power plot for observed and simulated total phosphorus at USGS 05317000 Cottonwood River -
calibration period (SWAT) G-76
Figure 65. Power plot for observed and simulated total phosphorus at USGS 05317000 Cottonwood River -
validation period (SWAT) G-77
Figure 66. Time series plot of total phosphorus concentration at USGS 05317000 Cottonwood River
(SWAT) G-77
Figure 67. Fit for monthly load of total nitrogen at USGS 05317000 Cottonwood River (SWAT) G-78
Figure 68. Power plot for observed and simulated total nitrogen at USGS 05317000 Cottonwood River -
calibration period (SWAT) G-79
Figure 69. Power plot for observed and simulated total nitrogen at USGS 05317000 Cottonwood River -
validation period (SWAT) G-80
Figure 70. Time series plot of total nitrogen concentration at USGS 05317000 Cottonwood River (SWAT).. G-80
G-6
-------
Water Body Characteristics
The Minnesota River (HUC 0702) constitutes 12 HUCSs, covering about 16,900 mi2, predominantly in the
Western Corn Belt ecoregion (Figure 1). The Minnesota River Basin is located primarily in southern Minnesota
with headwaters in South Dakota and is tributary to the Upper Mississippi River. Major cities include Mankato
and Minneapolis, MN.
Most of the watershed was originally native prairie and pothole wetlands. Intensive agricultural development
began in the mid to late 19th century, and the watershed is now part of the corn belt, with the majority of the land
area converted to corn-soybean rotation and other types of agriculture. Conversion of many parts of the watershed
to agriculture required enhancement of drainage through ditches and subsurface tile drains. These drainage ditches
and tile drains have resulted in a strong alteration of the hydrology by human modifications.
Precipitation, evapotranspiration, and air temperature exhibit a gradient from southwest to northeast, with a
warmer, wetter climate to the southeast and a colder, drier climate to the northwest. Topography is flat to gently
rolling, except in the area of the high bluffs adjoining the Minnesota River mainstem, created by glacial runoff.
The dominant land use in the watershed is row crop agriculture (72 percent; mostly in corn / soybean rotation),
with another 6 percent in pasture and hay. The surficial geology of the watershed consists of glacial till, moraines,
and lake deposits and in its natural state was poorly drained with numerous lakes and wetlands. This topography
was largely drained to establish agriculture and the use of tile drainage is now prevalent in the watershed.
The maximum streamflow occurs in spring and early summer as a result of rain and melting snow. Streamflow
variation is greatest during late summer and fall, when precipitation ranges from drought conditions to locally
heavy rains. Streamflow varies least during winter, when groundwater discharge to streams is dominant. Flow
from the upper portions of the Minnesota River is influenced by Lac qui Parle, a U.S. Army Corps of Engineers
impoundment of the Minnesota River near Montevideo, MN.
Water quality in the basin is affected by agricultural activities and point sources. The combination of extensive
corn production and tile drainage results in a high risk of nitrogen export. Erosion, sedimentation, and turbidity
problems are also frequent in the basin; however, analysis of radionuclide data suggests that only about a third of
the sediment transported in stream channels is derived from upland sheet and rill erosion, with the remainder
coming from gullies (often associated with tile drain outfalls), bank erosion, and bluff collapse.
The watershed does not contain major reservoirs. However, there are a number of smaller lakes and reservoirs that
influence flow in this low gradient terrain. As stated above, the Minnesota River's headwaters are located at Big
Stone Lake, a natural lake with multiple outlets. The river proceeds through a series of impoundments in the
upper reaches to Lac qui Parle, a US Army Corps of Engineers impoundment upstream of Montevideo,
Minnesota. Irrigation and groundwater pumping in the watershed are generally small (although irrigation is
somewhat more important in the western portions of the watershed). These factors are ignored for the purposes of
the 20 Watershed model.
G-7
-------
Hydrography
Water (Nat. Atlas Dataset)
US Census Populated Places
^B Municipalities (pop i 50.0001
County Boundaries
I I Watershed with HUCSs
North Dakota
(07020002) Cn.ppewa
(07020005)
Upper Minnesota
wk-Yellow Medicine
(07020004)
Redwood
(07020006)
Lower Minnesota
(070200012)
die Minneso
(07020007)
Cotton wood
(07020008)
South Dakota
Watonwan
(07020010)
Le Sueur
(07020011)
Blue Earth
(07020009)
GCRP Model Areas - Minnesota River Basin
Base Map
Figure 1. Location of the Minnesota River watershed.
G-8
-------
Soil Characteristics
Soils in the watershed, as described in STATSGO soil surveys, fall primarily into hydrologic soil groups (HSGs)
B (moderately high infiltration capacity) and D (low infiltration capacity). However, these designations are
believed to be somewhat misleading in the Minnesota River watershed, as most agricultural activity benefits from
subsurface tile drainage, which has been extensively installed since the 19th century. It is our impression that soils
in the basin have received a rating as HSG B primarily because of pre-existing drainage and would more likely be
classified as B/D (moderately high surface infiltration with a restricting layer) in its absence. This was
substantiated by a previously-developed HSPF model of the basin for TMDL development (Tetra Tech 2008).
The TMDL model obtained an excellent fit to basin hydrology without accounting for different HSGs, using
parameters typical of HSG D. Therefore, the 20 Watershed HSPF model was constructed with only a small
increase in infiltration capacity between group B and group D soils.
The SWAT model relies on a curve number approach rather than direct simulation of infiltration. SWAT uses
information drawn directly from the soils data layer to populate the model.
The soil survey data was also used to establish geographic distributions of infiltration rate and available soil water
capacity. These were used to index the spatial distribution of infiltration and lower zone soil nominal storage
capacities in the 20 Watershed model, as was done in the TMDL model.
Land Use Representation
Land use/cover in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage and is
predominantly row crop agriculture (Figure 2). Pasture and wetlands are more predominant in the upstream,
western portions of the watershed. A variety of small municipalities are present throughout the watershed;
however, major urban development is found only in the downstream Minneapolis-St. Paul metropolitan area.
G-9
-------
Interstate
j^H Water (Nat. Atlas Dataset)
3 County Boundaries
Watershed
2001 NLCD Land Use
| Open water
Developed, open space
Developed, low intensity
Developed, medium intensity
Developed, high intensity
Barren land
Deciduous forest
Evergreen forest
North Dakota
] Scrub/shrub
^ Grassland/hertiaceous
J Pasture/hay
J Cultivated crops
I \Afoodywetlands
GCRP Model Areas - Minnesota River Basin
Land Use Map
Figure 2. Land use in the Minnesota River watershed.
G-10
-------
National Land Cover Database (NLCD) land cover classes were aggregated according to the scheme shown in
Table 1 for representation in the 20 Watershed model, then overlain with the soils HSG grid. For HSPF, pervious
and impervious lands are specified separately, so only one developed pervious class is used, along with an
impervious class. HSPF simulates impervious land areas separately from pervious land. Impervious area
distributions were determined from the NLCD Urban Impervious data coverage. Specifically, percent impervious
area was calculated over the entire watershed for each of the four developed land use classes. These percentages
were then used to separate out impervious land. NLCD impervious area data products are known to underestimate
total imperviousness in rural areas; however, the model properly requires connected impervious area, not total
impervious area, and the NLCD tabulation is assumed to provide a reasonable approximation of connected
impervious area. In SWAT, different developed land classes are specified separately. In HSPF the WATER,
BARREN, DEVPERV, and WETLAND classes are not subdivided by HSG; SWAT uses the built-in HRU
overlay mechanism in the ArcSWAT interface.
Table 1.Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area
usually accounted for as
reach area
Deciduous
Evergreen
Mixed
Emergent & woody
wetlands
Aquatic bed wetlands (not
emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
HSPF (after processing)
WATER
BARREN, Assume HSG D
DEVPERV;
IMPERV
BARREN (D)
FOREST (A,B,C,D)
SHRUB (A,B,C,D)
GRASS (A,B,C,D), BARREN (D)
GRASS (A,B,C,D)
AGRI (A,B,C,D)
WETLAND, Assume HSG D
WATER
The distribution of land use in the watershed is summarized in Table 2.
G-ll
-------
Table 2. Land use distribution for the Minnesota River watershed (2001 NLCD) (mi2)
HUC8
watershed
Upper
Minnesota
07020001
Pomme De
Terre
07020002
Lac Qui
Parle
07020003
Hawk-
Yellow
Medicine
07020004
Chippewa
07020005
Redwood
07020006
Middle
Minnesota
07020007
Cottonwood
07020008
Blue Earth
07020009
Watonwan
07020010
Le Sueur
07020011
Lower
Minnesota
07020012
Total
Open
water
90.5
77.0
16.6
31.6
123.9
11.6
35.8
10.5
22.4
12.0
22.7
51.3
505.9
Developed9
Open
space
86.7
40.7
46.5
95.5
86.3
35.6
67.5
63.8
93.2
45.9
61.6
84.4
807.6
Low
density
10.6
5.8
3.0
20.7
15.2
6.8
19.0
8.3
11.5
6.1
7.4
88.7
203.3
Medium
density
3.0
1.2
0.8
4.4
1.9
2.1
7.9
2.1
2.8
1.6
1.6
45.8
75.1
High
density
0.7
0.3
0.3
1.1
0.6
0.6
2.8
0.4
0.9
0.3
0.6
21.8
30.4
Barren
land
2.6
0.7
0.6
2.6
1.4
0.9
4.7
0.7
1.0
0.4
0.6
0.7
16.9
Forest
36.0
47.2
9.7
35.5
94.2
6.9
59.3
16.5
13.0
9.6
16.1
144.2
488.2
Shru bland
313.8
40.7
114.4
56.7
61.2
43.4
17.1
26.9
32.3
7.9
26.9
45.6
786.8
Pasture/Hay
264.9
61.1
99.6
75.6
179.4
22.1
40.0
22.5
10.4
4.5
15.9
202.9
999.0
Cultivated
1,132.0
560.5
707.5
1,669.5
1,413.3
547.4
1,082.0
1,113.5
1,345.8
745.0
918.5
1,011.4
12,246.4
Wetland
175.8
54.6
75.9
92.5
110.6
21.9
87.0
45.2
39.7
26.6
38.5
61.3
829.6
Total
2,116.6
889.7
1,074.9
2,085.8
2,088.0
699.4
1,422.8
1,310.3
1,573.1
859.9
1,110.4
1,758.1
16,989.1
aThe percent imperviousness applied to each of the developed land uses is as follows: open space (6.59%), low density (29.20%), medium density (55.01%), and high
density (83.31%).
G-12
-------
The HSPF model is set up on a hydrologic response unit (HRU) basis. For HSPF, HRUs were formed
from an intersection of land use and hydrologic soil group, then further subdivided by precipitation
gage. Because slopes in the basin are relatively mild, HSPF HRUs were not further subdivided by slope.
However, average slopes (which tend to correlate with soils) were calculated for each HRU. The water
land use area was adjusted to prevent double counting with area described in HSPF reaches. SWAT
HRUs are formed from an intersection of land use and SSURGO major soils.
Point Sources
There are numerous point source discharges in the watershed, including approximately 70 mechanical wastewater
treatment plants and various industrial discharges. In addition, Minnesota PCA has identified approximately 70
stabilization ponds in the watershed that receive wastewater, primarily serving small communities, and that
discharge seasonally to the stream network.
For the purposes of 20 Watershed modeling, only the 13 major dischargers, with a design flow greater than 1
MGD are included in the simulation (Table 3 and Figure 3). The total of all discharges in the basin is believed to
be in the range of 100 MGD. The major dischargers account for about 80 percent of that total, so the effect of the
omitted sources distributed throughout the watershed will be relatively small, except during extreme low flow
conditions. The major dischargers are represented at long-term average flows, without accounting for changes
over time or seasonal variations.
Table 3. Major point source discharges in the Minnesota River watershed
NPDES ID
MN0022535
MN0030171
SD0020371
MN0020133
MN0022179
MN0057037
MN0030066
MN0030112
MN0025267
MN0024759
MN0024040
MN0029882
MN0030007
MN0020796
MN0020150
MN0025259
Name
SAINT PETER
MANKATO
MILBANK-CITYOF
MONTEVIDEO
MARSHALL
MINNESOTA CORN PROCESSORS
NEW ULM
FAIRMONT
WINNEBAGO
SAINT JAMES
MADELIA
METROPOLITAN COUNCIL-BLUE LAKE
METROPOLITAN COUNCIL-SENECA
WASECA
NEW PRAGUE
WILLMAR
Design flow
(MGD)
4.00
11.25
1.50
3.00
4.50
2.60
6.77
3.90
1.70
2.96
1.31
42.00
38.00
3.50
1.38
5.04
Observed flow
(MGD)
(1991-2006 average)
1.04
6.56
8.45
0.96
2.57
1.31
2.55
1.38
0.40
1.08
0.71
26.44
23.89
1.36
0.65
3.48
Most of these point sources have reasonably good monitoring for total phosphorus and total suspended solids
(TSS), but not for total nitrogen. In many cases, only ammonia nitrogen is monitored. The point sources were
G-13
-------
initially represented in the model with the median of reported values for total phosphorus and TSS and an
assumed total nitrogen concentration of 11.2 mg/L for secondary treatment facilities (Tetra Tech 1999).
Major Point Sources
Interstate
Water (Nat. Atlas Dataset)
US Census Populated Places
Municipalities (pop > 50,000)
North Dakota
I County Boundaries
T Watershed with HUCSs
MILBANK-CITYiOF
•X. . V
METRO. COUNCIL-SENECA1; I
METRO"COUNCIL-BLUE LAKE
NEW PRAGUE
NEWULM
SAIMIrffi&TcER
MANKATO—
MINNESOTA CORN PROCESSORS
*
Minnesota MADELIA/
FAIRMONT'
WINNEBAGO
GCRP Model Areas - Minnesota River Basin
Major Point Sources
G-14
-------
Figure 3. Major point sources in the Minnesota River watershed.
Meteorological Data
The required meteorological data series for the 20 Watershed study are precipitation, air temperature, and
potential evapotranspiration. The 20 Watershed model does not include water temperature or algal simulation and
uses a degree-day method for snowmelt. These meteorological data are drawn from the BASINS4 Meteorological
Database (USEPA 2008), which provides a consistent, quality-assured set of nationwide data with gaps filled and
records disaggregated. Scenario application will require simulation over 30 years, so the available stations are
those with a common 30-year period of record (or one that can be filled from an approximately co-located station)
that covers the year 2001. A total of 39 precipitation stations were identified for use in the Minnesota River model
with a common period of record of 10/1/1972-9/30/2002 (Table 4 and Figure 4). Temperature records are sparser;
where these are absent temperature is taken from nearby stations with an elevation correction. For each weather
station, Penman-Monteith reference evapotranspiration was calculated for use in HSPF using observed
precipitation and temperature coupled with SWAT weather generator estimates of solar radiation, wind
movement, cloud cover, and relative humidity.
For the 20 Watershed model applications, SWAT uses daily meteorological data, while HSPF requires hourly
data. It is important to note that a majority of the meteorological stations available for the Minnesota River
watershed are Cooperative Summary of the Day stations that do not report sub-daily data. The BASINS4 dataset
already has versions of the daily data that have been disaggregated to an hourly time step using template stations.
For each daily station, this disaggregation was undertaken in reference to a single disaggregation template.
Occasionally, this automated procedure provides undesirable results, particularly when the total rainfall for the
day is very different between the subject station and the disaggregation template. This yields a small number of
hourly precipitation intensity estimates that are unrealistically high (e.g., much greater than the 100-yr 1-hour
event for the region). This has only a small impact on the basin-scale hydrologic calibration as gages are
influenced by rainfall from multiple weather stations, but can introduce significant problems for the prediction of
erosion and sediment loads. Perhaps more importantly, past experience makes clear that the available precipitation
network is not sufficiently dense to accurately resolve watershed-scale precipitation depths, particularly during
summer convective storms (Tetra Tech 2008). This introduces an unavoidable level of uncertainty into the
hydrologic calibration.
Table 4. Precipitation stations for the Minnesota River watershed model
COOP ID
IA1 38270
MN210112
MN210287
MN210667
MN210981
MN211263
MN212698
MN212768
MN213076
MN213174
Name
TITONKA
ALEXANDRIA CHANDLER FL
ARTICHOKE LAKE
BENSON
BRICELYN
CAN BY
FAIRMONT
FERGUS FALLS
GAYLORD
GLENWOOD 2 WNW
Latitude
43.2353
45.8686
45.3783
45.3167
43.5514
44.7183
43.6447
46.2919
44.5564
45.6633
Longitude
-94.0417
-95.3942
-96.1542
-95.6167
-93.8481
-96.2697
-94.4656
-96.1172
-94.2206
-95.4442
Temperature
No
Yes
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
Elevation (ft)
1,170
1,416
1,075
1,040
1,170
1,243
1,187
1,250
1,018
1,198
G-15
-------
COOP ID
MN213311
MN214176
MN214546
MN214994
MN215073
MN215204
MN215400
MN215435
MN215563
MN215638
MN215887
MN216152
MN216835
MN217326
MN217405
MN217602
MN217907
MN218025
MN218323
MN218429
MN218520
MN218692
MN218808
MN219004
MN219046
SD391777
SD395536
SD397742
SD399337
Name
GRANITE FALLS
JORDAN 2E
LAMBERTON SW EXP STN
MADISON SEWAGE PLANT
MANKATO
MARSHALL
MILAN 1 NW
MINNEAPOLIS/ST PAUL AP
MONTEVIDEO 1 SW
MORRIS WC EXP STN
NEW ULM 2 SE
OLIVIA 3E
REDWOOD FALLS FAA ARPT
ST JAMES FILT PLANT
ST PETER
SHERBURN3WSW
SPRINGFIELD 1 NW
STEWART
TRACY
TYLER
VESTA
WASECA EXP STATION
WELLS
WILLMAR RTC
WINNEBAGO
CLEAR LAKE
MILBANK2SSW
SISSETON
WILMOT
Latitude
44.8108
44.6622
44.2394
45.0025
44.1556
44.4706
45.1219
44.8831
44.9364
45.5903
44.3006
44.7628
44.5472
43.9908
44.3222
43.6303
44.2469
44.7344
44.2394
44.2781
44.5069
44.0725
43.7333
45.1403
43.7689
44.7506
45.2061
45.6667
45.4081
Longitude
-95.5178
-93.5933
-95.3153
-96.1661
-94.0242
-95.7908
-95.9269
-93.2289
-95.7536
-95.8747
-94.4897
-94.9297
-95.0822
-94.6122
-93.9556
-94.7744
-94.9864
-94.3425
-95.6308
-96.1281
-95.4111
-93.5328
-93.7333
-95.0183
-94.1883
-96.6906
-96.6361
-97.0419
-96.8600
Temperature
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
No
Yes
No
Yes
Yes
Yes
Yes
Yes
No
Elevation (ft)
910
930
1,144
1,080
850
1,152
1,020
872
985
1,140
860
1,100
1,025
1,100
850
1,320
1,066
1,040
1,403
1,735
1,080
1,153
1,197
1,128
1,110
1,800
1,160
1,220
1,160
G-16
-------
»V*'*f'
Legend
Weather Stations
Interstate
•1 Water (Nat. Atlas Dataset)
| US Census Populated Places
HI Municipalities (pop > 50,000)
^] County Boundaries
~~ Watershed with HUCSs
Nortf* Dakota
Minneapolis
,^
-------
Watershed Segmentation
The Minnesota River basin was divided into 95 subwatersheds for the purposes of modeling (Figure 5). The initial
calibration watershed (Cottonwood River) is highlighted. The model encompasses the complete watershed and
does not require specification of any upstream boundary conditions for application. It should be noted, however,
that Big Stone Lake (subwatershed 63) is separated from nearby Lake Traverse (which is not part of the
watershed) by only a small dike, and Lake Traverse sometimes overflows into Big Stone Lake when winter ice
jams are present. This boundary phenomenon is not represented in the model.
The model subwatersheds approximate the HUC-10 scale, but are subdivided as needed to account for the
connection of tributaries and location of flow gages. The subwatersheds range in size from 10 to 436 mi2, with an
average size of 178 mi2.
In developing the HSPF simulation it was noted that the FTable simulation of Lac qui Parle did not perform well
for the model validation period. Therefore, for the purposes of HSPF hydrologic validation only, a separate
version of the model was created with outflow from Lac qui Parle set as a boundary condition.
G-18
-------
Legend
A USGS Gages
Interstate
Water (Nat. Atlas Dataset)
US Census Populated Places
Municipalities (pop > 50,000)
I I County Boundaries
] Model Subbasins
I I Initial Calibration Watershed
] Minnesota River Basin
USGS 05311000
Ivlankato
l(
USGS 05319500
USGS 05317000
USGS'05320500
USGS 05320000
GCRP Model Areas - Minnesota River
Model Segmentation
Figure 5. Model segmentation and USGS stations utilized for the Minnesota River
watershed.
G-19
-------
Note: SWAT subwatersheds numbering is shown; the HSPF model for this watershed uses the same subwatershed boundaries with an
alternative internal numbering scheme.
Calibration Data and Locations
The Minnesota River basin was selected as an early pilot site because of extensive previous experience in
modeling this watershed. The specific site chosen for initial calibration was the Cottonwood River at New Ulm, a
flow and water quality monitoring location that approximately coincides with the mouth of an 8-digit HUC at its
outflow to the Minnesota River. The Cottonwood watershed was selected for several reasons: 1) there is a good
set of flow and water quality data available, 2) previous modeling efforts were successful, and 3) the watershed
lacks major point sources and impoundments.
Previous experience in the watershed indicates that model fit is very sensitive to hydrologic parameter
specification - in part because precipitation and ET are, in general, balanced such that minor perturbations in soil
moisture persist for long periods of time. In addition, the Minnesota River watershed was an initial test case of
procedures; therefore, calibration and validation was pursued at multiple locations (Table 5). Parameters derived
on the Cottonwood were not fully transferable to other portions of the Minnesota River watershed, and additional
calibration was conducted at multiple gage locations.
Table 5. Calibration and validation locations in the Minnesota River basin
Station Name
Minnesota River at Montevideo, MN
Yellow Medicine River at Granite Falls, MN
Redwood River nr Redwood Falls, MN
Cottonwood River near New Ulm, MN
Watonwan River near Garden City, MN
Blue Earth River near Rapidan, MN
LeSueur River near Rapidan, MN
Minnesota River at Mankato, MN
Minnesota River near Jordan, MN
USGS ID
05311000
05313500
05316500
05317000
05319500
05320000
05320500
05325000
05330000
Drainage Area
(mi2)
6,180
664
629
1,300
851
2,410
1,110
14,900
16,200
Hydrology
Calibration
X
X
X
X
X
X
X
X
X
Water Quality
Calibration
X
X
X
X
X
X
X
X
The model hydrology calibration period was set to Water Years 1993-2002 (within the 30-year period of record
for modeling). The end date was constrained by the common period of the set of 20 Watershed meteorological
stations available for the watershed, and a ten year calibration period was desired. Calibration was done on the
later data, due to concerns that there have been significant changes in agricultural management practices and land
retirement programs since the 1980s. Hydrologic validation was then performed on Water Years 1983-1992.
Water quality calibration used calendar years 1993-2002, while validation used 1986-1992, as limited data were
available prior to 1986.
G-20
-------
A detailed HSPF model already exists for the portion of the basin between Lac qui Parle and Minnesota River at
Jordan, MN (Tetra Tech 2008). This model, which was developed through a number of iterations to support
TMDL development, is calibrated for flow, sediment, and nutrients. A particular focus of the calibration effort
was on sediment source attribution, using radionuclide data, including detailed calibration for loading from
ravines, in-channel processes, and contributions from bluff collapse where tributaries enter the mainstem at the
edges of the old glacial River Warren valley.
The existence of this earlier model provides a firm basis for parameter initialization, and is one of the reasons that
the Minnesota River was selected as a pilot site. There are significant differences between the 20 Watershed
model and the previous TMDL model. In general, the approach adopted for the 20 Watershed large-scale model
applications is intended to provide a basis for comparison across the country. Key characteristics of the 20
Watershed model include the following.
• The model is constrained to the land uses identified by NLCD for consistency with applications to
other basins. Supplementary refinements to the land use coverage to incorporate information on
conservation tillage and manure application were not used in the 20 Watershed model except as a
guide to general spatial trends in model parameters.
• The model makes use of the data present in the BASINS4 meteorological data set, with one station
assigned per model subwatershed. Patching and disaggregation of the BASINS4 precipitation data
sets generally used a single template station, occasionally resulting in very different interpretation of
peak events.
• The model uses Penman-Monteith reference evapotranspiration, based on measured precipitation and
temperature combined with solar radiation, wind movement, and relative humidity from the SWAT
weather generator.
• The model uses the degree-day approach to simulating snowmelt (because measured values of
insolation and wind movement are not available at all stations).
The TMDL model does provide important insights and parameter starting values that are incorporated into the 20
Watershed model. Of particular note are the following:
• Tile drainage is a significant component of hydrology in the basin and exhibits a spatial gradient, with
the most intensive tiling in the southeastern portion of the watershed. The TMDL model developed an
approach to represent tile drainage through the interflow component of HSPF, with values calibrated
by 8-digit HUC. This representation is carried forward into the 20 Watershed model.
• Channel hydraulics in the TMDL model (represented through FTables) was developed through use of
existing HEC-RAS models, where available. These channel characteristics were carried forward to
the 20 Watershed model (for areas covered by HEC-RAS models) by matching channel segments
between the two models and adjusting storage volumes to account for differences in reach length.
• Detailed work on sediment source calibration in the TMDL model was carried forward into the 20
Watershed model, including the representation of tillage and sediment loading from bluffs through
the HSPF SPECIAL ACTIONS programming capability.
Changes to Data Provided
No changes were made to the meteorological or land use base data. For the point source data it was known from
previous modeling that there is a rapid loss of phosphorus in the near field immediately downstream of various
wastewater treatment plant discharges, likely due to settling of particulate matter. This cannot be accurately
represented at the scale of the 20 Watershed model and indeed resulted in overprediction of phosphorus
concentrations under low flow conditions. Discounting total phosphorus concentration in the effluent by 50
percent resolved this problem during calibration.
G-21
-------
Assumptions
Flow in the upper portions of the Minnesota River is influenced by Lac qui Parle, a U.S. Army Corps of
Engineers impoundment near Montevideo, Minnesota. The hydrology of Lac qui Parle is complex, as there is a
diversion channel at Watson Sag on the Chippewa River (that naturally discharges to the Minnesota River at
Montevideo) that diverts high flows, as well as a portion of base flows, upstream into Lac qui Parle. For the
TMDL model, significant efforts were made to simulate the operations of the Watson Sag diversion, while Lac
qui Parle itself was taken as a boundary condition. For the 20 Watershed model, the basin upstream of Lac qui
Parle is simulated directly, while Watson Sag diversion is not explicitly simulated. The original intent was to
ignore Lac qui Parle altogether, as the operations of impoundments are difficult to extrapolate to future climate
conditions; however, this proved to be a significant detriment to the simulation of flows in the mainstem of the
Minnesota River downstream of Lac qui Parle. Therefore, a reservoir storage-discharge representation (FTable)
was constructed to approximate the hydraulic behavior of Lac qui Parle, based on reported lake storage and
downstream gaged flows. This FTable provides only an approximation of the actual operations of the Lac qui
Parle dam, and, together with the omission of the Watson Sag diversion, introduces uncertainty into the
representation of mainstem flows.
The other significant dam in the watershed (Rapidan Reservoir on Blue Earth River) is not represented in the 20
Watershed model because accurate data are lacking and the impoundment's storage capacity is greatly diminished
by sediment accumulation behind the dam. The many natural lakes in the watershed are also not explicitly
represented, although an approximation of their storage was introduced through representation of the water land
use. Specifically, the water "upland" land use was assigned characteristics that approximate storage in small
natural ponds by assigning a very low slope and a high surface storage capacity, equivalent to approximately 1
inch of runoff from the surrounding drainage area. Surface storage in HSPF is a function of the slope length
(SLSUR) and the roughness coefficient (NSUR). As the program limits NSURto "reasonable" ranges of
Mannings coefficient, additional surface storage capacity can be represented only be artificially increasing slope
length to a large value. It is also necessary to represent evaporation from these ponds, but HSPF does not include
evaporation from surface storage. This is achieved by specifying some upper zone storage capacity, but near zero
infiltration rates out of the upper zone. This effectively routes much of the water to evaporation, except when
large rainfall events exceed the surface storage plus upper zone storage capacity - which is how a pothole pond in
the Minnesota plains behaves.
Another important characteristic of the basin is the widespread presence of subsurface tile drainage.
Installation of tile drainage has converted what were predominantly glacial plain outwash depressional
wetlands into productive farmland. The presence of tile drains, which include both surface and
subsurface inlets, has radically altered the natural hydrology of the area. Surface inlet tile drains, in
particular, may also play a significant role in the transport of sediment and pollutants from agricultural
land to the river.
It is not feasible to simulate individual tile drain systems at the large basin scale. Further, neither the location nor
the total density of tile drainage is known throughout the basin. In most areas, only the public tile drains and
ditches are documented in spatial coverages, and the extent of private tile drains is known only for limited areas.
The HSPF model does not contain any routines for the explicit representation of tile drains. In typical applications
of HSPF, surface runoff represents the quick flow storm response; interflow represents an intermediate time-scale
hydrologic response; and groundwater discharge represents the base flow hydrologic response. In such
applications, interflow represents lateral movement of water through the shallow soil profile.
At a gross or basin scale, the net effect of tile drainage is to move water relatively rapidly out of surface storage
without direct surface drainage. Accordingly, it is to be expected that tile drainage is best represented in HSPF as
G-22
-------
an interflow component, with a response time that is somewhat slower than direct surface runoff, but quicker than
groundwater discharge, represented by a relatively fast recession coefficient. Accordingly, tile drainage is
represented through the interflow inflow and interflow recession parameters in HSPF, as was done for the TMDL
model (Tetra Tech 2008). USGS successfully implemented a similar approach for the heavily tiled Heron Lake
basin, just south of the Minnesota River drainage (Jones and Winterstein 2000).
The tile drain density has a generally decreasing trend from the southeast to the northwest portions of the basin.
Accordingly, interflow inflow rates are also scaled across the basin, using the calibrated values determined by
Tetra Tech (2008). These parameters, specified monthly, range from a high of 4 in the Le Sueur River basin to a
low of 1.1 in the northwestern portions of the watershed. As shown in Tetra Tech (2008), this results in the
models representing interflow ranging up to a maximum of 46 percent of total flow in the Le Sueur basin and
provides maximum interflow discharge rates that are consistent with typical drainage coefficients fertile drains.
Hydrology Calibration
As noted above, the starting point for calibration of hydrologic parameters was the existing TMDL model (Tetra
Tech 2008); however, differences between the models meant that not all parameters were directly transferable.
Therefore, the first focus of calibration was on areas of difference between the model formulations. This included
calibrating the degree-day snowmelt representation and adjusting factors related to the different representation of
PET. In addition, the model parameters were modified to reflect the HRU representation.
The TMDL model divided cropland into areas with conventional tillage, conservation tillage, and manure
application. The 20 Watershed model includes a single cropland class (divided by HSG). Therefore, parameter
estimates from the TMDL were converted to initial parameters for the 20 Watershed model by developing
weighted averages based on the crop management area distribution in the TMDL model. Similar analyses, based
on reported soil properties, were used to extend parameter initial values to the portions of the watershed not
covered by the TMDL model. Calibration adjustments focused on the following parameters:
• INFILT (nominal infiltration rate parameter): The TMDL model did not distinguish between HSGs, but
was adjusted based on watershed averages of reported soil survey infiltration rates. The resulting INFILT
values were generally typical of D soils. The 20 Watershed model includes both D and B soils (although,
as noted above, the B classification may be influenced by widespread pre-existing tile drainage). During
calibration, the D soils were kept at the values derived from the TMDL model while INFILT for the B
soils was adjusted upward as needed to match observations. The resulting final values for the B soils are
still generally less than are often cited for B soils (USEPA 2000).
• KMELT (degree-day melt factor): The 20 Watershed model switches to a degree-day approach for
snowmelt. This depends on the monthly values of KMELT. These were originally set to values
recommended by USAGE (1956), then adjusted during calibration.
• AGWRC (active groundwater recession constant): The preceding changes along with the modified form
of PET required compensating adjustments in AGWRC.
• PET factor: The 20 Watershed model uses Penman-Monteith reference ET estimates consistent with FAO
56 (Allen et al. 1998), whereas the TMDL model used Penman Pan evaporation. The FAO 56 reference
ET calculates ET for a well-watered actively growing grass surface and requires crop factors to convert to
actual PET. The reference ET is similar to, but generally less than Penman Pan evaporation, for which
pan factors, generally in the range around 0.6-0.8, are needed to convert to model PET. Factors on the
Penman-Monteith PET are thus expected to be needed in the model, but will be a little higher than those
determined in the previous calibration effort. The previous modeling also determined that these factors
tend to vary across the watershed, probably reflecting geographic trends in factors like cloud albedo and
opacity. Therefore, new PET factors were assigned during calibration on zonal basis, ranging from 0.71 to
0.935.
G-23
-------
Initial calibrations were performed for the Cottonwood River, comparing model results to data from USGS
05317000, and are summarized in Figure 6 through Figure 12 and Table 6 and Table 7. The fit is of high quality
for all except the very lowest flows and meets all the recommended criteria - although the fit in the more detailed
TMDL model is somewhat better. Potential problems with very low flows likely reflect a combination of factors,
including omission of minor point sources and simplified representation of the behavior of small ponds and
wetlands.
^•Avg Monthly Rainfall (in)
Avg Observed Flow (10/1/1992 to 9/30/2002 )
Avg Modeled Flow (Same Period)
25000
14
Apr-94
Oct-95
Apr-97 Oct-98
Date
Apr-00
Oct-01
Figure 6. Mean daily flow at USGS 05317000 Cottonwood River near New Dim, MN -
calibration period (HSPF).
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002 )
Avg Modeled Flow (Same Period)
0-01
Figure 7. Mean monthly flow at USGS 05317000 Cottonwood River near New Ulm, MN
- calibration period (HSPF).
G-24
-------
8000
• Avg Flow (10/1/1992 to 9/30/2002 )
Line of Equal Value
Best-Fit Line
6000
T3
JD
4000
D)
0
y = 0.9724X + 32.253
" = 0.9061
2000
100%
Avg Observed Flow (10/1/1992 to 9/30/2002 )
Avg Modeled Flow (10/1/1992 to 9/30/2002 )
-Line of Equal Value
T3
O
in
.a
O
0
o
c
_ro
ro
m
0
•a
2000 4000 6000 8000
Average Observed Flow (cfs)
O-92 A-94 O-95 A-97 O-98
Month
A-OO O-01
Figure 8. Mean monthly flow regression and temporal variation at USGS 05317000
Cottonwood River near New Dim, MN - calibration period (HSPF).
Avg Flow (10/1/1992 to 9/30/2002)
• Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002)
Avg Modeled Flow (Same Period)
3000
il 2000 -
-------
Average Monthly Rainfall (in)
•Median Observed Flow (10/1/1992 to 9/30/2002)
I Observed (25th, 75th)
Modeled (Median, 25th, 75th)
Figure 10. Seasonal medians and ranges at USGS 05317000 Cottonwood River near
New Ulm, MN - calibration period (HSPF).
Table 6. Seasonal summary at USGS 05317000 Cottonwood River near New Ulm, MN
- calibration period (HSPF)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
301 .38
350.75
264.09
120.42
183.37
885.87
2161.06
1320.39
1552.95
981.10
455.96
216.05
137.50
237.50
198.00
120.00
116.00
300.00
1395.00
1005.00
950.00
554.50
228.00
114.00
62.25
101.25
139.25
85.25
80.00
191.25
688.00
638.50
561 .25
336.00
138.00
65.00
499.00
562.50
369.50
169.00
174.50
755.25
2767.50
1507.50
1625.00
877.25
519.75
234.00
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
272.97
397.77
283.83
107.44
178.50
833.50
2514.03
1403.31
1395.52
868.51
476.75
207.22
200.96
307.86
187.17
91.30
65.86
373.98
1596.32
1018.84
792.72
502.40
385.70
164.22
89.69
98.12
104.90
39.28
24.23
150.34
730.83
477.76
392.66
306.70
237.43
67.12
372.11
615.44
307.50
163.40
144.26
1045.37
3500.30
1705.07
1387.65
853.99
643.06
280.86
G-26
-------
•Observed Flow Duration (10/1/1992 to 9/30/2002 )
•Modeled Flow Duration (10/1/1992 to 9/30/2002 )
o
0)
O)
CO
ro
Q
100000
10000
1000
100 --
10%
20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90% 100%
Figure 11. Flow exceedance at USGS 05317000 Cottonwood River near New Dim, MN
- calibration period (HSPF).
o
o
to
CO
CD
tO
CD
^
O
o
•a
CD
N
•Observed Flow Volume (10/1/1992 to 9/30/2002 )
•Modeled Flow Volume (10/1/1992 to 9/30/2002 )
120%
100% -
80% -
60% -
40% -
20% -
Oct-92
Apr-94
Oct-95
Apr-97
Oct-98
Apr-00
Oct-01
Figure 12. Flow accumulation at USGS 05317000 Cottonwood River near New Dim,
MN - calibration period (HSPF).
G-27
-------
Table 7. Summary statistics at USGS 05317000 Cottonwood River near New Dim, MN
- calibration period (HSPF)
HSPF Simulated Flow
REACH OUTFLOW FROM DSN 101
10-Year Analysis F^riod: 10/1/1992 - 9/30/2002
Flow volumes are (inches/year) for upstream drainage area
UpperMSd, mod Water; PET 0.9
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
7.79
4.09
0.67
1.37
0.84
0.98
4.60
2.61
0.44
Error Statistics
1.61
-0.73
4.26
-6.09
Observed Flow Gage
USGS 05317000 COTTONWOOD RIVER NEAR NEW ULM, MN
Hydrologic Unit Code: 7020008
Latitude: 44.291 351 77
Longitude: -94.4402495
Drainage Area (sq-rri): 1300
Total Observed In-stream Flow:
Total of Obsei-ved highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
4.07 » | 30
-5.95
5.54
0.86
-10.51
0.754
0.589
0.901
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
7.66
3.92
0.68
1.46
0.80
1.04
4.36
2.59
0.50
Clear [
Hydrology Validation
Validation for the Cottonwood watershed model was performed at the same location as calibration but for the
period 10/1/1982 through 9/30/1992. Results are presented in Figure 13 through Figure 19 and Table 8 and Table
9. The validation achieves a high coefficient of model fit efficiency, but is over on both total volume and 50% low
volume. Inspection of the figures and tables reveals that median flows are generally over-predicted through the
spring and summer.
It is important to recognize that the validation uses the 2001 land use and parameters that are calibrated to land
management practices of the 1990s. While the basin has remained largely agricultural, there are a number of
differences between the earlier and later periods. These differences include the following:
• Developed impervious surface areas have increased.
• The intensity of tile drainage has increased, with more tile lines with greater capacity installed.
• Cropped areas have changed, with a significant amount of land going out of production and into the
Conservation Reserve Program (CRP) and Conservation Reserve Enhancement Program (CREP).
• PET estimates for the 20 Watershed model use SWAT weather generator statistics for solar radiation,
cloud cover, wind, and relative humidity - essentially assuming that the central tendency of these factors
has not changed over time.
G-28
-------
All of these factors may contribute to an increase in the runoff rate for the more recent calibration period, leading
to an over-prediction of flows in the validation period.
The TMDL model (Tetra Tech 2008) also included an earlier period validation test, but used a separate land use
(ca. 1992) for the earlier period, which accounts for two of these factors, although information was not available
on the rate of change in tile drainage intensity. The TMDL model also calculated PET based on observed
meteorology, rather than using a weather generator for solar radiation, cloud cover, wind, and relative humidity.
Even with these changes it was found in that model that it was necessary to apply higher PET factors for the
earlier period to achieve a good hydrologic fit. That adjustment might be compensating for the change in tile drain
intensity or it might reflect actual changes in the relationship of actual PET to estimates obtained from solar
radiation and cloud cover. Similar discrepancies are found at most other gages in the basin.
Temporal modifications to land use, PET factors, and other parameters were not made for the 20 Watershed
model as its purpose is to provide a basis for comparison between current and potential future conditions, where
the current condition is characterized by 2001 land use and land management. Therefore, the discrepancies in the
validation test are not considered to present a significant bar to application of the model.
^•Avg Monthly Rainfall (in)
Avg Observed Flow (10/1/1982 to 9/30/1992 )
Avg Modeled Flow (Same Period)
JS.
14000
12000
10000
8000
6000
4000
2000
i
Oct-82 Apr-84 Oct-85 Apr-87 Oct-88
Date
Apr-90
Oct-91
4 §,
6 is
8 I
10
12
14
&
Figure 13. Mean daily flow at USGS 05317000 Cottonwood River near New Dim, MN -
validation period (HSPF).
4000
3000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1982 to 9/30/1992 )
Avg Modeled Flow (Same Period)
JS.
0-82
A-84
0-91
'ro
o:
G-29
-------
Figure 14. Mean monthly flow at USGS 05317000 Cottonwood River near New Dim,
MN - validation period (HSPF).
4000
• Avg Flow (10/1/1982 to 9/30/1992 )
Line of Equal Value
Best-Fit Line
3000
2000
o
c
_ro
ro
m
I
100%
90% -
80% -
70%
60%
50%
40%
30%
20%
10%
0%
Avg Observed Flow (10/1/1982 to 9/30/1992 )
Avg Modeled Flow (10/1/1982 to 9/30/1992 )
Line of Equal Value
1000 2000 3000 4000
Average Observed Flow (cfs)
O-82 A-84 O-85
A-87 O-88
Month
A-90 O-91
Figure 15. Monthly flow regression and temporal variation at USGS 05317000
Cottonwood River near New Dim, MN - validation period (HSPF).
Avg Flow (10/1/1982 to 9/30/1992)
• Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1982 to 9/30/1992)
Avg Modeled Flow (Same Period)
2000
1500 -
-------
Average Monthly Rainfall (in)
•Median Observed Flow (10/1/1982 to 9/30/1992)
I Observed (25th, 75th)
Modeled (Median, 25th, 75th)
i
o
1000
500
10 11 12 1
Month
Figure 17. Seasonal medians and ranges at USGS 05317000 Cottonwood River near
New Ulm, MN - validation period (HSPF).
Table 8. Seasonal summary at USGS 05317000 Cottonwood River near New Ulm, MN
- validation period (HSPF)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
311.49
246.34
170.13
98.82
158.34
1108.68
1354.28
1023.32
1051.98
558.86
256.15
454.86
75.50
112.00
89.00
75.00
75.00
525.50
637.50
623.00
553.50
374.00
151.50
84.50
40.00
52.00
31.25
11.00
15.50
162.50
355.00
266.50
222.25
160.25
71.00
46.00
467.25
380.00
300.00
172.00
166.00
1495.00
1957.50
1245.00
1300.00
736.50
311.50
446.00
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
409.43
317.71
193.45
139.56
212.14
1090.13
1565.20
1417.41
1071.80
626.37
296.15
455.65
139.27
151.23
135.84
87.56
86.22
1037.59
1023.14
949.90
665.65
514.21
231 .83
117.39
59.56
54.81
60.06
30.45
21.54
427.82
501 .22
445.09
289.05
302.17
116.57
63.86
734.36
281.15
269.57
197.78
223.51
1570.61
2414.57
1799.84
1470.56
804.68
409.98
512.03
G-31
-------
•Observed Flow Duration (10/1/1982 to 9/30/1992 )
•Modeled Flow Duration (10/1/1982 to 9/30/1992 )
o
0)
O)
CO
ro
Q
100000
10000
1000
100 --
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 18. Flow exceedance at USGS 05317000 Cottonwood River near New Dim, MN
- validation period (HSPF).
•Observed Flow Volume (10/1/1982 to 9/30/1992 )
•Modeled Flow Volume (10/1/1982 to 9/30/1992 )
140%
120%
100%
80%
o
o
CO
CO
•a
CD
CD
CO
CD
_3
O
o
rr 40%
•a
CD
N
m 20% -
E
o
Oct-82
60%
Apr-84
Oct-85
Apr-87
Oct-88
Apr-90
Oct-91
G-32
-------
Figure 19. Flow accumulation at USGS 05317000 Cottonwood River near New Dim,
MN - validation period (HSPF).
G-33
-------
Table 9. Summary statistics at USGS 05317000 Cottonwood River near New Dim, MN
- validation period (HSPF)
HSPF Simulated Flow
REACH OUTFLOW FROM DSN 101
10-Year Analysis F^riod: 10/1/1982 - 9/30/1992
Flow volumes are (inches/year) for upstream drainage area
UpperMSd, mod Water; PET 0.9
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
6.80
3.09
0.57
1.21
0.81
1.26
3.52
2.06
0.40
Error Statistics
Observed Flow Gage
USGS 05317000 COTTONWOOD RIVER NEAR NEW ULM, MN
Hydrologic Unit Code: 7020008
Latitude: 44.291 351 77
Longitude: -94.4402495
Drainage Area (sq-rri): 1300
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
14.78 10
51.89 10
-2.54 15
8.63 30
26.43 » | 30
5.27
18.42
-1.37
-11.91
0.779
0.587
0.856
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
-- --
5.92
3.17
0.38
1.11
0.64
1.20
2.97
2.09
0.45
Clear [
Hydrology Results for Larger Watershed
As described above, parameters determined for the Cottonwood gage were not fully transferable to other gages in
the watershed. Therefore, calibration was pursued at a total of nine gages throughout the watershed, including
seven gages at the outlet of 8-digit HUCs and two gages on the mainstem. Calibration results were acceptable at
all gages (Table 10). The close match between observed and predicted flow volumes at the most downstream
available gage (USGS 05330000, Minnesota River near Jordan) are shown in Figure 20 and Figure 21. Additional
calibration results are shown in Figures 22 through 26 and Table 11.
G-34
-------
Table 10. Summary statistics (percent error): all stations - calibration period (HSPF)
Station
Error in total
volume:
Error in 50%
lowest flows:
Error in 10%
highest
flows:
Seasonal
volume error
- Summer:
Seasonal
volume error
- Fall:
Seasonal
volume error
- Winter:
Seasonal
volume error
- Spring:
Error in
storm
volumes:
Error in
summer
storm
volumes:
Daily Nash-
Sutcliffe
Coefficient of
Efficiency, E:
Monthly
Nash-
Sutcliffe
Coefficient of
Efficiency, E:
05311000
Minnesota
River at
Montevideo
-7.76
-6.49
-5.98
-8.31
-9.09
-16.82
-5.42
1.81
37.17
0.784
0.886
05313500
Yellow
Medicine
River
-2.49
7.14
2.33
-19.71
-12.80
7.94
1.52
10.67
8.78
0.573
0.851
05316500
Redwood
River nr
Redwood
Falls
0.69
9.25
4.12
-4.24
-4.17
8.85
1.47
7.14
9.86
0.673
0.596
05317000
Cottonwood
River near
New Dim
1.61
-0.73
4.26
-6.09
4.07
-5.95
5.54
0.86
-10.51
0.754
0.885
05319500
Watonwan
River nr
Garden
City
0.88
5.46
-1.37
5.37
7.25
-12.48
0.90
5.12
-7.13
0.728
0.888
05320000
Blue Earth
River nr
Rapidan
-4.35
1.58
0.30
4.45
-12.99
-17.90
-3.44
8.79
12.38
0.811
0.939
05320500
LeSueur
River nr
Rapidan
-0.38
7.09
4.75
-9.04
1.53
33.10
-3.36
10.76
6.76
0.539
0.889
05325000
Minnesota
River at
Mankato
-3.80
-3.29
-0.94
-3.11
-5.15
-7.28
-3.17
12.41
14.66
0.899
0.954
05330000
Minnesota
River nr
Jordan
-4.25
-7.30
-1.25
-3.21
-6.26
-9.55
-3.33
8.85
7.85
0.916
0.953
G-35
-------
lAvg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002 )
Avg Modeled Flow (Same Period)
120000
100000
80000
60000
40000
20000
Oct-92
Apr-94
Oct-01
Figure 20. Mean daily flow simulation at USGS 05330000 Minnesota River near
Jordan, MN - calibration period (HSPF).
^•Avg Monthly Rainfall (in)
-•-Avg Observed Flow (10/1/1992 to 9/30/2002 )
Avg Modeled Flow (Same Period)
80000
60000
40000
20000
O-92
A-94
O-01
Figure 21. Mean monthly flow simulation at USGS 05330000 Minnesota River near
Jordan, MN - calibration period (HSPF).
G-36
-------
i
I
T3
-32
CD
T3
O
Best-Fit Line
60000 -
n j
y = 0.95
R2
•»
84x- 10.4
= 0.9546
•^
03
^^
,*
0 20000 40000 60000 80000
Average Observed Flow (cfs)
-o
o
.
ro
CQ
Avg Observed Flow (10/1/1992 to 9/30/2002 )
Avg Modeled Flow (10/1/1992 to 9/30/2002 )
-Line of Equal Value
O-92 A-94 O-95
A-97 O-98
Month
A-oo 0-01
Figure 22. Monthly flow regression and temporal variation at USGS 05330000
Minnesota River near Jordan, MN - calibration period (HSPF).
Avg Flow (10/1/1992 to 9/30/2002)
o
•— - onnnn
o
LL
-n isnnn
0
0
•Q
o innnn
0
D)
& ^nnn
5
n
y = 0
F
s
-'A Line
9803X -
2 = 0.9J
/
188.86
91
^
/
5000 10000 15000 20000 25000
Average Observed Flow (cfs)
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002)
Avg Modeled Flow (Same Period)
25000
10 11 12 1 2 3 4 5 6 7
Month
Figure 23. Seasonal regression and temporal aggregate at USGS 05330000
Minnesota River near Jordan, MN - calibration period (HSPF).
G-37
-------
Average Monthly Rainfall (in)
•Median Observed Flow (10/1/1992 to 9/30/2002)
I Observed (25th, 75th)
Modeled (Median, 25th, 75th)
o 15000
10 11 12
234
Month
Figure 24. Seasonal medians and ranges at USGS 05330000 Minnesota River near
Jordan, MN -calibration period (HSPF).
Table 11. Seasonal summary at USGS 05330000 Minnesota River near Jordan, MN
calibration period (HSPF)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
3732.17
4054.61
2858.86
1 757.39
1811.52
6364.40
23775.27
15802.45
15733.60
1 1 756.52
7020.71
3542.37
2190.00
3825.00
2900.00
1800.00
1500.00
3910.00
18700.00
13400.00
12800.00
9565.00
4225.00
2040.00
1082.50
1210.00
1702.50
1392.50
1272.50
2080.00
12550.00
9470.00
8805.00
5347.50
2377.50
1205.00
6237.50
5462.50
4000.00
2200.00
1900.00
8627.50
28925.00
20500.00
17900.00
12675.00
7782.50
3910.00
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
3081 .52
4059.50
2846.70
1474.05
1327.81
6154.83
23021.23
15392.60
15049.13
11613.34
6704.00
3279.92
2001.01
3234.97
2613.94
1435.78
989.91
4403.41
1 7473.36
12413.48
11811.63
9036.76
4601 .26
2191.42
1032.73
1403.38
1647.86
893.52
656.04
2529.67
9942.81
8451.15
8221 .43
6415.22
2608.84
1342.62
4762.72
6562.34
3784.83
1978.66
1398.81
8517.95
26689.60
19265.36
17020.20
12314.82
8692.04
4185.19
G-38
-------
•Observed Flow Duration (10/1/1992 to 9/30/2002 )
•Modeled Flow Duration (10/1/1992 to 9/30/2002 )
100000
10%
20% 30% 40% 50% 60% 70%
Percent of Time that Flow is Equaled or Exceeded
90%
100%
Figure 25. Flow exceedence at USGS 05330000 Minnesota River near Jordan, MN
calibration period (HSPF).
o
o
in
ro
T3
0
0
-------
For hydrologic validation, the FTable developed during calibration to represent Lac qui Parle dam did not appear
to provide realistic results for the earlier period. Therefore, the model was respecified using gaged flows below
Lac qui Parle as a boundary condition. Results of the validation exercise are summarized in Table 12. Problems
similar to those experienced on the Cottonwood River were seen at all tributary gages, with overprediction of
lower flows in summer. However, as noted above, this is likely due to the use of land use and model parameters
that are more reflective of current conditions and is not believed to present a bar to application of the model.
Table 12. Summary statistics: all stations - validation period (HSPF)
Station
Error in total
volume:
Error in 50%
lowest flows:
Error in 10%
highest
flows:
Seasonal
volume error
- Summer:
Seasonal
volume error
- Fall:
Seasonal
volume error
- Winter:
Seasonal
volume error
- Spring:
Error in
storm
volumes:
Error in
summer
storm
volumes:
Daily Nash-
Sutcliffe
Coefficient of
Efficiency, E:
Monthly
Nash-
Sutcliffe
Coefficient of
Efficiency, E:
05313500
Yellow
Medicine
River
-3.473
48.58
-11.24
-20.37
22.08
-8.64
-0.22
5.20
-15.69
0.317
0.714
05316500
Redwood
River nr
Redwood
Falls
8.55
74.44
-11.99
11.63
7.22
-1.04
11.95
-2.20
-11.47
0.491
0.671
05317000
Cottonwood
River near
New Dim
14.78
51.89
-2.54
8.63
26.43
5.27
18.42
-1.37
-11.91
0.779
0.856
05319500
Watonwan
River nr
Garden
City
13.31
75.61
-0.33
29.35
21.51
3.55
9.93
23.26
30.61
0.345
0.598
05320000
Blue Earth
River nr
Rapidan
5.11
67.23
-7.20
19.13
16.11
-6.63
2.58
12.85
19.48
0.712
0.829
05320500
LeSueur
River nr
Rapidan
21.93
59.34
18.94
34.92
39.00
9.97
19.17
29.98
61.17
0.374
0.717
05325000
Minnesota
River at
Mankato
-13.43
-2.50
-17.33
-17.07
-13.92
-12.26
-12.17
15.60
10.28
0.773
0.830
05330000
Minnesota
River nr
Jordan
-9.73
-0.75
-15.02
-13.91
-11.30
-9.88
-7.27
14.60
4.11
0.779
0.830
Water Quality Calibration and Validation
The 20 Watershed models are designed to provide water quality simulation for TSS, total nitrogen, and total
phosphorus. Total suspended solids is simulated with the standard HSPF approach (USEPA 2006), and takes
G-40
-------
advantage of detailed calibration efforts for the TMDL model (Tetra Tech 2008), which included radionuclide
attribution of sediment sources to field, ravine, and channel sources. However, the segmentation of the 20
Watershed model limits the ability to effectively transfer channel erosion and deposition parameters.
In contrast to TSS, total nitrogen and total phosphorus are simulated in this application in a simplistic fashion, as
HSPF general quality constituents (GQUALs) subject to an exponential decay rate during transport. This contrasts
with the approach in the TMDL model, where individual nutrient species are simulated along with kinetic
transformations and algal uptake/release in the stream reaches using the HSPF NUTRX routines. A significant
drawback of the GQUAL approach to nutrients is that it is not readily possible to account for the nutrient content
of channel bank erosion, which forms an important component of the total phosphorus load in the TMDL model.
The water quality calibration focuses on the replication of monthly loads, as specified in the project QAPP. Given
the approach to water quality simulation in the 20 Watershed model a close match to individual concentration
observations cannot be expected. However, comparison to monthly loads presents challenges, as monthly loads
are not observed. Instead, monthly loads must be estimated from scattered concentration grab samples and
continuous flow records. Such estimation inherently includes uncertainty because it depends on the degree and
form in which concentration and flow are correlated with one another. Further, the bulk of the load of sediment
and sediment-associated phosphorus is likely to move through the system in a limited number of high flow events,
which typically have not been monitored. As a result, the monthly load calibration is inevitably based on the
comparison of two uncertain numbers. Nonetheless, calibration is able to achieve a reasonable agreement. Further,
the load comparisons were supported by detailed examinations of the relationships of flows to loads and
concentrations and the distribution of concentration prediction errors versus flow, time, and season, as well as
standard time series plots.
For application on a nationwide basis, the 20 Watershed protocols assume that sediment and phosphorus loads
will likely exhibit a strong positive correlation to flow (and associated erosive processes), while total nitrogen
loads, which often have a dominant groundwater component, will not. Accordingly, sediment and phosphorus
loads were estimated from observations using a flow-stratified log-log regression approach, while total nitrogen
loads were estimated using a flow-stratified averaging estimator, consistent with the findings of Preston et al.
(1989).
As with hydrology, initial calibration and validation of water quality was done on the Cottonwood River, at USGS
gage 05317000, using 1993-2002 for calibration and 1986-1992 for validation. As with hydrology, calibration
was performed on the later period as this better reflects the land use included in the model. The start of the
validation period is constrained by data availability. Initial sediment parameters were transferred from the TMDL
model, with area-weighting to account for the change in subwatershed boundaries and the different representation
of land use in the 20 Watershed model. It was found that this approach resulted in overestimation of the peak
loading at high flows associated with ravine incision. On investigation, it was determined that this was caused by
the different methods of processing of rainfall data for the two models. In particular, the approach to
disaggregation of daily rainfall totals to hourly rainfall in the BASINS4 meteorological dataset results in greater
(and, in some cases, unrealistic) estimates of peak rainfall intensity. As ravine incision depends in a nonlinear
fashion on maximum runoff rates this component of the model is highly sensitive to rainfall intensity. This was
addressed by reducing the exponent on flow depth (JGER) in the 20 Watershed model and then adjusting the
coefficient (KGER) to achieve calibration. Channel scour and deposition critical shear stresses also needed to be
adjusted.
Once these changes were made, the sediment model performed well for both the calibration and validation
periods. Time series of simulated and estimated sediment loads at the Cottonwood gage for both periods are
shown in Figure 27 and statistics for the two periods are provided separately in Table 13. The key statistic in
Table 13 (consistent with the QAPP) is the relative percent error, which shows the error in the prediction of
monthly load normalized to the estimated load. The table also shows the relative average absolute error, which is
the average of the relative magnitude of errors in individual monthly load predictions. This number is inflated by
G-41
-------
outlier months in which the simulated and estimated loads differ by large amounts (which may be as easily due to
uncertainty in the estimated load due to limited data as to problems with the model) and the third statistic, the
relative median absolute error, is likely more relevant and shows good agreement.
TSS
1,000,000 -T
100,000 --
o
«
o
-Regression Loads
-Simulated Loads
Figure 27. Fit for monthly load of TSS at USGS 05317000 Cottonwood River (HSPF).
Table 13. Model fit statistics (observed minus predicted) for monthly TSS loads using
stratified regression (HSPF)
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1993-2002)
7.5%
54%
1.7%
Validation period
(1986-1992)
13.1%
79%
9.9%
A variety of other diagnostics were also pursued to ensure agreement between the model and observations. These
are available in full in the calibration spreadsheets, but a few examples are provided below. First, load-flow power
plots were compared for individual days (Figure 28 and Figure 29). These confirm that the relationship between
flow and load is consistent across the entire range of observed flows, for both the calibration and validation
periods.
G-42
-------
1000000
•o
«
c
o
•o
ra
o
(0
V)
0.1
Cottonwood River - New Dim
1993-2002
10000 100000
• Simulated A Observed
Power (Simulated)
Power (Observed)
Figure 28. Power plot for observed and simulated TSS at USGS 05317000
Cottonwood River - calibration period (HSPF).
^
"5>
•o
ra
o
CO
V)
Cottonwood River - New Dim
1986-1992
Simulated A Observed ^^™Rower (Simulated) ^^™Power (Observed)
10000
Figure 29. Power plot for observed and simulated TSS at USGS 05317000
Cottonwood River - validation period (HSPF).
G-43
-------
Standard time series plots (Figure 30) show that observed and simulated concentrations achieve good agreement,
although individual observations may deviate. Plots of concentration error versus flow and versus month (not
shown) were used to guard against hydrologic and temporal bias. Finally, statistics on concentration (Table 14)
show that low median errors are achieved.
Cottonwood River - New Dim
1993-2002
• Simulated A Observed
10000
1000
100
(0
(0
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Year
Figure 30. Time series plot of TSS concentration at USGS 05317000 Cottonwood
River (HSPF).
Table 14. Relative errors (observed minus predicted), TSS concentration at USGS
05317000 Cottonwood River (HSPF)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1993-2002)
121
-58.8%
0.41%
Validation period
(1986-1992)
75
13.1%
-2.7%
For simulation of total phosphorus and total nitrogen, the TMDL model parameters, which address individual
nutrient species separately, were converted to approximately equivalent parameters on total nutrients using area
weighting. The model simulates total phosphorus from the uplands as having sediment-associated (both with sheet
and rill erosion and ravine incision) and buildup-washoff components on the land surface along with monthly
variable interflow and groundwater components. The sediment-associated component of the surface load reflects
mineral phosphorus, while the buildup-washoff component addresses the organic phosphorus. Total nitrogen is
simulated with a buildup-washoff component for surface loading, plus monthly variable interflow and
groundwater components.
G-44
-------
The original parameter set derived in this way did not perform well for the Cottonwood River phosphorus
simulation - probably because the process of weighting the parameters related to different agricultural land uses
(manured land, conventional tillage, and conservation tillage) assumes linear additivity and independence of
hydrologic variation. Calibration was achieved by adjusting downward the sediment potency factors. A similar
approach was applied for other subwatersheds, maintaining the spatial variability in loading rates incorporated in
the TMDL model.
In-stream, total phosphorus is represented as a simple general quality component, subject to exponential decay.
Decay rates were adapted from the most recent version of the SPARROW model (Alexander et al. 2008), which
estimates decay coefficients as a function of stream depth, using typical depths for streams of different orders.
Monthly loading series for total phosphorus are shown in Figure 31 and load statistics are summarized in Table
15. In general, the observed and estimated total phosphorus loads attain a good match for both the calibration and
validation periods.
Total P
o
«
o
1000
100 -
10
0.1 -
0.01 4
Regression Loads
Simulated Loads
Figure 31. Fit for monthly load of total phosphorus at USGS 05317000 Cottonwood
River (HSPF).
Table 15. Model fit statistics (observed minus predicted) for monthly phosphorus
loads using stratified regression (HSPF)
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1993-2002)
23.0%
54%
2.5%
Validation period
(1986-1992)
15.8%
67%
13.5%
G-45
-------
As with suspended sediment, additional diagnostics for total phosphorus included flow-load power plots (Figure
32 and Figure 33), time series plots (Figure 34) and analysis of concentration errors (Table 16). All show good
agreement.
0.1
Cottonwood River - New Dim
1993-2006
10000
• Simulated A Observed
Row er (Simulated)
Row er (Observed)
100000
Figure 32. Power plot for observed and simulated total phosphorus at USGS
05317000 Cottonwood River- calibration period (HSPF).
G-46
-------
100
Cottonwood River - New Dim
1986-1992
10000
• Simulated A Observed ^^~Power (Simulated) ^^"Fbwer (Observed)
Figure 33. Power plot for observed and simulated total phosphorus at
05317000 Cottonwood River - validation period (HSPF).
USGS
E 0.1
0.01
0.001
Cottonwood River - New Dim
1993-2002
• Simulated A Observed
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Year
Figure 34. Time series plot of total phosphorus concentration at USGS 05317000
Cottonwood River (HSPF).
G-47
-------
Table 16. Relative errors (observed minus predicted), total phosphorus concentration
at USGS 05317000 Cottonwood River (HSPF)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1993-2002)
123
-52.9%
5.3%
Validation period
(1986-1992)
75
18.0%
-0.48%
For total nitrogen fewer data are available because many sampling events omitted one or more nitrogen species.
This increases the uncertainty of the comparison. However, development of the TMDL model also revealed that
there is large temporal variability in observed total nitrogen concentrations, likely related to seasonal differences
in the timing and amount of fertilizer application. (In this watershed, the primary fertilizer applications are of
subsurface anhydrous ammonia, which can occur in both spring and fall, along with animal manure.)
During calibration for total nitrogen the major change from the original parameter set was scaling down the
buildup-washoff factors. Subsurface concentrations, which represent the major loading pathway for nitrogen,
were generally acceptable as previously developed, except that the contribution of organic matter to groundwater
nitrogen loading was reduced. As with phosphorus, a similar procedure was applied across all model
subwatersheds, retaining the spatial variability in nitrogen loading that was identified in the development of the
TMDL model.
Results for total nitrogen are summarized in Figure 35 through Figure 38, Table 17, and Table 18, following the
same format as total phosphorus. The results are acceptable, although there is clearly greater uncertainty in the
prediction of total nitrogen than in the prediction of total phosphorus.
Total N
6,000
\\N\\\\\\\\\
-Averaging Loads
-Simulated Loads
Figure 35. Fit for monthly load of total nitrogen at USGS 05317000 Cottonwood River
(HSPF).
G-48
-------
Table 17. Model fit statistics (observed minus predicted) for monthly total nitrogen
loads using averaging estimator (HSPF)
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1993-2002)
15.4%
35%
5.4%
Validation period
(1986-1992)
16.2%
43%
14.5%
1000
0.001
0.1
Cottonwood River - New Dim
1993-2006
10 100 1000 10000 100000
Flow, cfs
» Simulated A Observed ^^™ Power (Simulated) ^^™ Power (Observed)
Figure 36. Power plot for observed and simulated total nitrogen at USGS 05317000
Cottonwood River - calibration period (HSPF).
G-49
-------
1000
100
•o
ra
o
0.001
Cottonwood River - New Dim
1986-1992
10
100
Flow, cfs
1000
10000
Simulated A Observed
Row er (Simulated)
Power (Observed)
Figure 37. Power plot for observed and simulated total nitrogen at USGS 05317000
Cottonwood River - validation period (HSPF).
Cottonwood River - New Dim
1993-2002
- Simulated A Observed
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Year
Figure 38. Time series plot of total nitrogen concentration at USGS 05317000
Cottonwood River (HSPF).
G-50
-------
Table 18. Relative errors (observed minus predicted), total nitrogen concentration at
USGS 05317000 Cottonwood River (HSPF)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1993-2002)
20
36.6%
39.4%
Validation period
(1986-1992)
75
25.6%
22.2%
Water Quality Results for Larger Watershed
As with hydrology, the Cottonwood parameters for water quality are not directly transferable to other portions of
the watershed. It is well established that there are strong spatial gradients in ravine and bank erosion of solids, soil
test phosphorus, and subsurface loading of nitrogen, with the highest rates generally in the Blue Earth and Le
Sueur basins and the lowest rates in the western watersheds. However, a consistent procedure for translating the
parameters of the more detailed TMDL model (Tetra Tech 2008) to the 20 Watershed model provided good
results, requiring only relatively minor modifications.
Summary statistics for the water quality calibration and validation at other stations in the watershed are provided
in Table 19 and Table 20. The relative percent error on the monthly loads is within 26 percent for all parameters at
all stations during the calibration period, with the exception of the mainstem station at Mankato. This station is
immediately below the confluence of the Minnesota River and the Blue Earth River (a major source of loading)
and it is believed that concentration measurements there are influenced by incomplete mixing, which seems to be
borne out by much better fit downstream at Jordan. In contrast, the validation tests underestimate total solids and
total phosphorus loads at a number of stations. This is likely due to changes in land use and management over
time (including aggressive efforts to increase conservation tillage and decrease erosion), coupled with propagation
of errors in the hydrologic simulation.
Table 19. Summary statistics for water quality: all stations - calibration period 1993-
2002 (HSPF)
Station
Relative
Percent Error
TSS Load
TSS
Concentration
Median Error
Relative
Percent Error
TP Load
TP
Concentration
Median Error
Relative
Percent Error
TN Load
05313500
Yellow
Medicine
River
-3.8%
6.5%
6.9%
7.1%
-21.0%
05316500
Redwood
River nr
Redwood
Falls
2.2%
8.7%
10.8%
21.2%
-7.7%
05317000
Cottonwood
River near
New Dim
7.5%
0.4%
23.0
5.3%
15.4
05319500
Watonwan
River nr
Garden
City
1 1 .4%
19.5%
2.7%
23.2%
-7.2%
05320000
Blue
Earth
River nr
Rapidan
-21.6%
4.1%
1.5%
0.53%
-9.0%
05320500
LeSueur
River nr
Rapidan
2.6%
-1.7%
-0.1%
1.8%
14.9%
05325000
Minnesota
River at
Mankato
-3.7%
41.7%
-52.7%
-6.6%
44.1%
05330000
Minnesota
River nr
Jordan
6.4%
25.8%
1.3%
15.0%
6.5%
G-51
-------
TN
Concentration
Median Error
19.1%
16.7%
39.4%
-4.9%
-10.2%
6.5%
-23.9%
3.1%
Table 20. Summary statistics for water quality: all stations - validation period 1986-
1992 (HSPF)
Station
Relative
Percent Error
TSS Load
TSS
Concentration
Median Error
Relative
Percent Error
TP Load
TP
Concentration
Median Error
Relative
Percent Error
TN Load
TN
Concentration
Median Error
05313500
Yellow
Medicine
River
-6.4%
3.0%
-38.1%
3.6%
-5.0%
-43.1%
05316500
Redwood
River nr
Redwood
Falls
-17.0%
-5.2%
-21.1%
-1.7%
-4.9%
-12.7%
05317000
Cottonwood
River near
New Dim
13.1%
-2.7%
15.8%
-0.48%
6.2%
22.2%
05319500
Watonwan
River nr
Garden
City
-46.4%
12.7%
-35.0%
17.6%
-14.8%
3.2%
05320000
Blue
Earth
River nr
Rapidan
-25.2%
ND
-14.0%
ND
-19.5%
ND
05320500
LeSueur
River nr
Rapidan
-36.5%
11.9%
-31.7%
-51.9%
8.7%
29.6%
05325000
Minnesota
River at
Mankato
-37.1%
11.3%
-85.3%
-5.1%
38.6%
32.2%
05330000
Minnesota
River nr
Jordan
-6.8%
9.1%
-27.3%
29.1%
-1 .2%
-4.9%
G-52
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a si
The SWAT model for the Minnesota River basin was set up with the ArcSWAT interface using the same
subwatersheds and other geospatial coverages described above for the HSPF model. The SWAT model also uses
the same weather data, but at a daily, rather than hourly timestep.
No changes were made to the meteorological or land use base data for the SWAT model.
Three major reservoirs occur in the upper portion of the Minnesota River basin. These are Swan Lake, Lac Qui
Parle Dam and Big Stone Lake of which only the Lac Qui Parle dam was modeled. Pertinent reservoir
information including surface area and storage at principal (normal) and emergency spillway levels for the
reservoirs modeled were obtained from the National Inventory of dams (NID) database. The SWAT model
provides four options to simulate reservoir outflow: measured daily outflow, measured monthly outflow, average
annual release rate for uncontrolled reservoir, and controlled outflow with target release. Keeping in view, the 20
Watershed climate change impact evaluation application, it was assumed that the best representation of the
reservoirs was to simulate them without supplying time series of outflow records. Therefore, target release
approach was used in the GCRP-SWAT model.
Another important characteristic of the watershed is the widespread presence of subsurface tile drainage.
Installation of tile drainage has converted what were predominantly glacial plain outwash depressional wetlands
into productive farmland. The presence of tile drains, which include both surface and subsurface inlets, has
radically altered the natural hydrology of the area. Surface inlet tile drains, in particular, may also play a
significant role in the transport of sediment and pollutants from agricultural land to the river. It is not feasible to
simulate individual tile drain systems at the large basin scale. Further, neither the location nor the total density of
tile drainage is known throughout the watershed. In most areas, only the public tile drains and ditches are
documented in spatial coverages, and the extent of private tile drains is known only for limited areas.
The SWAT model allows for some representation of tile drains in the form of three parameters: depth to the tile
drains, time to drain soil to field capacity and tile drain lag time.
ire
A spatial calibration approach was not adopted for GCRP-SWAT modeling for Upper Mississippi River basin,
unlike the HSPF application. However, a systematic adjustment of parameters has been adopted and some
adjustments are applied throughout the basin. Most of the calibration efforts were geared towards getting a closer
match between simulated and observed flows at the outlet of calibration focus area.
A 5/10/5 percent threshold was used for land use/soil/slope in the SWAT model while defining the HRUs. The
cropland HRUs were split into corn and soybean in the ratio 1:1. Further these classes and the urban (including
current and future urban class types) classes were exempt from applying the thresholds.
The calibration focus area (Cottonwood River) represents 7 subwatersheds, which together consists of 349 HRUs.
The calibration focus area well represented the general land use characteristics of the overall watershed. Since the
___^^
-------
Minnesota River basin has predominantly an agricultural land use, there is essentially one set of parameters for
the entire watershed.
The parameters were adjusted within the practical range to obtain reasonable fit between the simulated and
measured flows in terms of Nash-Sutcliffe modeling efficiency and the high flow and low flow components as
well as the seasonal flows.
The water balance of the whole Minnesota River basin predicted by the SWAT model is as follows:
PRECIP = 689.0 MM
SNOW FALL = 102.36 MM
SNOW MELT = 98.01 MM
SUBLIMATION = 4.58 MM
SURFACE RUNOFF Q = 50.96 MM
LATERAL SOIL Q = 1.12 MM
TILE Q = 31.59 MM
GROUNDWATER (SHAL AQ) Q = 67.14 MM
REVAP (SHAL AQ => SOIL/PLANTS) = 3.08 MM
DEEP AQ RECHARGE = 3.70 MM
TOTAL AQ RECHARGE = 73.91 MM
TOTAL WATER YLD = 148.19 MM
PERCOLATION OUT OF SOIL = 71.32 MM
ET = 533.9 MM
PET = 1239.2MM
TRANSMISSION LOSSES = 2 . 62 MM
As is consistent with earlier studies (Tetra Tech, 2008), the baseflow (i.e., the groundwater and the tile Q)
component accounts for more than 50 percent of the total water yield.
Calibration adjustments focused on the following parameters:
• FFCB
• SURLAG (surface runoff lag coefficient)
• CNCOEFF
• Baseflow factor
• GW_DELAY (groundwater delay time)
• Manning's "n" value for main channels, and tributary channels
• Sol_AWC (available water capacity of the soil layer, mm water/mm of soil)
• Heat Units to maturity for corn and soybean
• Depth to impervious surface
• BLAI for corn
• Snow parameters SMTMP, SMFMX and SMFMN
• Tile drain parameters (DDRAIN, TDRAIN and GDRAIN)
Initial calibrations were performed for the Cottonwood River, comparing model results to data from USGS
05317000, and are summarized in Figures 39 through 45 and Table 21 and Table 22.
G-54
-------
I Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002 )
Avg Modeled Flow (Same Period)
25000
20000
15000
£ 10000
5000
Oct-01
Figure 39. Mean daily flow at USGS 05317000 Cottonwood River near New Dim, MN
calibration period (SWAT).
8000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002 )
-Avg Modeled Flow (Same Period)
0-92
A-94
0-01
Figure 40. Mean monthly flow at USGS 05317000 Cottonwood River near New Dim,
MN - calibration period (SWAT).
G-55
-------
8000
• Avg Flow (10/1/1992 to 9/30/2002)
Line of Equal Value
Best-Fit Line
6000
4000
CD
O)
ro
2000
y = 0.8927x + 38.418
= 0.9137
2000 4000 6000 8000
Average Observed Flow (cfs)
100%
T3
o
+
w
O
CD
O
m
ro
m
CD
"ro
Avg Observed Flow (10/1/1992 to 9/30/2002)
Avg Modeled Flow (10/1/1992 to 9/30/2002 )
-Line of Equal Value
O-92 A-94 O-95 A-97 O-98 A-00 O-01
Month
Figure 41. Monthly flow regression and temporal variation at USGS 05317000
Cottonwood River near New Dim, MN - calibration period (SWAT).
• Avg Flow (10/1 /1992 to 9/30/2002)
Best-Fit Line
¥
^9nnn
o
LL
T3 1500
.cu
T3
o -innn
^
80.885
74
.•"^
^
.-•'S
ir
,'*
^*
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002)
Avg Modeled Flow (Same Period)
2500
500 1000 1500 2000 2500
Average Observed Flow (cfs)
10 11 12 1 23456789
Figure 42. Seasonal regression and temporal aggregate at USGS 05317000
Cottonwood River near New Dim, MN - calibration period (SWAT).
G-56
-------
Average Monthly Rainfall (in)
-Median Observed Flow (10/1/1992 to 9/30/2002)
I Observed (25th, 75th)
Modeled (Median, 25th, 75th)
3000 -r
f
I
Figure 43. Seasonal medians and ranges at USGS 05317000 Cottonwood River near
New Dim, MN - calibration period (SWAT).
Table 21. Seasonal summary at USGS 05317000 Cottonwood River near New Dim, MN
- calibration period (SWAT)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
301.38
350.75
264.09
120.42
183.37
885.87
2161.06
1320.39
1552.95
981.10
455.96
216.05
137.50
237.50
198.00
120.00
116.00
300.00
1395.00
1005.00
950.00
554.50
228.00
114.00
62.25
101.25
139.25
85.25
80.00
191.25
688.00
638.50
561 .25
336.00
138.00
65.00
499.00
562.50
369.50
169.00
174.50
755.25
2767.50
1507.50
1625.00
877.25
519.75
234.00
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
381 .46
359.14
218.04
101.28
167.59
639.79
1819.87
1222.77
1406.00
1064.17
599.42
333.14
227.46
199.53
138.01
67.56
45.34
164.53
1015.47
788.58
747.96
641.49
477.98
275.54
126.33
125.15
90.43
44.06
27.41
53.07
316.98
430.93
448.05
445.58
304.94
142.37
377.43
508.97
328.58
168.76
141.13
478.34
2365.55
1623.50
1379.83
1123.71
792.46
463.77
G-57
-------
o
-------
Table 22. Summary statistics at USGS 05317000 Cottonwood River near New Dim, MN
- calibration period (SWAT)
HSPF Simulated Flow
REACH OUTFLOW FROM DSN 6001
10-Year Analysis F^riod: 10/1/1992 - 9/30/2002
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9)
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
7.25
3.66
0.68
1.76
0.84
0.79
3.85
2.42
0.43
Error Statistics
-5.41
0.30
-6.65
20.65
Observed Flow Gage
USGS 05317000 COTTONWOOD RIVER NEAR NEW ULM, MN
Hydrologic Unit Code: 7020008
Latitude: 44.291 351 77
Longitude: -94.4402495
Drainage Area (sq-rri): 1300
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
4.66 » | 30
-23.83
-11.60
-6.50
-13.60
0.794
0.580
0.912
30
30
20
50
Model accuracy increases
as E or E' approaches 1 .0
7.66
3.92
0.68
1.46
0.80
1.04
4.36
2.59
0.50
Clear [
Hydrology Validation
Validation for the Cottonwood model was performed at the same location but for the period 10/1/1982 through
9/30/1992. Results are presented in Figures 46 through 52 and Tables 23 and 24. The validation achieves a high
coefficient of model fit efficiency, but is under on 50 percent low volume and over on seasonal volumes for
summer and fall.
Factors that may have contributed to the difference in the flows between the calibration and validation period are:
• Increase in urban impervious surface areas.
• Increase in the intensity of tile drainage.
• Cropped areas have changed.
• PET estimates for the 20 Watershed model use SWAT weather generator statistics for solar radiation,
cloud cover, wind, and relative humidity - essentially assuming that the central tendency of these factors
has not changed over time.
G-59
-------
I Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1982 to 9/30/1991 )
Avg Modeled Flow (Same Period)
14000
12000
10000
8000
Oct-82 Oct-83 Oct-84 Oct-85
Oct-86 Oct-87
Date
Oct-88 Oct-89 Oct-90
Figure 46. Mean daily flow at USGS 05317000 Cottonwood River near New Dim, MN
validation period (SWAT).
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1982 to 9/30/1991 )
Avg Modeled Flow (Same Period)
4000
_, 3000
I
I 2000
1000
O-82
O-83 O-84
O-85
O-86
O-87
O-88
O-89
O-90
Month
Figure 47. Mean monthly flow at USGS 05317000 Cottonwood River near New Dim,
MN - validation period (SWAT).
G-60
-------
• Avg Flow (10/1/1982 to 9/30/1991 )
Line of Equal Value
Best-Fit Line
Avg Observed Flow (10/1/1982 to 9/30/1991 )
Avg Modeled Flow (10/1/1982 to 9/30/1991 )
-Line of Equal Value
i
o
4000
3000
CD
T3
o
CD
O)
ro
CD
2000
1000
0 1000 2000 3000 4000
Average Observed Flow (cfs)
O-82 O-83 O-84 O-85 O-86 O-87 O-88 O-89 O-90
Month
Figure 48. Monthly flow regression and temporal variation at USGS 05317000
Cottonwood River near New Dim, MN - validation period (SWAT).
• Avg Flow (10/1/1982 to 9/30/1991)
Line of Equal Value
Best-Fit Line
1500
EZ 1000
T3
<0
CD
500
i
o
Avg Monthly Rainfall (in)
-•-Avg Observed Flow (10/1/1982 to 9/30/1991)
Avg Modeled Flow (Same Period)
1500 -r T T T T T T T T T T ' T 0
1000
500
500
1000 1500
Average Observed Flow (cfs)
10 11 12 1 234567
Month
Figure 49. Seasonal regression and temporal aggregate at USGS 05317000
Cottonwood River near New Dim, MN - validation period (SWAT).
G-61
-------
Average Monthly Rainfall (in)
•Median Observed Flow (10/1/1982 to 9/30/1991)
I Observed (25th, 75th)
Modeled (Median, 25th, 75th)
2500
Figure 50. Seasonal medians and ranges at USGS 05317000 Cottonwood River near
New Dim, MN - validation period (SWAT).
Table 23. Seasonal summary at USGS 05317000 Cottonwood River near New Dim, MN
- validation period (SWAT)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
325.44
236.81
151.12
78.47
138.70
1049.35
1393.56
1075.69
1094.19
518.17
205.90
441.19
70.00
97.00
85.00
63.00
56.00
399.00
602.50
662.00
576.50
302.00
134.00
68.50
30.00
49.00
26.50
11.00
13.00
128.00
293.75
244.50
198.75
146.50
62.50
43.00
530.00
378.25
231.00
148.00
158.00
1165.00
2057.50
1330.00
1317.50
684.00
260.00
305.75
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
483.16
279.00
150.35
71.41
87.25
719.49
1024.55
1064.03
985.89
656.53
396.61
725.22
117.35
68.79
50.61
31.16
44.67
419.19
354.91
667.09
767.21
600.00
412.12
225.34
29.73
11.44
14.60
8.80
5.29
30.22
168.44
67.73
77.69
116.11
87.85
49.80
828.48
434.99
212.47
118.92
78.17
1029.60
1513.23
1529.65
1199.11
942.20
556.21
556.82
G-62
-------
•Observed Flow Duration (10/1/1982 to 9/30/1991 )
•Modeled Flow Duration (10/1/1982 to 9/30/1991 )
100000
0.1
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 51. Flow exceedance at USGS 05317000 Cottonwood River near New Ulm, MN
- validation period (SWAT).
•Observed Flow Volume (10/1/1982 to 9/30/1991 )
Modeled Flow Volume (10/1/1982 to 9/30/1991 )
8
in
ro
T3
-------
Table 24. Summary statistics at USGS 05317000 Cottonwood River near New Dim, MN
- validation period (SWAT)
HSPF Simulated Flow
REACH OUTFLOW FROM DSN 6001
9-Year Analysis F^riod: 10/1/1982 - 9/30/1991
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9)
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3^:
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
5.80
2.94
0.22
1.56
0.80
0.77
2.67
1.89
0.44
Error Statistics
-0.84
-29.79
-10.88
52.47
Observed Flow Gage
USGS 05317000 COTTONWOOD RIVER NEAR NEW ULM, MN
Hydrologic Unit Code: 7020008
Latitude: 44.291 351 77
Longitude: -94.4402495
Drainage Area (sq-rri): 1300
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
28.02 » | 30
-30 60
-13.60
-9.94
6.14
0.740
0.599
0.831
30
30
20
50
Model accuracy increases
as E or E' approaches 1 .0
5.85
3.30
0.32
1.02
0.63
1.11
3.09
2.10
0.41
Clear [
Hydrology Results for Larger Watershed
As described above, parameters determined for the Cottonwood gage were fully transferable to other gages in the
watershed. In addition, calibration and validation was pursued at a total of nine gages throughout the watershed,
including seven gages at the outlet of 8-digit HUCs and two gages on the mainstem. Calibration results were
acceptable at most gages, as summarized in Table 25. The match between observed and predicted flow volumes at
the most downstream available gage (USGS 05330000, Minnesota River near Jordan) are shown in Figures 53
through 58 and Tables 26 and 27.
G-64
-------
Table 25. Summary statistics (percent error): all stations - calibration period (SWAT)
Station
Error in total
volume:
Error in 50%
lowest flows:
Error in 10%
highest
flows:
Seasonal
volume error
- Summer:
Seasonal
volume error
- Fall:
Seasonal
volume error
- Winter:
Seasonal
volume error
- Spring:
Error in
storm
volumes:
Error in
summer
storm
volumes:
Daily Nash-
Sutcliffe
Coefficient of
Efficiency, E:
Monthly
Nash-
Sutcliffe
Coefficient of
Efficiency,
E::
05311000
Minnesota
River at
Montevideo
-7.70
22.21
-15.93
30.78
15.65
-19.05
-23.84
-11.85
23.67
0.637
0.801
05313500
Yellow
Medicine
River
19.10
98.49
9.44
59.05
37.01
26.86
4.67
13.55
-8.32
0.638
0.810
05316500
Redwood
River nr
Redwood
Falls
25.84
48.26
23.39
56.54
53.30
26.65
10.93
43.78
31.94
0.641
0.798
05317000
Cottonwood
River near
New Dim
-5.41
0.30
-6.65
20.65
4.66
-23.83
-11.60
-6.50
-13.60
0.794
0.912
05319500
Watonwan
River nr
Garden
City
-2.88
-30.14
19.60
20.56
11.22
-9.86
-12.94
72.02
34.92
0.381
0.825
05320000
Blue Earth
River nr
Rapidan
-6.46
-1.22
-0.02
25.26
1.28
-20.62
-18.60
18.65
6.89
0.724
0.885
05320500
LeSueur
River nr
Rapidan
12.11
38.42
8.82
25.23
32.44
52.98
-4.87
20.92
0.02
0.688
0.845
05325000
Minnesota
River at
Mankato
16.69
40.12
14.57
55.02
32.84
25.76
-2.83
48.13
45.88
0.653
0.841
05330000
Minnesota
River nr
Jordan
7.89
21.60
8.10
38.77
21.31
18.03
-9.25
43.49
35.70
0.633
0.882
G-65
-------
80000
60000
40000
20000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002 )
Avg Modeled Flow (Same Period)
O-92 A-94 O-95 A-97 O-98
Month
A-00
O-01
Figure 53. Monthly flow simulation: USGS 05330000 Minnesota River near Jordan,
MN - calibration period (SWAT).
Avg Flow (10/1/1992 to 9/30/2002 )
• Line of Equal Value
-Best-Fit Line
i
o 60000
iZ
T3
CD
^ 40000 -
o
CD
O) 20000
ro
CD
£
n
y = 0.96
R2
I
• .
&
74X + 910
= 0.8923
9
\^r
«
97
S
,*
0 20000 40000 60000 80000
Average Observed Flow (cfs)
Avg Observed Flow (10/1/1992 to 9/30/2002)
Avg Modeled Flow (10/1/1992 to 9/30/2002 )
, Line of Equal Value
O-92 A-94 O-95 A-97 O-98 A-00 O-01
Month
Figure 54. Monthly flow regression and temporal variation at USGS 05330000
Minnesota River near Jordan, MN - calibration period (SWAT).
G-66
-------
• Avg Flow (10/1 /1992 to 9/30/2002)
Line of Equal Value
Best-Fit Line
&
^
LL
_0)
to
20000
i15000
!10000
y = 0,8349x-ij 1996.3
K2 =
5000 10000 15000 20000 25000
Average Observed Flow (cfs)
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002)
-Avg Modeled Flow (Same Period)
25000
10 11 12 1 23456789
Month
Figure 55. Seasonal regression and temporal aggregate at USGS 05330000
Minnesota River near Jordan, MN - calibration period (SWAT).
Average Monthly Rainfall (in)
-Median Observed Flow (10/1/1992 to 9/30/2002)
35000 -i
I Observed (25th, 75th)
Modeled (Median, 25th, 75th)
Figure 56. Seasonal medians and ranges at USGS 05330000 Minnesota River near
Jordan, MN - calibration period (SWAT).
G-67
-------
Table 26. Seasonal summary at USGS 05330000 Minnesota River near Jordan, MN
calibration period (SWAT).
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
3732.17
4054.61
2858.86
1757.39
1811.52
6364.40
23775.27
15802.45
15733.60
11756.52
7020.71
3542.37
2190.00
3825.00
2900.00
1800.00
1500.00
3910.00
18700.00
13400.00
12800.00
9565.00
4225.00
2040.00
1082.50
1210.00
1702.50
1392.50
1272.50
2080.00
12550.00
9470.00
8805.00
5347.50
2377.50
1205.00
6237.50
5462.50
4000.00
2200.00
1900.00
8627.50
28925.00
20500.00
17900.00
12675.00
7782.50
3910.00
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
5253.33
4640.70
3010.99
1555.88
3090.56
7164.26
20056.78
13244.86
16930.06
15032.51
9987.30
5986.83
2954.43
2973.49
2304.81
1265.15
2277.97
3503.74
11800.40
10601.46
12234.77
11823.35
7453.16
4391.38
1425.65
1405.61
1684.51
892.75
1606.91
1705.43
6084.72
6478.48
8519.66
8634.44
5505.56
3145.48
5515.27
7576.76
3881.96
1712.94
3129.50
8215.07
25252.64
17027.85
18846.55
15728.27
10963.44
7497.30
•Observed Flow Duration (10/1/1992 to 9/30/2002 )
•Modeled Flow Duration (10/1/1992 to 9/30/2002 )
1000000
-------
o
o
in
ro
T3
0
0
-------
Table 27. Summary statistics at USGS 05330000 Minnesota River near Jordan, MN
calibration period (SWAT).
HSPF Simulated Flow
REACH OUTFLOW FROM DSN 6001
10-Year Analysis F^riod: 10/1/1992 - 9/30/2002
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9)
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
7.41
3.01
0.99
2.19
0.91
0.82
3.49
2.06
0.44
Error Statistics
7.89
21.60
8.10
38.77
Observed Flow Gage
USGS 05330000 MINNESOTA RIVER NEAR JORDAN, MN
Hydrologic Unit Code: 7020012
Latitude: 44.69301 845
Longitude: -93.641 902
Drainage Area (sq-rri): 16200
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
21.31 » | 30
18.03
-9J5_
35~70~
0.633
0.502
0.882
30
30 _
50
Model accuracy increases
as E or E' approaches 1 .0
6.87
2.79
0.81
1.58
0.75
0.70
3.85
1.43
0.32
Clear [
Results of the validation exercise are summarized in Table 28. Problems similar to those experienced on the
Cottonwood gage were seen at all the tributary gages, with overprediction of seasonal flows in summer and fall.
However, as noted above, this is likely due to the use of land use and model parameters that are more reflective of
current conditions and is not believed to present a bar to application of the model.
G-70
-------
Table 28. Summary statistics: all stations - validation period (SWAT)
Station
Error in total
volume:
Error in 50%
lowest flows:
Error in 10%
highest
flows:
Seasonal
volume error
- Summer:
Seasonal
volume error
- Fall:
Seasonal
volume error
- Winter:
Seasonal
volume error
- Spring:
Error in
storm
volumes:
Error in
summer
storm
volumes:
Daily Nash-
Sutcliffe
Coefficient of
Efficiency, E:
Monthly
Nash-
Sutcliffe
Coefficient of
Efficiency, E:
05313500
Yellow
Medicine
River
11.24
20.06
-4.84
71.72
54.63
-25.13
-0.75
8.05
2.96
0.517
0.738
05316500
Redwood
River nr
Redwood
Falls
31.84
65.19
10.80
127.37
43.60
1.06
15.22
45.17
62.49
0.439
0.668
05317000
Cottonwood
River near
New Dim
-0.84
-29.79
-10.88
52.47
28.02
-30.60
-13.60
-9.94
6.14
0.740
0.831
05319500
Watonwan
River nr
Garden
City
-1.10
-68.25
23.45
59.14
22.81
-18.95
-18.56
90.61
135.92
-0.245
0.440
05320000
Blue Earth
River nr
Rapidan
-2.66
13.13
-5.64
50.15
23.23
-24.39
-16.18
25.30
36.86
0.636
0.828
05320500
LeSueur
River nr
Rapidan
32.01
48.38
24.34
92.70
42.94
11.05
19.79
45.02
59.70
0.513
0.657
05325000
Minnesota
River at
Mankato
38.66
60.68
28.42
98.32
57.68
26.20
15.14
77.61
81.05
0.423
0.656
05330000
Minnesota
River nr
Jordan
28.58
34.93
22.70
72.78
40.65
18.12
10.04
74.72
60.81
0.421
0.723
Water Quality Calibration and Validation
Initial calibration and validation of water quality was done on the Cottonwood River (USGS 05317000), using
1993-2002 for calibration and 1986-1992 for validation. As with hydrology, calibration was performed on the
later period as this better reflects the land use included in the model. The start of the validation period is
constrained by data availability.
Calibration adjustments for sediment focused on the following parameters:
• BIOMIX
G-71
-------
• SPCON (Linear parameters for estimating maximum amount of sediment that can be re-entrained during
channel sediment routing)
• CH_COV (Channel cover factor)
• CH_EROD (Channel erodibility factor)
• USLE-C (Land surface cover factor).
Simulated and estimated sediment loads at the Cottonwood station for both the calibration and validation periods
are shown in Figures 59 through 62 and statistics for the two periods are provided separately in Tables 29 and 30.
The key statistic in Table 29 is the relative percent error, which shows the error in the prediction of monthly load
normalized to the estimated load. The table also shows the relative average absolute error, which is the average of
the relative magnitude of errors in individual monthly load predictions. This number is inflated by outlier months
in which the simulated and estimated loads differ by large amounts (which may be as easily due to uncertainty in
the estimated load due to limited data as to problems with the model) and the third statistic, the relative median
absolute error, is likely more relevant and shows good agreement.
TSS
10,000,000
1,000,000
100.000
10,000
o
I 1,000
I 100
10 -
-Regression
Loads
COCOCOOOO)O)O>O)O)O)O>G>O>G)
c
TO
c
03
C
TO
C
TO
C
ro
c
TO
TO
C C
TO to
TO
C
TO
C
TO
TO
C
TO
C
TO
Figure 59. Fit for monthly load of TSS at USGS 05317000 Cottonwood River (SWAT).
Table 29. Model fit statistics (observed minus predicted) for monthly sediment loads
using stratified regression at USGS 05317000 Cottonwood River (SWAT)
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1993-2002)
9.2%
36%
9.1%
Validation period
(1986-1992)
9.0%
65%
14.3%
G-72
-------
COTTONWOOD RIVER NEAR NEW ULM, MN 1993-2002
1000000
100000 -
> 10000 -
ra
1 1000 H
•o
ra
o
(0
V)
100 -
10 -
1 -
0.1 -
0.01
1
10
100 1000
Flow, cfs
10000
100000
» Simulated
Observed
Power(Simulated)
Power (Observed)
Figure 60. Power plot for observed and simulated TSS at USGS 05317000
Cottonwood River - calibration period (SWAT).
COTTONWOOD RIVER NEAR NEW ULM, MN 1986-1992
ra
•o
ra
o
(0
V)
10000
» Simulated A Observed
Power (Simulated)
Power (Observed)
Figure 61. Power plot for observed and simulated TSS at USGS 05317000
Cottonwood River - validation period (SWAT).
G-73
-------
Cotton wood River - New Dim
1993-2002
-Simulated
A Observed
O)
eo
CO
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Year
Figure 62. Time series plot of TSS concentration at USGS 05317000 Cottonwood
River (SWAT).
Table 30. Relative errors (observed minus predicted), TSS concentration at USGS
05317000 Cottonwood River (SWAT)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1993-2002)
121
27.69%
-2.52%
Validation period
(1986-1992)
75
13.61%
-4.04%
Calibration adjustments for total phosphorus and total nitrogen focused on the following parameters:
• PHOSKD (Phosphorus soil partitioning coefficient)
• RS4
• PSP
• BC3 and BC4
• SOL_CBN1 (Organic carbon in the first soil layer)
• Michaelis-Menton half-saturation constant for nitrogen and phosphorus
• MUMAX
Results for the phosphorus simulation are shown in Figures 63 through 66 and Tables 31 and 32. Results for the
nitrogen simulation are shown in Figures 67 through 70 and Tables 33 and 34. The SWAT fit is generally good,
with calibration and validation error statistics similar to those obtained from the HSPF model.
G-74
-------
Total P
1000
100 -
o
E
In
10 -
B 1 -
0.1 -
0.01
opapapapcpcpcpcpcpcpcpcpcpcpcpcpcp
ccccccccccccccccc
-Regression Loads
-Simulated Loads
Figure 63. Fit for monthly load of total phosphorus at USGS 05317000 Cottonwood
River (SWAT).
Table 31. Model fit statistics (observed minus predicted) for monthly phosphorus
loads using stratified regression at USGS 05317000 Cottonwood River (SWAT)
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1993-2002)
9.3%
46%
1 1 .2%
Validation period
(1986-1992)
-21.6
80%
9.3%
G-75
-------
COTTONWOOD RIVER NEAR NEW ULM, MN 1993-2002
ra
1000
100 -
10 -
1 -
0.1 -
0.01 -
0.001 -
0.0001
1
10
100 1000
Flow, cfs
10000
100000
» Simulated -* Observed ^^— Power (Simulated)
Power(Observed)
Figure 64. Power plot for observed and simulated total phosphorus at USGS
05317000 Cottonwood River - calibration period (SWAT).
COTTONWOOD RIVER NEAR NEW ULM, MN 1986-1992
100
» Simulated A Observed
Power (Simulated)
Power (Observed)
10000
G-76
-------
Figure 65. Power plot for observed and simulated total phosphorus at USGS
05317000 Cottonwood River - validation period (SWAT).
O)
E
Cottonwood River - New Dim
1993-2002
0.01
0.001
-Simulated A Observed
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Year
Figure 66. Time series plot of total phosphorus concentration at USGS 05317000
Cottonwood River (SWAT).
Table 32. Relative errors (observed minus predicted), total phosphorus concentration
at USGS 05317000 Cottonwood River (SWAT)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1993-2002)
123
-20
-0
45%
25%
Validation period
(1986-1992)
75
-707.00%
-87.32%
G-77
-------
Total N
6,000
5,000
4,000
3,000
2,000
1,000 -•«•-•
-Averaging Loads
•Simulated Loads
Figure 67. Fit for monthly load of total nitrogen at USGS 05317000 Cottonwood River
(SWAT).
Table 33. Model fit statistics (observed minus predicted) for monthly total nitrogen
loads using averaging estimator at USGS 05317000 Cottonwood River (SWAT)
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1993-2002)
-8.9%
54%
24.4%
Validation period
(1986-1992)
-1.3%
65%
28.7%
G-78
-------
COTTONWOOD RIVER NEAR NEW ULM, MN 1993-2002
0.001
10
100 1000
Flow, cfs
10000
100000
• Simulated A Observed ^^—Power(Simulated) ^^— Power (Observed)
Figure 68. Power plot for observed and simulated total nitrogen at USGS 05317000
Cottonwood River- calibration period (SWAT).
COTTONWOOD RIVER NEAR NEW ULM, MN 1986-1992
1000
10000
» Simulated A Observed ^^™Power(Simulated) ^^™ Power (Observed)
G-79
-------
Figure 69. Power plot for observed and simulated total nitrogen at USGS 05317000
Cottonwood River-validation period (SWAT).
Cotton wood River - New Dim
1993-2002
-Simulated
A Observed
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Year
Figure 70. Time series plot of total nitrogen concentration at USGS 05317000
Cottonwood River (SWAT).
Table 34. Relative errors (observed minus predicted), total nitrogen concentration at
USGS 05317000 Cottonwood River (SWAT)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1993-2002)
20
12.65%
31.42%
Validation period
(1986-1992)
75
-78.90%
12.33%
Water Quality Results for Larger Watershed
Water quality results from the larger watershed from the SWAT model appear to be much less precise than those
obtained with the HSPF model. This is believed to be largely a result of the calibration strategy adopted for the
SWAT application: As with hydrology, the Cottonwood River watershed SWAT model parameters for water
quality were directly transferred to other portions of the watershed. In contrast, the HSPF model used a spatial
calibration approach. Application of the SWAT model without spatial adjustments resulted in relatively large
errors in predicting loads and concentrations at some stations.
G-80
-------
Summary statistics for the SWAT water quality calibration and validation at other stations in the watershed are
provided in Tables 35 and 36.
G-81
-------
Table 35. Summary statistics for water quality at all stations - calibration period 1993-
2002 (SWAT)
Station
Relative
Percent Error
TSS Load
TSS
Concentration
Median Error
Relative
Percent Error
TP Load
TP
Concentration
Median Error
Relative
Percent Error
TN Load
TN
Concentration
Median Error
05313500
Yellow
Medicine
River
-97.8%
7.10%
-38.3%
19.53%
-22.9%
29.17%
05316500
Redwood
River nr
Redwood
Falls
-96.0%
-3.0%
-84.0%
-52.87%
-44.3%
19.87%
05317000
Cottonwood
River near
New Dim
9.2%
-2.84%
9.3%
-0.25%
-8.9%
31.42%
05319500
Watonwan
River nr
Garden
City
-166.7%
-11.14%
-58.1%
-16.19%
-17.9%
38.06%
05320000
Blue
Earth
River nr
Rapidan
-145.1%
-31.85%
-54.7%
-23.38%
-10.9%
31.66%
05320500
LeSueur
River nr
Rapidan
-139.8%
-28.91%
-65.7%
-13.37%
13.4%
28.4%
05325000
Minnesota
River at
Mankato
-73.1%
-41.2%
-13.1%
-5.65%
9.5%
42.39%
05330000
Minnesota
River nr
Jordan
-40.7%
-20.48%
-5.0%
-5.99%
18.20%
47.64%
Table 36. Summary statistics for water quality at all stations - validation period 1986-
1992 (SWAT)
Station
Relative
Percent Error
TSS Load
TSS
Concentration
Median Error
Relative
Percent Error
TP Load
TP
Concentration
Median Error
Relative
Percent Error
TN Load
TN
Concentration
Median Error
05313500
Yellow
Medicine
River
-56.8%
-8.60%
-27.4%
10.04%
-28.0%
24.09%
05316500
Redwood
River nr
Redwood
Falls
-75.1%
-14.87%
-1142.6%
-275.32%
-68.7%
-3.37%
05317000
Cottonwood
River near
New Dim
9.0%
-4.04%
-21.6%
-87.32%
-1.3%
12.33%
05319500
Watonwan
River nr
Garden
City
-227.1%
-24.78%
-174.3%
-74.48%
-69.9%
21.14%
05320000
Blue
Earth
River nr
Rapidan
-136.3%
-65.75%
-60.4%
-80.91%
-15.3%
10.54%
05320500
LeSueur
River nr
Rapidan
-199.8%
-51.33%
-143.7%
-91.5%
-43.1%
5.29%
05325000
Minnesota
River at
Mankato
-95.1%
-91.48%
-39.6%
-76.07%
-4.8%
10.57%
05330000
Minnesota
River nr
Jordan
-43.2%
-39.50%
-31.0%
-80.93%
4.2%
12.33%
G-82
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
Alexander, R.B., R.A. Smith, G.E. Schwarz, E.W. Boyer, J.V. Nolan, and J.W. Brakebill. 2008. Differences in
phosphorus and nitrogen delivery to the Gulf of Mexico from the Mississippi River Basin. Environmental
Science and Technology. 42: 822-830.
Allen, R.G., L.S. Pereira, D. Raes, and M. Smith. 1998. Crop Evapotranspiration: Guidelines for Computing
Crop Water Requirements. Irrigation and Drainage Paper 56. Food and Agriculture Organization of the United
Nations, Rome.
Jones, P.M. and T.A. Winterstein. 2000. Characterization of Rainfall-Runoff Response and Estimation of the
Effect of Wetland Restoration on Runoff, Heron Lake Basin, Southwestern Minnesota, 1991-97. Water-Resources
Investigations Report 00-4095. U.S. Geological Survey, Mounds View, MN.
Preston, S.D., V.J. Bierman, Jr., and S.E. Silliman. 1989. An evaluation of methods for the estimation of tributary
mass loads. Water Resources Research, 25(6): 1379-1389.
Tetra Tech. 2008. Minnesota River Basin Turbidity TMDL and Lake Pepin Excessive Nutrient TMDL, Model
Calibration and Validation Report. Prepared for Minnesota Pollution Control Agency, St. Paul, MN by Tera
Tech, Inc., Research Triangle Park, NC.
Tetra Tech. 1999. Improving Point Source Loadings Data for Reporting National Water Quality Indicators. Final
Technical Report prepared for U.S. Environmental Protection Agency, Office of Waste water Management,
Washington, DC, by Tetra Tech, Inc., Fairfax, VA.
U.S. Army Corps of Engineers. 1956. Snow Hydrology. North Pacific Division, Corps of Engineers, Portland,
OR.
USEPA. 2008. Using the BASINS Meteorological Database (Version 2006). BASINS Technical Note 10.
Office of Water, U.S. Environmental Protection Agency, Washington, DC.
http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf (Accessed June,
2009).
USEPA. 2006. Sediment Parameter and Calibration Guidance for HSPF. BASINS Technical Note 8. Office of
Water, U.S. Environmental Protection Agency, Washington, DC.
http://water.epa.gov/scitech/datait/models/basins/upload/2006_02_02_BASINS_tecnote8.pdf (Accessed June,
2009).
USEPA. 2000. Estimating Hydrology and Hydraulic Parameters for HSPF. BASINS Technical Note 6. EPA-
823-ROO-012. Office of Water, U.S. Environmental Protection Agency, Washington, DC.
http://water.epa.gov/scitech/datait/models/basins/upload/2000_08_14_BASINS_tecnote6.pdf (Accessed June,
2009).
G-83
-------
Appendix H
Model Configuration, Calibration and
Validation
Basin: Willamette River (Willa)
H-l
-------
Contents
Watershed Background H-7
Water Body Characteristics H-7
Soil Characteristics H-9
Land Use Representation H-9
Point Sources H-13
Meteorological Data H-16
Watershed Segmentation H-19
Calibration Data and Locations H-21
HSPF Modeling H-22
Changes Made to Base Data Provided H-22
Assumptions H-23
Hydrology Calibration H-25
Hydrology Validation H-31
Hydrology Results for Larger Watershed H-36
Water Quality Calibration and Validation H-43
Water Quality Results for Larger Watershed H-52
SWAT Modeling H-54
Changes Made to Base Data Provided H-54
Assumptions H-54
Hydrology Calibration H-55
Hydrology Validation H-60
Hydrology Results for Larger Watershed H-66
Water Quality Calibration H-72
Water Quality Results for Larger Watershed H-81
References H-82
H-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes H-ll
Table 2. Land use distribution for the Willamette River watershed (2001 NLCD mi2) H-12
Table 3. Major point source discharges in the Willamette River watershed H-14
Table 4. Precipitation stations for the Willamette River watershed model H-16
Table 5. Calibration and validation locations in the Willamette River watershed H-21
Table 6. Reservoirs represented in the Willamette River watershed model H-25
Table 7. Seasonal summary at USGS 14207500 Tualatin River at West Linn, OR - calibration period
(HSPF) H-28
Table 8. Summary statistics at USGS 14207500 Tualatin River at West Linn, OR - calibration period
(HSPF) H-30
Table 9. Seasonal summary at USGS 14207500 Tualatin River at West Linn - validation period (HSPF) H-33
Table 10. Summary statistics at USGS 14207500 Tualatin River at West Linn, OR - validation period
(HSPF) H-35
Table 11. Seasonal summary at USGS 14191000 Willamette River at Salem, OR - calibration period
(HSPF) 39
Table 12. Summary statistics at USGS 14191000 Willamette River at Salem, OR - calibration period
(HSPF) 41
Table 13. Summary statistics (percent error) for all stations - calibration period (HSPF) H-42
Table 14. Summary statistics (percent error) for all stations - validation period (HSPF) H-42
Table 15. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression (HSPF) H-44
Table 16. Relative errors (observed minus predicted), TSS concentration, at USGS 14207500 Tualatin
River at West Linn, OR (HSPF) H-46
Table 17. Model fit statistics (observed minus predicted) for monthly total phosphorus loads using stratified
regression (HSPF) H-47
Table 18. Relative errors (observed minus predicted), total phosphorus concentration at USGS 14207500
Tualatin River at West Linn, OR (HSPF) H-49
Table 19. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using averaging
estimator (HSPF) H-50
Table 20. Relative errors (observed minus predicted), total nitrogen concentration, USGS 14207500 Tualatin
River at West Linn, OR (HSPF) H-52
Table 21. Summary statistics for water quality for all stations (observed minus predicted) (HSPF) H-5 3
Table 22. Seasonal summary at USGS 14207500 Tualatin River at West Linn, OR - calibration period
(SWAT) H-58
Table 23. Summary statistics: USGS 14207500 Tualatin River at West Linn, OR - calibration period
(SWAT) H-60
Table 24. Seasonal summary at USGS 14207500 Tualatin River at West Linn, OR - validation period
(SWAT) H-63
Table 25. Summary statistics at USGS 14207500 Tualatin River at West Linn, OR - validation period
(SWAT) H-65
Table 26. Seasonal summary at USGS 14191000 Willamette River at Salem, OR - calibration period
(SWAT) H-68
Table 27. Summary statistics at USGS 14191000 Willamette River at Salem, OR - calibration period
(SWAT) H-70
Table 28. Summary statistics (percent error) for all stations - calibration period (SWAT) H-71
Table 29. Summary statistics (percent error) for all stations - validation period (SWAT) H-71
Table 30. Model fit statistics (observed minus predicted) for monthly TSS loads using stratified regression
(SWAT) H-73
Table 31. Relative errors (observed minus predicted), TSS concentration, at USGS 14207500 Tualatin
River at West Linn, OR (SWAT) H-75
H-3
-------
Table 32. Model fit statistics (observed minus predicted) for monthly total phosphorus loads using stratified
regression (SWAT) H-76
Table 33. Relative errors (observed minus predicted), total phosphorus concentration, at USGS 14207500
Tualatin River at West Linn, OR (SWAT) H-78
Table 34. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using averaging
estimator (SWAT) H-78
Table 35. Relative errors (observed minus predicted), total nitrogen concentration, at USGS 14207500
Tualatin River at West Linn, OR (SWAT) H-80
Table 36. Summary statistics for water quality: all stations (observed minus predicted) (SWAT) H-81
Figures
Figure 1. Location ofthe Willamette River watershed H-8
Figure 2. Land use in the Willamette River watershed H-10
Figure 3. Major point sources in the Willamette River watershed H-15
Figure 4. Weather stations for the Willamette River watershed model H-18
Figure 5. Model segmentation and USGS stations utilized for the Willamette River watershed H-20
Figure 6. Dams and reservoirs in the Willamette River watershed (Source: USAGE 2009) H-24
Figure 7. Mean daily flow at USGS 14207500 Tualatin River at West Linn, OR - calibration period
(HSPF) H-26
Figure 8. Mean monthly flow at USGS 14207500 Tualatin River at West Linn, OR - calibration period
(HSPF) H-26
Figure 9. Monthly flow regression and temporal variation at USGS 14207500 Tualatin River at West Linn,
OR-calibration period (HSPF) H-27
Figure 10. Seasonal regression and temporal aggregate at USGS 14207500 Tualatin River at West Linn, OR -
calibration period (HSPF) H-27
Figure 11. Seasonal medians and ranges at USGS 14207500 Tualatin River at West Linn, OR - calibration
period (HSPF) H-28
Figure 12. Flow exceedence at USGS 14207500 Tualatin River at West Linn, OR - calibration period (HSPF)....
H-29
Figure 13. Flow accumulation at USGS 14207500 Tualatin River at West Linn, OR - calibration period
(HSPF) H-29
Figure 14. Mean daily flow at USGS 14207500 Tualatin River at West Linn, OR - validation period
(HSPF) H-31
Figure 15. Mean monthly flow at USGS 14207500 Tualatin River at West Linn, OR - validation period
(HSPF) H-31
Figure 16. Monthly flow regression and temporal variation at USGS 14207500 Tualatin River at West Linn,
OR-validation period (HSPF) H-32
Figure 17. Seasonal regression and temporal aggregate at USGS 14207500 Tualatin River at West Linn,
OR-validation period (HSPF) H-32
Figure 18. Seasonal medians and ranges at USGS 14207500 Tualatin River at West Linn, OR - validation
period (HSPF) H-33
Figure 19. Flow exceedence at USGS 14207500 Tualatin River at West Linn, OR - validation period
(HSPF) H-34
Figure 20. Flow accumulation at USGS 14207500 Tualatin River at West Linn, OR - validation period
(HSPF) H-34
Figure 21. Mean daily flow at USGS 14191000 Willamette River at Salem, OR - calibration period
(HSPF) H-36
Figure 22. Mean monthly flow at USGS 14191000 Willamette River at Salem, OR - calibration period
(HSPF) H-37
H-4
-------
Figure 23. Monthly flow regression and temporal variation at USGS 14191000 Willamette River at Salem,
OR-calibration period (HSPF) H-37
Figure 24. Seasonal regression and temporal aggregate at USGS 14191000 Willamette River at Salem, OR -
calibration period (HSPF) H-38
Figure 25. Seasonal medians and ranges at USGS 14191000 Willamette River at Salem, OR - calibration
period (HSPF) H-38
Figure 26. Flow exceedence at USGS 14191000 Willamette River at Salem, OR - calibration period
(HSPF) H-39
Figure 27. Flow accumulation at USGS 14191000 Willamette River at Salem, OR - calibration period
(HSPF) H-40
Figure 28. Fit for monthly load of TSS at USGS 14207500 Tualatin River at West Linn, OR (HSPF) H-44
Figure 29. Power plot for observed and simulated TSS at USGS 14207500 Tualatin River at West Linn,
OR-calibration period (HSPF) H-45
Figure 30. Power plot for observed and simulated TSS at USGS 14207500 Tualatin River at West Linn, OR -
validation period (HSPF) H-45
Figure 31. Time series plot of TSS concentration at USGS 14207500 Tualatin River at West Linn, OR -
(HSPF) H-46
Figure 32. Fit for monthly load of total phosphorus at USGS 14207500 Tualatin River at West Linn, OR
(HSPF) H-47
Figure 33. Power plot for observed and simulated total phosphorus at USGS 14207500 Tualatin River at
West Linn, OR - calibration period (HSPF) H-48
Figure 34. Power plot for observed and simulated total phosphorus at USGS 14207500 Tualatin River at
West Linn, OR-validation period (HSPF) H-48
Figure 35. Time series plot of total phosphorus concentration at USGS 14207500 Tualatin River at West Linn,
OR (HSPF) H-49
Figure 36. Fit for monthly load of total nitrogen at USGS 14207500 Tualatin River at West Linn, OR
(HSPF) H-50
Figure 37. Power plot for observed and simulated total nitrogen at USGS 14207500 Tualatin River at West
Linn, OR-calibration period (HSPF) H-51
Figure 38. Power plot for observed and simulated total nitrogen at USGS 14207500 Tualatin River at West
Linn, OR-validation period (HSPF) H-51
Figure 39. Time series plot of total nitrogen concentration at USGS 14207500 Tualatin River at West Linn,
OR (HSPF) H-52
Figure 40. Mean daily flow at USGS 14207500 Tualatin River at West Linn, OR - calibration period
(SWAT) H-56
Figure 41. Mean monthly flow at USGS 14207500 Tualatin River at West Linn, OR - calibration period
(SWAT) H-56
Figure 42. Monthly flow regression and temporal variation at USGS 14207500 Tualatin River at West Linn,
OR-calibration period (SWAT) H-57
Figure 43. Seasonal regression and temporal aggregate at USGS 14207500 Tualatin River at West Linn,
OR-calibration period (SWAT) H-57
Figure 44. Seasonal medians and ranges at USGS 14207500 Tualatin River at West Linn, OR - calibration
period (SWAT) H-58
Figure 45. Flow exceedence at USGS 14207500 Tualatin River at West Linn, OR - calibration period
(SWAT) H-59
Figure 46. Flow accumulation at USGS 14207500 Tualatin River at West Linn, OR - calibration period
(SWAT) H-59
Figure 47. Mean daily flow at USGS 14207500 Tualatin River at West Linn, OR - validation period
(SWAT) H-61
Figure 48. Mean monthly flow at USGS 14207500 Tualatin River at West Linn, OR - validation period
(SWAT) H-61
Figure 49. Monthly flow regression and temporal variation at USGS 14207500 Tualatin River at West Linn,
OR-validation period (SWAT) H-62
H-5
-------
Figure 50. Seasonal regression and temporal aggregate at USGS 14207500 Tualatin River at West Linn, OR -
validation period (SWAT) H-62
Figure 51. Seasonal medians and ranges at USGS 14207500 Tualatin River at West Linn, OR - validation
period (SWAT) H-63
Figure 52. Flow exceedence at USGS 14207500 Tualatin River at West Linn, OR - validation period
(SWAT) H-64
Figure 53. Flow accumulation at USGS 14207500 Tualatin River at West Linn, OR - validation period
(SWAT) H-64
Figure 54. Mean daily flow at USGS 14191000 Willamette River at Salem, OR - calibration period
(SWAT) H-66
Figure 55. Mean monthly flow at USGS 14191000 Willamette River at Salem, OR - calibration period
(SWAT) H-66
Figure 56. Monthly flow regression and temporal variation at USGS 14191000 Willamette River at Salem,
OR-calibration period (SWAT) H-67
Figure 57. Seasonal regression and temporal aggregate at USGS 14191000 Willamette River at Salem, OR -
calibration period (SWAT) H-67
Figure 58. Seasonal medians and ranges at USGS 14191000 Willamette River at Salem, OR - calibration
period (SWAT) H-68
Figure 59. Flow exceedence at USGS 14191000 Willamette River at Salem, OR - calibration period
(SWAT) H-69
Figure 60. Flow accumulation at USGS 14191000 Willamette River at Salem, OR - calibration period
(SWAT) H-69
Figure 61. Fit for monthly load of TSS at USGS 14207500 Tualatin River at West Linn, OR (SWAT) H-72
Figure 62. Power plot for observed and simulated TSS at USGS 14207500 Tualatin River at West Linn, OR -
calibration period (SWAT) H-73
Figure 63. Power plot for observed and simulated TSS at USGS 14207500 Tualatin River at West Linn, OR -
validation period (SWAT) H-74
Figure 64. Time series plot of TSS concentration at USGS 14207500 Tualatin River at West Linn, OR
(SWAT) H-74
Figure 65. Fit for monthly load of total phosphorus at USGS 14207500 Tualatin River at West Linn, OR
(SWAT) H-75
Figure 66. Power plot for observed and simulated total phosphorus at USGS 14207500 Tualatin River at
West Linn, OR - calibration period (SWAT) H-76
Figure 67. Power plot for observed and simulated total phosphorus at USGS 14207500 Tualatin River at West
Linn, OR-validation period (SWAT) H-77
Figure 68. Time series plot of total phosphorus concentration at USGS 14207500 Tualatin River at West
Linn, OR (SWAT) H-77
Figure 69. Fit for monthly load of total nitrogen at USGS 14207500 Tualatin River at West Linn, OR
(SWAT) H-78
Figure 70. Power plot for observed and simulated total nitrogen at USGS 14207500 Tualatin River at West
Linn, OR-calibration period (SWAT) H-79
Figure 71. Power plot for observed and simulated total nitrogen at USGS 14207500 Tualatin River at West
Linn, OR-validation period (SWAT) H-79
Figure 72. Time series plot of total nitrogen concentration at USGS 14207500 Tualatin River at West Linn,
OR (SWAT) H-80
H-t
-------
The Willamette River basin is located in northwestern Oregon. The model study area is within HUC 1709,
consisting of 11 HUCSs and covering about 11,200 mi2. The Willamette River is the 13th largest river in the
conterminous U.S. in terms of streamflow and produces more runoff per unit area than any of the larger rivers. It
discharges to the Columbia River, which flows west to the Pacific Ocean along Oregon's northern border
(Figure 1).
The basin is bordered on the west by the Coast Range, where elevations exceed 4,000 ft, and on the east by the
Cascade Range, with several peaks higher than 10,000 ft. The Willamette Valley, with elevations near sea level,
lies between the two ranges (USGS, 2001). Forested land covers approximately 70 percent of the watershed and
dominates the foothills and mountains of the Coast and Cascade Ranges. Agricultural land, mostly cropland,
comprises 22 percent of the basin and is located predominantly in the Willamette Valley. About one-third of the
agricultural land is irrigated, and most of this is adjacent to the main stem Willamette River in the southern basin
or scattered throughout the northern valley. Urban land comprises 6 percent of the watershed and is located
primarily in the valley along the main stem Willamette River. The Willamette River flows through Portland,
Oregon's largest metropolitan area, before entering the Columbia River.
The Willamette basin is characterized by cool, wet winters and warm, dry summers. About 70-80 percent of the
annual precipitation falls from October through March. Most precipitation falls as snow above about the 5,000 ft
level of the Cascades; however, the Coast Range and Willamette Valley receive relatively little snow. Mean
monthly air temperatures in the valley range from about 3-5° C during January to 17-20° C during August.
Although annual precipitation averages 62 inches in the Willamette basin, topography strongly influences its
distribution. Yearly amounts range from 40-50 inches in the valley to as much as 200 inches near the crests of the
Coast and Cascade Ranges.
More than three-fourths of the water used in the Willamette watershed is surface water. The largest single use is
for the irrigation of crops. Public water supply (serving cities, towns, mobile home parks, apartment complexes) is
the second largest use. Public supply consists mostly of withdrawals from Cascade streams, including the Bull
Run in the Sandy River watershed and the Clackamas, Santiam, and McKenzie Rivers. The small amount of
groundwater used for public supply (-10% of the total) comes predominantly from alluvial aquifers located along
Cascade streams or along the main stem Willamette River. Most commercial water use is by fish hatcheries, and
most industrial use is by pulp-and-paper mills.
Water Body Characteristics
The Willamette River is the 13th largest river in the conterminous U.S. in terms of streamflow and produces more
runoff per mi2 than any of the larger rivers. The Sandy River watershed includes the Bull Run watershed, which is
Portland's primary drinking water supply. The Willamette and Sandy Rivers are tributary to the Columbia River,
which flows west to the Pacific Ocean along Oregon's northern border. The Willamette River flows through
Portland, Oregon's largest metropolitan area, before entering the Columbia River.
Streamflow in the Willamette basin reflects the seasonal distribution of precipitation, with 60-85 percent of runoff
occurring from October through March, but less than 10 percent occurring during July and August. Releases from
13 tributary reservoirs are managed for water quality enhancement by maintaining a flow of 6,000 cfs in the
Willamette River at Salem during summer months. Flows in the lower Willamette River watershed are dominated
by the effects of 13 reservoirs and their associated dams operated by the U.S. Army Corps of Engineers for water
supply, flood control, and navigation. These reservoirs control much of the runoff from the southern and eastern
mountainous portions of the watershed where precipitation and snow fall are highest. Incorporation of the
reservoirs in the model was a significant part of the model development effort.
H-7
-------
Hydrography
Water (Nat. Atlas Dataset)
US Census Populated Places
^B Municipalities (pop £ 50.000)
County Boundaries
I I Watershed with HUCBs
Tualatin
(17090010)
Yamhill
(17090008)
ladle
Willamette
1709000
Molalla-Pudding
(17090009)
Clackamas
(17090011)
pper Willamette
(17090003)
South Santiam
(17090006)
North Santiam
17090005)
Mckenzie
(17090004)
Middle Fork
Willamette
(17090001)
Coastal For
Willamette
GCRP Model Areas - Willamette River Basin
Base Map
Figure 1. Location of the Willamette River watershed.
-------
Soil Characteristics
One of the most important characteristics of soils for watershed modeling is their hydrologic soil group (HSG).
The 20 Watershed study utilized STATSGO soil survey HSG information during model set-up. Soils are
classified into four hydrologic groups (SCS 1986), separated by runoff potential, as follows:
A Low runoff potential and high infiltration rates even when thoroughly wetted. Chiefly deep, well
to excessively drained sands or gravels. High rate of water transmission (> 0.75 cm/hr).
B Moderate infiltration rates when thoroughly wetted. Chiefly moderately deep to deep, moderately
well to well drained soils with moderately fine to moderately coarse textures. Moderate rate of
water transmission (0.40—0.75 cm/hr).
C Low infiltration rates when thoroughly wetted. Chiefly soils with a layer that impedes downward
movement of water, or soils with moderately fine to fine texture. Low rate of water transmission
(0.15—0.40 cm/hr).
D High runoff potential. Very low infiltration rates when thoroughly wetted. Chiefly clay soils with
a high swelling potential, soils with a permanent high water table, soils with a claypan or clay
layer at or near the surface, or shallow soils over nearly impervious material. Very low rate of
water transmission (0—0.15 cm/hr).
The Willamette River watershed contains all four HSGs, but consists of mostly B, C, and D soils with a
dominance of C soils.
Land Use Representation
Land use/cover in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage and is
predominantly forest (Figure 2). Most of the developed areas of the watershed are found along the Willamette
River with the major urban development near the mouth of the river at the city of Portland.
H-9
-------
V • y.- •;?'•/>• //*. ' If
/Z.. ;—4
M
Sf-.- - • v*
Interstate
^^— Hydrography
^H Water (Nat. Atlas Dataset)
^| County Boundaries
State Boundaries
Watershed
2001 NLCD Land Use
I Open water
^ Developed, open space
| Developed, low intensity
I Developed, medium intensity
I Developed, high intensity
^ Barren land
I Deciduous forest
Jj Evergreen forest
| | Mixed forest
^ Scrub/shrub
^ Grassland/hertaceous
~j PastureAiay
| Cultivated crops
| Woody wetlands
~j Emergent herbaceous wetlands
GCRP Model Areas - Willamette River Basin
Land Use Map
Figure 2. Land use in the Willamette River watershed.
H-10
-------
NLCD land cover classes were aggregated according to the scheme shown in Table 1 for representation in the 20
Watershed model, and then overlain with the soils HSG grid. For HSPF, pervious and impervious lands are
specified separately, so only one developed pervious class is used, along with an impervious class. HSPF
simulates impervious land areas separately from pervious land. Impervious area distributions were determined
from the NLCD Urban Impervious data coverage. Specifically, percent impervious area was calculated over the
entire watershed for each of the four developed land use classes. These percentages were then used to separate out
impervious land. NLCD impervious area data products are known to underestimate total imperviousness in rural
areas; however, the model properly requires connected impervious area, not total impervious area, and the NLCD
tabulation is assumed to provide a reasonable approximation of connected impervious area. In SWAT, different
developed land classes are specified separately. In HSPF the WATER, BARREN, DEVPERV, and WETLAND
classes are not subdivided by HSG; SWAT uses the built-in HRU overlay mechanism in the ArcSWAT interface.
Table 1. Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area
usually accounted for as
reach area
Deciduous
Evergreen
Mixed
Emergent & woody
wetlands
Aquatic bed wetlands (not
emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
HSPF (after processing)
WATER
BARREN, Assume HSG D
DEVPERV;
IMPERV
BARREN (D)
FOREST (A,B,C,D)
SHRUB (A,B,C,D)
GRASS (A,B,C,D), BARREN (D)
GRASS(A,B,C,D)
AGRI (A,B,C,D)
WETLAND, Assume HSG D
WATER
The distribution of land use in the watershed is summarized in Table 2.
H-ll
-------
Table 2. Land use distribution for the Willamette River watershed (2001 NLCD mi2)
HUC8
watershed
17090001
17090002
17090003
17090004
17090005
17090006
17090007
17090008
17090009
17090010
17090011
17090012
Total
Open
water
26.4
4.0
13.4
7.2
8.7
8.6
11.7
0.8
1.8
1.9
5.3
6.4
96.0
Snow/
Ice
0.0
0.0
0.0
2.5
1.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
3.5
Developed9
Open
space
8.2
14.2
79.3
10.0
9.2
10.0
34.2
26.7
18.4
30.1
15.4
11.5
267.1
Low
density
2.5
5.1
76.6
3.0
4.7
5.0
63.1
17.3
23.1
69.3
11.0
41.0
321.7
Medium
density
0.9
2.1
34.4
1.5
1.0
1.3
28.5
4.3
9.4
38.5
3.8
33.7
159.2
High
density
0.2
0.8
14.1
0.3
0.3
0.3
9.9
1.6
3.0
11.2
1.7
15.4
58.8
Barren
land
12.8
1.1
16.7
41.1
8.5
2.9
2.5
7.4
1.7
8.1
1.1
0.4
104.3
Forest
1,111.6
464.6
615.0
1,059.3
558.0
673.6
112.1
333.2
366.1
261.1
716.8
26.8
6,298.4
Shru bland
174.2
113.6
214.7
177.0
93.1
183.6
32.9
93.3
99.1
79.3
118.2
1.6
1,380.6
Pasture/Hay
16.6
42.7
519.7
15.6
38.3
105.9
215.3
134.4
190.5
79.2
42.2
5.9
1,406.3
Cultivated
10.7
10.4
220.9
13.8
33.5
42.8
168.5
132.2
138.2
114.6
25.0
3.7
914.1
Wetland
2.9
8.5
70.6
3.6
8.4
6.0
33.3
21.2
22.7
16.2
3.6
2.0
199.1
Total
1,367.0
667.2
1,875.5
1,334.8
764.4
1,040.0
712.0
772.3
874.0
709.6
944.0
148.4
11,209.1
The percent imperviousness applied to each of the developed land uses is as follows: open space (9.56%), low density (32.31%), medium density (61.49%), and high
density (88.94%).
H-12
-------
The HSPF model is set up on a hydrologic response unit (HRU) basis. For HSPF, HRUs were formed from an
intersection of land use and hydrologic soil group, and then further subdivided by precipitation gage. Average
slopes (which tend to correlate with soils) were calculated for each HRU. The water land use area was adjusted to
prevent double counting with area described in HSPF reaches. SWAT HRUs are formed from an intersection of
land use and SSURGO major soils.
Point Sources
Facilities permitted under the National Pollutant Discharge Elimination System (NPDES) are, by definition,
considered point sources. It was assumed that minor dischargers (below 1.0 MOD) were insignificant, therefore,
not included in the model setup and simulation. Data were sought from the PCS database for the major
dischargers in the Willamette River watershed (Table 3 and Figure 3). Facilities that were missing total nitrogen,
total phosphorus, or total suspended solids (TSS) concentrations were filled with atypical pollutant concentration
value from literature based on SIC classification (Tetra Tech 1999). Constant point source flows and
concentrations were assumed for each major discharge facility in the watershed for the entire simulation period.
During the water quality calibration it was noticed that assumptions used for total phosphorus, at some facilities,
were too high. An investigation into the point sources that had assumed values for total phosphorus was
conducted. A new assumed value was supplied for these facilities. The modifications made to the total
phosphorus values are described in the "Changes to the Base Data" section of this report. The new assumed value
was also applied to the SWAT simulation. Both the HSPF and SWAT models used the same flows and
concentrations for each of the major point sources included in the simulations for the Willamette River watershed.
H-13
-------
Table 3. Major point source discharges in the Willamette River watershed
NPDES ID
OR0026891
OR0026140
OR0026221
OR0030589
OR0031259
OR00005663
OR00007873
OR0020214
OR0028118D
OR0029777C
OR0023345a
OR00201680
OR0020001
OR0022764
OR0032352
OR00005583
OR0026409
OR0034002
OR0020737
OR0020818
OR0028801
OR0001112
OR0000442
OR0020427
OR0020346
OR0026361
OR0001716
OR00334053
OR00010743
OR00005153
OR0031224
OR0020559
OR0020656
Name
PORTLAND,
OAK LODGE
CLACKAMAS
SILTRONIC
TRI-CITY
BLUE HERO
WEST LINN
CAN BY, Cl
CLEAN WAT
CLEAN WAT
CLEAN WAT
CLEAN WAT
WOODBURN,
WILSONVIL
NEWBERG,
VIRGINIA
SALEM, Cl
MCMINNVIL
DALLAS, C
LEBANON,
ALBANY, C
TOY INDUS
WEYERHAEU
STAYTON,
SWEET HOM
CORVALLIS
OREGON ME
FORT JAME
CASCADE P
WEYERHAEU
METROPOLI
COTTAGE G
SILVERTON
Design flow
(MGD)
8.30
4.00
2.66
—
0.00
—
—
2.00
22.60
39.00
7.50
5.00
3.33
2.25
4.00
—
35.00
5.60
2.00
3.00
8.70
—
—
1.90
1.38
9.70
—
—
—
—
49.00
1.20
2.50
Observed flow
(MGD)
(1991-2006 average)
15.42
6.29
19.82
0.88
17.85
—
—
1.31
22.60
25.00
5.02
3.74
4.09
19.03
48.85
—
51.60
68.53
5.47
3.93
7.28
2.50
—
3.47
3.69
13.57
0.73
—
—
—
36.77
2.78
2.84
Paper/pulp mills; discharge was ignored as their withdrawal and discharge are about the same
bDue to the upgrading of the treatment plant, total phosphorus concentration in the effluent value
to 1992 and 0.07 mg/L for 1992 and onward.
cDue to the upgrading of the treatment plant, total phosphorus concentration in the effluent value
to 1992 and 0.07 mg/L for 1992 and onward.
dDoes not discharge to the river in summer
considered is 3.6 mg/L prior
considered is 2.1 mg/L prior
H-14
-------
EVERGREEN
PORTLAND
BOISE WHITE PAPER, LLC.
SALMON CR'STP
/ PORTSIDE LAGOON
VANCOUVER^^AND LANDFILL
SILTRONIC
CORP.
CLEAN WATER
MAR NE PARK WATER RECL.
GEORGIA-PACIFIC
-TRTDUTDALE
CKAMAS OAK LODGE
TR -C TYvSERV. D SIR CT
BLUE-HER^ON PAPER CO.
WILSONVILLE
SALEM
DALLAS
CANBY
WOODBURN,
WEYERHAEUSER
TDYINDUST.
SILVERTON
STAYTON
/ALBANY
REGON METAIIS
FORT JAMES
OPERATING CO.
CASCADE-PACIFIC
PULP
METROPOLITAN
WASTEWAIER
WEYERHAEUSER
COTTAGE GROVE
Legend
Point Sources
Hydrography
Interstate
Water (Nat. Atlas Dataset)
US Census Populated Places
Municipalities (pop > 50,000)
U County Boundaries
l~~l Watershed with HUCSs
GCRP Model Areas - Willamette River Basin
Major Point Sources
NAD_1983_Albers_meters
Map produced 2-8-2010 - B. Tucker
Figure 3. Major point sources in the Willamette River watershed.
H-15
-------
Meteorological Data
The required meteorological data series for the 20 Watershed study are precipitation, air temperature, and
potential evapotranspiration. The 20 Watershed model does not include water temperature or algal simulation and
uses a degree-day method for snowmelt. These are drawn from the BASINS4 Meteorological Database (USEPA
2008), which provides a consistent, quality-assured set of nationwide data with gaps filled and records
disaggregated. Scenario application will require simulation over 30 years, so the available stations are those with
a common 30-year period of record (or one that can be filled from an approximately co-located station) that
covers the year 2001. A total of 40 precipitation stations were identified for use in the Willamette River watershed
model with a common period of record (Table 4 and Figure 4). Temperature records are sparser; where these are
absent, temperature is taken from nearby stations with an elevation correction. For each weather station, Penman-
Monteith reference evapotranspiration was calculated for use in HSPF using observed precipitation and
temperature coupled with SWAT weather generator estimates of solar radiation, wind movement, cloud cover,
and relative humidity.
For the 20 Watershed model applications, SWAT uses daily meteorological data, while HSPF requires hourly
data. It is important to note that a majority of the meteorological stations available for the Willamette River
watershed are Cooperative Summary of the Day stations that do not report sub-daily data. The BASINS4 dataset
already has versions of the daily data that have been disaggregated to an hourly time step using template stations.
For each daily station, this disaggregation was undertaken in reference to a single disaggregation template.
Occasionally, this automated procedure provides undesirable results, particularly when the total rainfall for the
day is very different between the subject station and the disaggregation template. This yields a small number of
hourly precipitation intensity estimates that are unrealistically high (e.g., much greater than the 100-yr 1-hour
event for the region). This has only a small impact on the watershed-scale hydrologic calibration as gages are
influenced by rainfall from multiple weather stations, but can introduce significant problems for the prediction of
erosion and sediment loads.
Table 4. Precipitation stations for the Willamette River watershed model
COOP ID
350595
350652
351222
351433
351735
351862
351877
351902
351914
352112
352292
352345
352374
352493
352693
352709
352805
352997
353047
353705
353971
Name
OR350595
OR350652
OR351222
OR351433
OR351735
OR351862
OR351877
OR351902
OR351914
OR352112
OR352292
OR352345
OR352374
OR352493
OR352693
OR352709
OR352805
OR352997
OR353047
OR353705
OR353971
Latitude
45.4548
44.2868
45.6864
44.3981
45.1701
44.6333
44.5087
43.7178
44.1331
44.9464
44.7243
43.7078
43.7823
45.2743
45.2690
44.1279
44.8578
45.5244
44.4139
45.3122
44.3525
Longitude
-122.8200
-122.0380
-123.1910
-122.4850
-122.4330
-123.1890
-123.4580
-123.0570
-122.2500
-123.2910
-122.2540
-122.7390
-122.9630
-122.2010
-122.3180
-123.2200
-123.4300
-123.1030
-122.6720
-123.3510
-122.7840
Temperature
Yes
Yes
No
Yes
No
Yes
Yes
Yes
No
Yes
Yes
No
Yes
No
Yes
Yes
Yes
Yes
Yes
No
No
Elevation (ft)
269
2,152
157
860
679
226
591
830
384
289
1,220
1,217
820
925
449
354
420
180
551
755
610
H-16
-------
COOP ID
354606
354811
355050
355213
355221
355384
356151
356213
356334
356749
357127
357500
357631
357809
357823
358095
358466
359083
359372
Name
OR354606
OR354811
OR355050
OR355213
OR355221
OR355384
OR356151
OR356213
OR356334
OR356749
OR357127
OR357500
OR357631
OR357809
OR357823
OR358095
OR358466
OR359083
OR359372
Latitude
44.6254
44.1001
43.9145
44.1707
44.6125
45.2215
45.2818
43.7429
45.3553
45.5181
45.3037
44.9051
44.9469
44.8734
45.0051
44.7895
45.1250
44.5000
45.0832
Longitude
-122.7180
-122.6880
-122.7600
-122.8710
-121.9480
-123.1620
-122.7510
-122.4430
-122.6050
-122.6890
-122.9140
-123.0010
-122.5240
-122.6480
-122.7730
-122.8140
-122.0720
-122.8190
-123.4890
Temperature
Yes
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
Yes
No
No
Elevation (ft)
518
676
712
545
2,474
154
151
1,276
167
157
515
203
2,316
1,348
407
427
1,119
436
384
H-17
-------
Legend
Weather Stations
Interstate
Hydrography
• Water (Nat. Atlas Dataset)
^H US Census Populated Places
^H Municipalities (pop > 50,000)
^] County Boundaries
Watershed with HUCSs
Pa c i f i c
Ocean
OR352997
OR357127
OR353705
OR350595v Portland
OR352693
OR352493
OR357823
OR357500 I
U 'QR357631
A OR357809
OR358095
OR359083
OR353047
OR35521
OR354811
GCRP Model Areas - Willamette River Basin
Weather Stations
NAD_1983_Albers_meters
Map produced 11-17-2009- B. Tucker
Figure 4. Weatherstations for the Willamette River watershed model.
H-18
-------
Watershed Segmentation
The Willamette River watershed was divided into 75 subwatersheds for the purposes of modeling (Figure 5). The
initial calibration watershed (Tualatin HUC) is highlighted. Each of the subwatershed delineations represents
roughly a HUC 10 scale watershed. Each of the major reservoirs in the Willamette watershed was delineated so
that each dam outlet represents an individual watershed outlet. The delineations were done this way to ensure that
any individual lake was contained in one watershed and that the watershed was only represented by one outlet.
The Willamette 20 Watershed model is set for the complete Willamette watershed without any inflow from
outside and thus does not require specification of any boundary conditions for application.
H-19
-------
Legend
A USGS Gages
— Hydrography
= Interstate
^H Water (Nat. Atlas Dataset)
^B US Census Populated Places
^H Municipalities (pop > 50,000)
| | County Boundaries
] Model Subbasins
| | Initial Calibration Watershed
T Minnesota River Basin
GCRP Model Areas - Willamette River Basin
Model Segmentation
NAD_1983_Albers_meters
Map produced 2-11-2010 - B. Tucker
15 30
TETRATECH
Figure 5. Model segmentation and USGS stations utilized for the Willamette River watershed.
Note: SWAT subwatershed numbering is shown; the HSPF model for this watershed uses the same
subwatershed boundaries with an alternative internal numbering scheme.
H-20
-------
Calibration Data and Locations
Each of the twelve HUCSs in the watershed was considered as a candidate representative subwatershed for
calibration. The objective was to find a HUC8 that resembled the overall characteristics of the Willamette
watershed with respect to land use, precipitation and terrain. The USGS gages on the Tualatin River at West Linn,
OR (USGS 14207500) and the Pudding River at Aurora, OR (USGS 14202000) were chosen as the primary
hydrology and water quality calibration locations. Additional tributary hydrology calibration was performed on
the South Yamhill River at McMinnville, OR (USGS 14194150) and the Mohawk (McKenzie) River near
Springfield (USGS 14165000). A hydrology calibration check was performed at the USGS gage on the
Willamette River at Salem, OR (USGS 14191000), which is the most downstream gage in the watershed that does
not include tidal effects. At this location, 43 percent of the tributary area is controlled by the major dams.
Therefore, calibration at this location would have been of limited use in developing model parameters. Table 5
presents the calibration and validation locations chosen for the Willamette River watershed.
Table 5.
Calibration and validation locations in the Willamette River watershed
Station name
Willamette River at Salem, Oregon
Pudding River at Aurora, Oregon
Tualatin River at West Linn, Oregon
Mohawk River near Springfield, Oregon
South Yamhill River at McMinnville
USGS ID
14191000
14202000
14207500
14165000
14194150
Drainage area
(mi2)
7280
479
706
177
528
Hydrology
calibration
X
X
X
X
X
Water quality
calibration
X
X
The model calibration period varied based on the availability of data. In general a calibration period of water
years 1996 through 2005 was used and a period from water years 1986 through 1995 was used for validation.
Water quality data were very limited and the period of coverage was not consistent between the two gages used
for water quality calibration in this study. A calibration period of water years 1996 to 2002 and validation period
of water years 1993 to 1995 were used for the Pudding gage; whereas, a calibration period of water years 1994-
1995 and validation period of water years 1986 to 1993 were used for the Tualatin gage.
H-21
-------
Flows in the lower Willamette River watershed are dominated by the effects of 13 reservoirs and their associated
dams operated by the U.S. Army Corps of Engineers for water supply, flood control, and navigation. These
reservoirs control much of the runoff from the southern and eastern mountainous portions of the watershed where
precipitation and snow fall are highest. Incorporation of the reservoirs in the model was a significant part of the
model development effort. For the calibration model, the reservoirs were not modeled; the flow at the nearest
USGS gage downstream of each reservoir was used as a boundary flow. One main stem gage was used as a
hydrology calibration check; however, the flow at this gage was significantly affected by the boundary flows,
particularly in the summer months when the boundary flows resulted in a significant overprediction of the
observed flow. For the scenario model, the boundary conditions were removed, and the reservoirs were modeled
by replacing the default HSPF FTABLEs with more realistic estimates of volume surface area, and spillway
outflow rate. U.S. Army Corps of Engineer data were analyzed to develop seasonal storages and minimum flow
time series which are specified as inputs to the model. During the simulation, each reservoir receives inflows from
upstream areas, and the program computes outflows consisting of the minimum releases plus any water necessary
to maintain the storage at or below the seasonal target storage.
Initial hydrologic parameterization for the Willamette calibration focus area came from the King County,
Washington HSPF Modeling (Green River Water Quality Assessment and Sammamish-Washington Analysis and
Modeling Program) (Bicknell et al. 2005). The King County hydrologic models have been under development for
many years, and under the Green River/Sammamish-Washington project were extended to most of the watershed.
Calibrated parameters from the Tualatin River (USGS 14207500) and South Yamhill River (USGS 14194150)
were applied to the eastern portions of the study area, while calibration adjustments for the Pudding River (USGS
14202000) were applied to the upper and western sections of the study area. Parameters from the Mohawk River
(USGS 14165000) were applied to the southern and southwestern portions of the study area.
Once the hydrology calibration was complete for the entire Willamette watershed, the focus turned to sediment
and water quality representation. The starting water quality parameters were again taken from the King
County/Seattle HSPF models.
Changes to Data Provided
No changes were made to the meteorological or land use base data. However, one of the rainfall stations
(OR357631) was not used because of unrealistic rainfall in 1996 that significantly skewed the water quality
calibration results in the Pudding River subwatershed. A number of changes were made to the point sources. The
flow from the six paper mills was changed to zero, since they draw water from the same rivers that they discharge
to. These point sources are: OR0000442, OR0000515, OR0000558, OR0000566, OR0000787, and OR0001074.
Also, several other point sources discharge to the calibration watersheds. Some of the parameters were determined
to be erroneous, and since they caused obvious problems in the calibration in the Tualatin River and Pudding
River sub-watersheds, were modified based on information obtained from other sources. Three point sources
(OR0020001, OR0020168, and OR0023345) do not discharge to surface waters during the summer, so these three
were modified to turn off the discharges between June and September. The discharges of three point sources
(OR0028118, OR0029777, and OR0034002) were found to be overestimated by using the "Observed Flow", and
were changed to the "Design Flow". The total phosphorus concentrations of two point sources were reduced as a
result of the use of advanced treatment methods; these are OR0020168 and OR0029777. Summer loads of total
phosphorus were substantially overpredicted with the higher values.
H-22
-------
Assumptions
Reservoirs
There are 13 dams and 11 major reservoirs in the study area. Figure 6 shows the locations of the dams and the
reservoirs in the watershed and Table 6 presents the 11 reservoirs that were included in the HSPF model. Two of
the dams (Big Cliff and Dexter) are re-regulation dams that allow the Corps to adjust the downstream flow more
smoothly than the releases from the upstream reservoir. The primary tributary calibration sites were chosen in
order to avoid effects of these dams. The main stem calibration site on the Willamette at Salem, OR is affected by
all of the major dams, so it was only used to check the calibration. The model used for calibration was modified
from the original model to include specification of boundary inflows at the USGS gage downstream from each
reservoir that provides flow to the Willamette main stem. The final model used for climate scenarios was
modified by improving the hydraulic representation of each reservoir, and including a simplified representation of
reservoir operation. Fortunately, all of the major reservoirs are operated by the U.S. Army Corps of Engineers,
and they are operated in a relatively consistent manner. Seasonally varying target storages and minimum releases
were programmed into the model using input time series. HSPF computes the reservoir outflows as the sum of the
minimum releases and sufficient water to maintain the actual storage at or below the target storage. While one
would assume the reservoirs influence the flow and water quality exiting the Willamette River at the outlet, for
this model the impacts of these reservoirs are assumed to be implicitly represented through the modified
FTABLES and the simplified operations, which should be applicable under future conditions.
Withdrawals
Because nobody knows what water withdrawals, by municipal and industrial facilities, will look like in the future
they were not included in the 20 Watershed model application.
Irrigation
Irrigation is not being explicitly modeled in the Willamette River watershed.
Snow Simulation
The Willamette HPSF model includes snow simulation using the degree-day method for snowmelt. It is modeled
in the subwatersheds that have a large area at high elevations, generally above 2,500 feet. The parameter values
were extracted from other applications, and minor adjustments were made to ensure that the snow depths and
duration were reasonable. No further calibration was performed for snow.
H-23
-------
US Army Corps
of Engineers -
Portland District
Figure 6. Dams and reservoirs in the Willamette River watershed (Source: USAGE 2009).
H-24
-------
Table 6. Reservoirs represented in the Willamette River watershed model
Dam Name
Fall Creek
Dorena
Lookout Point
Green Peter
Foster
Hills Creek
Detroit
Cottage Grove
Cougar
Blue River
Fern Ridge
Other Name
Fall Creek Lake
Dorena Lake
Lookout Point Lake
Green Peter
Foster Lake
Hills Creek Lake
Detroit Lake
Cottage Grove Lake
Cougar Lake
Blue River Lake
Fern Ridge Lake
River
Fall Creek
Row River
Middle Fork-Willamette
River
Middle Santiam River
South Santiam River
Middle Fork-Willamette
River
North Santiam River
Coast Fork-Willamette
River
South Fork-McKenzie
River
Blue River-McKenzie
River
Long Tom River
Owner
USAGE
USAGE
USAGE
USAGE
USAGE
USAGE
USAGE
USAGE
USAGE
USAGE
USAGE
Hydrology Calibration
As mentioned above, the starting parameters for this Willamette River HSPF model came from the King County,
Washington HSPF models. After the starting parameters were inserted into the model input files, average annual
potential evapotranspiration values were computed and compared to published values. Through this process it was
determined the input potential evapotranspiration time series should be reduced by multipliers, since the
computation of these time series produced more PET on an average annual basis than the published values
indicate. The default multipliers used for PET were 0.80; however, some of the multipliers were adjusted slightly
during the hydrology calibration. Calibration adjustments focused on the following parameters:
• LZSN (lower zone nominal storage): LZSN was reduced to shift flows to the wet period and reduce them
in the summer. It was also used to increase total runoff.
• INFILT (index to mean soil infiltration rate): Infiltration was generally decreased from the initial values
to increase storm peaks and reduce low flows.
• DEEPFR (fraction of groundwater inflow that will enter deep groundwater): small values of DEEPFR
were used to attempt to reduce low flows and to reduce total flow volume.
• BASETP (ET by riparian vegetation): Slightly increasing the BASETP value provided some ET by
riparian vegetation and improved the simulation of low flows.
• LZETP (lower zone E-T parameter): LZETP was generally increased to reduce flow, particularly the low
flows, and to reduce total volumes.
• AGWRC (Groundwater recession rate)
Initial calibration was performed at the USGS gage on the Tualatin River at West Linn, OR (USGS 14207500),
and is summarized in Figures 7 through 13 and Tables 7 and 8. The model fit is of good quality overall, but
simulates slightly high on the storm flows as is indicated by the Error in Storm Volumes metric. The model
calibration period was set to the 10 water years from 10/01/1995 to 09/30/2005.
H-25
-------
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1995 to 9/30/2005 )
Avg Modeled Flow (Same Period)
40000
« 25000
o
o
0
F
It
J*^\
o
- 2
4 £
- 6 -2
8 2
10
- 12
Oct-95 Apr-97 Oct-98 Apr-00 Oct-01 Apr-03 Oct-04
Date
Figure 7. Mean daily flow at USGS 14207500 Tualatin River at West Linn, OR - calibration period
(HSPF).
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1995 to 9/30/2005 ;
Avg Modeled Flow (Same Period)
O-95
A-97
O-98 A-00 O-01
Month
A-03
O-04
Figure 8. Mean monthly flow at USGS 14207500 Tualatin River at West Linn, OR - calibration period
(HSPF).
H-26
-------
Avg Flow (10/1/1995 to 9/30/2005 )
•Line of Equal Value
Best-Fit Line
t
o
•a
_CD
a>
•a
CD
O)
co
CD
10000
y = 0.8952x+ 111.22
5000 -
•a
o
co
.a
O
CD
o
c
_co
to
00
t_
CD
100% -,
90% -
80% -
70%
60% -
50% -
40% -
30% -
20%
10%
0%
Avg Observed Flow (10/1/1995 to 9/30/2005 )
I Avg Modeled Flow (10/1/1995 to 9/30/2005 )
-Line of Equal Value
5000 10000
Average Observed Flow (cfs)
O-95 A-97 O-98
A-00 O-01
Month
A-03 O-04
Figure 9. Monthly flow regression and temporal variation at USGS 14207500 Tualatin River at West
Linn, OR - calibration period (HSPF).
Avg Flow (10/1 /1995 to 9/30/2005)
• Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1995 to 9/30/2005)
Avg Modeled Flow (Same Period)
5000
5000
4000
3000 -
t
I
Hr 2000 - --
1000 2000 3000 4000
Average Observed Flow (cfs)
5000
1000 - --
8 9
Figure 10. Seasonal regression and temporal aggregate at USGS 14207500 Tualatin River at West
Linn, OR - calibration period (HSPF).
cc
OH
H-27
-------
Average Monthly Rainfall (in)
•Median Observed Flow (10/1/1995 to 9/30/2005)
I Observed (25th, 75th)
Modeled (Median, 25th, 75th)
7000 -i
6000
5000
^ 4000
o 3000
LJ_
2000
1000
:'
o
10 11 12 1
3 4
Month
6 7
0
1
2
3
4
5
6
- 7
8
9
10
Figure 11. Seasonal medians and ranges at USGS 14207500 Tualatin River at West Linn, OR •
calibration period (HSPF).
to
or
Table 7. Seasonal summary at USGS 14207500 Tualatin River at West Linn, OR -
calibration period (HSPF)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
399.00
1305.46
3817.72
4175.40
4193.04
2763.99
1503.13
929.54
434.30
236.99
215.16
258.55
309.00
534.00
3420.00
3675.00
3440.00
2340.00
1165.00
681.00
385.50
227.00
197.50
232.00
234.00
305.75
1565.00
2297.50
1955.00
1155.00
855.00
497.25
295.00
187.00
171.25
197.75
436.75
1460.00
5747.50
5340.00
5485.00
3962.50
1760.00
1140.00
485.00
275.50
233.75
286.25
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
454.72
1323.58
3432.50
3905.86
4018.44
2572.21
1551.10
984.03
529.68
251.49
207.51
215.52
329.52
697.37
2985.78
3369.21
2892.20
2415.34
1296.01
762.86
479.68
222.59
170.68
156.46
193.65
389.21
1690.84
2197.11
1907.41
1238.50
858.61
557.19
342.37
182.91
149.20
135.63
592.38
1674.78
4827.79
5211.28
4798.64
3501.83
1860.24
1243.80
655.71
294.57
208.77
203.93
H-28
-------
•Observed Flow Duration (10/1/1995 to 9/30/2005 )
Modeled Flow Duration (10/1/1995 to 9/30/2005 )
100000
o
O)
(0
10000
1000
100
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Percent of Time that Flow is Equaled or Exceeded
Figure 12. Flow exceedence at USGS 14207500 Tualatin River at West Linn, OR - calibration period
(HSPF).
^—Observed Flow Volume (10/1 /1995 to 9/30/2005 )
^—Modeled Flow Volume (10/1/1995 to 9/30/2005 )
o
o
en
(0
.0
O
0)
E
_2
o
o
T3
0)
N
15
120%
100% -
80% -
60% -
40% -
20% -
Oct-95
Apr-97
Oct-98
Apr-00
Oct-01
Apr-03
Oct-04
Figure 13. Flow accumulation at USGS 14207500 Tualatin River at West Linn, OR - calibration period
(HSPF).
H-29
-------
Table 8. Summary statistics at USGS 14207500 Tualatin River at West Linn, OR
calibration period (HSPF)
HSPF Simulated Flow
REACH OUTFLOW FROM DSN 14
10-Year Analysis F^riod: 10/1/1995 - 9/30/2005
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
_Nash;Sjjteliflej:;pj5f^^
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
30.99
13.03
3.06
1.09
8.44
16.57
4.89
7.59
0.20
Error Statistics
-3.92
7.28
-6.64
-4.95
Observed Flow Gage
USGS 14207500 TUALATIN RIVERAT WEST LINN, OR
Hydrologic Unit Code: 17090010
Latitude: 45.35067559
Longitude: -122.6762044
Drainage Area (sq-rri): 706
Total Obsei-ved In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring Flow VolumeJ4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
-5.69 » | 30
-5.76
6.89
22.92
27.87
0.799
0.731
0.965
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
32.25
13.96
2.85
1.15
8.95
17.58
4.58
6.18
0.16
Clear [
H-30
-------
Hydrology Validation
Validation for the Willamette River watershed calibration focus area was performed at the same location (Tualatin
River) but for water years 10/01/1985 to 09/30/1995. Results are presented in Figures 14 through 20 and Tables 9
and 10. Similar to the calibration years, the validation years' model fit is of good quality, although the validation
shows oversimulation of low flows and summer seasonal flows, and undersimulation of the 10 percent highest
flows. The rest of the metrics fall within the acceptable range set for the 20 Watershed study.
lAvg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1985 to 9/30/1995 ;
Avg Modeled Flow (Same Period)
12000
t3
10000 -
8000 -
iyr
-
0
Oct-85 Apr-87
Oct-88 Apr-90 Oct-91 Apr-93
Date
Oct-94
Figure 14. Mean daily flow at USGS 14207500 Tualatin River at West Linn, OR - validation period
(HSPF).
• Avg Monthly Rainfall (in)
-»-Avg Observed Flow (10/1/1985 to 9/30/1995 ;
Avg Modeled Flow (Same Period)
8000
i
o
6000 -
4000 -
2000 -
O-85
A-87
O-88
A-90 O-91
Month
A-93
O-94
Figure 15. Mean monthly flow at USGS 14207500 Tualatin River at West Linn, OR - validation period
(HSPF).
H-31
-------
Avg Flow (10/1/1985 to 9/30/1995 )
•Line of Equal Value
Best-Fit Line
8000
6000 -\
"O
o
<
H-32
-------
Average Monthly Rainfall (in)
•Median Observed Flow (10/1/1985 to 9/30/1995)
(Observed (25th, 75th)
Modeled (Median, 25th, 75th)
5000
10 11 12 1
Figure 18. Seasonal medians and ranges at USGS 14207500 Tualatin River at West Linn, OR -
validation period (HSPF).
Table 9. Seasonal summary at USGS 14207500 Tualatin River at West Linn - validation
period (HSPF)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
178.35
663.27
1762.91
2892.62
3115.09
2472.13
1520.16
618.69
313.98
173.94
131.78
140.05
132.00
369.00
1345.00
2725.00
2670.00
2220.00
1035.00
505.50
242.00
164.50
126.50
126.50
100.00
198.00
786.50
1460.00
1262.50
1380.00
731.25
343.75
182.75
127.25
102.50
107.75
193.50
792.00
2382.50
3985.00
4737.50
3377.50
1855.00
788.25
382.50
211.75
158.00
162.25
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
257.87
742.01
1625.28
2350.73
2488.82
1980.44
1363.78
733.54
468.21
244.29
166.56
173.33
148.30
539.33
1334.70
2010.52
1947.93
1691.04
1124.25
644.97
380.64
210.25
150.77
137.78
134.63
271.12
897.16
1285.46
1325.47
1233.37
798.68
488.42
293.42
169.95
131.70
127.92
250.42
940.92
2002.84
2745.93
3171.78
2442.64
1542.90
903.26
555.62
296.58
182.07
167.28
H-33
-------
•s
I
O)
CO
'CD
Q
•Observed Flow Duration (10/1/1985 to 9/30/1995 )
Modeled Flow Duration (10/1/1985 to 9/30/1995 )
100000
10000
1000
100
10%
20%
30%
40%
50%
60%
70%
80%
Percent of Time that Flow is Equaled or Exceeded
90%
100%
Figure 19. Flow exceedence at USGS 14207500 Tualatin River at West Linn, OR - validation period
(HSPF).
o
o
CD
T3
o
_o
LJ_
T3
N
"CD
•Observed Flow Volume (10/1/1985 to 9/30/1995 )
Modeled Flow Volume (10/1/1985 to 9/30/1995 )
120%
100%
80%
60%
40%
20%
Oct-85
Apr-87
Oct-88
Apr-90
Oct-91
Apr-93
Oct-94
Figure 20. Flow accumulation at USGS 14207500 Tualatin River at West Linn, OR - validation period
(HSPF).
H-34
-------
Table 10. Summary statistics at USGS 14207500 Tualatin River at West Linn, OR
validation period (HSPF)
HSPF Simulated Flow
REACH OUTFLOW FROM DSN 14
10-Year Analysis F^riod: 10/1/1985 - 9/30/1995
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
_NashJ5uteliffe^pj5ffi^^
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
20.06
7.44
2.53
0.94
4.25
10.77
4.09
4.52
0.14
Error Statistics
-9.80
38.11
-20.60
31.12
Observed Flow Gage
USGS 14207500 TUALATIN RIVER AT WEST LINN, OR
Hydrologic Unit Code: 17090010
Latitude: 45.35067559
Longitude: -122.6762044
Drainage Area (sq-rri): 706
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Obsei-ved Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
0.70 » | 30
-19.56
4.71
-1.77
22.38
0.811
0.702
0.922
30
30
20
50
Model accuracy increases
as E or E' approaches 1 .0
22.24
9.37
1.83
0.72
4.22
13.39
3.91
4.60
0.12
Clear [
H-35
-------
Hydrology Results for Larger Watershed
Since the Tualatin River calibration location represents only a small portion of the drainage area for this project,
results near the outlet of the entire watershed were examined at the Willamette River at Salem, OR (USGS
14191000). This gage is downstream of the large reservoirs in the Willamette River and more than 40 percent of
the area at Salem is controlled by dams. Results are presented in Figures 21 through 27 and Tables 11 and 12. The
results at the Salem gage look fairly good as well, but are being strongly determined by the input boundary
inflows at the dams, particularly during the summer. The simulated output is quite high during the summer, which
is manifested in overprediction of the metrics for 50 percent lowest flows, seasonal summer volume, and summer
storm volumes. Summer storms are small in this region, and the summer storm volumes are also small; therefore,
an error of 0.6 inches produces a large percent difference. The overall storm volumes are overpredicted, which
results in exceedance of the metric by a small amount. The remainder of the metrics fall within the acceptable
range set for the 20 Watershed study including a daily Nash-Sutcliffe of 0.88 at the Salem gage. Tables 13 and 14
show a summary of the hydrology calibration and validation results for all five locations. In general, the
hydrology calibration results on the tributaries were quite good, largely as a result of the calibration efforts at each
station. The calibrated parameters were transferred from the tributaries to other non-calibrated portions of the
watershed based on location. Since the Salem mainstem gage is so heavily influenced by the reservoirs (as
described above), the results at that location are reasonable, but not very useful in concluding that the calibrated
parameters are transferrable to the entire watershed.
lAvg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1995 to 9/30/2005
Avg Modeled Flow (Same Period)
250000
^ 100000 --
Oct-95 Apr-97 Oct-98 Apr-00 Oct-01 Apr-03 Oct-04
14
Figure 21. Mean daily flow at USGS 14191000 Willamette River at Salem, OR- calibration period
(HSPF).
H-36
-------
o
150000
100000
50000 -
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1995 to 9/30/2005
Avg Modeled Flow (Same Period)
co
or
10 -^1
£=
O
0-95
A-97 0-98 A-00 O-01
Month
A-03
0-04
Figure 22. Mean monthly flow at USGS 14191000 Willamette River at Salem, OR- calibration period
(HSPF).
Avg Flow (10/1/1995 to 9/30/2005 )
•Line of Equal Value
Best-Fit Line
Avg Observed Flow (10/1/1995 to 9/30/2005 )
Avg Modeled Flow (10/1/1995 to 9/30/2005 )
IOUUUU -
I
g
o
U- 1 00000 -
•a
CD
CD
•a
o
^
^ 50000 -
CD
O)
CO
t_
CD
0 -
— 7
y = 0.8129x + 5024.1
R2 = 0.9535 X
I ii*
" ,T~
•J^l
*"J^
*Jff
jj&t*
^
O
5
(/5
.Q
O
CD
O
TO
"co
00
CD
-1—'
1
0 50000 100000 150000
Average Observed Flow (cfs)
1 UU /O ~|
90% -
80%
70% -
60%
cr\f\l
40% -
30% -
20% \
10% J
O-95 A-97 O-98 A-00 O-01 A-03 O-04
Month
Figure 23. Monthly flow regression and temporal variation at USGS 14191000 Willamette River at
Salem, OR - calibration period (HSPF).
H-37
-------
Avg Flow (10/1/1995 to 9/30/2005)
• Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1995 to 9/30/2005)
Avg Modeled Flow (Same Period)
60000
t
40000 --
T3
O
20000 --
D)
ro
y = 0.7483x +6556.6
R2 = 0.9587
60000
40000
20000 -
20000 40000 60000
Average Observed Flow (cfs)
10 11 12 1 23456789
Month
Figure 24. Seasonal regression and temporal aggregate at USGS 14191000 Willamette River at Salem,
OR - calibration period (HSPF).
Average Monthly Rainfall (in)
-Median Observed Flow (10/1/1995 to 9/30/2005)
80000
70000
60000 -
50000 -
40000 -
E 30000 -
20000 -
10000 -
I Observed (25th, 75th)
Modeled (Median, 25th, 75th)
£
o^
o
0 -I
M
M
J
J
p~ ~r~ 'i~ ~r" 'i~ ~r" "i~
II
fr* [tt^Uti^iti
0
- 2
4 *
-6 |
8 t
o
10 S
12
10 11 12 1
345
Month
8 9
Figure 25. Seasonal medians and ranges at USGS 14191000 Willamette River at Salem, OR •
calibration period (HSPF).
H-38
-------
Table 11. Seasonal summary at USGS 14191000 Willamette River at Salem, OR-
calibration period (HSPF)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
12717.87
25500.23
50807.87
50369.00
36872.79
27664.39
23458.33
20733.55
12950.87
7675.19
7159.26
8761.37
11800.00
15500.00
41900.00
49350.00
28300.00
23950.00
20000.00
16950.00
12250.00
7430.00
7110.00
8450.00
8752.50
12375.00
22175.00
26325.00
19050.00
14950.00
16600.00
15100.00
9807.50
6780.00
6602.50
7210.00
14875.00
26525.00
74950.00
70900.00
43500.00
34500.00
24825.00
23675.00
14825.00
8227.50
7395.00
9847.50
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
17880.47
29822.48
47121.03
44675.55
30886.30
23206.99
21472.71
22704.18
15686.14
11749.44
12563.64
13918.00
15883.04
19065.62
40022.34
42552.42
23372.96
18325.25
16424.03
18536.53
14097.00
11749.37
12010.44
12760.73
12873.57
13575.24
20587.66
24029.88
12456.04
11562.67
13542.39
15626.60
11904.53
10807.98
11030.47
11221.73
20553.78
30711.68
67465.45
65317.06
38808.34
29087.93
25476.06
25378.76
18484.91
12544.74
12990.64
15139.36
•Observed Flow Duration (10/1/1995 to 9/30/2005 )
Modeled Flow Duration (10/1/1995 to 9/30/2005 )
1000000
100000
O)
ro
-^ 1 0000
'ro
Q
1000
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
Percent of Time that Flow is Equaled or Exceeded
100%
Figure 26. Flow exceedence at USGS 14191000 Willamette River at Salem, OR - calibration period
(HSPF).
H-39
-------
o
o
CO
T3
.Q
O
I
T3
0)
N
"CD
•Observed Flow Volume (10/1/1995 to 9/30/2005 )
Modeled Flow Volume (10/1/1995 to 9/30/2005 )
120%
100% -
80% -
60%
40%
20% -
Oct-95
Apr-97
Oct-98
Apr-00
Oct-01
Apr-03
Oct-04
Figure 27. Flow accumulation at USGS 14191000 Willamette River at Salem, OR- calibration period
(HSPF).
H-40
-------
Table 12. Summary statistics at USGS 14191000 Willamette River at Salem, OR
calibration period (HSPF)
HSPF Simulated Flow
REACH OUTFLOW FROM DSN 12
10-Year Analysis F^riod: 10/1/1995 - 9/30/2005
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulate^jTighe^MO%_flpws^__
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3^:
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% higjpest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutclifle Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
45.35
14.60
11.01
5.98
14.86
15.21
9.29
13.95
0.75
Error Statistics
2.58
25.19
-6.12
62.06
Observed Flow Gage
USGS 14191000 WILLAMETTE RIVER AT SALEM, OR
Hydrologic Unit Code: 17090007
Latitude: 44.9442863
Longitude: -123.0428742
Drainage Area (sq-rri): 7280
Total Observed In-stream Flow:
_j£t^l_of_p^s^r\«djTigjTe^tJ^%JlOT/s^__
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
6.42 » | 30
-13.98
4.82
21.60
364.63
0.879
0.662
0.932
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
Cle
44.21
15.55
8.79
3.69
13.97
17.69
8.86
11.48
0.16
fit'
H-41
-------
Table 13. Summary statistics (percent error) for all stations - calibration period (HSPF)
Station
Calibration Period:
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient of
Efficiency, E:
Monthly Nash-Sutcliffe Coefficient
of Efficiency, E:
14191000
Salem
WY 96-05
2.58
25.19
-6.12
62.06
6.42
-13.98
4.82
21.60
364.63
0.879
0.932
14207500
West Linn
WY 96-05
-3.92
7.28
-6.64
-4.95
-5.69
-5.76
6.89
22.92
27.87
0.799
0.965
14202000
Aurora
WY 03-05
6.08
-13.76
2.42
-2.94
0.36
7.69
8.69
-7.94
-45.66
0.912
0.970
14194150
McMinnville
WY 00-05
-0.74
-12.16
3.88
1.19
9.45
-4.63
-7.70
6.83
-41.57
0.711
0.947
14165000
Springfield
WY 99-05
-6.41
-11.20
2.78
9.35
0.24
-8.75
-11.07
33.01
-38.36
0.674
0.879
Table 14. Summary statistics (percent error) for all stations - validation period (HSPF)
Station
Calibration Period:
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient of
Efficiency, E:
Monthly Nash-Sutcliffe Coefficient
of Efficiency, E:
14191000
Salem
WY 86-95
5.04
27.56
-0.59
65.84
7.98
-14.25
7.81
24.68
298.14
0.835
0.880
14207500
West Linn
WY 86-95
-9.80
38.11
-20.60
31.12
0.70
-19.56
4.71
-1.77
22.38
0.811
0.922
14202000
Aurora
WY 94-97
7.48
12.54
8.10
38.31
2.36
10.21
6.68
0.34
3.37
0.886
0.973
14194150
McMinnville
WY 95-99
-4.52
-19.80
-4.74
1.81
-2.46
-5.28
-7.98
-3.10
-51.72
0.720
0.968
14165000
Springfield
WY 88-97
-4.70
-4.88
4.68
13.91
18.51
-14.93
-13.54
39.72
-39.06
0.467
0.710
H-42
-------
Water Quality Calibration and Validation
The 20 Watershed models are designed to provide water quality simulation for total suspended solids (TSS), total
nitrogen, and total phosphorus. TSS is simulated with the standard HSPF approach (USEPA 2006). In contrast to
sediment, total nitrogen and total phosphorus are simulated in this application in a simplistic fashion, as HSPF
general quality constituents (GQUALs) subject to an exponential decay rate during transport.
The water quality calibration focuses on the replication of monthly loads, as specified in the project QAPP. Given
the approach to water quality simulation in the 20 Watershed model, a close match to individual concentration
observations cannot be expected. Comparison to monthly loads presents challenges, as monthly loads are not
observed. Instead, monthly loads must be estimated from scattered concentration grab samples and continuous
flow records. As a result, the monthly load calibration is inevitably based on the comparison of two uncertain
numbers. Nonetheless, calibration is able to achieve a reasonable agreement. Further, the load comparisons were
supported by detailed examinations of the relationships of flows to loads and concentrations and the distribution
of concentration prediction errors versus flow, time, and season, as well as standard time series plots.
For application on a nationwide basis, the 20 Watershed protocols assume that sediment and total phosphorus
loads will likely exhibit a strong positive correlation to flow (and associated erosive processes), while total
nitrogen loads, which often have a dominant groundwater component, will not. Accordingly, TSS and total
phosphorus loads were estimated from observations using a flow-stratified log-log regression approach, while
total nitrogen loads were estimated using a flow-stratified averaging estimator, consistent with the findings of
Preston etal. (1989).
Water quality calibration was done on the Tualatin River at West Linn, OR, comparing model results to data from
USGS 14207500. Calibration and validation were performed for the period with available water quality data,
which was 1986-1995. The 1991-1995 time period was used for calibration, and the 1986-1990 period was used
for validation. TSS calibration was performed by adjusting the coefficients in the soil detachment (KRER) and
soil washoff (KSER) equations along with changes to the seasonal vegetation COVER. Results of the TSS
calibration are generally acceptable. The results are shown in Figures 28 through 31 and the statistics of TSS
loads and concentrations are shown in Tables 15 and 16, respectively. Visually, the model is roughly simulating
the trends contained in the observed data.
H-43
-------
100,000
10,000
o
E
1
I
1,000
TSS
• Regression Loads
•Simulated Loads
CDCDi^i^ooooo>a>
oooooooooooooooo
C"=1C"=1C"=1C"=1
(0-2(0-2(0-2(0.2
—>~—>~—>~—>~
CM CM CO CO
c
CO
c
CO
c
CO
c
CO
-3
c
CO
-3
c
CO
-3
Figure 28. Fit for monthly load of TSS at USGS 14207500 Tualatin River at West Linn, OR (HSPF).
Table 15. Model fit statistics (observed minus predicted) for monthly sediment loads
using stratified regression (HSPF)
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1991-1995)
3%
47%
13.3%
Validation period
(1986-1990)
5%
53%
15.9%
H-44
-------
Tualatin River at West Linn
1991-1995
10000
1000
ro
•o
ro
o
_i
CO
CO
10
0.1
10
100
1000
Flow, Cfs
10000
100000
Simulated i Observed
Power (Simulated)
Power (Observed)
Figure 29. Power plot for observed and simulated TSS at USGS 14207500 Tualatin River at West Linn,
OR - calibration period (HSPF).
Tualatin River at West Linn
1986-1990
10000
1000
re
g 100 J
ro
o
CO
CO
10 -
1
0.1
*J£di&E#
10
100
1000
Flow, cfs
10000
100000
• Simulated _ Observed
-Power (Simulated)
Power (Observed)
Figure 30. Power plot for observed and simulated TSS at USGS 14207500 Tualatin River at West Linn,
OR - validation period (HSPF).
H-45
-------
Tualatin River at West Linn, OR
• Simulated A Observed
1000
100
O)
eo
CO
1986 1987 1988 1989 1990 1991 1992
Year
Figure 31. Time series plot of TSS concentration at USGS 14207500 Tualatin River at West Linn, OR
(HSPF).
Table 16. Relative errors (observed minus predicted), TSS concentration, at USGS
14207500 Tualatin River at West Linn, OR (HSPF)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1991-1995)
35
-40.4%
-7.8%
Validation period
(1986-1990)
29
21 .9%
10.0%
The total phosphorus calibration performed well at the Tualatin River location. Adjustments were made to the
potency factors and the subsurface concentrations. In general, the observed and simulated total phosphorus loads
attain an acceptable match for the simulation period (Figure 32 and Table 17). As with TSS, additional
diagnostics for total phosphorus included flow-load power plots (Figures 33 and 34), a time series plot of
concentrations (Figure 35), and statistics (Table 18). All show acceptable agreement.
H-46
-------
o
E
In
I
100
Total P
-Regression Loads
-Simulated Loads
Figure 32. Fit for monthly load of total phosphorus at USGS 14207500 Tualatin River at West Linn, OR
(HSPF).
Table 17. Model fit statistics (observed minus predicted) for monthly total phosphorus
loads using stratified regression (HSPF)
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1991-1995)
-1%
31%
23.0%
Validation period
(1986-1990)
-9%
32%
21.8%
H-47
-------
10
•c
re
0.1
0.01
Tualatin River at West Linn
1991-1995
10
100
1000
Flow, cfs
10000
100000
Simulated
Observed
Power (Simulated)
Power (Observed)
Figure 33. Power plot for observed and simulated total phosphorus at USGS 14207500 Tualatin River
at West Linn, OR - calibration period (HSPF).
10
•c
re
0.1
0.(
10
Tualatin River at West Linn
1 986-1 990
100
1000
Flow, cfs
10000
100000
Simulated
- Observed
Power (Simulated)
Power (Observed)
Figure 34. Power plot for observed and simulated total phosphorus at USGS 14207500 Tualatin River
at West Linn, OR - validation period (HSPF).
H-48
-------
Tualatin River at West Linn, OR
1986 1987 1988 1989 1990
Year
1991
1992
Figure 35. Time series plot of total phosphorus concentration at USGS 14207500 Tualatin River at
West Linn, OR (HSPF).
Table 18. Relative errors (observed minus predicted), total phosphorus concentration at
USGS 14207500 Tualatin River at West Linn, OR (HSPF)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1991-1995)
35
-78.8%
-59.7%
Validation period
(1986-1990)
19
26.3%
26.5%
Nitrogen adjustments were made to the seasonally varying accumulation/washoff and subsurface concentrations.
Results for total nitrogen are summarized in Figures 36 through 39 and Tables 19 and 20. The results are
acceptable, and generally better than those for total phosphorus. This is because nitrogen is not sediment-
associated, therefore, problems with sediment are not reflected in the calibration for total nitrogen. A summary of
the water quality statistics at the two locations (Tualatin River and Pudding River) are shown in Table 21.
H-49
-------
Total N
2,000
-Avsraging Loads
-Simulated Loads
O5O5OOi-i-CMCMCOCO'^-'^-l£)l£)
opopopopopopopopopopopopopop
Figure 36. Fit for monthly load of total nitrogen at USGS 14207500 Tualatin River at West Linn, OR
(HSPF).
Table 19. Model fit statistics (observed minus predicted) for monthly total nitrogen loads
using averaging estimator (HSPF)
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1991-1995)
2%
21%
15.5%
Validation period
(1986-1990)
-6%
20%
17.0%
H-50
-------
Tualatin River at West Linn
1991-1995
100
I 10
U)
o
TJ
ro
o
O.I
10
100
1000
Flow, cfs
10000
100000
* Simulated - Observed
-Power (Simulated)
• Power (Observed)
Figure 37. Power plot for observed and simulated total nitrogen at USGS 14207500 Tualatin River at
West Linn, OR - calibration period (HSPF).
Tualatin River at West Linn
1986-1990
100
I 10 H
c
o
•d
o
0.1
10
100
1000
Flow, cfs
10000
100000
Simulated - Observed Power (Simulated)
•Power (Observed)
Figure 38. Power plot for observed and simulated total nitrogen at USGS 14207500 Tualatin River at
West Linn, OR - validation period (HSPF).
H-51
-------
Tualatin River at West Linn, OR 1986-1995
1986 1987 1988 1989 1990
Year
1991
1992
Figure 39. Time series plot of total nitrogen concentration at USGS 14207500 Tualatin River at West
Linn, OR (HSPF).
Table 20. Relative errors (observed minus predicted), total nitrogen concentration, USGS
14207500 Tualatin River at West Linn, OR (HSPF)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1991-1995)
35
-33.5%
-16.8%
Validation period
(1986-1990)
20
-25.6%
-19.2%
Water Quality Results for Larger Watershed
The Tualatin River water quality parameters were transferred to the Pudding River (Aurora) watershed, and
further calibration was necessary in the Pudding. A combination of the parameters sets from the Tualatin and
Pudding watersheds was transferred to the remaining portions of the watershed. Since there are no other water
quality data available in the Willamette watershed, it was not possible to determine whether the parameter set was
applicable to the entire watershed. However, the calibration at the two locations was fairly good with respect to
the loads (Table 21). As expected, the concentration errors are larger.
H-52
-------
Table 21. Summary statistics for water quality for all stations (observed minus predicted)
(HSPF)
Station
Relative Percent Error
TSS Load
TSS Concentration
Median Percent Error
Relative Percent Error
TP Load
TP Concentration
Median Percent Error
Relative Percent Error
TN Load
TN Concentration
Median Percent Error
14207500
West Linn
Calibration
3%
-7.8%
-1%
-59.7%
2%
-16.8%
14207500
West Linn
Validation
5%
10.0%
-9%
26.5%
-6%
-19.2%
14202000
Aurora
Calibration
1%
22.1%
4%
-18.2%
0%
16.0%
14202000
Aurora
Validation
20%
-10.3%
-28%
38.3%
11%
-12.2%
H-53
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
The USGS gages on the Tualatin River at West Linn, OR (USGS 14207500) and the Pudding River at Aurora,
OR (USGS 14202000) were used as the primary hydrology and water quality calibration locations. Additional
tributary hydrology calibration was performed on the South Yamhill River at McMinnville, OR (USGS
14194150) and the Mohawk (McKenzie) River near Springfield (USGS 14165000). A hydrology calibration
check was performed at the USGS gage on the Willamette River at Salem, OR (USGS 14191000), which is the
most downstream gage in the watershed that does not include tidal effects.
to
No changes were made to the input data provided for the SWAT model except for point sources. The flow from
the six paper mills was changed to zero, since they draw water from the same rivers that they discharge to. These
point sources are: OR0000442, OR0000515, OR0000558, OR0000566, OR0000787, and OR0001074. Several
other point sources discharge to the calibration watersheds. Some of the parameters were determined to be
erroneous, and since they caused obvious problems in the calibration in the Tualatin and Pudding subwatersheds,
were modified based on information obtained from other sources. Three point sources (OR0020001, OR0020168,
and OR0023345) do not discharge to surface waters during the summer, so these three were modified to turn off
the discharges between June and September. The discharges of three point sources (OR0028118, OR0029777, and
OR0034002) were found to be overestimated by using the "Observed Flow" and were changed to the "Design
Flow". The total phosphorus concentrations of two point sources were reduced as a result of the use of advanced
treatment methods; these are OR0020168 and OR0029777. Summer loads of total phosphorus were substantially
overpredicted with the higher values.
Fall Creek Lake, Dorena Lake, Lookout Point Lake, Green Peter Lake, Foster Lake, Hills Creek Lake, Detroit
Lake, Cottage Grove Lake, Cougar Lake, and Fern Ridge Lake (Table 6) were represented in the Willamette
River watershed SWAT model. Pertinent reservoir information including surface area and storage at principal
(normal) and emergency spillway levels for the reservoirs modeled were obtained from the National Inventory of
dams (NID) database (USAGE 1982). The SWAT model provides four options to simulate reservoir outflow: 1)
measured daily outflow, 2) measured monthly outflow, 3) average annual release rate for uncontrolled reservoir,
and 4) controlled outflow with target release. Keeping in view the 20 Watershed climate change impact evaluation
application, it was assumed that the best representation of the reservoirs was to simulate them without supplying
time series of outflow records. Therefore, a target release approach was used in the GCRP-SWAT model. The
number of days to reach target storage was assumed to be 50 days and an average release rate of 50 m3/s was
assumed for all lakes.
No withdrawals, either by municipal and industrial facilities, were included in the 20 Watershed model
application.
Irrigation was not explicitly modeled in the Willamette River watershed.
H-54
-------
Hydrology Calibration
The SWAT model setup for the Willamette River watershed 20 Watershed project was set up fresh, with no prior-
existing SWAT model for the watershed.
Most of the calibration efforts were geared toward getting a closer match between simulated and observed flows
at the outlet of the calibration focus area. Initially, the parameters set for this area were applied across the
watershed and the model performance was verified at other stations. This resulted in model performance that was
not the same as in the calibration focus area, mostly because of dominance of different land uses in different parts
of the watershed. In response to the variations in spatial characteristics of the subwatersheds, a systematic
adjustment of parameters, individually, by land use type was adopted and the same adjustment was applied
throughout the watershed.
It can be acknowledged that a hydrologic/water quality model can be precisely calibrated, given the degree of
freedom, resources, time, and data. Keeping in view the interests of this project, which are to study the land use
change and climate change impacts on flow and water quality, a site-specific calibration was deliberately not
attempted. To some extent, the limitation of this approach is that the local differences in soil, weather,
management, and hydrology is not thoroughly accounted for. This approach will provide an idea of the model
performance when it is not spatially-tightly calibrated and what to expect when transferring the parameters to
other ungaged watersheds or to watersheds where detailed modeling is not practical due to limited resources.
While adjusting the hydrology and water quality parameters for calibration, crop yields were also checked. The
crop yields for wheat, corn, and hay were found to be reasonably close to the reported yield values in the National
Agricultural Statistics Service (NASS) database.
Land Use/Soil/Slope Definition
A 5/10/5 percent threshold was used for land use/soil/slope in the SWAT model while defining the HRUs. The
cropland HRUs were simulated as 2-year winterwheat-corn-winterwheat rotation with every other year fallow
during summer. The hay HRUs were simulated as hay every year with the fourth year being fallow. The urban
(including current and future urban class types) classes were exempt from applying the thresholds.
The calibration focus area represents 3 subwatersheds, which together consist of 195 HRUs. The parameters were
adjusted within the practical range to obtain a reasonable fit between the simulated and measured flows in terms
of Nash-Sutcliffe modeling efficiency and the high flow and low flow components as well as the seasonal flows.
The calibration focus area well-represented the general land use characteristics of the overall watershed. The
predominantly forested subwatershed (Mohawk River near Springfield, OR; USGS 14165000) was chosen to set
the parameters for forest, which were then applied across the entire watershed. There is essentially one set of
parameters for a land use type for the entire watershed.
Once the hydrology calibration was complete for the entire Willamette watershed, the water quality calibration
was pursued. Similar to hydrology, there is a single set of water quality parameters for the entire Willamette River
watershed.
Hydrology calibration adjustments focused on the following parameters:
• Curve numbers (varied systematically by land use)
• ESCO (soil evaporation compensation factor)
• SURLAG (surface runoff lag coefficient)
• Baseflow factor
• GW_DELAY (groundwater delay time)
• Sol_AWC (available water capacity of the soil layer, mm water/mm of soil)
H-55
-------
Initial hydrology calibrations were performed for the Tualatin River at West Linn, OR (USGS 14207500) and are
summarized in Figures 40 through 46 and Tables 22 and 23. The model calibration period was set to the 10 water
years from 10/01/1995 to 09/30/2005. As evidenced through the time series plot, the model performed well in
simulating the timing at various seasons. The model overpredicted seasonal spring, summer, and overall storm
volumes indicated by the Error in Storm Volumes metric.
I
I
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1995 to 9/30/2005 )
•Avg Modeled Flow (Same Period)
40000
35000
30000
25000
20000
15000
10000
Oct-04
Figure 40. Mean daily flow at USGS 14207500 Tualatin River at West Linn, OR - calibration period
(SWAT).
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1995 to 9/30/2005 )
Avg Modeled Flow (Same Period)
18
I
0-95
A-97
0-04
Figure 41. Mean monthly flow at USGS 14207500 Tualatin River at West Linn, OR - calibration period
(SWAT).
H-56
-------
• Avg Flow (10/1 /1995 to 9/30/2005 )
• - - - • Line of Equal Value
Best-Fit Line
2
LL
0)
E
0)
O)
CO
10000
8000
6000
4000
2000
2000 4000 6000 8000 10000
Average Observed Flow (cfs)
Avg Observed Flow (10/1/1995 to 9/30/2005)
Avg Modeled Flow (10/1/1995 to 9/30/2005 )
-Line of Equal Value
O
_
ro
m
O-95 A-97 O-98
A-00 O-01
Month
A-03 O-04
Figure 42. Monthly flow regression and temporal variation at USGS 14207500 Tualatin River at West
Linn, OR - calibration period (SWAT).
.0
0
T3
iSOOO
2000
1000
Avg Flow (10/1/1995 to 9/30/2005)
Best- Fit Line
y = C
,
f^\
.7664x i
R2 = 09
•*f
-313.64
541
,*^^^
tjr
st
^^
,*
*
1000 2000 3000 4000
Average Observed Flow (cfs)
5000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1995 to 9/30/2005)
Avg Modeled Flow (Same Period)
5000
g
"co
TO
or
£=
O
Figure 43. Seasonal regression and temporal aggregate at USGS 14207500 Tualatin River at West
Linn, OR - calibration period (SWAT).
H-57
-------
Average Monthly Rainfall (in)
•Median Observed Flow (10/1/1995 to 9/30/2005)
I Observed (25th, 75th)
Modeled (Median, 25th, 75th)
7000
c
'ro
a:
Figure 44. Seasonal medians and ranges at USGS 14207500 Tualatin River at West Linn, OR -
calibration period (SWAT).
Table 22. Seasonal summary at USGS 14207500 Tualatin River at West Linn, OR - calibration
period (SWAT)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
399.00
1305.46
3817.72
4175.40
4193.04
2763.99
1503.13
929.54
434.30
236.99
215.16
258.55
309.00
534.00
3420.00
3675.00
3440.00
2340.00
1165.00
681 .00
385.50
227.00
197.50
232.00
234.00
305.75
1565.00
2297.50
1955.00
1155.00
855.00
497.25
295.00
187.00
171.25
197.75
436.75
1460.00
5747.50
5340.00
5485.00
3962.50
1760.00
1140.00
485.00
275.50
233.75
286.25
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
451 .80
1766.52
2950.48
3356.79
3557.28
2594.23
1822.10
1307.67
749.55
347.93
186.90
177.81
233.45
591.34
1891.63
2452.25
2384.45
2270.38
1663.14
1187.81
705.59
325.30
138.33
98.44
133.30
265.80
1053.44
1333.92
1397.75
1469.18
1182.51
835.28
519.57
183.15
85.99
78.28
502.09
2302.69
3756.60
4517.63
3944.65
3195.54
2135.92
1596.66
927.19
468.27
222.95
155.79
H-58
-------
•Observed Flow Duration (10/1/1995 to 9/30/2005 )
Modeled Flow Duration (10/1/1995 to 9/30/2005)
100000
o
10000
ro' 1000
100 - =
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 45. Flow exceedence at USGS 14207500 Tualatin River at West Linn, OR - calibration period
(SWAT).
o
o
ro
T3
.a
O
_3
o
ro
•Observed Flow Volume (10/1/1995 to 9/30/2005 )
Modeled Flow Volume (10/1/1995 to 9/30/2005)
120%
100% -
80%
60%
40% -
20%
Oct-95
Apr-97
Oct-98
Apr-00
Oct-01
Apr-03
Oct-04
Figure 46. Flow accumulation at USGS 14207500 Tualatin River at West Linn, OR - calibration period
(SWAT).
H-59
-------
Table 23. Summary statistics: USGS 14207500 Tualatin River at West Linn, OR - calibration
period (SWAT)
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 5
10-Year Analysis F^riod: 10/1/1995 - 9/30/2005
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
_Jotak3fj3irrujlatejdjT^^
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9)
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3^:
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
E rronrM05yTighesHlowsj__
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring;
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
30.72
13.37
3.17
1.22
8.33
14.99
6.18
10.59
0.24
Error Statistics
-4.76
11.08
-4.19
6.04
Observed Flow Gage
USGS 14207500 TUALATIN RIVER AT WEST LINN, OR
Hydrologic Unit Code: 17090010
Latitude: 45.35067559
Longitude: -122.6762044
Drainage Area (sq-rri): 706
Total Observed In-stream Flow:
Total of Observed highesMO%Jows:_^
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
-6.91 » | 30
-14.74
35.01
71.50
52.98
0.489
0.505
0.885
30
30
20
50
Model accuracy increases
as E or E' approaches 1 .0
32.25
13.96
2.85
1.15
8.95
17.58
4.58
6.18
0.16
Clear [
Hydrology Validation
Consistent with HSPF modeling efforts, validation for the Tualatin River calibration focus area was performed at
the same location but for the water years from 10/01/1985 to 09/30/1995. Results are presented in Figures 47
through 53 and Tables 24 and 25. Although, the Nash-Sutcliffe modeling efficiency is not as good as it was for
the calibration period, the model performance was adequate for the validation period. The model underestimates
total flow volumes while it overestimates low flows and storm volumes. The rest of the metrics fall within the
acceptable range set for the 20 Watershed study.
H-60
-------
I
o
Avg Daily Rainfall (in)
-Avg Observed Flow (10/1/1986 to 9/30/1995 )
•Avg Modeled Flow (Same Period)
14000
12000
10000
Oct-87 Oct-88 Oct-89
Oct-90 Oct-91
Date
Oct-92 Oct-93 Oct-94
14
Figure 47. Mean daily flow at USGS 14207500 Tualatin River at West Linn, OR - validation period
(SWAT).
^H Avg Monthly Rainfall (in)
-»-Avg Observed Flow (10/1/1986 to 9/30/1995 )
Avg Modeled Flow (Same Period)
8000
6000
4000 -
2000 -
0-86 0-87 0-88 O-89 O-90 O-91 O-92 O-93 O-94
Figure 48. Mean monthly flow at USGS 14207500 Tualatin River at West Linn, OR - validation period
(SWAT).
H-61
-------
Avg Flow (10/1/1986 to 9/30/1995 )
• Line of Equal Value
Best-Fit Line
8000
&
u^
2
LL
T3
jl>
0)
E
0)
O) 2000
CD
o5
^
y = 0.6554X + 261
R2 = 0.8731
.48
.a
O
_
ro
m
100% -,
90% -
80% -
70% -
60% -
50%
Avg Observed Flow (10/1/1986 to 9/30/1995)
Avg Modeled Flow (10/1/1986 to 9/30/1995 )
-Line of Equal Value
I, I
2000 4000 6000
Average Observed Flow (cfs)
8000
O-86 O-87 O-88 O-89 O-90 O-91 O-92 O-93 O-94
Month
Figure 49. Monthly flow regression and temporal variation at USGS 14207500 Tualatin River at West
Linn, OR - validation period (SWAT).
• Avg Flow (10/1/1986 to 9/30/1995)
• • • • • Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1986 to 9/30/1995)
•Avg Modeled Flow (Same Period)
4000
3000
2000
-------
Average Monthly Rainfall (in)
- Median Observed Flow (10/1/1986 to 9/30/1995)
[Observed (25th, 75th)
Modeled (Median, 25th, 75th)
10 11 12 1
234
Month
c
'ro
or
>,
Figure 51. Seasonal medians and ranges at USGS 14207500 Tualatin River at West Linn, OR -
validation period (SWAT).
Table 24. Seasonal summary at USGS 14207500 Tualatin River at West Linn, OR - validation
period (SWAT)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
176.23
684.20
1849.41
2939.52
3060.41
2478.49
1615.37
612.33
328.74
173.48
137.55
138.14
133.00
324.00
1470.00
2900.00
2580.00
2200.00
1110.00
491.00
248.00
165.00
132.00
127.00
103.00
193.00
873.00
1460.00
1150.00
1335.00
791 .00
339.50
199.00
130.50
112.00
111.00
187.50
875.00
2450.00
4055.00
4450.00
3380.00
1917.50
770.50
410.25
211.50
159.50
158.00
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
253.63
887.67
1595.11
2234.76
2025.35
1906.07
1560.53
915.70
599.41
258.61
112.11
90.80
92.74
455.03
996.23
1393.87
1429.89
1506.88
1272.74
882.87
547.55
233.54
89.66
74.92
83.59
200.88
507.82
880.57
865.74
1014.94
1025.63
667.62
367.63
149.65
72.13
66.13
157.15
1100.85
2060.43
2590.33
2438.83
2270.91
1620.06
1053.08
731.63
336.11
125.44
97.54
H-63
-------
o
D)
ro
Q
•Observed Flow Duration (10/1/1986 to 9/30/1995 )
Modeled Flow Duration (10/1/1986 to 9/30/1995)
100000
10000
1000 - =
100 - =
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Percent of Time that Flow is Equaled or Exceeded
Figure 52. Flow exceedence at USGS 14207500 Tualatin River at West Linn, OR - validation period
(SWAT).
•Observed Flow Volume (10/1/1986 to 9/30/1995 )
Modeled Flow Volume (10/1/1986 to 9/30/1995)
o
o
ro
T3
_a
O
|
"o
^3
LL
T3
N
15
120%
100%
80%
60%
40%
20%
Oct-86 Oct-87 Oct-i
Oct-89 Oct-90 Oct-91 Oct-92 Oct-93 Oct-94
Figure 53. Flow accumulation at USGS 14207500 Tualatin River at West Linn, OR - validation period
(SWAT).
H-64
-------
Table 25. Summary statistics at USGS 14207500 Tualatin River at West Linn, OR - validation
period (SWAT)
REACH OUTFLOW FROM OUTLET 5
9-Year Analysis F^riod: 10/1/1986 - 9/30/1995
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9)
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Srjring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
_Naj3h^3ut£liffeJDpj5^
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
19.85
8.29
2.28
0.75
4.42
9.77
4.91
6.95
0.12
Error Statistics
-12.10
24.62
-12.81
3.12
USGS 14207500 TUALATIN RIVER AT WEST LINN, OR
Hydrologic Unit Code: 17090010
Latitude: 45.35067559
Longitude: -122.6762044
Drainage Area (sq-rri): 706
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
0.74 » | 30
-27.05
20.54
47.27
4.49
0.394
0.459
0.807
30
30
20
50
Model accuracy increases
as E or E' approaches 1 .0
22.59
9.50
1.83
0.73
4.39
13.40
4.07
4.72
0.11
Clear [
H-65
-------
Hydrology Results for Larger Watershed
Since the Tualatin River calibration location represents only a small portion of the entire Willamette River
watershed, the results near the outlet of the entire watershed were examined at the Willamette River at Salem, OR
(USGS 14191000). This gage is downstream of the large reservoirs in the Willamette River watershed. Greater
than 40 percent of the area at Salem is controlled by dams. The results are presented in Figures 54 through 60 and
Tables 26 and 27. Summer storms are small in this region, and the summer storm volumes are also small;
therefore, an error of 0.2 inches produces a large percent difference. Underestimation of low flow is manifested in
underestimation of seasonal summer volumes. The remainder of the metrics fall within the acceptable range set
for the 20 Watershed study including a daily Nash-Sutcliffe of 0.67 at the Salem gage. Tables 28 and 29 show a
summary of the hydrology calibration and validation results for all five locations, respectively.
^M Avg Monthly Rainfall (in)
Avg Observed Flow (10/1/1995 to 9/30/2005 )
Avg Modeled Flow (Same Period)
250000
200000
4? 150000
2 100000
50000
Oct-95 Apr-97
Oct-98 Apr-00 Oct-01
Date
Apr-03
Oct-04
Figure 54. Mean daily flow at USGS 14191000 Willamette River at Salem, OR - calibration period
(SWAT).
^H Avg Monthly Rainfall (in)
-•-Avg Observed Flow (10/1/1995 to 9/30/2005 )
Avg Modeled Flow (Same Period)
150000
t
I
100000
50000
O-95
Figure 55. Mean monthly flow at USGS 14191000 Willamette River at Salem, OR - calibration period
(SWAT).
H-66
-------
Avg Flow (10/1/1995 to 9/30/2005 )
• Line of Equal Value
Best-Fit Line
o
0)
E
0)
O)
CO
150000
100000
y = 0.9287X
R2 = 0.^606
+ 588.7
50000
0 50000 100000 150000
Average Observed Flow (cfs)
to
JD
O
_
ro
m
100% -,
90% -
80% -
70% -
60% -
50%
Avg Observed Flow (10/1/1995 to 9/30/2005)
Avg Modeled Flow (10/1/1995 to 9/30/2005 )
-Line of Equal Value
40% -
30% -
20% -
10% -
0%
O-95 A-97 O-98
A-00 O-01
Month
A-03 O-04
Figure 56. Monthly flow regression and temporal variation at USGS 14191000 Willamette River at
Salem, OR - calibration period (SWAT).
• Avg Flow (10/1/1995 to 9/30/2005)
• • • • • Line of Equal Value
Best-Fit Line
60000
0 20000 40000 60000
Average Observed Flow (cfs)
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1995 to 9/30/2005)
•Avg Modeled Flow (Same Period)
I
o
60000
40000
20000
10 11 12 1 23456789
Month
Figure 57. Seasonal regression and temporal aggregate at USGS 14191000 Willamette River at Salem,
OR - calibration period (SWAT).
H-67
-------
Average Monthly Rainfall (in)
• Median Observed Flow (10/1/1995 to 9/30/2005)
I Observed (25th, 75th)
Modeled (Median, 25th, 75th)
80000
70000
60000
10 11 12 1
12
c
'ro
or
>,
Figure 58. Seasonal medians and ranges at USGS 14191000 Willamette River at Salem, OR
calibration period (SWAT).
Table 26. Seasonal summary at USGS 14191000 Willamette River at Salem, OR - calibration
period (SWAT)
MONTH
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
OBSERVED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
12717.87
25500.23
50807.87
50369.00
36872.79
27664.39
23458.33
20733.55
12950.87
7675.19
7159.26
8761 .37
11800.00
15500.00
41900.00
49350.00
28300.00
23950.00
20000.00
16950.00
12250.00
7430.00
7110.00
8450.00
8752.50
12375.00
22175.00
26325.00
19050.00
14950.00
16600.00
15100.00
9807.50
6780.00
6602.50
7210.00
14875.00
26525.00
74950.00
70900.00
43500.00
34500.00
24825.00
23675.00
14825.00
8227.50
7395.00
9847.50
MODELED FLOW (CFS)
MEAN MEDIAN 25TH 75TH
5439.92
20933.37
41930.27
45283.04
45938.63
37075.54
28696.06
22035.69
13294.29
6400.42
2621.14
1757.98
2240.19
10921.06
35365.87
42977.95
44496.48
36462.39
26883.29
2001 1 .06
12840.41
6098.84
2353.02
1243.78
905.47
5974.36
21415.70
29260.85
30854.42
28114.89
21591.39
15987.83
10054.09
4428.46
1650.52
715.39
6665.64
22100.80
55488.17
59355.13
55461 .68
45953.21
34482.12
26090.48
16113.20
8161.22
3287.97
1834.86
H-68
-------
•Observed Flow Duration (10/1/1995 to 9/30/2005 )
Modeled Flow Duration (10/1/1995 to 9/30/2005)
1000000
100000
o
ro' 10000
ro
Q
1000
100
10% 20%
30%
40% 50%
60%
70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 59. Flow exceedence at USGS 14191000 Willamette River at Salem, OR - calibration period
(SWAT).
o
o
ro
T3
_a
O
|
"o
^3
LL
T3
N
15
•Observed Flow Volume (10/1/1995 to 9/30/2005 )
Modeled Flow Volume (10/1/1995 to 9/30/2005)
120%
100%
80%
60%
40%
20%
0% n
Oct-95
Apr-97
Oct-98
Apr-00
Oct-01
Apr-03
Oct-04
Figure 60. Flow accumulation at USGS 14191000 Willamette River at Salem, OR- calibration period
(SWAT).
H-69
-------
Table 27. Summary statistics at USGS 14191000 Willamette River at Salem, OR - calibration
period (SWAT)
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 18
10-Year Analysis F^riod: 10/1/1995 - 9/30/2005
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
_Jotak3fj3irrujlatejdjT^^
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9)
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3^:
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
42.01
13.73
5.44
1.70
10.71
19.68
9.92
10.96
0.34
Error Statistics
-4.96
-38.08
-11.71
-54.01
Observed Flow Gage
USGS 14191000 WILLAMETTE RIVER AT SALEM, OR
Hydrologic Unit Code: 17090007
Latitude: 44.9442863
Longitude: -123.0428742
Drainage Area (sq-rri): 7280
Total Observed In-stream Flow:
Total of Observed highesMO%Jows:_^
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
-23.33 » | 30
11.28
11.98
-4.51
109.03
0.669
0.449
0.852
30
30
20
50
Model accuracy increases
as E or E' approaches 1 .0
44.21
15.55
8.79
3.69
13.97
17.69
8.86
11.48
0.16
Clear [
H-70
-------
Table 28. Summary statistics (percent error) for all stations - calibration period (SWAT)
Station
Calibration Period:
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient of
Efficiency, E:
Monthly Nash-Sutcliffe Coefficient of
Efficiency, E:
14191000
Salem
WY 96-05
-4.96
-38.08
-11.71
-54.01
-23.33
11.28
11.98
-4.51
109.03
0.669
0.852
14207500
West Linn
WY 96-05
-4.76
9.00
-4.03
0.65
-6.72
-14.57
35.37
71.00
52.22
0.489
0.885
14202000
Aurora
WY 03-05
-1.06
31.57
-14.73
71.04
3.75
-12.99
11.82
5.57
49.05
0.691
0.910
14194150
McMinnville
WY 00-05
-31.37
-17.54
-38.63
-9.98
-39.84
-35.67
-0.70
-35.70
5.47
0.448
0.729
14165000
Springfield
WY 99-05
-18.37
-34.33
-15.69
-60.43
0.45
-23.92
-23.90
-0.30
2.75
0.663
0.820
Table 29. Summary statistics (percent error) for all stations - validation period (SWAT)
Station
Calibration Period:
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient of
Efficiency, E:
Monthly Nash-Sutcliffe Coefficient of
Efficiency, E:
14191000
Salem
WY 86-95
-5.67
-37.51
-7.99
-51.91
-24.97
12.23
8.57
-5.50
79.12
0.534
0.798
14207500
West Linn
WY 86-95
-12.10
24.62
-12.81
3.12
0.74
-27.05
20.54
47.27
4.49
0.394
0.807
14202000
Aurora
WY 94-97
1.56
63.52
-10.57
144.28
-11.14
-7.96
48.39
9.52
180.98
0.699
0.904
14194150
McMinnville
WY 95-99
-30.25
18.59
-39.11
64.02
-43.35
-32.93
15.41
-37.44
25.45
0.451
0.713
14165000
Springfield
WY 88-97
-19.42
-30.89
-16.44
-57.17
10.29
-30.79
-29.57
2.05
3.44
0.486
0.667
H-71
-------
Water Quality Calibration
Initial water quality calibration and validation was performed for the Tualatin River at Linn, OR
(USGS14207500) using water years 1991-1995 for calibration and water years 1986-1990 for validation. As with
hydrology, water quality calibration was performed on the later period as this better reflects the land use included
in the model. The start of the validation period is constrained by data availability.
Calibration adjustments for TSS focused on the following parameter:
• RSDCO (Residue decomposition coefficient)
Time series of simulated and estimated TSS loads at the Tualatin River gage for both the calibration and
validation periods are shown in Figure 61. Statistics for the two periods are provided separately in Table 30. The
key statistic in Table 30 is the relative percent error, which shows the error in the prediction of monthly load
normalized to the estimated load. Table 30 also shows the relative average absolute error, which is the average of
the relative magnitude of errors in individual monthly load predictions. This number is inflated by outlier months
in which the simulated and estimated loads differ by large amounts (which may be as easily due to uncertainty in
the estimated load due to limited data as to problems with the model) and the third statistic, the relative median
absolute error, is likely more relevant and shows good agreement. Additional diagnostics for TSS included flow-
load power plots (Figures 62 and 63), a time series plot of concentrations (Figure 64), and statistics (Table 31).
TSS
100,000
-Regression Loads
-Simulated Loads
opopopopopopopopcpcpcpcpcpcpcpcpcpcpcpcp
" " " " " " " " " "
Figure 61. Fit for monthly load of TSS at USGS 14207500 Tualatin River at West Linn, OR (SWAT).
H-72
-------
Table 30. Model fit statistics (observed minus predicted) for monthly TSS loads using
stratified regression (SWAT)
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1991-1995)
-12
47
17.2
Validation period
(1986-1990)
-7
40
10.5
ra
•o
«
c
o
•o
ra
o
CO
V)
TUALATIN RIVER AT WEST LINN, OR 1991-1995
10000
1000
100
10
1
0.1
WP
10
100 1000
Flow, cfs
10000
100000
Simulated
A Observed
Power (Simulated)
Power (Observed]
Figure 62. Power plot for observed and simulated TSS at USGS 14207500 Tualatin River at West Linn,
OR - calibration period (SWAT).
H-73
-------
TUALATIN RIVER AT WEST LINN, OR 1986-1990
1
o
•o
ra
o
(0
V)
10000
1000
100
10
0.1
10
100 1000
Flow, cfs
10000
100000
» Simulated A Observed
Power (Simulated)
•Power (Observed^
Figure 63. Power plot for observed and simulated TSS at USGS 14207500 Tualatin River at West Linn,
OR - validation period (SWAT).
TUALATIN RIVER AT WEST LINN, OR 1986-1995
1000
100
O)
eo
1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996
Year
Figure 64. Time series plot of TSS concentration at USGS 14207500 Tualatin River at West Linn, OR
(SWAT).
H-74
-------
Table 31. Relative errors (observed minus predicted), TSS concentration, at USGS
14207500 Tualatin River at West Linn, OR (SWAT)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1991-1995)
35
-30.3%
10.13%
Validation period
(1986-1990)
19
0.06%
-22.31%
Calibration adjustments for total phosphorus and total nitrogen focused on the following parameters:
• PPERCO (phosphorus percolation coefficient)
• NPERCO (nitrogen percolation coefficient)
• PHOSKD (phosphorus soil partitioning coefficient)
• HLIFE_NGW (half life of nitrate in the shallow aquifer)
• SOL_CBN1 (organic carbon in the first soil layer)
• QUAL2E parameters such as algal, organic nitrogen, and organic phosphorus settling rate in the reach,
benthic source arte for dissolved phosphorus and NH4-N in the reach, fraction of algal biomass that is
nitrogen and phosphorus, Michaelis-Menton half-saturation constant for nitrogen and phosphorus
The time series of observed and simulated total phosphorus loads is shown in Figure 65 (monthly loads) and
Table 32 (load statistics). As with TSS, additional diagnostics for total phosphorus included flow-load power
plots (Figures 66 and 67), a time series plot of concentrations (Figure 68), and statistics (Table 33). In general,
total phosphorus for the Willamette River watershed was overestimated by the SWAT model.
Total P
1000
100 -.
o
In
I
-Regression Loads
-Simulated Loads
Figure 65. Fit for monthly load of total phosphorus at USGS 14207500 Tualatin River at West Linn, OR
(SWAT).
H-75
-------
Table 32. Model fit statistics (observed minus predicted) for monthly total phosphorus loads
using stratified regression (SWAT)
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1991-1995)
-114%
118%
51.7%
Validation period
(1986-1990)
-105%
109%
79.7%
0.001
TUALATIN RIVER AT WEST LINN, OR 1991-1995
10
100 1000
Flow, cfs
10000 100000
• Simulated A Observed
Power (Simulated) ^^^"Power (Observed]
Figure 66. Power plot for observed and simulated total phosphorus at USGS 14207500 Tualatin River
at West Linn, OR - calibration period (SWAT).
H-76
-------
TUALATIN RIVER AT WEST LINN, OR 1986-1990
-------
Table 33. Relative errors (observed minus predicted), total phosphorus concentration, at
USGS 14207500 Tualatin River at West Linn, OR (SWAT)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1991-1995)
35
-312.95%
-163.58%
Validation period
(1986-1990)
19
-127.9%
-109.9%
Results for total nitrogen are summarized in Figures 69 through 72 and Tables 34 and 35. Again, total nitrogen
loads are overestimated, but are generally better than those for total phosphorus. A summary of the water quality
statistics at the two locations (Tualatin River and Pudding River) are shown in Table 36.
Total N
2,000
-Averaging Loads
-Simulated Loads
CO CO O> O>
CO CO CO CO CO CO
=s £ 3 £ 3 § 3
co
—
co
—
O O i-
0)0)0)
c "^ c
CO -2 CO
a
CO CO
O) O)
co 1
s s
LO LO
O5 O5
§ =
Figure 69. Fit for monthly load of total nitrogen at USGS 14207500 Tualatin River at West Linn, OR
(SWAT).
Table 34. Model fit statistics (observed minus predicted) for monthly total nitrogen loads
using averaging estimator (SWAT)
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1991-1995)
-72%
86%
68.2%
Validation period
(1986-1990)
-66%
86%
66.3%
H-78
-------
PUDDING RIVER AT AURORA, OR 1991-1995
1000
100
•o
«
c
o
•o
ra
o
10
100 1000
Flow, cfs
10000
100000
• Simulated A Observed
Power (Simulated)
Power (Observed]
Figure 70. Power plot for observed and simulated total nitrogen at USGS 14207500 Tualatin River at
West Linn, OR - calibration period (SWAT).
PUDDING RIVER AT AURORA, OR 1986-1990
1000
100
5
"5>
•o
ra
o
10
100 1000
Flow, cfs
10000
100000
• Simulated A Observed
Power (Simulated)
Power (Observed]
Figure 71. Power plot for observed and simulated total nitrogen at USGS 14207500 Tualatin River at
West Linn, OR - validation period (SWAT).
H-79
-------
PUDDING RIVER AT AURORA, OR 1986-1995
1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996
Year
Figure 72. Time series plot of total nitrogen concentration at USGS 14207500 Tualatin River at West
Linn, OR (SWAT).
Table 35. Relative errors (observed minus predicted), total nitrogen concentration, at
USGS 14207500 Tualatin River at West Linn, OR (SWAT)
Statistic
Count
Concentration Average Error
Concentration Median Error
Calibration period
(1991-1995)
35
-251 .26%
-137.89%
Validation period
(1986-1990)
20
-265.41%
-160.42%
H-80
-------
Water Quality Results for Larger Watershed
As with hydrology, the Tualatin River watershed parameters for water quality were directly transferred to other
portions of the watershed. This approach resulted in relatively large errors in predicting loads and concentrations
at some stations. Summary statistics for the water quality calibration and validation at other stations in the
watershed are provided in Table 36.
Table 36. Summary statistics for water quality: all stations (observed minus predicted)
(SWAT)
Station
Relative Percent Error
TSS Load
TSS Concentration
Median Percent Error
Relative Percent Error
TP Load
TP Concentration
Median Percent Error
Relative Percent Error
TN Load
TN Concentration
Median Percent Error
14207500
West Linn
Calibration
-12%
-10.13%
-114%
-163%
-72%
-137.89%
14207500
West Linn
Validation
-7%
-22.31%
-105%
-109.9%
-66%
-160.42%
14202000
Aurora
Calibration
30%
-13.27%
-30%
-106.01%
-11%
-42.13%
14202000
Aurora
Validation
-100%
-43.55%
-373%
-94.96%
-218%
-165.16%
H-81
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
Bicknell, B.R., J.J. Burkey, and R.A. Dusenbury. 2005. Modeling Water Quality in Urban Northwest Watersheds.
Presented at ASCE Watershed Management 2005, Williamsburg, VA.
Preston, S.D., V.J. Bierman, Jr., and S.E. Silliman. 1989. An evaluation of methods for the estimation of tributary
mass loads. Water Resources Research, 25(6): 1379-1389.
Tetra Tech. 1999. Improving Point Source Loadings Data for Reporting National Water Quality Indicators. Final
Technical Report prepared for U.S. Environmental Protection Agency, Office of Waste water Management,
Washington, DC, by Tetra Tech, Inc., Fairfax, VA.
USAGE (U.S. Army Corps of Engineers). 1982. National inventory of dams data base in card format, available
from National Technical Information Service, Springfield, VA 22162, #ADA 118670.
USAGE (U.S. Army Corps of Engineers). 2009. Available at http://www.nwp.usace.army.mi1/op/v/wvmap.asp.
USEPA (United States Environmental Protection Agency). 2008. Using the BASINS Meteorological Database
(Version 2006). BASINS Technical Note 10. Office of Water, U.S. Environmental Protection Agency,
Washington, DC. http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf
(Accessed June, 2009).
USEPA (United States Environmental Protection Agency). 2006. Sediment Parameter and Calibration Guidance
for HSPF. BASINS Technical Note 8. Office of Water, U.S. Environmental Protection Agency, Washington,
DC. http://water.epa.gov/scitech/datait/models/basins/upload/2006_02_02_BASINS_tecnote8.pdf (Accessed
June, 2009).
USGS (United States Geological Survey). 2001. National Water Quality Assessment (NAWQA) Program:
Willamette Basin Study Unit. http://or.water.usgs.gov/projs_dir/pn366/nawqa.html (Accessed June, 2009).
H-82
-------
Appendix I
Model Configuration, Calibration and
Validation
Basin: Lake Pontchartrain Drainage
(LPont)
1-1
-------
Contents
Watershed Background 1-4
Water Body Characteristics 1-4
Soil Characteristics 1-6
Land Use Representation 1-6
Point Sources 1-10
Meteorological Data 1-11
Watershed Segmentation 1-12
Calibration Data and Locations 1-14
SWAT Modeling 1-15
Assumptions 1-15
Hydrology Calibration 1-15
Hydrology Validation 1-19
Hydrology Results for Larger Watershed I-22
Water Quality Calibration and Validation I-24
References I-28
1-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes 1-8
Table 2. Land use distribution for the Pontchartrain watershed (2001 NLCD) (mi2) 1-9
Table 3. Major point source discharges in the Pontchartrain watershed 1-10
Table 4. Precipitation stations for the Pontchartrain watershed model 1-11
Table 5. Calibration and validation locations in the Pontchartrain watershed 1-14
Figure 8. Summary statistics at USGS 07378500 Amite River near Denham Springs, LA - calibration
period 1-19
Table 6. Summary statistics at USGS 07378500 Amite River near Denham Springs, LA - validation
period 1-22
Table 7. Summary statistics (percent error): all stations - calibration period 1-23
Table 8. Summary statistics: all stations - validation period 1-23
Table 9. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression 1-25
Table 10. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression 1-26
Table 11. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator 1-27
Figures
Figure 1. Location of the Pontchartrain watershed 1-5
Figure 2. Land use in the Pontchartrain watershed 1-7
Figure 3. Model segmentation and USGS stations utilized for the Pontchartrain watershed 1-13
Figure 4. Mean monthly flow at USGS 07378500 Amite River near Denham Springs, LA - calibration
period 1-16
Figure 5. Seasonal regression and temporal aggregate at USGS 07378500 Amite River near Denham
Springs, LA - calibration period 1-17
Figure 6. Seasonal medians and ranges at USGS 07378500 Amite River near Denham Springs, LA -
calibration period 1-17
Figure 7. Flow exceedance at USGS 07378500 Amite River near Denham Springs, LA - calibration
period 1-18
Figure 9. Mean monthly flow at USGS 07378500 Amite River near Denham Springs, LA - validation
period 1-20
Figure 10. Seasonal regression and temporal aggregate at USGS 07378500 Amite River near Denham
Springs, LA - validation period 1-20
Figure 11. Seasonal medians and ranges at USGS 07378500 Amite River near Denham Springs, LA -
validation period 1-21
Figure 12. Flow exceedance at USGS 07378500 Amite River near Denham Springs, LA - validation
period 1-21
Figure 13. Fit for monthly load of TSS at USGS 07378500 Amite River near Denham Springs, LA 1-24
Figure 14. Fit for monthly load of total phosphorus at USGS 07378500 Amite River near Denham
Springs, LA 1-26
Figure 15. Fit for monthly load of total nitrogen at USGS 07378500 Amite River near Denham Springs,
LA 1-27
1-3
-------
The Lake Pontchartrain drainage study area was selected as one of the 15 non-pilot application watersheds for the
20 Watershed study. Watershed modeling for the non-pilot areas is accomplished using the SWAT model only,
and model calibration and validation results are presented in abbreviated form.
Water Body Characteristics
The Acadian-Pontchartrain NAWQA study area encompasses 26,408 mi2 in the southern half of Louisiana and
includes downstream portions of major rivers, such as the Mississippi with drainage areas far larger than the
target size for this project. Therefore, the focus of modeling in this study was the Pontchartrain portion of the
study area, including the rivers that drain to Lake Pontchartrain and the cities of New Orleans and Baton Rouge
(Figure 1). The resulting model area encompasses over 5,800 mi2 and seven HUC8s within HUCs 0807 and
0809.The watershed includes the Calcasieu, Mermentau, Vermilion-Teche, Grosse Tete/Verret, Terrebonne,
Barataria, and Pontchartrain basins (USGS 2002).
The entire model area is near sea level and frequently impacted by tropical storms from the Gulf of Mexico. The
climate is classified as humid subtropical, with an average annual temperature around 70 °F and average annual
precipitation of 64 inches per year (USGS, 2002).
Ecosystems and communities in the watershed include cypress-tupelo swamp; freshwater marsh; saltwater marsh;
wet prairie; oak cheniers; bottomland hardwood forest; Piney Hills; and longleaf pine savanna. The coastal zone
of the watershed is affected by the ocean and its tides. Different wetland types are determined by the salinity of
the water in them, which may infiltrate naturally through bayous or reach further inland through canals.
Land uses include a mixture of urban and rapidly urbanizing/industrial areas (12 percent), large areas of mixed
forest and pasture (34 percent), wetlands (32 percent) and areas of rice and sugarcane crops (5 percent).
Population is rapidly increasing on the north shore of Lake Pontchartrain and surrounding Baton Rouge, causing
changes in rainfall-runoff characteristics and quality. Urban streams in the Baton Rouge area are usually
channelized, and cleared of woody vegetation to speed drainage during high water.
Surface water in the watershed includes the lower Mississippi delta and wet prairie streams as well as upland
streams. Lower Mississippi delta and wet prairie streams tend to have very slow flow, and water can also be
pushed upstream by tides or wind causing generally stagnant, backwater conditions. Wetlands develop naturally
in poorly drained areas. Streams in the uplands have a moderate flow gradient and sandy, shifting beds that are
reshaped quickly in the fast water that is usual for flood conditions.
Modifications to flow include levees, and canals and drainage. Levees are created both naturally during the
flooding process (sediment drops out of floodwater next to the waterbody) and by man along many bayous and
rivers to reduce floods and to maintain a deeper channel for shipping.
1-4
-------
Hydrography
Water (Nat. Atlas Datasel)
US Census Populated Places
^B Municipalities (pop i 50.000!
^ County Boundaries
| | Watershed with HUCBs
Mississippi
f—f-^
Louisiana
Bayou Sara-
Thompson/
(08070201}
Tangipahoa
08070205)
Amite
(08070202)
Tickpaw
08070203
Liberty Bayou-
Tchefuncta
(08090201)
Baton Rouge;
Lake
Pomchartrain
Lake Maurepas
08070204)
Eastern Louisiana
Coastal
(08090203)
New Orleans
GCRP Model Areas - Lake Pontchartrain Drainage
Base Map
NACM &S3_Alberi._meters
N
A
0 10 20
0 10
40
20
40
Figure 1. Location of the Pontchartrain watershed
1-5
-------
Soil Characteristics
Soils in the watershed, as described in STATSGO soil surveys, fall primarily into hydrologic soil groups (HSGs)
C (moderately low infiltration capacity, 58%) and D (low infiltration capacity, 31%). SWAT uses information
drawn directly from the soils data layer to populate the model.
Land Use Representation
Land use/cover in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage and is
predominantly wetland (32 %) and forest (Figure 2). NLCD land cover classes were aggregated according to the
scheme shown in Table 1 for representation in the 20 Watershed model. SWAT uses the built-in hydrologic
response unit (HRU) overlay mechanism in the ArcSWAT interface. SWAT HRUs are formed from an
intersection of land use and SSURGO major soils. The distribution of land use in the watershed is summarized in
Table 2.
1-6
-------
Hydrography
Interstate
I I County Boundaries
2001 NLCD Land Use
I Open water
I Developed, open space
| Developed, low intensity
| Developed, medium intensity
^B Developed, high intensity
I Barren land
B Deciduous forest
| Evergreen forest
I I Mixed forest
I I Scrub/shrub
I I Grassland/herbaceous
Pasture/hay
^ Cultivated crops
] Woody wetlands
n Emergent herbaceous wetlands
Pontchartrain
GCRP Model Areas - Pontchartrain Basin
Land Use Map
NAD 1983 Albers meters - Map produced 03-31-2011 - P. Cada
Figure 2. Land use in the Pontchartrain watershed
1-7
-------
Table 1. Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area usually
accounted for as reach area
Deciduous
Evergreen
Mixed
Emergent & woody wetlands
Aquatic bed wetlands (not
emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
-------
Table 2. Land use distribution for the Pontchartrain watershed (2001 NLCD) (mi )
HUC 8 watershed
Liberty Bayou-
Tchefuncta. Louisiana.
08090201
Bayou Sara-Thompson.
Louisiana, Mississippi.
08070201
Amite. Louisiana,
Mississippi. 08070202
Tickpaw. Louisiana,
Mississippi. 08070203
Lake Maurepas.
Louisiana. 08070204
Tangipahoa. Louisiana,
Mississippi. 08070205
Eastern Louisiana
Coastal. Louisiana.
08090203
Total
Open
water
7.9
15.7
3.1
100.3
7.0
11.6
50.0
195.6
Developed9
Open
space
28.0
113.1
46.3
22.1
38.7
63.9
3.6
315.7
Low
density
10.7
76.7
15.9
41.7
11.8
23.0
69.8
249.7
Medium
density
6.0
29.2
3.4
7.0
2.5
8.4
30.4
86.9
High
density
3.3
5.9
0.8
4.7
0.7
1.7
18.5
35.7
Barren
land
4.2
11.0
1.7
0.6
1.7
1.4
0.5
21.1
Forest
285.4
455.9
188.8
8.7
201.9
213.0
0.6
1,354.3
Shru bland/
Grassland
77.6
296.1
159.9
11.8
152.6
138.2
2.1
838.3
Pasture/
Hay
72.9
246.5
66.0
30.5
123.5
63.1
0.7
603.1
Cultivated
45.7
82.5
16.7
59.8
46.2
9.8
2.9
263.6
Wetland
153.1
551.1
223.5
425.8
188.9
160.2
185.6
1,888.0
Total
694.9
1,883.8
726.0
712.9
775.5
694.3
364.7
5,852.0
The percent imperviousness applied to each of the developed land uses is as follows: open space (7.53%),
density (88.08%).
low density (32.91%), medium density (60.11%), and high
1-9
-------
Point Sources
There are numerous point source discharges in the watershed. Only the major dischargers, with a design or
observed flow greater than 1 MGD are included in the simulation (Table 3). The major dischargers are
represented at long-term average flows, without accounting for changes over time or seasonal variations.
Table 3. Major point source discharges in the Pontchartrain watershed
NPDES ID
LA0000841
LA0044695
LA0002933
LA0003191
LA0004090
LA0005401
LA0005479
LA0005851
LA0046361
LA0045730
LA0004464
LA0050962
LA0003522
LA0003280
LA0005355
LA0045446
LA0052256
LA0041718
LA0000914
LA0006149
LA0032328
LA0038431
LA0064092
LA0047180
LA0068730
LA0084336
Name
EXXON CORP-BATON ROUGE
RESIN
PONCHATOULA, CITY OF
OCCIDENTAL CHEMICAL CORP. FKA
ENTERGY LOUISIANA.LLC- LITTLE
ETHYL CORP-BATON ROUGE
EXXON CHEM CO-BATON ROUGE
EXXON CHEMICAL AMERICAS
ENTERGY GULF STATES, INC-
WILLO
TAMINCO HIGHER AMINES, INC.
DENHAM SPRINGS, CITY OF
EXIDE CORP-SCHUYLKILL METALS
SHELL CHEMICAL LP-NORCO
CYPRES
MOTIVA ENTERPRISES, LLC
AIR PROD & CHEM INC-NEW
ORLEAN
EXXON CHEM CO-BATON ROUGE
COAST WATERWORKS-EDNE ISLES
LOCKHEED MARTIN CORPORATION
UOP LLC
LION COPOLYMER, LLC
FORMOSA PLASTICS-BATON
ROUGE
HAMMOND CITY OF SOUTH POND
AMITE CITY, TOWN OF
ST JOHN THE BAPTIST PAR-SD #2L
SLIDELL, CITY OF
H2O SYSTEMS, INC- GREENLEAVES
COVINGTON, CITY OF
Design flow
(MGD)
2.00
1.00
1.31
11.11
8.68
11.03
3.90
8.40
9.31
2.90
3.36
5.11
6.41
1.17
15.23
0.96
13.82
4.21
2.70
44.92
2.50
0.80
3.30
6.00
4.92
1.75
Observed flow
(MGD)
(1991-2006 average)
0.18
1.19
1.99
0.36
1.79
5.98
2.61
1.71
0.31
1.58
0.27
0.57
11.37
0.64
1.20
0.85
0.21
1.70
2.90
5.19
2.16
1.20
0.66
4.10
0.52
1.62
Most of these point sources have reasonably good monitoring for total suspended solids (TSS), but often lack
detailed nutrient monitoring. The point sources without nutrient monitoring were represented in the model with
typical nutrient concentrations by SIC code (Tetra Tech 1999).
1-10
-------
Meteorological Data
The required meteorological time series for the 20 Watershed SWAT simulations are precipitation and air
temperature. The 20 Watershed simulations do not include water temperature simulation and use a degree-day
method for snowmelt. SWAT estimates Penman-Monteith potential evapotranspiration using a statistical weather
generator for inputs other than temperature and precipitation. These meteorological time series are drawn from the
BASINS4 Meteorological Database (USEPA 2008), which provides a consistent, quality-assured set of
nationwide data with gaps filled and records disaggregated. Scenario application requires simulation over 30
years, so the available stations are those with a common 30-year period of record (or one that can be filled from
an approximately co-located station) that covers the year 2001. A total of 22 precipitation stations were identified
for use in the Minnesota River model with a common period of record of 10/1/1973-9/30/2005 (Table 4).
Temperature records are sparser; where these are absent temperature is taken from nearby stations with an
elevation correction.
Table 4. Precipitation stations for the Pontchartrain watershed model
COOP ID
LA1 66686
MS229793
LA1 60549
MS225070
MS221578
LA161899
LA166911
LA1 63867
LA1 65620
LA1 64034
LA1 64859
LA1 67304
LA1 60205
LA1 62534
LA1 67767
MS225614
LA162151
LA1 68539
LA1 66660
LA1 66666
LA168108
LA1 60021
Name
NEW ROADS 5 NE
WOODVILLE 4 ESE
BATON ROUGE METRO AP
LIBERTY 5 W
CENTREVILLE
CLINTON 5 SE
OAKNOLIA 2 N
GREENWELL SPRINGS
LSU BEN HUR FARM
HAMMOND
KENTWOOD
PINE GROVE FIRE TOWER
AMITE
DONALDSONVILLE 4 SW
RESERVE
MCCOMB AIRPORT
COVINGTON 4 NNW
SLIDELL
NEW ORLEANS AP
NEW ORLEANS ALGIERS
ST BERNARD
ABITA SPRINGS FIRE TOWER
Latitude
30.7268
31.0929
30.5372
31.1632
31.0943
30.8178
30.7531
30.5590
30.3644
30.4839
30.9434
30.7111
30.7094
30.0717
30.0565
31.1829
30.5273
30.2651
29.9934
29.9519
29.8722
30.4397
Longitude
-91.3671
-91.2327
-91.1469
-90.8944
-91.0686
-90.9732
-90.9938
-90.9856
-91.1671
-90.4731
-90.5117
-90.7519
-90.5250
-91.0275
-90.5802
-90.4707
-90.1114
-89.7697
-90.2510
-90.0502
-89.8299
-90.0464
Temperature
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Elevation (ft)
14
122
20
105
113
61
46
18
6
27
70
58
52
9
5
126
12
3
1
1
2
9
1-11
-------
Watershed Segmentation
The Pontchartrain watershed was divided into 37 subwatersheds for the purposes of modeling (Figure 3). The
initial calibration watersheds correspond to the USGS gages shown on the figure. The model encompasses only
complete watersheds that drain to Lake Pontchartrain or the Mississippi and does not require specification of any
upstream boundary conditions for application.
1-12
-------
Legend
USGS gages
— Hydrography
^^= Interstate
^H Water (Nat. Atlas Dataset)
US Census Populated Places
I I County Boundaries
T Model Subbasins
19 / 18 I
Pontchartrain
USGS 07378500
GCRP Model Areas - Pontchartrain Basin
Model Segmentation
NAD 1983 Albers meters - Map produced 03-31-2011 - P. Cada
Figure 3. Model segmentation and USGS stations utilized for the Pontchartrain watershed
1-13
-------
Calibration Data and Locations
Only limited flow gaging is available in the watershed. The specific site chosen for initial calibration was the
Amite River near Denham Springs (USGS 07378500). The Amite watershed was selected because there is a good
set of flow and water quality data available and the watershed lacks major point sources and impoundments.
Calibration and validation was ultimately pursued at several locations (Table 5). Parameters derived on the Amite
were not fully transferable to other portions of the watershed, and additional refinements to the calibration were
conducted at the other gage locations.
Table 5. Calibration and validation locations in the Pontchartrain watershed
Station name
Amite River near Denham Springs, LA
Tangipahoa River at Robert, LA
Tickfaw River at Holden, LA
USGS ID
07378500
07375500
07376000
Drainage area
(mi2)
1,280
646
247
Hydrology
calibration
X
X
X
Water quality
calibration
X
X
The model hydrology calibration period was set to Water Years 1995-2004 (within the 30-year period of record
for modeling). Hydrologic validation was then performed on Water Years 1985-1994. Water quality calibration
used available data for calendar years 1984-1999. Insufficient water quality data were available for a separate
validation time period.
1-14
-------
SWAT Modeling
Assumptions
The modeled portion of the Pontchartrain watershed does not contain major dams or impoundments. It does,
however, contain large amounts of low lying land with swamp land cover and high water tables which may not be
a good fit for SWAT's curve number approach to hydrology. The watershed is also characterized by the presence
of many canals and distributary streams that complicate the flow of water. The USGS gages are located in more
upland areas where these issues are of less importance, but the extrapolation of calibration parameters to
downstream, swampy areas may be suspect.
Due to the flat topography, the boundary between land and water is often ill-defined in this watershed and is
changing over time in response to storms and sea level rise. This modeling exercise does not address the changes
in topography and hydrology that have occurred or will occur in the basin; instead, fixed conditions associated
with the 2001 NLCD are assumed.
Hydrology Calibration
A partial spatial calibration approach was adopted for GCRP-SWAT modeling for the Pontchartrain watershed,
with calibration at three USGS gages with long periods of record. The majority of the calibration effort was
geared toward getting a closer match between simulated and observed flows at the outlet of calibration focus area.
The calibration focus area (Amite River) includes seven subwatersheds and is representative of the general land
use characteristics of the more upstream portions of the watershed. The parameters were adjusted within the
recommended ranges to obtain reasonable fit between the simulated and measured flows in terms of Nash-
Sutcliffe modeling efficiency and the high flow and low flow components as well as the seasonal flows.
The average annual water balance of the entire Pontchartrain watershed predicted by the SWAT model over the
32-year simulation period is as follows:
PRECIP = 1658.2 MM
SNOW FALL = 8.87 MM
SNOW MELT = 8.74 MM
SUBLIMATION = 0.12 MM
SURFACE RUNOFF Q = 699.14 MM
LATERAL SOIL Q = 2.03 MM
TILE Q = 0.00 MM
GROUNDWATER (SHAL AQ) Q = 158.64 MM
REVAP (SHAL AQ => SOIL/PLANTS) = 29.16 MM
DEEP AQ RECHARGE = 181.07 MM
TOTAL AQ RECHARGE = 368.91 MM
TOTAL WATER YLD = 639.74 MM
PERCOLATION OUT OF SOIL = 155.42 MM
ET = 790. 9 MM
PET = 1522.1MM
TRANSMISSION LOSSES = 220.07 MM
Hydrologic calibration adjustments focused on the following parameters:
• CN2 (initial SCS runoff curve number for moisture condition II)
• ESCO (soil evaporation compensation factor)
• SURLAG (surface runoff lag coefficient)
1-15
-------
SOL_AWC (available water capacity of the soil layer, mm water/mm of soil)
ALPHA_BF (baseflow alpha factor, days)
GW_DELAY (groundwater delay time, days)
GWQMIN (threshold depth of water in the shallow aquifer required for return flow to occur, mm)
GW_REVAP (groundwater "revap" coefficient)
CH_N1 (Manning's "n" value for tributary channels)
CH_N2 (Manning's "n" value for main channels)
CH_K1 (Effective hydraulic conductivity in tributary channel alluvium)
CH_K2 (Effective hydraulic conductivity in main channel alluvium)
SFTMP (Snowfall temperature)
SMTMP (Snowmelt base temperature)
SMFMX (Maximum melt rate for snow during the year)
SMFMN (Minimum melt rate for snow during the year)
The same general area was modeled with SWAT by Wu and Xu (2006). While the 20 Watershed model did not
adopt parameter values directly from this paper, the results and quality of model fit are generally similar.
Calibration was performed for the period of water year 1995 - 2003. Results for the Amite River are summarized
in the following figures and table (Figure 4, Figure 5, Figure 6, Figure 7, and Table 6). The overall quality of fit is
good to excellent.
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1994 to 9/30/2004)
Avg Modeled Flow (Same Period)
10000 ---
I
"- 5000 - -
O-94
O-03
Figure 4. Mean monthly flow at USGS 07378500 Amite River near Denham Springs, LA - calibration
period
1-16
-------
• Avg Flow (10/1 /1994 to 9/30/2004)
Line of Equal Value
Best-Fit Line
5000
y = Oi8727x HJ 234.46
R2 = 0.9905
.j • j.
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1994 to 9/30/2004)
Avg Modeled Flow (Same Period)
4000
Figure 5.
1000 2000 3000 4000 5000 10 11 12 1 234567
Average Observed Flow (cfs) Month
Seasonal regression and temporal aggregate at USGS 07378500 Amite River near Denham
Springs, LA - calibration period
• Observed (25th, 75th)
•Median Observed Flow (10/1/1994 to 9/30/2004)
Average Monthly Rainfall (in)
Modeled (Median, 25th, 75th)
5000 i
4000
10 11
to
a:
Figure 6. Seasonal medians and ranges at USGS 07378500 Amite River near Denham Springs, LA -
calibration period
1-17
-------
•Observed Flow Duration (10/1/1994 to 9/30/2004 ;
Modeled Flow Duration (10/1/1994 to 9/30/2004 )
10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 7. Flow exceedance at USGS 07378500 Amite River near Denham Springs, LA - calibration period
1-18
-------
Figure 8. Summary statistics at USGS 07378500 Amite River near Denham Springs, LA - calibration
period
REACH OUTFLOW FROM OUTLET(S) 15, 21
10-Year Analysis Period: 10/1/1994 - 9/30/2004
Flow/volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
21.98
12.53
2.62
2.57
4.02
9.21
6.19
12.81
0.97
Error Statistics
-1.61
0.29
-4.83
7.32
10.13 »
-6.17
-4.61
-6.92
4.05
0.789
0.619
0.950
Observed Flow Gage
USGS 07378500 Amite River near
Hydrologic Unit Code: 8070202
Latitude: 30.464079
Longitude: -90.99038
Drainage Area (sq-mi): 1280
Denham Springs, LA
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3J:
Observed Spring Flow Volume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30_
30
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
22.34
13.16
2.62
2.39
3.65
9.82
6.49
13.76
0.93
Clear [
Hydrology Validation
Hydrology validation for the Amite River was performed for the period 10/1/1984 through 9/30/1994. Results are
presented in Figure 9, Figure 10, Figure 11, Figure 12, and Table 6. The validation also achieves a high quality of
fit, but does overpredict the average flows in summer. This is apparently associated with several tropical storms.
1-19
-------
20000
15000
10000
5000
S-84
M-86
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1984 to 9/30/1994)
Avg Modeled Flow (Same Period)
S-87
M-89
Month
S-90
M-92
S-93
Figure 9. Mean monthly flow at USGS 07378500 Amite River near Denham Springs, LA - validation
period
• Avg Flow (10/1 /1984 to 9/30/1994)
Line of Equal Value
Best-Fit Line
8000
6000
T3
•f 4000
T3
O
y = 0.8067x+ 482.05
. V
2000 4000 6000
Average Observed Flow (cfs)
8000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1984 to 9/30/1994)
Avg Modeled Flow (Same Period)
8000
6000
•e
s"^*n-i
1
4000 -
2000
Sepi Oct\No\A,Dec\Jan\Feb\Mar\Apr\May\Jun\ Jul \
Figure 10. Seasonal regression and temporal aggregate at USGS 07378500 Amite River near Denham
Springs, LA - validation period
1-20
-------
• Observed (25th, 75th) Average Monthly Rainfall (in) - Median Observed Flow (9/1/1984 to 9/30/1994) Modeled (Median, 25th, 75th)
8000
Figure 11. Seasonal medians and ranges at USGS 07378500 Amite River near Denham Springs, LA -
validation period
•Observed Flow Duration (9/30/1984 to 9/30/1994
Modeled Flow Duration (9/30/1984 to 9/30/1994 )
284
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 12. Flow exceedance at USGS 07378500 Amite River near Denham Springs, LA - validation period
1-21
-------
Table 6. Summary statistics at USGS 07378500 Amite River near Denham Springs, LA - validation
period
REACH OUTFLOW FROM OUTLET(S) 15, 21
10-Year Analysis Period: 10/1/1984 - 9/30/1994
Flow/volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
27.52
14.04
3.38
4.53
5.04
12.19
5.76
15.89
2.25
Error Statistics
-0.93
-7.29
-5.54
32.16
Observed Flow Gage
USGS 07378500 Amite River near
Hydrologic Unit Code: 8070202
Latitude: 30.464079
Longitude: -90.99038
Drainage Area (sq-mi): 1280
Denham Springs, LA
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall FlowVolume (10-12):
Observed Winter FlowVolume (1-3):
Observed Spring FlowVolume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
6.20 » | 30
-7.94
-9.49
-4.31
43.74
0.693
0.567
0.899
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
___pBH
27.78
14.86
3.65
3.43
4.74
13.24
6.37
16.60
1.57
^ .
Clear [
Hydrology Results for Larger Watershed
In addition to Amite River, calibration and validation was pursued at a two other gages in the watershed. Similar
to the Amite River, these are in upland areas; there is no gaging of a single pour point of the watershed.
Calibration results were acceptable at all gages (Table 7).
Results of the validation exercise are summarized in Table 8. Problems similar to those experienced on the Amite
River gage were seen at all the tributary gages, with overprediction of seasonal flows in summer. However, as
noted above, this is likely due to the use of land use and model parameters that are more reflective of current
conditions and is not believed to present a bar to application of the model.
1-22
-------
Table 7. Summary statistics (percent error): all stations - calibration period
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error -
Summer:
Seasonal volume error - Fall:
Seasonal volume error -
Winter:
Seasonal volume error -
Spring:
Error in storm volumes:
Error in summer storm
volumes:
Nash-Sutcliffe Coefficient of
Efficiency, E:
Monthly Nash-Sutcliffe
Coefficient
07378500 Amite River near
Denham Springs, LA
-1.61
0.29
-4.83
7.32
10.13
-6.17
-4.61
-6.92
4.05
0.789
0.950
07375500 Tangipahoa River
at Robert, LA
2.89
-8.14
-5.50
20.46
4.34
-0.45
-4.40
-12.99
18.66
0.644
0.900
07376000 Tickfaw River at
Holden, LA
-0.81
-9.73
-6.86
33.29
13.05
-9.84
-10.75
4.38
82.23
0.481
0.778
Table 8. Summary statistics: all stations - validation period
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error -
Summer:
Seasonal volume error - Fall:
Seasonal volume error -
Winter:
Seasonal volume error -
Spring:
Error in storm volumes:
Error in summer storm
volumes:
Nash-Sutcliffe Coefficient of
Efficiency, E:
Monthly Nash-Sutcliffe
Coefficient
07378500 Amite River near
Denham Springs, LA
-0.93
-7.29
-5.54
32.16
6.20
-7.94
-9.49
-4.31
43.74
0.693
0.899
07375500 Tangipahoa River
at Robert, LA
4.45
-4.83
-4.59
20.94
15.67
0.24
-7.16
-11.28
23.91
0.564
0.850
07376000 Tickfaw River at
Holden, LA
-6.39
-1.25
-13.26
24.25
1.25
-15.19
-8.84
-2.76
55.01
0.589
0.840
1-23
-------
Water Quality Calibration and Validation
Initial calibration of water quality was done on the Amite River (USGS 07378500), using data from 1984 - 1994.
Insufficient data were available to support a separate validation period. Instead, the performance of the calibration
parameters was checked against limited additional data collected from the Tangipahoa River.
Calibration adjustments for sediment focused on the following parameters:
• SPCON (linear parameter for estimating maximum amount of sediment that can be re-entrained during
channel sediment routing)
• SPEXP (exponential parameter for estimating maximum amount of sediment that can be re-entrained
during channel sediment routing)
• CH_COV (channel cover factor)
• CH_EROD (channel erodibility factor)
• USLE_P (USLE support practice factor)
Simulated and estimated sediment loads at the Amite River station are shown in Figure 13 and statistics for the
two stations are provided in Table 9. The key statistic in Table 10 is the relative percent error, which shows the
error in the prediction of monthly load normalized to the estimated load. Table 10 also shows the relative average
absolute error, which is the average of the relative magnitude of errors in individual monthly load predictions.
This number is inflated by outlier months in which the simulated and estimated loads differ by large amounts
(which may be as easily due to uncertainty in the estimated load due to limited data as to problems with the
model) and the third statistic, the relative median absolute error, is likely more relevant and shows better
agreement. Overall, the model appears to somewhat underpredict TSS due to the estimated loads associated with
major storm events.
TSS
-Regression Loads
-Simulated Loads
1 4
-*t in to
CO CO CO
O) O
co q>
Figure 13. Fit for monthly load of TSS at USGS 07378500 Amite River near Denham Springs, LA
1-24
-------
Table 9. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
07378500 Amite River near
Denham Springs, LA (1984-1994)
9.2%
36%
9.1%
07375500 Tangipahoa River at
Robert, LA (1984-1 999)
9.0%
65%
14.3%
Calibration adjustments for total phosphorus and total nitrogen focused on the following parameters:
• RHOQ (algal respiration rate at 20° C)
• PHOSKD (phosphorus soil partitioning coefficient)
• PSP (phosphorus availability index)
• RSI (Local algal settlement rate in the reach at 20° C)
• AL1 (Fraction of algal biomass that is nitrogen)
• AL2 (Fraction of algal biomass that is phosphorus)
• MUMAX (Rate of oxygen uptake per unit NO2-N oxidation at 20° C)
• RHOQ (Algal respiration rate at 20° C)
• RS2 (benthic source rate for dissolved P in the reach at 20° C)
• RS3 (Benthic source rate for NFLpN in the reach at 20° C)
• RS5 (organic P settling rate in the reach at 20° C)
• BC4 (rate constant for mineralization of organic P to dissolved P in the reach at 20° C)
• RS4 (rate coefficient for organic N settling in the reach at 20° C)
• CH_ONCO (Channel organic nitrogen concentration)
• CH_OPCO (Channel organic phosphorus concentration)
• SDNCO (Denitrification threshold water content)
• CDN (Denitrification exponential rate constant)
Results for the phosphorus simulation are shown in Figure 14 and Table 10. Results for the nitrogen simulation
are shown in Figure 15 and Table 11. The model fit is generally reasonable for nutrients at both stations.
1-25
-------
1000
100
Total P
- Regression Loads
-Simulated Loads
Figure 14. Fit for monthly load of total phosphorus at USGS 07378500 Amite River near Denham Springs,
LA
Table 10. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
07378500 Amite River near
Denham Springs, LA (1984-1994)
2.4%
46%
13.0%
07375500 Tangipahoa River at
Robert, LA (1984-1 999)
-31.2
84%
9.3%
1-26
-------
10,000
1,000
o
Total N
88888888
-Averaging Loads
-Simulated Loads
Figure 15. Fit for monthly load of total nitrogen at USGS 07378500 Amite River near Denham Springs, LA
Table 11. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
07378500 Amite River near
Denham Springs, LA (1984-1994)
-8.9%
54%
24.4%
07375500 Tangipahoa River at
Robert, LA (1984-1 999)
-1.3%
65%
28.7%
1-27
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
Tetra Tech. 1999. Improving Point Source Loadings Data for Reporting National Water Quality Indicators. Final
Technical Report prepared for U.S. Environmental Protection Agency, Office of Waste water Management,
Washington, DC, by Tetra Tech, Inc., Fairfax, VA.
USEPA. 2008. Using the BASINS Meteorological Database (Version 2006). BASINS Technical Note 10.
Office of Water, U.S. Environmental Protection Agency, Washington, DC.
http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf (Accessed June,
2009).
USGS (United States Geological Survey). 2002. National Water Quality Assessment (NAWQA) Program: The
Acadian-Pontchartrain Study Unit http://la.water.usgs.gov/nawqa/ (Accessed June, 2009).
Wu, K. and Y.J. Xu. 2006. Evaluation of the applicability of the SWAT model for coastal watersheds in
southeastern Louisiana. Journal of the American Water Resources Association, 42 (5): 1247-1260.
1-28
-------
Appendix J
Model Configuration, Calibration and
Validation
Basin: Tar and Neuse Rivers (TarNeu)
-------
Contents
Watershed Background J-4
Soil Characteristics J-5
Land Use Representation J-5
Point Sources J-9
Meteorological Data J-10
Watershed Segmentation J-11
Calibration Data and Locations J-13
SWAT Modeling J-14
Assumptions J-14
Hydrology Calibration J-14
Hydrology Validation J-18
Hydrology Results for Larger Watershed J-21
Water Quality Calibration and Validation J-23
Water Quality Results for Larger Watershed J-25
References J-27
J-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes J-6
Table 2. Land use distribution for the Albemarle-Pamlico basin (2001 NLCD) (mi2) J-8
Table 3. Major point source discharges in the Tar andNeuse River basin J-9
Table 4. Precipitation stations for the Tar andNeuse River watershed model J-ll
Table 5. Calibration and validation locations in the Tar andNeuse River basin J-13
Table 6. Summary statistics at USGS 02091500 Contentnea Creek near Hookerton, NC - calibration
period J-18
Table 7. Summary statistics at USGS 02091500 Contentnea Creek near Hookerton, NC - validation
period J-21
Table 8. Summary statistics (percent error): all stations - calibration period J-22
Table 9. Summary Statistics (percent error): All Stations - Validation Period J-22
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 02091500 Contentnea Creek near Hookerton, NC J-24
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 02091500 Contentnea Creek near Hookerton, NC J-25
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 02091500 Contentnea Creek near Hookerton, NC J-25
Table 13. Summary statistics for water quality at all stations - calibration period 1993-2003 J-26
Table 14. Summary statistics for water quality at all stations- validation period 1983-1993 J-26
Figures
Figure 1. Location of the Tar andNeuse River basins J-5
Figure 2. Land use in the Tar andNeuse River basin J-6
Figure 3. Model segmentation and USGS stations utilized for the Tar and Neuse River basin J-12
Figure 4. Mean monthly flow at USGS 02091500 Contentnea Creek near Hookerton, NC - calibration
period J-15
Figure 5. Seasonal regression and temporal aggregate at USGS 02091500 Contentnea Creek near
Hookerton, NC - calibration period J-16
Figure 6. Seasonal medians and ranges at USGS 02091500 Contentnea Creek near Hookerton, NC -
calibration period J-16
Figure 7. Flow exceedance at USGS 02091500 Contentnea Creek near Hookerton, NC - calibration
period J-17
Figure 8. Mean monthly flow at USGS 02091500 Contentnea Creek near Hookerton, NC - validation
period J-19
Figure 9. Seasonal regression and temporal aggregate at USGS 02091500 Contentnea Creek near
Hookerton, NC -validation period J-19
Figure 10. Seasonal medians and ranges at USGS 02091500 Contentnea Creek near Hookerton, NC -
validation period J-20
Figure 11. Flow exceedance at USGS 02091500 Contentnea Creek near Hookerton, NC - validation
period J-20
Figure 12. Fit for monthly load of TSS at USGS 02091500 Contentnea Creek near Hookerton, NC J-23
Figure 13. Fit for monthly load of total phosphorus at USGS 02091500 Contentnea Creek near
Hookerton, NC J-24
Figure 14. Fit for monthly load of total nitrogen at USGS 02091500 Contentnea Creek near Hookerton,
NC J-25
J-3
-------
Watershed Background
The Tar and Neuse River basins, within the Albemarle-Pamlico NAWQA study area, were selected as one of the
15 non-pilot application watersheds for the 20 Watershed study. Watershed modeling for the non-pilot areas is
accomplished using the SWAT model only, and model calibration and validation results are presented in
abbreviated form.
The Tar and Neuse River drainages are located entirely within North Carolina (Figure 1) and drain to two
important estuaries (Pamlico and Neuse Estuaries) that have been impacted by excess nutrient loads. The
watershed covers an area of 9,972 mi2 in 8 HUCSs, all within HUC 0302. The watershed is divided between the
Piedmont, and Coastal Plain physiographic provinces. Land-surface elevations range from about 885 feet above
sea level in the Piedmont northwest of Durham to sea level in the eastern Coastal Plain (McMahon and Lloyd,
1995). Streams descend through the Piedmont province to the Coastal Plain Province (Spruill et al. 1998).
The watershed as a whole is dominated by forested (34 percent) and agricultural crop and pasture land (29
percent). Agricultural land in the study area is used primarily for growing crops (soybeans, corn, wheat, peanuts,
tobacco, and cotton) and raising livestock (chickens, turkeys, hogs, and cattle.)
Less than 10 percent of the watershed consists of developed land, primarily in and around the cities of Raleigh,
Durham, and Greenville, NC are prominent in the eastern third of the watershed and occupy 13 percent of the
study area.
Average annual temperatures in the watershed range from about 58 °F in the western headwaters to slightly more
than 62° F along Pamlico Sound in the eastern part of the Coastal Plain. Average annual precipitation ranges from
about 44 to about 55 inches per year, but can be much greater in years impacted by tropical storms. The highest
average monthly streamflow typically occurs during the months that include the non-growing season when
temperatures are low and evapotranspiration rates are low. The lowest average monthly streamflow occurs during
the growing season when evapotranspiration rates are high. Groundwater is a significant component of the total
water discharged to the Albemarle-Pamlico estuarine system.
The greatest uses of surface water in the Tar and Neuse River drainage basin are for public water supplies and
thermoelectric power. Domestic groundwater use and agricultural surface water use are comparable in size, and
both are slightly less than groundwater use for public water supplies. Surface water use is highest in areas with
large urban populations served by surface water diversions for public water supplies (e.g., Neuse River basin) and
in areas with large commercial, industrial, or mining water users (e.g., the Tar-Pamlico River basin). Groundwater
use is generally highest in the Coastal Plain.
J-4
-------
Legend
— Hydrography
£d| Water (Nat. Atlas Dataset)
US Census Populated Places
^H Municipalities (pop > 50.QOO)
County Boundaries
I I Watershed With HUCSs
Middle Neuse
(03020202)
GCRP Model Areas-Tar and Neuse River Basins
Base Map
Figure 1. Location of the Tar and Neuse River basins.
Soil Characteristics
Soils in the watershed are described in STATSGO soil surveys. SWAT uses information drawn directly from the
soils data layer to populate the model.
Land Use Representation
Land use/cover in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage (Figure
2). NLCD land cover classes were aggregated according to the scheme shown in Table 1 for representation in the
20 Watershed model. SWAT uses the built-in hydrologic response unit (HRU) overlay mechanism in the
ArcSWAT interface. SWAT HRUs are formed from an intersection of land use and SSURGO major soils. The
distribution of land use in the watershed is summarized in Table 2.
J-5
-------
Legend
Hydrography
nterstate
^ County Boundaries
2001 NLCD Land Use
I | Op en water
| Developed, open space
| Developed, low intensity
j^H Developed, medium intensity
^^| Developed, high intensity
I | Barren land
| Deciduous forest
j^H Evergreen forest
I I Mixed forest
| | Scrub/shrub
^ Grassland/herbaceous
I | Pasture/hay
Cultivated crops
Woody wetlands
Emergent herbaceous wetlands
GCRP Model Areas - Tar and Neuse River Basins
Land Use Map
Figure 2. Land use in the Tar and Neuse River basin.
Table 1. Aggregation of NLCD land cover classes
NLCD class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
Comments
Water surface area usually
accounted for as reach area
Deciduous
Evergreen
Mixed
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
J-6
-------
NLCD class
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Emergent & woody wetlands
Aquatic bed wetlands (not emergent)
SWAT class
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
J-7
-------
Table 2. Land use distribution for the Albemarle-Pamlico basin (2001 NLCD) (mi )
HUC8
watershed
Upper Tar
03020101
Fishing
03020102
Lower Tar
03020103
Pamlico
03020104
Upper Neuse
03020201
Middle Neuse
03020202
Content nea
03020203
Lower Neuse
03020204
Total
Open
water
11.6
2.7
6.5
184.5
40.1
10.2
10.2
182.6
448.3
Developed9
Open
space
84.5
42.1
53.0
37.9
244.9
59.7
69.8
47.5
639.4
Low
density
22.8
4.6
17.3
6.3
96.2
19.0
21.8
14.0
201.9
Medium
density
7.9
0.6
6.4
1.3
37.0
6.5
6.8
4.3
70.7
High
density
2.8
0.1
2.0
0.2
9.4
2.2
1.9
1.5
20.3
Barren
land
2.5
1.8
0.4
9.7
4.4
0.8
1.0
0.4
21.0
Forest
599.6
448.3
257.4
259.5
908.0
276.7
274.6
312.6
3,336.6
Shru bland
115.3
76.1
125.6
129.0
200.2
126.9
105.5
114.0
992.6
Pasture/Hay
212.9
75.7
27.0
6.6
288.6
28.6
82.3
4.9
726.4
Cultivated
157.1
153.7
330.5
262.4
287.0
332.0
412.1
170.7
2,105.4
Wetland
87.7
88.4
134.1
277.7
156.4
202.9
156.6
305.6
1 ,409.6
Total
1 ,304.8
894.0
960.2
1,175.2
2,272.1
1,065.4
1,142.6
1,158.1
9,972.4
aThe percent imperviousness applied to each of the developed land uses is as follows: open space (7.17%), low density (30.90%), medium density (61.05%), and high
density (87.31%).
J-8
-------
Point Sources
There are numerous point source discharges in the watershed. Only the major dischargers, generally defined as
those with a design flow greater than 1 MGD are included in the simulation (Table 3). The major dischargers are
represented at long-term average flows, without accounting for changes over time or seasonal variations.
Table 3. Major point source discharges in the Tar and Neuse River basin
NPDES ID
NC0001881
NC0003191
NC0003255
NC0003417
NC0003760
NC0003816
NC0020231
NC0020389
NC0020605
NC0020648
NC0020834
NC0023841
NC0023906
NC0023931
NC0024236
NC0025054
NC0025348
NC0025453
NC0026042
NC0026433
NC0029033
NC0029572
NC0030317
NC0030716
NC0030759
NC0032077
NC0048879
Name
PHILLIPS PLATING COMPANY INC
WEYERHAEUSER N R CO NEW BERN C
PCS PHOSPHATE CO INC AURORA Ml
CP&L CO DBA PROG ENRG CAROLINA
E I DUPONT DE NEMOURS E I DUPO
US MCAS CHERRY PT MCALF ATLANT
LOUISBURGWWTP
BENSON WWTP
TARBORO WWTP
WASHINGTON WWTP
WARRENTON WWTP
DURHAM NORTH DURHAM WRF
WILSON WWTP
GREENVILLE UTIL COMMISSION GUC
KINSTON REGWTR RECLAMATION FA
OXFORD WWTP
NEW BERN WWTP
CLAYTON LITTLE CREEK WWTP
ROBERSONVILLE WWTP
HILLSBOROUGH WWTP
RALEIGH NEUSE RIVER WWTP
FARMVILLEWWTP
ROCKY MOUNT TAR RIVER REG WWTP
JOHNSTON CO DEPARTMENT OF PUBL
RALEIGH SMITH CREEK WWTP
CONTENTNEA METRO SWRG DIS CONT
GARY NORTH GARY WRF
Design flow
(MGD)
0.10
32.00
NA
1.40
3.60
3.50
1.37
1.50
5.00
3.65
2.00
20.00
14.00
17.50
11.85
3.50
7.00
2.50
1.80
3.00
75.00
3.50
21.00
7.00
2.40
2.85
12.00
Observed flow
(MGD)
(1991-2006 average)
0.02
19.74
62.83
2.38
1.58
1.99
0.79
1.50
2.10
1.76
0.36
5.90
10.52
9.08
1.73
1.30
3.26
1.10
1.22
1.15
47.24
2.05
11.76
4.15
0.87
1.49
2.45
J-9
-------
NPDES ID
NC0064050
NC0065102
NC0066516
NC0069311
Name
APEX WATER RECLAMATION FAC
GARY SOUTH GARY WRF
FUQUAYVARINA TERRIBLE CRKWWT
FRANKLIN CO PUBLIC UTIL FRANKL
Design flow
(MGD)
3.60
12.80
1.00
1.00
Observed flow
(MGD)
(1991-2006 average)
4.58
7.69
0.18
0.41
Most of these point sources have reasonably complete monitoring for total phosphorus and total suspended solids
(TSS). Many dischargers in the Tar and Neuse River basin also report total nitrogen (unlike other study areas)
because of concerns over nitrogen impacts on the coastal estuaries. The point sources were initially represented in
the model with the median of reported values for total phosphorus, TSS and total nitrogen.
Meteorological Data
The required meteorological time series for the 20 Watershed SWAT simulations are precipitation and air
temperature. The 20 Watershed simulations do not include water temperature and uses a degree-day method for
snowmelt. SWAT estimates Penmann-Monteith potential evapotranspiration using a statistical weather generator
for inputs other than temperature and precipitation. These meteorological time series are drawn from the
BASINS4 Meteorological Database (USEPA 2008), which provides a consistent, quality-assured set of
nationwide data with gaps filled and records disaggregated. Scenario application requires simulation over 30
years, so the available stations are those with a common 30-year period of record (or one that can be filled from
an approximately co-located station) that covers the year 2001. A total of 40 precipitation stations were identified
for use in the Tar and Neuse River watershed model with a common period of record of 10/1/1973-9/30/2004
(Table 4). Temperature records are sparser; where these are absent temperature is taken from nearby stations with
an elevation correction.
MO
-------
Table 4. Precipitation stations for the Tar and Neuse River watershed model
COOP ID
NC310241
NC310576
NC310674
NC311241
NC311285
NC311606
NC311677
NC311820
NC311881
NC312500
NC312515
NC312827
NC313232
NC313510
NC313555
NC313638
NC313969
NC314684
NC314689
NC314962
NC315123
NC315830
NC316108
NC316135
NC316853
NC317069
NC317074
NC317079
NC317319
NC317395
NC317400
NC317499
NC317516
NC317994
NC318500
NC318706
NC319100
NC319440
NC319476
VA444414
Latitude
36.2912
35.1500
35.4993
36.1279
36.1414
34.9833
35.9086
35.6408
35.0248
35.3247
36.0425
36.1686
36.1050
35.3445
36.0504
35.6401
36.3482
35.1967
35.2975
36.1326
36.1029
34.7337
35.0667
35.4486
35.8722
35.8707
35.7283
35.7945
36.4783
35.9100
35.8936
36.2119
36.3469
35.5164
35.8848
35.0667
35.5554
35.8529
35.6939
36.6003
Longitude
-77.9822
-76.7167
-76.6864
-79.4068
-78.7736
-76.2999
-79.0794
-78.4633
-78.2758
-78.6881
-78.9625
-77.6749
-78.4591
-77.9646
-79.3727
-77.3983
-78.4119
-77.5433
-77.5721
-77.1707
-78.3038
-76.7357
-77.0499
-76.2108
-76.6591
-78.7864
-78.6843
-78.6988
-77.6717
-77.8892
-77.6805
-78.8568
-78.8858
-78.3457
-77.5386
-77.3499
-77.0721
-77.0306
-77.9455
-78.3011
Temperature
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Elevation (m)
101
3
2
195
108
2
152
91
48
61
122
34
114
33
201
10
146
7
18
15
79
3
5
1
6
127
128
122
64
40
34
165
216
46
11
9
3
6
34
76
Watershed Segmentation
The Tar and Neuse River basin was divided into 71 subwatersheds for the purposes of modeling (Figure 3). The
initial calibration watershed (Contentnea Creek) is highlighted. The model encompasses the complete watershed
and does not require specification of any upstream boundary conditions for application.
Ml
-------
USGS Gages
— Hydrography
Interstate
^H Water (Nat. Atlas Dataset)
US Census Populated Places
| | County Boundaries
I I Initial Calibration Watershed
~~| Model Subbasins
GCRP Model Areas - Tar and Neuse River Basins
Model Segmentation
NAD_1983_Albers_meters-Map produced 12-23-2010- P Cada
TETRATECH
Figure 3. Model segmentation and USGS stations utilized for the Tar and Neuse River basin.
J-12
-------
Calibration Data and Locations
The specific site chosen for initial calibration was the Contentnea Creek near Hookerton (USGS 02091500), a
flow and water quality monitoring location that approximately coincides with the mouth of an 8-digit HUC at its
outflow to the Neuse River. The Contentnea Creek watershed was selected because there is a good set of flow and
water quality data available and the watershed lacks major point sources and impoundments. Additional
calibration and validation was pursued at multiple locations (Table 5). Parameters derived on the Contentnea
Creek were not fully transferable to other portions of the Tar and Neuse River basin, and additional calibration
was conducted at multiple gage locations.
Table 5. Calibration and validation locations in the Tar and Neuse River basin
Station name
Contentnea Creek near Hookerton, NC
Neuse River near Falls, NC
Neuse River near Goldsboro, NC
Neuse River at Kinston, NC
Tar River at Tarboro, NC
USGS ID
02091500
02087183
02089000
02089500
02083500
Drainage area
(mi2)
733
771
2399
2692
2183
Hydrology
calibration
X
X
X
X
X
Water quality
calibration
X
X
X
The model hydrology calibration period was set to Water Years 1993-2003 (within the 32-year period of record
for modeling). Hydrologic validation was then performed on Water Years 1983-1993. Water quality calibration
used calendar years 1993-2003, while validation used 1983-1993.
J-13
-------
SWAT Modeling
Assumptions
Falls Lake reservoir is the major impoundment in the Tar and Neuse River study area that was sufficiently large
enough to represent in the model. It is located on the Neuse River. Pertinent reservoir information including
surface area and storage at principal (normal) and emergency spillway levels for the reservoir was obtained from
the US Army Corps of Engineers. The SWAT model provides four options to simulate reservoir outflow:
measured daily outflow, measured monthly outflow, average annual release rate for uncontrolled reservoir, and
controlled outflow with target release. Keeping in view the 20 Watershed climate change impact evaluation
application to future climate scenarios, it was assumed that the best representation of the reservoir was to simulate
it without supplying time series of outflow records. Therefore, the target release approach was used in the GCRP-
SWAT model.
Hydrology Calibration
A spatial calibration approach was adopted for GCRP-SWAT modeling for the Tar and Neuse River basin. A
systematic adjustment of parameters has been adopted and some adjustments are applied throughout the basin.
Most of the calibration efforts were geared towards getting a closer match between simulated and observed flows
at the outlet of calibration focus area.
Land Use/Soil/Slope Definition
A 5/10/5 percent threshold was used for land use/soil/slope in the SWAT model while defining the HRUs. Urban
land use classes were exempted from the HRU overlay thresholds.
The calibration focus area (Contentnea Creek) includes five subwatersheds and is generally representative of the
general land use characteristics of the overall watershed with the exception of a higher percentage of cultivated
lands. The parameters were adjusted within the practical range to obtain reasonable fit between the simulated and
measured flows in terms of Nash-Sutcliffe modeling efficiency and the high flow and low flow components as
well as the seasonal flows.
The water balance of the whole Tar and Neuse River basin predicted by the SWAT model over the 32-year
simulation period is as follows:
PRECIP = 1234.6 MM
SNOW FALL = 40.57 MM
SNOW MELT = 40.17 MM
SUBLIMATION = 0.39 MM
SURFACE RUNOFF Q = 170.61MM
LATERAL SOIL Q = 18.20 MM
TILE Q = 0.00 MM
GROUNDWATER (SHAL AQ) Q = 223.76MM
REVAP (SHAL AQ => SOIL/PLANTS) = 40.29 MM
DEEP AQ RECHARGE = 13.90 MM
TOTAL AQ RECHARGE = 277.94 MM
TOTAL WATER YLD = 410.46 MM
PERCOLATION OUT OF SOIL = 276.02 MM
ET = 763.0 MM
PET = 1433.2MM
TRANSMISSION LOSSES = 2.10 MM
J-14
-------
Hydrologic calibration adjustments focused on the following parameters:
• CN2 (initial SCS runoff curve number for moisture condition II)
• ESCO (soil evaporation compensation factor)
• SURLAG (surface runoff lag coefficient)
• SOL_AWC (available water capacity of the soil layer, mm water/mm of soil)
• ALPHA_BF (baseflow alpha factor, days)
• GW_DELAY (groundwater delay time, days)
• GWQMIN (threshold depth of water in the shallow aquifer required for return flow to occur, mm)
• GW_REVAP (groundwater "revap" coefficient)
• CH_N2 (Manning's "n" value for main channels)
Calibration results for the Contentnea Creek are summarized in Figure 4, Figure 5, Figure 6, Figure 7 and Table 6.
Avg Monthly Rainfall (in)
- Avg Observed Flow (10/1/1993 to 9/30/2003 )
Avg Modeled Flow (Sarre Period)
30
O-93 A-95 O-96 A-98 O-99
Month
A-01
O-02
Figure 4. Mean monthly flow at USGS 02091500 Contentnea Creek near Hookerton, NC - calibration
period.
J-15
-------
AvgFlow (10/1/1993 to 9/30/2003)
• Line of Equal Value
-Best-Fit Line
2500
—2000 -
^
11500
CD
T3
liooo -
CD
D)
ro
S> 500 •}
y = 0.9056x +48.446
Avg Monthly Rainfall (in)
- Avg Observed Flow (10/1/1993 to 9/30/2003)
Avg Modeled Flow (Sarre Period)
500 1000 1500 2000 2500
Average Observed Flow (cfs)
10 11 12 1 23456789
Month
Figure 5. Seasonal regression and temporal aggregate at USGS 02091500 Contentnea Creek near
Hookerton, NC - calibration period.
• Observed (25th, 75th) Average Monthly Rainfall (in) -Median Observed Flow (10/1/1993 to 9/30/2003) Modeled (Median, 25th, 75th)
2500
2000
500
Jan Feb Mar Apr May
10 11 12 1
2
- 3 jo
Figure 6. Seasonal medians and ranges at USGS 02091500 Contentnea Creek near Hookerton, NC -
calibration period.
J-16
-------
•Observed Flow Duration (10/1/1993 to 9/30/2003 )
Modeled Flow Duration (10/1/1993 to 9/30/2003 )
100000
0% 10% 20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90% 100%
Figure 7. Flow exceedance at USGS 02091500 Contentnea Creek near Hookerton, NC - calibration
period.
J-17
-------
Table 6.
period
Summary statistics at USGS 02091500 Contentnea Creek near Hookerton, NC - calibration
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 35
10-Year Analysis Period: 10/1/1993 - 9/30/2003
Flow/volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
15.98
7.16
1.78
4.86
3.79
5.48
1.85
3.81
1.51
Error Statistics
-3.98
-9.98
2.15
30.50
Observed Flow Gage
USGS 02091500 CONTENTNEA CREEK AT HOOKERTON, NC
Hydrologic Unit Code: 3020203
Latitude: 35.42888889
Longitude: -77.5825
Drainage Area (sq-mi): 733
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring Flow Volume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
15.05 » f 30
-16.48
-39.56
-12.39
2.17
0.678
0.459
0.859
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
16.64
7.01
1.97
3.72
3.29
6.56
3.06
4.35
1.48
••
Clear [
Hydrology Validation
Hydrology validation for Contentnea Creek was performed for the period 10/1/1983 through 9/30/1993. Results
are presented in Figures 8 through 11 and Table 7. The validation achieves a moderately high coefficient of model
fit efficiency, but is over on 10 percent highest flow volume, and summer and fall seasonal volumes (Figure 8,
Figure 9, Figure 10, Figure 11 and Table 7).
J-18
-------
Avg Monthly Rainfall (in)
- Avg Observed Flow (10/1/1983 to 9/30/1993 )
Avg Modeled Flow (Sarre Period)
4000
3000 -
1000 -
2000 -
O-83
A-85
O-86
A-88
O-89
A-91
O-92
Month
Figure 8. Mean monthly flow at USGS 02091500 Contentnea Creek near Hookerton, NC - validation
period.
• Avg Flow (10/1/1983 to 9/30/1993)
Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1983 to 9/30/1993)
Avg Modeled Flow (Sarre Period)
2000
•6
I
1500 -
T3
|iooo
o
500 -
2000
500 1000 1500 2000
Average Observed Flow (cfs)
10 11 12 1 2345
Month
67
Figure 9. Seasonal regression and temporal aggregate at USGS 02091500 Contentnea Creek near
Hookerton, NC - validation period.
J-19
-------
• Observed (25th, 75th) Average Monthly Rainfall (in) -Median Observed Flow (10/1/1983 to 9/30/1993) Modeled (Median, 25th, 75th)
2500
2000
Jan Feb Mar Apr May
10 11 12 1
Figure 10. Seasonal medians and ranges at USGS 02091500 Contentnea Creek near Hookerton, NC
validation period.
•Observed Flow Duration (10/1/1983 to 9/30/1993 )
Modeled Flow Duration (10/1/1983 to 9/30/1993 )
10000
0% 10% 20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90% 100%
Figure 11. Flow exceedance at USGS 02091500 Contentnea Creek near Hookerton, NC - validation
period.
J-20
-------
Table 7. Summary statistics at USGS 02091500 Contentnea Creek near Hookerton, NC - validation
period
REACH OUTFLOW FROM OUTLET 35
10-Year Analysis Period: 10/1/1983 - 9/30/1993
Flow/volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
14.24
5.76
1.79
3.65
2.88
4.91
2.80
3.19
0.99
Error Statistics
-1.18
Error in 50% lowest flows: 12.17
Error in 10% highest flows: -3.49
Seasonal volume error - Summer:
Seasonal volume error - Fall:
39.76
35.01 >
Seasonal volume error - Winter: -15.83
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
-26.94
-13.04
2.93
0.635
0.479
0.742
Observed Flow Gage
USGS 02091500 CONTENTNEA CRI
Hydrologic Unit Code: 3020203
Latitude: 35.42888889
Longitude: -77.5825
Drainage Area (sq-mi): 733
EEK AT HOOKERTON, NC
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall FlowVolume (10-12):
Observed Winter FlowVolume (1-3):
Observed Spring FlowVolume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
> f 30
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
14.41
5.97
1.59
2.61
2.13
5.84
3.83
3.66
0.96
^ .
Clear |~
Hydrology Results for Larger Watershed
As described above, parameters determined for the gage at Contentnea Creek were initially transferred to other
gages in the watershed. However, changes to subbasin level parameter were required to fit the model to the
observed flows. In all, calibration and validation was pursued at a total of five gages throughout the watershed,
including one gage at the outlet of an 8-digit HUC, one gage at the outfall of the Falls Lake reservoir and three
gages on the mainstem. Results of the calibration and validation exercise are summarized in Table 8 and Table 9,
respectively. Calibration and validation results were acceptable at most gages.
J-21
-------
Table 8. Summary statistics (percent error): all stations - calibration period
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient of Efficiency,
E:
Monthly Nash-Sutcliffe Coefficient
02091500
Contentnea
Creek near
Hookerton,
NC
-3.98
-9.98
2.15
30.50
15.05
-16.48
-39.56
-12.39
2.17
0.678
0.859
02087183
Neuse River
near Falls
Lake, NC
5.36
-11.43
-11.68
30.71
21.62
3.54
-20.48
-61.90
-29.11
0.417
0.719
02089000
Neuse River
near
Goldsboro,
NC
-3.95
-2.44
-1.26
20.87
7.54
-11.38
-26.07
-14.75
3.89
0.736
0.864
02089500
Neuse River
at Kinston,
NC
-3.00
-0.14
1.23
22.20
8.74
-9.95
-26.52
-10.38
2.19
0.732
0.859
02083500
Tar River at
Tarboro, NC
0.23
0.92
-6.55
18.69
29.36
-11.81
-21.93
-33.68
-15.34
0.754
0.894
Table 9. Summary Statistics (percent error): All Stations - Validation Period
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient of Efficiency,
E:
Monthly Nash-Sutcliffe Coefficient
02091500
Contentnea
Creek near
Hookerton,
NC
-1.18
12.17
-3.49
39.76
35.01
-15.83
-26.94
-13.04
2.93
0.635
0.742
02087183
Neuse River
near Falls
Lake, NC
-1.02
-29.70
-26.58
5.54
29.41
-7.93
-7.10
-63.68
-46.92
0.620
0.845
02089000
Neuse River
near
Goldsboro,
NC
-7.60
0.60
-20.17
14.56
16.15
-14.24
-22.37
-28.64
-22.43
0.775
0.849
02089500
Neuse River
at Kinston,
NC
-8.55
-2.55
-20.42
14.73
12.09
-14.46
-24.12
-27.30
-20.82
0.767
0.832
02083500
Tar River at
Tarboro, NC
-4.88
-9.07
-20.23
39.72
26.83
-16.29
-22.16
-37.61
-7.61
0.688
0.808
J-22
-------
Water Quality Calibration and Validation
Initial calibration and validation of water quality was done on the Contentnea Creek (USGS 02091500), using
1993-2003 for calibration and 1983-1993 for validation. As with hydrology, water quality calibration was
performed on the later period as this better reflects the land use included in the model.
Calibration adjustments for sediment focused on the following parameters:
• SPCON (linear parameter for estimating maximum amount of sediment that can be re-entrained during
channel sediment routing)
• SPEXP (exponential parameter for estimating maximum amount of sediment that can be re-entrained
during channel sediment routing)
• CH_COV (channel cover factor)
• CH_EROD (channel erodibility factor)
• USLE_P (USLE support practice factor)
Simulated and estimated sediment loads at the Contentnea Creek station for both the calibration and validation
periods are shown in Figure 12 and statistics for the two periods are provided separately in Table 10. The key
statistic in Table 10 is the relative percent error, which shows the error in the prediction of monthly load
normalized to the estimated load. Table 10 also shows the relative average absolute error, which is the average of
the relative magnitude of errors in individual monthly load predictions. This number is inflated by outlier months
in which the simulated and estimated loads differ by large amounts (which may be as easily due to uncertainty in
the estimated load due to limited data as to problems with the model) and the third statistic, the relative median
absolute error, is likely more relevant and shows better agreement.
TSS
100,000
-Regression Loads
-Simulated Loads
Figure 12. Fit for monthly load of TSS at USGS 02091500 Contentnea Creek near Hookerton, NC.
J-23
-------
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 02091500 Contentnea Creek near Hookerton, NC
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1993-2003)
-19.9%
81%
34.7%
Validation period
(1983-1993)
9.9%
61%
33.1%
Calibration adjustments for total phosphorus and total nitrogen focused on the following parameters:
• RHOQ (algal respiration rate at 20° C)
• PHOSKD (phosphorus soil partitioning coefficient)
• PSP (phosphorus availability index)
• RS2 (benthic source rate for dissolved P in the reach at 20° C)
• RS5 (organic P settling rate in the reach at 20° C)
• BC4 (rate constant for mineralization of organic P to dissolved P in the reach at 20° C)
• RS4 (rate coefficient for organic N settling in the reach at 20° C)
Results for the phosphorus simulation are shown in Figure 13 and Table 11. Results for the nitrogen simulation
are shown in Figure 14 and Table 12. The model fit is generally acceptable.
Total P
1000
100 -
10 -
0.1 -
0.01
-Regression Loads
Simulated Loads
Figure 13. Fit for monthly load of total phosphorus at USGS 02091500 Contentnea Creek near Hookerton,
NC.
J-24
-------
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 02091500 Contentnea Creek near Hookerton, NC
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1993-2003)
15.9%
69%
49.3%
Validation period
(1983-1993)
5.3%
66%
39.3%
o
I
1,400
Total N
-Averaging Loads
-Simulated Loads
0°
0°
0°
v v v v
0° O^ 0° 0° 0° 0° 0° 0°
v v v
0° O^ 0°
0°
Figure 14. Fit for monthly load of total nitrogen at USGS 02091500 Contentnea Creek near Hookerton,
NC.
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 02091500 Contentnea Creek near Hookerton, NC
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1993-2003)
-5.6%
56%
24.3%
Validation period
(1983-1993)
5.3%
57%
31.6%
Water Quality Results for Larger Watershed
As with hydrology, a spatial calibration approach was adopted. Contentnea Creek watershed SWAT model
parameters for water quality were transferred to other portions of the watershed with necessary changes to
subwatershed level parameters. Summary statistics for the SWAT water quality calibration and validation at other
stations in the watershed are provided in Table 13 and Table 14.
J-25
-------
Table 13. Summary statistics for water quality at all stations - calibration period 1993-2003
Station
Relative Percent Error TSS Load
Relative Percent Error TP Load
Relative Percent Error TN Load
02091500
Contentnea Creek
near Hookerton, NC
-19.9%
15.9%
-5.6%
02089500
Neuse River
at Kinston, NC
-6.7%
-10.5%
-15%
02083500
Tar River
at Tarboro, NC
-5.1%
-0.4%
-33.8%
Table 14. Summary statistics for water quality at all stations - validation period 1983-1993
Station
Relative Percent Error TSS Load
Relative Percent Error TP Load
Relative Percent Error TN Load
02091500
Contentnea Creek
near Hookerton, NC
9.9%
5.3%
5.3%
02089500
Neuse River
at Kinston, NC
17.3%
2.5%
6.7%
02083500
Tar River
at Tarboro, NC
22.9%
13.1%
13.5%
J-26
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
Spruill, T.B., D.A. Harned, P.M. Ruhl, J.L. Eimers, G. McMahon, K.E. Smith, D.R. Galeone, and M.D.
Woodside. 1998. Water Quality in the Albemarle-Pamlico Drainage Basin, North Carolina and Virginia, 1992-
95. Circular 1157. U.S. Geological Survey, Denver, CO.
USEPA. 2008. Using the BASINS Meteorological Database (Version 2006). BASINS Technical Note 10.
Office of Water, U.S. Environmental Protection Agency, Washington, DC.
http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf (Accessed June,
2009).
J-27
-------
Appendix K
Model Configuration, Calibration and
Validation
Basin: Nebraska: Loup and Elkhorn
Rivers (Neb)
K-l
-------
Contents
Watershed Background K-4
Water Body Characteristics K-4
Soil Characteristics K-5
Land Use Representation K-5
Point Sources K-9
Meteorological Data K-9
Watershed Segmentation K-11
Calibration Data and Locations K-12
SWAT Modeling K-14
Assumptions K-14
Hydrology Calibration K-14
Hydrology Validation K-18
Hydrology Results for Larger Watershed K-21
Water Quality Calibration and Validation K-23
Water Quality Results for Larger Watershed K-26
References K-28
K-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes K-7
Table 2. Land use distribution for the Loup and Elkhorn River basins (2001 NLCD) (mi2) K-8
Table 3. Major point source discharges in the Loup and Elkhorn River basins K-9
Table 4. Precipitation stations for the Loup and Elkhorn River basins model K-9
Table 5. Calibration and validation locations in the Loup and Elkhorn River basins K-13
Table 6. Summary statistics at USGS 06800500 Elkhorn River at Waterloo, NE - calibration period .. K-18
Table 7. Summary statistics at USGS 06800500 Elkhorn River at Waterloo, NE - validation period.... K-21
Table 8. Summary statistics (percent error): all stations - calibration period K-22
Table 9. Summary statistics (percent error): all stations - validation period K-23
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 06800500 Elkhorn River at Waterloo, NE K-24
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 06800500 Elkhorn River at Waterloo, NE K-25
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 06800500 Elkhorn River at Waterloo, NE K-26
Table 13. Summary statistics for water quality at all stations - calibration period 1993 -2002 K-27
Figures
Figure 1. Location of the Loup and Elkhorn River basins K-5
Figure 2. Land use in the Loup and Elkhorn River basins K-6
Figure 3. Model segmentation and USGS stations utilized for the Loup and Elkhorn River basins K-12
Figure 4. Mean monthly flow at USGS 06800500 Elkhorn River at Waterloo, NE - calibration
period K-16
Figure 5. Seasonal regression and temporal aggregate at USGS 06800500 Elkhorn River at Waterloo,
NE - calibration period K-16
Figure 6. Seasonal medians and ranges at USGS 06800500 Elkhorn River at Waterloo, NE - calibration
period K-17
Figure 7. Flow exceedance at USGS 06800500 Elkhorn River at Waterloo, NE - calibration period K-17
Figure 8. Mean monthly flow at USGS 06800500 Elkhorn River at Waterloo, NE - validation period . K-19
Figure 9. Seasonal regression and temporal aggregate at USGS 06800500 Elkhorn River at Waterloo,
NE - validation period K-19
Figure 10. Seasonal medians and ranges at USGS 06800500 Elkhorn River at Waterloo, NE - validation
period K-20
Figure 11. Flow exceedance at USGS 06800500 Elkhorn River at Waterloo, NE - validation period K-20
Figure 12. Fit for monthly load of TSS at USGS 06800500 Elkhorn River at Waterloo, NE K-24
Figure 13. Fit for monthly load of total phosphorus at USGS 06800500 Elkhorn River at Waterloo,
NE K-25
Figure 14. Fit for monthly load of total nitrogen at USGS 06800500 Elkhorn River at Waterloo, NE K-26
K-3
-------
The Loup and Elkhorn River basins within the Central Nebraska NAWQA study area were selected as one of the
15 non-pilot application watersheds for the 20 Watershed study. Watershed modeling for the non-pilot areas is
accomplished using the SWAT model only, and model calibration and validation results are presented in
abbreviated form.
Water Body Characteristics
The Loup and Elkhorn River basins are tributary to the Platte River (Huntzinger and Ellis, 1993). Together they
include 14 HUCSs within HUC 1021 and 1022 and cover approximately 22,100 mi2 (Figure 1). The Loup River
basin includes the North Loup, Middle Loup, and South Loup Rivers, as well as Calamus River, Cedar River,
Dismal River, and Mud Creek. Major tributaries of the Elkhorn River include the North Fork Elkhorn River and
Logan Creek.
The watersheds are located in the Central Plains ecoregion (Huntzinger and Ellis, 1993). The Loup River and its
major tributaries originate in the Nebraska Sandhills, a region of steep grass-covered dunes, and then flow
through dissected plains with broad valleys. Permeable soils and subsurface materials in the Loup River basin
provide flows sustained by shallow groundwater and little if any runoff. The Elkhorn River, in the eastern and
northeastern part of the watershed, flows through rolling hills and well-defined valleys of stable glacial material in
the Western Corn Belt Plains except where it originates in the Sandhills. Runoff in the Elkhorn basin is the largest
in the watershed because of the steeper slopes and fine-grained soils. The city of Omaha lies just outside the
watershed. The portion of the watershed along the eastern boundary is influenced by the Omaha suburban area
and is located near the mouth of the Platte River. Most of the water in the watershed is consumed by irrigation or
used for power generation and returned to the stream for reuse. The water used for irrigation is primarily from
groundwater. The few urban areas within the watershed use groundwater as a municipal water supply. The city of
Omaha obtains part of its water supply from wells in the Elkhorn and Platte River Valleys.
The watersheds are dominated by rural areas. The land use is predominantly pasture and rangeland (66 percent)
and croplands of row-cropped feed grains (27 percent). Groundwater development for irrigation has increased the
productivity of agriculture in the valleys and uplands. Large areas have soils well suited to cultivated crops
whereas other large areas are not suited to crops but to productive grasslands. Counties that are primarily cropland
agriculture without urban areas have population densities of 50 persons per square mile or less. Areas in the west
that are primarily rangeland have population densities of less than five persons per square mile.
The central Nebraska climate ranges from semiarid in the northwest to subhumid in the east. Hot summers, cold
winters, and large daily and annual variations in temperature are typical. Precipitation is greatest in May and June.
Mean annual precipitation varies from about 18 inches in the western part of the watershed to about 30 inches in
the eastern part. Most of the study unit has at least 20 inches of annual precipitation, and more than one-half
occurs during the growing season, April through September. Snowfall is a dominant climatic characteristic of
central Nebraska. Mean annual snowfall ranges from about 25 inches in the southeast to about 35 inches in the
northwest.
K-4
-------
Legend
- Hydrography
BB Water (Nat. Alias Datasel)
US Census Populated Places
^H Municipalities (pop > 50.000)
I I County Boundaries
I I Watershed with HUCSs
Upper Middle Loup__
(10210001)
GCRP Model Areas - Nebraska. Loup and Elkhom River Basins
Base Map
NACJK3.Altars f™«r*
Figure 1. Location of the Loup and Elkhorn River basins
Soil Characteristics
Soils in the watershed, as described in STATSGO soil surveys, fall primarily into hydrologic soil groups (HSGs)
A (high infiltration capacity) and B (moderately high infiltration capacity). SWAT uses information drawn
directly from the soils data layer to populate the model.
Land Use Representation
Land use/cover in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage and is
predominantly rangeland in the northwest and row crop agriculture in the south and east (Figure 2). NLCD land
cover classes were aggregated according to the scheme shown in Table 1 for representation in the 20 Watershed
model. SWAT uses the built-in hydrologic response unit (HRU) overlay mechanism in the ArcSWAT interface.
SWAT HRUs are formed from an intersection of land use and SSURGO major soils. The distribution of land use
in the watershed is summarized in Table 2.
K-5
-------
Hydrography
nterstate
| | County Boundaries
2001 NLCD Land Use
| | Open water
^ Developed, open space
Developed, low intensity
j^H Developed, medium intensity
^^| Developed, high intensity
I | Barren land
^^| Deciduous forest
^^| Evergreen forest
I I Mixed forest
I | Scrub/shrub
^ Grassland/herbaceous
| Pasture/hay
^ Cultivated crops
I | Woody wetlands
n Emergent herbaceous wetJands
GCRP Model Areas - Central Nebraska River Basins
Land Use Map
Figure 2. Land use in the Loup and Elkhorn River basins.
K-6
-------
Table 1. Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area usually
accounted for as reach area
Deciduous
Evergreen
Mixed
Emergent & woody wetlands
Aquatic bed wetlands (not
emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
K-7
-------
Table 2. Land use distribution for the Loup and Elkhorn River basins (2001 NLCD) (mi )
HUC8
watershed
Upper
Elkhorn
10220001
North Fork
Elkhorn
10220002
Lower
Elkhorn
10220003
Logan
10220004
Upper
Middle Loup
10210001
Dismal
10210002
Lower
Middle Loup
10210003
South Loup
10210004
Mud
10210005
Upper North
Loup
10210006
Lower North
Loup
10210007
Calamus
10210008
Loup
10210009
Cedar
10210010
Total
Open
water
25.2
2.9
17.6
2.7
25.9
8.8
19.6
4.1
0.3
25.7
9.8
22.3
12.4
6.5
183.8
Developed9
Open
space
71.8
34.6
86.9
40.5
9.0
2.8
48.7
42.3
26.4
10.5
32.1
2.7
51.1
28.5
488.0
Low
density
10.9
7.6
22.2
10.1
3.4
0.3
9.4
5.5
8.1
0.9
6.1
0.1
9.9
3.1
97.4
Medium
density
2.1
1.5
4.4
0.9
0.3
0.0
1.1
0.3
1.3
0.0
0.7
0.0
1.3
0.2
14.3
High
density
1.1
0.8
1.6
0.4
0.1
0.0
0.2
0.1
0.3
0.0
0.2
0.0
0.6
0.1
5.5
Barren
land
2.6
0.1
0.2
0.0
0.7
1.6
1.0
0.8
0.3
0.9
0.4
1.0
1.6
1.2
12.5
Forest
22.5
17.0
45.2
7.1
4.7
14.8
21.7
14.8
9.4
3.1
31.4
1.3
26.0
14.7
233.8
Shru bland/
Grassland
1,724.7
198.0
311.1
111.2
1 ,955.0
1,711.6
1,310.2
1,133.3
488.8
2,181.3
643.8
904.9
678.1
897.7
14,249.7
Pasture/Hay
129.4
1.2
20.4
10.9
3.3
4.0
10.9
10.5
7.0
3.3
10.6
1.1
28.6
9.2
250.4
Cultivated
719.9
580.8
1,667.5
866.1
2.5
6.2
347.9
335.1
195.0
20.4
224.2
7.1
663.5
226.2
5,862.5
Wetland
182.6
4.2
29.4
3.1
85.7
40.3
38.7
34.0
3.9
103.4
18.5
50.7
60.5
42.6
697.4
Total
2,892.8
848.6
2,206.6
1,053.1
2,090.6
1,790.5
1,809.5
1,580.8
740.8
2,349.4
977.9
991.3
1,533.7
1,229.9
22,095.4
aThe percent imperviousness applied to each of the developed land uses is as follows: open space (8.34%), low density (29.68%), medium density (60.14%), and high
density (86.59%).
K-8
-------
Point Sources
There are several point source discharges in the watershed. Only the major dischargers, primarily those with a
design flow greater than 1 MGD, are included in the simulation (Table 3). The major dischargers are represented
at long-term average flows, without accounting for changes over time or seasonal variations.
Table 3. Major point source discharges in the Loup and Elkhorn River basins
NPDES ID
NE0000761
NE0001392
NE0028363
NE 0031381
NE0033421
NE0035025
NE01 11287
Name
TYSON FRESH MEATS INC W POINT
TYSON FRESH MEATS, INC.
TYSON FRESH MEAT INC. MADISON
FREMONT WWTF
NORFOLK WWTF
COLUMBUS WWTF
NUCOR STEEL NORFOLK
Design flow
(MGD)
1.3
3.6
1.2
10.5
3.47
2.6
0.118
Observed flow
(MGD)
(1991-2006 average)
0.79
3.36
0.56
4.05
3.75
4.31
0.46
Most of these point sources have reasonably good monitoring data available for total suspended solids (TSS), but
not for nutrients. The point sources were thus represented in the model with the median of reported values for
TSS and nutrient concentrations set to representative values by SIC code (Tetra Tech 1999).
Meteorological Data
The required meteorological time series for the 20 Watershed SWAT simulations are precipitation and air
temperature. The 20 Watershed simulations do not include water temperature simulation and use a degree-day
method for snowmelt. SWAT estimates Penman-Monteith potential evapotranspiration using a statistical weather
generator for inputs other than temperature and precipitation. These meteorological time series are drawn from the
BASINS4 Meteorological Database (USEPA 2008), which provides a consistent, quality-assured set of
nationwide data with gaps filled and records disaggregated. Scenario application requires simulation over 30
years, so the available stations are those with a common 30-year period of record (or one that can be filled from
an approximately co-located station) that includes the year 2001, if possible. A total of 57 precipitation stations
were identified for use in the Loup and Elkhorn River basins model with a common period of record of
10/1/1968-12/31/1999 (Table 4). Due to the discontinuance of many stations a simulation period ending slightly
prior to 2001 was chosen. Temperature records are sparser; where these are absent temperature is taken from
nearby stations with an elevation correction.
Table 4. Precipitation stations for the Loup and Elkhorn River basins model
COOP ID
NE250070
NE250180
NE250245
NE250320
NE250355
Name
ALBION
AMELIA 2W
ANSELMO 2 SE
ARCADIA
ARNOLD
Latitude
41 .6842
42.2347
41.5975
41 .4244
41 .4242
Longitude
-98.0033
-98.9506
-99.8258
-99.1231
-100.193
Temperature
X
X
X
Elevation (ft)
546
668
794
658
838
K-9
-------
COOP ID
NE250365
NE250385
NE250420
NE250525
NE250680
NE251130
NE251200
NE251345
NE251590
NE251660
NE251776
NE251835
NE252380
NE252595
NE252645
NE252647
NE252770
NE252805
NE253050
NE253075
NE253185
NE253425
NE253630
NE254986
NE255050
NE255080
NE255250
NE255370
NE255525
NE255702
NE255830
Name
ARTHUR
ASHTON
ATKINSON
BARTLETT 4S
BEEMER
BREWSTER
BROKEN BOW 2 W
BURWELL
CHAMBERS
CLARKSON
COLERIDGE
COMSTOCK
DODGE
ELGIN
ELLSWORTH
ELLSWORTH 15NNE
ERICSON 6 WNW
EWING
FREMONT
FULLERTON
GENOA 2 W
GREELEY
HARTINGTON
LOUPCITY6NNE
LYONS
MADISON 2W
MASON CITY
MEADOW GROVE
MILLER
MULLEN 21 NW
NELIGH
Latitude
41.5697
41.2481
42.5342
41.8278
41.9325
41.9375
41.4083
41.7769
42.2031
41.7239
42.5056
41.5569
41.7233
41.9872
42.0631
42.2647
41.7986
42.2611
41.43
41.3594
41.4514
41.5461
42.6167
41.3611
41.9378
41.8306
41.2231
42.0292
40.9283
42.2506
42.1303
Longitude
-101.691
-98.7989
-98.9783
-98.5494
-96.8108
-99.8628
-99.675
-99.1433
-98.7467
-97.1256
-97.2086
-99.2372
-96.8828
-98.0747
-102.283
-102.214
-98.7842
-98.3417
-96.4669
-97.9761
-97.7644
-98.5336
-97.2608
-98.9222
-96.4789
-97.49
-99.3008
-97.7386
-99.3886
-101.336
-98.0275
Temperature
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Elevation (ft)
1067
620
643
652
415
760
762
663
649
472
488
687
427
590
1190
1210
642
564
360
503
485
616
418
677
390
511
689
497
704
1055
536
K-10
-------
COOP ID
NE255925
NE256040
NE256135
NE256167
NE256290
NE256386
NE256395
NE256630
NE256720
NE256735
NE256970
NE257040
NE257515
NE257685
NE258025
NE258110
NE258455
NE258480
NE259050
NE259200
NE259262
Name
NEWPORT
NORTH LOUP
OAKDALE
OCONTO
ONEILL
OSHKOSH 10NE
OSMOND
PENDER
PIERCE
PILGER
PURDUM
RAVENNA
SAINT PAUL 4 N
SCRIBNER
SPALDING
STANTON
TAYLOR
TEKAMAH
WAYNE 4 NW
WEST POINT
WHITMAN 4 E
Latitude
42.6008
41.4933
42.0678
41.1439
42.4594
41.5
42.3569
42.1153
42.1958
42.0067
42.065
41.0333
41 .2686
41.6678
41.6031
41.9564
41.7708
41.7861
42.295
41.845
42.0828
Longitude
-99.3333
-98.7747
-97.9675
-99.7633
-98.6564
-102.183
-97.5969
-96.7058
-97.5206
-97.0561
-100.247
-98.9142
-98.4697
-96.6689
-98.3483
-97.2222
-99.3814
-96.2264
-97.0569
-96.7142
-101.431
Temperature
X
X
X
X
X
X
X
X
X
X
X
X
X
Elevation (ft)
680
597
521
786
607
327
503
408
485
429
820
625
541
382
578
469
692
338
457
399
1093
Watershed Segmentation
The Loup and Elkhorn River basins was divided into 114 subwatersheds for the purposes of modeling (Figure 3).
The initial calibration watershed was selected as Elkhorn River at Waterloo (USGS 06800500). The area modeled
encompasses complete watersheds and does not require specification of any upstream boundary conditions for
application.
K-ll
-------
USGS 06786000
Sioux City
USGS 06799000
USGS 06775900
USGS 06794000
06792000
USGS 06800500
i) USGS 06790500
Legend
A USGS gages
— Hydrography
= Interstate
Water (Nat Alias Datasel)
US Census Populated Places
I | County Boundaries
Model Subbasms
USGS 06784000
GCRP Model Areas - Central Nebraska River Basins
Model Segmentation
Figure 3. Model segmentation and USGS stations utilized for the Loup and Elkhorn River basins
Calibration Data and Locations
The specific site chosen for initial calibration was the Elkhorn River at Waterloo, NE, a flow and water quality
monitoring location at the Elkhorn River outflow to the Platte River. The drainage area for this gage is somewhat
larger than those selected for most other 20 Watershed study areas, but is the only station on the Elkhorn that
provides both flow and TSS monitoring over long periods of time. The Elkhorn River watershed was selected for
calibration focus because of the difficulties in obtaining model fit to the Sandhills area - both in this project and
in the earlier USGS modeling effort (Strauch and Linard 2009). Calibration and validation were then pursued at
multiple locations (Table 5), including multiple stations such as Dismal River that are entirely within the
Sandhills. Parameters derived on the Elkhorn River were not fully transferable to other portions of the watershed;
therefore, additional calibration was conducted at multiple gage locations.
K-12
-------
Table 5. Calibration and validation locations in the Loup and Elkhorn River basins
Station name
Dismal River near Thedford, NE
South Loup River at Saint Michael, NE
Middle Loup River at Saint Paul, NE
North Loup River at Taylor, NE
North Loup River near Saint Paul, NE
Cedar River near Fullerton, NE
Beaver Creek at Genoa, NE
Elkhorn River at Norfolk, NE
Elkhorn River at Waterloo, NE
USGS ID
06775900
06784000
06785000
06786000
06790500
06792000
06794000
06799000
06800500
Drainage area
(mi2)
966
2,320
8,075
2,350
4,302
1,220
677
2,790
6,900
Hydrology
calibration
X
X
X
X
X
X
X
X
X
Water quality
calibration
X
X
X
The model hydrology calibration period was set to Water Years 1989-1999 (within the 30-year period of record
for modeling). Hydrologic validation was then performed on Water Years 1978-1988. Water quality data
availability is somewhat low for the watershed, and water quality calibration used calendar years 1990-1995,
while validation used 1986-1989.
K-13
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
The Sandhills present a major challenge for hydrologic simulation. Previous attempts by USGS (Strauch and
Linard, 2009) found that it was very difficult to achieve a good fit to observed flows in the Sandhills region using
the SWAT model. Flow in this region of highly permeable soils tends to maintain steady rates driven by
groundwater discharge and much of the effort of calibration was focused on obtaining a reasonable representation
of this behavior.
There is one major reservoir in the Loup River basin - the Calamus Reservoir, which was included in the model.
Two smaller reservoirs (less than 100,000 AF storage) - Sherman Reservoir and Davis Creek Reservoir - were
not explicitly modeled. Pertinent information on Calamus Reservoir including surface area and storage at
principal (normal) and emergency spillway levels for the reservoirs modeled were obtained from the United Sates
Bureau of Reclamation website. The SWAT model provides four options to simulate reservoir outflow: measured
daily outflow, measured monthly outflow, average annual release rate for uncontrolled reservoir, and controlled
outflow with target release. Keeping in view the 20 Watershed climate change impact evaluation application, it
was assumed that the best representation of the reservoirs was to simulate them without supplying time series of
outflow records. Therefore, target release approach was used in the GCRP-SWAT model.
The Loup River system has a major water withdrawal (Loup River Power Canal) just before the point of entry
into the North Platte River. This withdrawal is represented in the model by monthly average rates and results in
substantially lower flows in the Loup River at the mouth than upstream.
A spatial calibration approach was adopted for GCRP-SWAT modeling for the Loup and Elkhorn River basins. In
particular, a distinctly different set of parameters was needed to simulate hydrology in the Sandhills area.
The initial calibration focus area (Elkhorn River) includes 39 subwatersheds, of which about half (the western
portion) are in the Sandhills with the remainder more representative of typical plains land use. The Loup River
basin also originates in the Sandhills and has similar downstream soils and land uses. The model parameters were
adjusted to obtain reasonable fit between the simulated and measured flows in terms of Nash-Sutcliffe modeling
efficiency and the high flow and low flow components as well as the seasonal flows.
The water balance was evaluated separately for the Elkhorn and Loup River watersheds. For the Elkhorn, the
water balance predicted by the SWAT model over the 32-year simulation period is as follows:
PRECIP = 675.6 MM
SNOW FALL = 80.60 MM
SNOW MELT = 79.20 MM
SUBLIMATION = 1.40 MM
SURFACE RUNOFF Q = 0.48 MM
LATERAL SOIL Q = 57.86 MM
TILE Q = 0.00 MM
GROUNDWATER (SHAL AQ) Q = 26.08MM
REVAP (SHAL AQ => SOIL/PLANTS) = 30.22 MM
DEEP AQ RECHARGE = 8.24 MM
TOTAL AQ RECHARGE = 82.42 MM
TOTAL WATER YLD = 84.42 MM
K-14
-------
PERCOLATION OUT OF SOIL = 84.99 MM
ET = 530.1 MM
PET = 1750.5MM
TRANSMISSION LOSSES = 0.00 MM
The water balance for the Loup watershed is summarized as follows:
PRECIP = 579.4 MM
SNOW FALL = 77.00 MM
SNOW MELT = 76.25 MM
SUBLIMATION = 0.76 MM
SURFACE RUNOFF Q = 1.15 MM
LATERAL SOIL Q = 10.02 MM
TILE Q = 0.00 MM
GROUNDWATER (SHAL AQ) Q = 45.46MM
REVAP (SHAL AQ => SOIL/PLANTS) = 25.69 MM
DEEP AQ RECHARGE = 66.73 MM
TOTAL AQ RECHARGE = 156.04 MM
TOTAL WATER YLD = 56.63 MM
PERCOLATION OUT OF SOIL = 158.39 MM
ET = 408 . 0 MM
PET = 1489.7MM
TRANSMISSION LOSSES = 0.00 MM
Hydrologic calibration adjustments focused on the following parameters:
• CN2 (initial SCS runoff curve number for moisture condition II)
• ESCO (soil evaporation compensation factor)
• SURLAG (surface runoff lag coefficient)
• SOL_AWC (available water capacity of the soil layer, mm water/mm of soil)
• ALPHA_BF (baseflow alpha factor, days)
• GW_DELAY (groundwater delay time, days)
• GWQMIN (threshold depth of water in the shallow aquifer required for return flow to occur, mm)
• GW_REVAP (groundwater "revap" coefficient)
• CH_N1 (Manning's "n" value for tributary channels)
• CH_N2 (Manning's "n" value for main channels)
• CH_K1 (Effective hydraulic conductivity in tributary channel alluvium)
• CH_K2 (Effective hydraulic conductivity in main channel alluvium)
• SFTMP (Snowfall temperature)
• SMTMP (Snowmelt base temperature)
• SMFMX (Maximum melt rate for snow during the year)
• SMFMN (Minimum melt rate for snow during the year)
• SOL_CRK (Crack volume potential of soil)
Calibration was performed for water years 1990-1999. Results for the Elkhorn River are summarized in Figure 4,
Figure 5, Figure 6, Figure 7, and Table 6. The quality of fit is generally adequate, except that late winter/early
spring high flow events tend to be underpredicted.
K-15
-------
15000
10000 --
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1989 to 9/30/1999)
Avg Modeled Flow (Same Period)
o
5000 -
O-89
A-91 O-92 A-94 O-95
Month
A-97
O-98
Figure 4. Mean monthly flow at USGS 06800500 Elkhorn River at Waterloo, NE - calibration period
• Avg Flow (10/1 /1989 to 9/30/1999)
Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1989 to 9/30/1999)
Avg Modeled Flow (Same Period)
5000
y = 0.9998x-63.412
R2 = 0.7971
s
5000
OcfjA/ovjDecjJanjFebjMarj/lp
0 2000 4000
Average Observed Flow (cfs)
10 11 12 1 23456789
Month
Figure 5. Seasonal regression and temporal aggregate at USGS 06800500 Elkhorn River at Waterloo, NE
- calibration period
K-16
-------
i Observed (25th, 75th)
•Median Observed Flow (10/1/1979 to 9/30/1989)
Average Monthly Rainfall (in)
Modeled (Median, 25th, 75th)
Figure 6. Seasonal medians and ranges at USGS 06800500 Elkhorn River at Waterloo, NE - calibration
period
100000
•Observed Flow Duration (10/1/1989 to 9/30/1999)
Modeled Flow Duration (10/1/1989 to 9/30/1999 )
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 7. Flow exceedance at USGS 06800500 Elkhorn River at Waterloo, NE - calibration period
K-17
-------
Table 6. Summary statistics at USGS 06800500 Elkhorn River at Waterloo, NE - calibration period
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 1
10-Year Analysis Period: 10/1/1989 - 9/30/1999
Flow/volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
4.32
1.43
0.91
1.30
0.66
0.52
1.83
0.97
0.31
Error Statistics
-2.59
2.09
-12.56
16.30
Observed Flow Gage
USGS 06800500 Elkhorn River at Waterloo, Nebr.
Hydrologic Unit Code: 10220003
Latitude: 41.2933333
Longitude: -96.2838889
Drainage Area (sq-mi): 6900
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring FlowVolume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
6.89 » [ 30
-42.36
2.58
-35.47
-30.24
0.416
0.337
0.642
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
4.43
1.64
0.89
1.12
0.62
0.91
1.78
1.50
0.44
Clear f
. ~^j
mmmmm
Hydrology Validation
Hydrology validation for the Elkhorn River was performed for the period water years 1980-1989. The validation
achieves a reasonable coefficient of model fit efficiency, but again appears to underpredict winter/spring storm
events (Figure 8, Figure 9, Figure 10, Figure 11, and Table 7).
K-18
-------
15000
10000 ---
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1979 to 9/30/1989)
Avg Modeled Flow (Same Period)
I
5000
O-79
A-81
O-88
Figure 8. Mean monthly flow at USGS 06800500 Elkhorn River at Waterloo, NE - validation period
• Avg Flow (10/1 /1979 to 9/30/1989)
• Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1979 to 9/30/1989)
•Avg Modeled Flow (Same Period)
4000
4000
3000
2000
1000
0 1000 2000 3000 4000
Average Observed Flow (cfs)
10 11 12 1 23456789
Month
Figure 9. Seasonal regression and temporal aggregate at USGS 06800500 Elkhorn River at Waterloo, NE
- validation period
K-19
-------
• Observed (25th, 75th)
•Median Observed Flow (10/1/1979 to 9/30/1989)
Average Monthly Rainfall (in)
Modeled (Median, 25th, 75th)
to
o:
Figure 10. Seasonal medians and ranges at USGS 06800500 Elkhorn River at Waterloo, NE - validation
period
99000
•Observed Flow Duration (10/1/1979 to 9/30/1989)
Modeled Flow Duration (10/1/1979 to 9/30/1989)
99
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 11. Flow exceedance at USGS 06800500 Elkhorn River at Waterloo, NE - validation period
K-20
-------
Table 7. Summary statistics at USGS 06800500 Elkhorn River at Waterloo, NE - validation period
REACH OUTFLOW FROM OUTLET 1
10-Year Analysis Period: 10/1/1979 - 9/30/1989
Flow/volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
3.19
1.32
0.57
0.68
0.54
0.53
1.44
0.76
0.18
Error Statistics
-8.81
-1.66
-12.33
33.85
USGS 06800500 Elkhorn River at Waterloo, Nebr.
Hydrologic Unit Code: 10220003
Latitude: 41.2933333
Longitude: -96.2838889
Drainage Area (sq-mi): 6900
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring FlowVolume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
3.94 » [ 30
-43.05
-6.46
-38.41
3.39
0.518
0.405
0.701
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
3.49
1.50
0.58
0.51
0.52
0.93
1.54
1.23
0.17
Clear [
Hydrology Results for Larger Watershed
As described above, parameters determined for the Elkhorn gage were not fully transferable to other gages in the
watershed, particularly in the Sandhills area. Calibration and validation was pursued at a total often gages
throughout the watershed. Calibration results are summarized in Table 8. Those watersheds that are dominantly in
the Sandhills area tend to show very low Nash-Sutcliffe coefficients of model fit efficiency for daily flows, but
high values for monthly flow volumes, reflecting the complex, groundwater-dominated nature of flow in this area.
The quality of fit is comparable to that obtained by Strauch and Linard (2009).
Results of the validation exercise are summarized in Table 9. Results are very similar to those obtained during the
calibration period, although total volume is underpredicted at several stations.
K-21
-------
Table 8. Summary statistics (percent error): all stations - calibration period
Station
Error in total
volume:
Error in 50%
lowest flows:
Error in 1 0%
highest
flows:
Seasonal
volume error
- Summer:
Seasonal
volume error
- Fall:
Seasonal
volume error
- Winter:
Seasonal
volume error
- Spring:
Error in
storm
volumes:
Error in
summer
storm
volumes:
Daily Nash-
Sutcliffe
Coefficient of
Efficiency, E:
Monthly
Nash-
Sutcliffe
Coefficient
06800500
Elkhorn
River at
Waterloo
-2.59
2.09
-12.56
16.30
6.89
-42.36
2.58
-35.47
-30.24
0.416
0.642
067995000
Elkhorn
River at
Norfolk
8.47
28.72
6.49
46.11
20.94
-24.17
2.54
-1.22
41.15
0.441
0.664
06775900
Dismal
River nr
Thedford,
NE
-4.65
-8.78
1.51
-4.90
-2.06
-6.22
-5.45
-37.60
4.52
-1.595
-2.111
06784000
South Loup
River at St.
Michael
2.43
17.14
-12.61
26.54
19.97
-13.83
-10.55
-48.32
-23.62
0.239
0.451
06785000
Middle
Loup River
at St. Paul
-3.78
28.06
-28.75
43.95
-12.90
-24.49
-0.68
-51.06
-21.44
0.206
0.252
06786000
North
Loup
River at
Taylor
-3.45
-4.99
6.93
34.75
-8.31
-21.70
-6.25
-35.46
-8.13
-0.214
-0.345
06790500
North
Loup
River nr
St. Paul
-2.07
-5.43
-11.58
0.36
-6.61
-19.54
17.45
-44.46
-24.44
-0.025
-0.412
06792000
Cedar
River nr
Fullerton
5.07
22.91
-12.09
26.71
15.37
-16.91
-0.34
-29.99
-22.99
0.140
0.383
06794000
Beaver
Creek at
Genoa
-3.41
36.21
-22.87
20.98
30.25
-18.95
-23.38
-18.33
1.85
0.032
0.356
K-22
-------
Table 9. Summary statistics (percent error): all stations - validation period
Station
Error in total
volume:
Error in 50%
lowest flows:
Error in 10%
highest flows:
Seasonal
volume error
- Summer:
Seasonal
volume error
- Fall:
Seasonal
volume error
- Winter:
Seasonal
volume error
- Spring:
Error in storm
volumes:
Error in
summer
storm
volumes:
Daily Nash-
Sutcliffe
Coefficient of
Efficiency, E:
Monthly
Nash-
Sutcliffe
Coefficient
06800500
Elkhorn
River at
Waterloo
-8.81
-1.66
-12.33
33.85
3.94
-43.05
-6.46
-38.41
3.39
0.518
0.701
067995000
Elkhorn
River at
Norfolk
-15.83
-7.47
-8.08
55.96
-2.64
-40.10
-25.65
-11.57
125.83
0.368
0.472
06775900
Dismal
River nr
Thedford,
NE
2.93
2.36
3.36
4.73
1.79
-0.34
5.58
-43.85
-26.84
-0.779
-1.232
06784000
South Loup
River at St.
Michael
-27.91
-29.93
-28.29
-12.53
-20.98
-37.75
-32.18
-51.73
-46.64
0.217
0.169
06785000
Middle
Loup River
at St. Paul
-21.34
3.64
-40.03
28.95
-28.56
-40.63
-17.24
-53.82
-25.11
0.105
0.022
06786000
North
Loup
River at
Taylor
-4.72
-0.85
-0.84
35.58
-15.38
-23.08
-0.91
-41.29
-16.61
-0.148
-0.218
06790500
North Loup
RivernrSt.
Paul
-5.74
0.74
-15.43
17.91
-8.56
-22.61
-2.78
-39.38
-14.87
0.031
-0.210
06792000
Cedar
River nr
Fullerton
-10.92
-17.86
5.04
24.55
-8.54
-27.14
-20.96
-9.49
28.81
0.071
0.086
06794000
Beaver
Creek at
Genoa
-20.87
-4.78
-28.94
1.07
-2.33
-36.39
-30.28
-6.55
-9.23
-0.051
0.149
Water Quality Calibration and Validation
Initial calibration and validation of water quality was done on the Elkhorn River at Waterloo (USGS 06800500),
using 1990-1995 for calibration and 1979-1989 for validation. As with hydrology, calibration was performed on
the later period as this better reflects the land use included in the model. The start of the validation period is
constrained by data availability.
K-23
-------
Calibration adjustments for sediment focused on the following parameters:
• SPCON (linear parameter for estimating maximum amount of sediment that can be re-entrained during
channel sediment routing)
• SPEXP (exponential parameter for estimating maximum amount of sediment that can be re-entrained
during channel sediment routing)
• CH_COV (channel cover factor)
• CH_EROD (channel erodibility factor)
• USLE_P (USLE support practice factor)
Simulated and estimated sediment loads at the Elkhorn River station for both the calibration and validation
periods are shown in Figure 12 and statistics for the two periods are provided separately in Table 10. The key
statistic in the Table 10 is the relative percent error, which shows the error in the prediction of monthly load
normalized to the estimated load. Table 10 also shows the relative average absolute error, which is the average of
the relative magnitude of errors in individual monthly load predictions. This number is inflated by outlier months
in which the simulated and estimated loads differ by large amounts (which may be as easily due to uncertainty in
the estimated load due to limited data as to problems with the model) and the third statistic, the relative median
absolute error, is likely more relevant and shows better agreement. Overall, TSS loads seem to be somewhat
underpredicted due to the representation of scattered spring high flow events. This likely reflects the
underprediction of winter/spring storm flow peaks as noted under the hydrology calibration. Elevated TSS loads
during these events is likely attributable to primarily channel scour.
TSS
100,000,000
-Regression Loads
-Simulated Loads
hp 00 00 00 00 00 00
opopopq>q>q>q>q>
Figure 12. Fit for monthly load of TSS at USGS 06800500 Elkhorn River at Waterloo, NE
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 06800500 Elkhorn River at Waterloo, NE
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1990-1995)
59.6%
73.9%
14.3%
Validation period
(1979-1989)
66.8%
79.5%
7.2%
K-24
-------
Calibration adjustments for total phosphorus and total nitrogen focused on the following parameters:
• PHOSKD (phosphorus soil partitioning coefficient)
• PSP (phosphorus availability index)
• RS2 (benthic source rate for dissolved P in the reach at 20° C)
• RS3 (Benthic source rate for NFLpN in the reach at 20° C)
• RS5 (organic P settling rate in the reach at 20° C)
• BC4 (rate constant for mineralization of organic P to dissolved P in the reach at 20° C)
• RS4 (rate coefficient for organic N settling in the reach at 20° C)
• CH_ONCO (Channel organic nitrogen concentration)
• CH_OPCO (Channel organic phosphorus concentration)
Results for the phosphorus simulation are shown in Figure 13 and Table 11. Results for the nitrogen simulation
are shown in Figure 14 and Table 12. The model fit reproduces observed seasonal trends, but appears to
underpredict total load, reflecting the underprediction of TSS load.
Total P
10000 T
1000
100
-Regression Loads
-Simulated Loads
O-!-(NCO^-mCOI^OOO>O-!-(NCO^-
opopopopopopopopopopq>q>q>q>q>
§§§§§§§§§§§§§§§§
Figure 13. Fit for monthly load of total phosphorus at USGS 06800500 Elkhorn River at Waterloo, NE
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 06800500 Elkhorn River at Waterloo, NE
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1993-2002)
24.2%
39.8%
26.5%
Validation period
(1986-1992)
34.9%
49.7%
15.6%
K-25
-------
Total N
10,000
1,000 ---«•
o
«
10 -•-
-Averaging Loads
-Simulated Loads
r-pooopopopopopopopopopcpcpcpcpq)
§§§§§§§§§§§§§§§§
Figure 14. Fit for monthly load of total nitrogen at USGS 06800500 Elkhorn River at Waterloo, NE
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 06800500 Elkhorn River at Waterloo, NE
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1993-2002)
28.1%
39%
22.3%
Validation period
(1986-1992)
18.1%
38%
19.2%
Water Quality Results for Larger Watershed
The Elkhorn River watershed SWAT model parameters for water quality were directly transferred to other
portions of the watershed. Only very limited amounts of water quality data are readily available for the remainder
of the watershed. Comparison to the data that are available suggests the model may underpredict loads associated
with large flow events at other stations as well. Summary statistics for the SWAT water quality calibration other
stations in the watershed are provided in Table 13. Insufficient monitoring data were readily available to provide
additional validation tests.
K-26
-------
Table 13. Summary statistics for water quality at all stations - calibration period 1993-2002
Station
Relative Percent Error TSS
Load
Relative Percent Error TP
Load
Relative Percent Error TN
Load
06800500
Elkhorn River at
Waterloo
59.6%
24.2%
28.1%
06799000
Elkhorn River at
Norfolk
ND
35.8%
35.5%
06794000
Beaver Creek at Genoa
ND
54.4%
25.3%
K-27
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
Huntzinger, T.L. and M.J. Ellis. 1993. Central Nebraska River Basins, Nebraska. Water Resources Bulletin. 29
(4):533-574.
Strauch, K.R. and J.I. Linard. 2009. Streamflow Simulations and Percolation Estimates Using the Soil and Water
Assessment Tool for Selected Basins in North-Central Nebraska, 1940-2005. Scientific Investigations Report
2009-5075. U.S. Geological Survey, Reston, VA.
Tetra Tech. 1999. Improving Point Source Loadings Data for Reporting National Water Quality Indicators. Final
Technical Report prepared for U.S. Environmental Protection Agency, Office of Waste water Management,
Washington, DC, by Tetra Tech, Inc., Fairfax, VA.
USEPA. 2008. Using the BASINS Meteorological Database (Version 2006). BASINS Technical Note 10.
Office of Water, U.S. Environmental Protection Agency, Washington, DC.
http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf (Accessed June,
2009).
K-28
-------
Appendix L
Model Configuration, Calibration and
Validation
Basin: Cook Inlet (Cook)
L-l
-------
Contents
Watershed Background L-4
Water Body Characteristics L-4
Soil Characteristics L-5
Land Use Representation L-5
Point Sources L-9
Meteorological Data L-9
Watershed Segmentation L-10
Calibration Data and Locations L-10
SWAT Modeling L-12
Assumptions L-12
Hydrology Calibration L-12
Hydrology Validation L-16
Hydrology Results for Larger Watershed L-20
Water Quality Calibration and Validation L-21
Water Quality Results for Larger Watershed L-24
References L-26
L-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes L-7
Table 2. Land use distribution for the Cook Inlet watershed (2001 NLCD) (mi2) L-8
Table 3. Major point source discharges in the Cook Inlet watershed L-9
Table 4. Precipitation stations for the Cook Inlet watershed model L-9
Table 5. Calibration and validation locations in the Cook Inlet watershed L-10
Table 6. Summary statistics at USGS 15266300 Kenai River at Soldotna, AK - calibration period L-16
Table 7. Summary statistics at USGS 15266300 Kenai River at Soldotna, AK - validation period L-19
Table 8. Summary statistics (percent error): all stations - calibration period L-20
Table 9. Summary statistics: all stations - validation period L-21
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 152927000 Talkeetna River near Talkeetna, AK L-22
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 152927000 Talkeetna River near Talkeetna, AK L-23
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 152927000 Talkeetna River near Talkeetna, AK L-24
Table 13. Summary statistics for water quality at all stations - calibration period 1985-2001 L-25
Table 14. Summary statistics for water quality at all stations - validation period 1971-1984 L-25
Figures
Figure 1. Location ofthe Cook Inlet watershed L-5
Figure 2. Land use in the Cook Inlet watershed L-6
Figure 3. Model segmentation and USGS stations utilized for the Cook Inlet watershed L-10
Figure 4. Mean monthly flow at USGS 15266300 Kenai River at Soldotna, AK - calibration period L-13
Figure 5. Seasonal regression and temporal aggregate at USGS 15266300 Kenai River at Soldotna,
AK - calibration period L-14
Figure 6. Seasonal medians and ranges at USGS 15266300 Kenai River at Soldotna, AK - calibration
period L-14
Figure 4. Flow exceedance at USGS 15266300 Kenai River at Soldotna, AK - calibration period L-15
Figure 8. Mean monthly flow at USGS 15266300 Kenai River at Soldotna, AK - validation period L-17
Figure 9. Seasonal regression and temporal aggregate at USGS 15266300 Kenai River At Soldotna,
AK - validation period L-17
Figure 10. Seasonal medians and ranges at USGS 15266300 Kenai River at Soldotna, AK - validation
period L-18
Figure 5. Flow exceedance at USGS 15266300 Kenai River at Soldotna, AK - validation period L-18
Figure 6. Fit for monthly load of TSS at USGS 152927000 Talkeetna River near Talkeetna, AK L-22
Figure 7. Fit for monthly load of total phosphorus at USGS 152927000 Talkeetna River near
Talkeetna, AK L-23
Figure 8. Fit for monthly load of total nitrogen at USGS 152927000 Talkeetna River near Talkeetna,
AK L-24
-------
The Cook Inlet watershed was selected as one of the 15 non-pilot application watersheds for the 20 Watershed
study. Watershed modeling for the non-pilot areas was accomplished using the SWAT model only and model
calibration and validation results are presented in abbreviated form.
Water Body Characteristics
The Cook Inlet stretches 180 miles (290 km) from the Gulf of Alaska to Anchorage in south-central Alaska. The
watershed draining to Cook Inlet covers 47,000 square miles east of the Aleutian Range and south of the Alaska
Range including the drainage area of Mount McKinley (Figure 1). The model area includes seven HUCSs within
HUC 1902, encompassing about 22,200 mi2 of the Cook Inlet watershed. The Cook Inlet watershed receives
water from its tributaries the Kenai, the Susitna and Matanuska rivers from the melting snow and ice from Mount
McKinley, the Chugach Mountains, and the Aleutian Range. Cook Inlet branches into the Knik Arm and
Turnagain Arm at its northern end, almost surrounding Anchorage.
The watershed is dominated by igneous rocks in the mountains and by continental shelf and alluvial deposits in
the lowlands. Glaciation has dramatically altered the landscape and glaciers are extensive on the southeastern and
northwestern boundaries of the watershed. Five physiographic regions - grading from plains and lowlands to
extremely high rugged mountains - are represented in the watershed. Altitude ranges from sea level to 20,320 ft at
the highest point in North America, Mount McKinley. Rugged mountains surround Cook Inlet and include four
active volcanoes on the western side of the inlet. Precipitation is closely associated with altitude and ranges from
about 15 to more than 200 inches annually (USGS, 2008b).
Numerous river systems drain the watershed, including the Susitna, Matanuska, and Kenai Rivers. The largest
river, the Susitna, drains about half of the watershed. Most rivers have relatively small drainages but yield large
quantities of water because of substantial snowfall in the mountains. Many streams are fed by glaciers and have
different physical characteristics than streams that do not have glacial contributions. Glacier-fed streams have
periods of sustained high flow during summers and are more turbid than streams lacking glacial contributions.
Numerous wetlands and lakes also influence the physical and chemical characteristics of streams by moderating
peak flows and trapping sediment and nutrients.
Land cover in the model area is dominated by forests (24 percent), shrubland (38 percent), and barren land (19
percent). Glaciers cover 8 percent of the area, and lakes and wetlands cover another 10 percent. Less than 1
percent of the basin is used for agricultural purposes. The Municipality of Anchorage dominates the urban and
residential features of the basin; however, the total urban and residential land cover is less than 1 percent of the
basin. More than half of the state's population lives in the metropolitan Anchorage area. Expansion of suburban
areas continues to the north of Anchorage and residential density is increasing throughout the municipality. The
remainder of the basin is largely unpopulated; however, native villages exist at a number of locations.
Watersheds of the Cook Inlet basin are largely undeveloped and contain parts of four national parks totaling about
6,300 mi2. Nearly 1,800 mi2 of the Chugach National Forest and the 3,000 mi2 Kenai National Wildlife Refuge
also are within the boundaries of the watershed.
L-4
-------
Legend
— Hydrography
•I Water (Nal. Alias Datasel)
US Census Populated Places
^H Municipalities (pop > 50.000)
I I County Boundaries
I I Watershed with HUCSs
V"
Upper Susitna
River
(19020501)
CohOO\ upper Kenai
Peninsula
(19020302)
Anchor Point
Figure 1. Location of the Cook Inlet watershed.
Soil Characteristics
Soils in the watershed, as described in STATSGO soil surveys, fall primarily into hydrologic soil groups (HSGs)
B (moderately high infiltration capacity) and D (low infiltration capacity). SWAT uses information drawn directly
from the soils data layer to populate the model.
Land Use Representation
Land use/cover in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage and is
predominantly rangeland (Figure 2). NLCD land cover classes were aggregated according to the scheme shown in
Table 1 for representation in the 20 Watershed model. SWAT uses the built-in hydrologic response unit (HRU)
overlay mechanism in the ArcSWAT interface. SWAT HRUs are formed from an intersection of land use and
SSURGO major soils. The distribution of land use in the watershed is summarized in Table 2.
L-5
-------
Legend
Hydrography
= Interstate
| | County Boundaries
2001 NLCD Land Use
I | Open water
H Perennial Ice/Snow
[ ] Developed, open space
| Developed, low intensity
MB Developed, medium intensity
BB Developed, high intensity
^] Barren land
^H Deciduous forest
MB Evergreen forest
I ! Mixed forest
| | Scrub/shrub
| Grassland/herbaceous
L | Pasture/hay
^] Cultivated crops
I | Woody wetlands
I ] Emergent herbaceous wetlands
Gulf of
Alaska
GCRP Model Areas - Cook Inlet River Basins
Land Use Map
0 20 -W
20 40
TETRATECH
NAD 19E3 Albers
- Map produced 05-03-3011 - P Cada
Figure 2. Land use in the Cook Inlet watershed.
L-6
-------
Table 1. Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area usually
accounted for as reach area
Deciduous
Evergreen
Mixed
Emergent & woody wetlands
Aquatic bed wetlands (not
emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
L-7
-------
:2\
Table 2. Land use distribution for the Cook Inlet watershed (2001 NLCD) (mi )
HUC8
watershed
Upper Kenai
Peninsula
19020302
Anchorage
19020401
Matansuka
19020402
Upper Susitna
River
19020501
Chulitna River
19020502
Talkeetna
River
19020503
Lower Susitna
River
19020505
Total
Open
water
173.02
8.44
46.94
174.21
37.54
16.67
110.38
567.20
Snow/Ice
267.71
115.51
622.93
171.92
258.50
258.25
17.59
1,712.40
Developed3
Open
space
12.15
20.60
6.85
0.82
0.56
0.76
21.47
63.21
Low
density
19.40
35.73
8.65
1.57
1.17
0.21
22.71
89.44
Medium
density
3.51
15.43
1.07
0.01
0.06
0.03
2.17
22.26
High
density
1.45
6.13
0.17
0.00
0.01
0.01
0.36
8.13
Barren
land
323.25
161.42
1 ,323.82
951.59
780.26
492.99
186.08
4,219.40
Forest
1 ,439.93
202.26
384.32
1,119.99
330.32
384.80
1 ,499.68
5,361.30
Shrubland/
Herbaceous
1,049.19
430.64
1,018.62
3,399.62
862.14
1,141.99
555.86
8,458.06
Pasture/Hay
0.77
0.67
5.36
0.01
0.00
0.80
2.69
10.30
Cultivated
2.20
0.67
1.57
0.00
0.00
0.43
19.16
24.02
Wetland
353.91
32.80
65.32
244.20
72.82
28.20
890.55
1,687.80
Total
3,646.48
1,030.29
3,485.63
6,063.93
2,343.37
2,325.13
3,328.70
22,223.53
aThe percent imperviousness applied to each of the developed land uses is as follows: open space (10.11%), low density (29.79%), medium density (61.48%), and high
density (87.17%).
L-8
-------
Point Sources
There are only two point source discharges in the watershed. Only the major dischargers, with a design flow
greater than 1 MGD are included in the simulation (Table 3). The major dischargers are represented at long-term
average flows, without accounting for changes overtime or seasonal variations.
Table 3. Major point source discharges in the Cook Inlet watershed
NPID
AK0022543
AK0047856
NAME
ANCHORAGE, MUNICIPALITY OF
ANCHORAGE, MUNICIPALITY OF
Design Flow
(MGD)
2.5
0.6
Observed
Flow (MGD)
1.843066667
0.3767
The point sources were initially represented in the model with the median of reported values for TSS and an
assumed total nitrogen concentration of 11.2 mg/L and assumed total phosphorus concentration of 7.0 mg/L for
secondary treatment facilities (TetraTech 1999).
Meteorological Data
The required meteorological time series needed for the 20 Watershed SWAT simulations are precipitation and air
temperature. The 20 Watershed simulations do not include water temperature simulation and use a degree-day
method for snowmelt. SWAT estimates Penman-Monteith potential evapotranspiration using a statistical weather
generator for inputs other than temperature and precipitation. These meteorological time series are drawn from the
BASINS4 Meteorological Database (USEPA 2008), which provides a consistent, quality-assured set of
nationwide data with gaps filled and records disaggregated. Scenario application requires simulation over 30
years, so the available stations are those with a common 30-year period of record (or one that can be filled from
an approximately co-located station) that covers the year 2001. A total of 14 precipitation stations were identified
for use in the Cook Inlet model with a common period of record of 10/1/1972-9/30/2002 (Table 4). Temperature
records are sparser; where these are absent temperature was taken from nearby stations with an elevation
correction.
Table 4. Precipitation stations for the Cook Inlet watershed model
ID
500243
500280
500302
500707
501926
502144
503299
504546
505733
506870
508371
508594
508976
509790
Name
AK500243
AK500280
AK500302
AK500707
AK501926
AK502144
AK503299
AK504546
AK505733
AK506870
AK508371
AK508594
AK508976
AK509790
Latitude
60.9584
61.1954
61.6245
61.5678
62.8293
60.3925
61.1001
60.5798
61.5665
61.4222
60.1040
62.0303
62.3201
61.7067
Longitude
-149.1100
-150.0030
-149.3390
-149.1380
-149.8960
-149.6660
-149.6930
-151.2390
-149.2540
-149.0990
-149.4430
-146.6920
-150.0940
-149.9970
Elevation
83
40
140
46
433
154
689
28
52
67
34
701
107
82
Temperature
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
L-9
-------
Watershed Segmentation
The Cook Inlet watershed was divided into 116 subwatersheds for the purposes of modeling (Figure 3). The
model encompasses the complete watershed and does not require specification of any upstream boundary
conditions for application.
Legend
USGS gages
Hydrography
Interstate
•I Water (Nat. Atlas Dataset)
US Census Populated Places
I I County Boundaries
Model Subbasins
Gulf of
Alaska
GCRP Model Areas - Cook Inlet River Basins
Model Segmentation
NAD 1983 Alters meters • Map produced 05-03-2011 - P Cada
Figure 3. Model segmentation and USGS stations utilized for the Cook Inlet watershed.
Calibration Data and Locations
The specific site chosen for initial calibration was at the USGS station at the Kenai River at Soldotna, AK.
Calibration and validation were pursued at two locations (Table 5).
Table 5. Calibration and validation locations in the Cook Inlet watershed
Station name
Kenai River at Soldotna
Talkeetna River near Talkeetna
USGS ID
15266300
15292700
Drainage area
(mi2)
1951
1996
Hydrology
calibration
X
X
Water quality
calibration
X
X
L-10
-------
The model hydrology calibration period was set to Water Years 1992-2001 (within the 30-year period of record
for modeling). Hydrologic validation was then performed on Water Years 1982-1991. Water quality calibration
used calendar years 1985-2001, while validation used 1972-1984. However, there was some variation to this time
period across the monitoring stations depending on the availability of monitored data.
L-ll
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
Two major reservoirs occur in the Cook Inlet watershed. Pertinent reservoir information including surface area
and storage at principal (normal) and emergency spillway levels for the reservoirs modeled were obtained from
the National Inventory of Dams (NID) database. The SWAT model provides four options to simulate reservoir
outflow: measured daily outflow, measured monthly outflow, average annual release rate for uncontrolled
reservoir, and controlled outflow with target release. Keeping in view the 20 Watershed climate change impact
evaluation application, it was assumed that the best representation of the reservoirs was to simulate them without
supplying time series of outflow records. Therefore, the target release approach was used in the GCRP-SWAT
model.
Elevation bands were also created in the subwatersheds where elevation was above 500 m to account for the
impact of higher elevation. Additionally, regions of permafrost were identified within the watershed and were
accounted for by adding initial snow water content in the elevation bands.
Calil
A spatial calibration approach was adopted for GCRP-SWAT modeling for the Cook Inlet watershed; however,
adjustments to specific subwatersheds were kept as minimal as possible. Moreover, a systematic adjustment of
parameters was adopted and some adjustments were applied throughout the watershed. Most of the calibration
efforts were geared toward getting a closer match between simulated and observed flows at one of the USGS
gaging stations in the watershed.
A 5/10/5 percent threshold was used for land use/soil/slope in the SWAT model while defining the HRUs. Urban
land use classes were exempted from the HRU overlay thresholds.
The parameters were adjusted within the practical range at the calibration focus area to obtain reasonable fit
between the simulated and measured flows in terms of Nash-Sutcliffe modeling efficiency and the high flow and
low flow components as well as the seasonal flows.
The water balance of the whole Cook Inlet watershed predicted by the SWAT model over the 30-year simulation
period is as follows:
PRECIP = 653.3 MM
SNOW FALL = 351.55 MM
SNOW MELT = 544.51 MM
SUBLIMATION = 54.10 MM
SURFACE RUNOFF Q = 99.35 MM
LATERAL SOIL Q = 310.36 MM
TILE Q = 0.00 MM
GROUNDWATER (SHAL AQ) Q = 225.36MM
REVAP (SHAL AQ => SOIL/PLANTS) = 6.71 MM
DEEP AQ RECHARGE = 26.93 MM
TOTAL AQ RECHARGE = 269.25 MM
TOTAL WATER YLD = 634.47 MM
L-12
-------
PERCOLATION OUT OF SOIL
ET = 187.1 MM
PET = 405.8MM
TRANSMISSION LOSSES =
= 269.04 MM
0.60 MM
Hydrologic calibration adjustments focused on the following parameters:
• Snow parameters SMTMP, SMFMX, SMFMN, TIMP
• Sol_AWC (available water capacity of the soil layer, mm water/mm of soil)
• Baseflow factor
• GW_DELAY (groundwater delay time)
• GWQMN (threshold depth of water in the shallow aquifer required for return flow to occur)
• SHALLST (Initial depth of water in the shallow aquifer)
• RevapMN (threshold depth of water in the shallow aquifer required for "revap" or percolation to the deep
aquifer to occur
• Rchrg_DP
• CH_K2 (channel hydraulic conductivity)
• NDTarg
• Curve Number
• Temperature Lapse Rate
• Precipitation Lapse Rate
Calibration results for the Cook Inlet at Kenai River near Soldotna are summarized in the following Figures 4
through 7 and Table 6. In general, the model captured the timing of the peaks well but tends to underestimate both
the high flows and the base flows resulting in overall underestimation of total volume by 18 percent (Figure 4,
and Table 6).
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2001 )
Avg Modeled Flow (Same Period)
20000
I
o
O-92 O-93 O-94 O-95 O-96 O-97 O-98 O-99 O-OO
Figure 4. Mean monthly flow at USGS 15266300 Kenai River at Soldotna, AK - calibration period.
L-13
-------
Avg Flow (10/1 /1992 to 9/30/2001)
• Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2001)
-Avg Modeled Flow (Same Period)
15000
•6
10000
-------
•Observed FlowDuration (10/1/199210 9/30/2001)
Modeled Flow Duration (10/1/1992 to 9/30/2001)
100000
10000
1000
1
LL
0
0)
TO
0)
ro
Q
0.1
0% 10% 20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90% 100%
Figure 4. Flow exceedance at USGS 15266300 Kenai River at Soldotna, AK - calibration period.
L-15
-------
Table 6. Summary statistics at USGS 15266300 Kenai River at Soldotna, AK - calibration period
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 74
9-Year Analysis F^riod: 10/1/1992 - 9/30/2001
Flow volumes are (inches/year) for upstream drainage area
Observed Flow Gage
USGS 15266300 Kenai Rat Soldotna, AK
Hydrologic Unit Code: 19020302
Latitude: 60.4775
Longitude:-151.0738
Drainage Area (sq-rri): 1951
Total Simulated In-stream Flow:
21.27
Total Observed In-stream Flow:
26.24
hest_10%_flows:
Total of Simulated lowest 50% flows:
7.52
3.87
_Simu]ated_Summe£Fjpw_yolume_(months 7J9)
_Simulated_FaN Flow Volume (mo_nths
_Simulated_Winter Flow Volijme (months J_-3^:
_Simulated_Spring Flow Volume (mcinths 4-6):
1J.52
3.22
pbsen/ed Summer _Flqw Volume (7-9):
Observed Fjll_Flow_yolumeJ10-12):_
14.95
4.40
1-29
5.24
1.62
5.28
Spring_Flow_yq[ume_(4-6):_
Total Simulated Storm Volume:_
Simulated Summer Storm Volume (7-9):
3.41
Total Observed Storm Volume:
2.29
1.85
Observed Summer Storm Volume (7-9):
1.22
Errors (Simulated-Observed)
Error Statistics
Recommended Criteria
-20.69
-16-10
-22.93
-26.68
-20.76
Error in total volume:
-18.96
_Erro£ I r^ 50%J ow est _f I ow s:
_Erro£ ir^ 10%highest_flows:
Seasonal volume error - Summer:
Seasonaj_volume_errpr^ F_all:_
Seasnavolumeerrr Wirrter
Seasonal volume error - Spring:
Error jn _stqrrn_volumes:_
Error in summer storm volumes:
Baseline adjusted coefficient (Garrick),
0.684
0.592
Model accuracy increases
as E or E' approaches 1.0
Monthly NSE
0.800
Hydrology Validation
Hydrology validation for the Cook Inlet was performed for the period 10/1/1983 through 9/30/1992. Results are
presented in Figures 8 through 11 and Table 7. The validation achieves a high coefficient of model fit efficiency,
but is over predicted on 50 percent low volume, fall and winter volume and thereby the total flow is also
overpredicted (Figure 5, Table 7).
L-16
-------
Avg Monthly Rainfall (in) —•—Avg Observed Flow (10/1/1983 to 9/30/1992 ) Avg Modeled Flow (Same Period)
10000
1
O-83 O-84 O-85 O-86
O-87 O-88
Month
O-89 O-90 O-91
Figure 8. Mean monthly flow at USGS 15266300 Kenai River at Soldotna, AK - validation period.
Avg Flow (10/1 /1983 to 9/30/1992)
• Line of Equal Value
Best-Fit Line
15000
•6
10000
-------
To Lower Bound Average Monthly Rainfall (in) -Median Observed Flow (10/1/1983 to 9/30/1992) Modeled (Median, 25th, 75th)
15000
10000
f
o
5000
Jan Feb
i
Mar
Apr
May
Jun
Jul
Aug
Sep
Dec
2
- 3
- 3
CD
or
10 11 12 1 23456789
Month
Figure 10. Seasonal medians and ranges at USGS 15266300 Kenai River at Soldotna, AK -validation
period.
•Observed FlowDuratlon (10/1/1983to 9/30/1992)
Modeled Flow Duration (10/1/1983 to 9/30/1992)
100000
10000
1000 -
1
LL
0>
0)
5
o>
ro
Q
0% 10% 20% 30% 40%, 50%. 60% 70% 80%. 90%, 100%,
Percent of Time that Flow is Equaled or Exceeded
Figure 5. Flow exceedance at USGS 15266300 Kenai River at Soldotna, AK - validation period.
L-18
-------
Table 7. Summary statistics at USGS 15266300 Kenai River at Soldotna, AK - validation period
REACH OUTFLOW FROM OUTLET 74
9-Year Analysis F^riod: 10/1/1983 - 9/30/1992
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Srjring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error- Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sjjt£liffeJDpj5fficjejT^^
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
21.61
6.15
3.40
11.28
3.99
1.47
4.87
3.57
1.81
Error Statistics
19.49
106.86
2.31
19.94
USGS 15266300 Kenai Rat Soldotna, AK
Hydrologic Unit Code: 19020302
Latitude: 60.4775
Longitude: -151. 0738
Drainage Area (sq-rri): 1951
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
94.13 » | 30
107.45
-17.71
6.05
2.50
0.554
0.487
0.749
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
18.08
6.01
1.64
9.41
2.06
0.71
5.91
3.37
1.77
L-19
-------
Hydrology Results for Larger Watershed
As described above, parameters determined for the Soldotna gage were fully transferable to other gages in the
watershed. In addition, calibration and validation was pursued at 2 gages in the watershed. Calibration results
were acceptable at both gages (Table 8). Results of the validation exercise are summarized in Table 9.
Table 8. Summary statistics (percent error): all stations - calibration period
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error- Summer:
Seasonal volume error- Fall:
Seasonal volume error - Winter:
Seasonal volume error- Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient of
Efficiency, E:
Baseline adjusted coefficient
(Garrick), E':
Monthly Nash-Sutcliffe Coefficient of
Efficiency, E:
USGS 15266300 Kenai R
at Soldotna, AK
-18.96
-20.69
-16.10
-22.93
-26.68
-20.76
-0.72
48.87
51.66
0.684
0.592
0.800
USGS 152927000 Talkeetna R nr
Talkeetna, AK
-18.84
-38.31
16.27
-25.08
-54.40
-31.58
1.23
146.88
125.74
0.240
0.427
0.762
L-20
-------
Table 9. Summary statistics: all stations - validation period
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error -
Summer:
Seasonal volume error- Fall:
Seasonal volume error - Winter:
Seasonal volume error- Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient
of Efficiency, E:
Baseline adjusted coefficient
(Garrick), E':
Monthly Nash-Sutcliffe
Coefficient of Efficiency, E:
USGS 15266300 Kenai R at
Soldotna, AK
19.49
106.86
2.31
19.94
94.13
107.45
-17.71
6.05
2.50
0.554
0.487
0.749
USGS 152927000 Talkeetna R nr
Talkeetna, AK
2.40
-9.55
35.16
2.03
-39.48
-9.75
25.47
143.61
137.59
0.174
0.431
0.739
Water Quality Calibration and Validation
Initial calibration and validation of water quality was done on Cook Inlet at USGS 152927000 Talkeetna River
near Talkeetna, AK, using 1985-2001 for calibration and 1972-1984 for validation. As with hydrology, calibration
was performed on the later period as this better reflects the land use included in the model. The start of the
validation period is constrained by data availability.
Calibration adjustments for sediment focused on the following parameters:
• SPCON (Linear parameters for estimating maximum amount of sediment that can be re-entrained during
channel sediment routing)
• CH_COV (Channel cover factor)
• CH_EROD (Channel erodibility factor)
Simulated and estimated sediment loads at the USGS 152927000 Talkeetna River station for both the calibration
and validation periods are shown in Figure 6 and statistics for the two periods are provided separately in Table 10.
The key statistic in Table 10 is the relative percent error, which shows the error in the prediction of monthly load
normalized to the estimated load. Table 10 also shows the relative average absolute error, which is the average of
the relative magnitude of errors in individual monthly load predictions. This number is inflated by outlier months
in which the simulated and estimated loads differ by large amounts (which may be as easily due to uncertainty in
the estimated load due to limited data as to problems with the model) and the third statistic, the relative median
absolute error, is likely more relevant and shows better agreement.
L-21
-------
TSS
1,000,000
*i'* ::?
- Regression Loads
-Simulated Loads
Figure 6. Fit for monthly load of TSS at USGS 152927000 Talkeetna River near Talkeetna, AK.
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 152927000 Talkeetna River near Talkeetna, AK
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1985-2001)
66.4%
69%
3.4%
Validation period
(1972-1984)
64.1%
68%
3.1%
Calibration adjustments for total phosphorus and total nitrogen focused on the following parameters:
• PHOSKD (phosphorus soil partitioning coefficient)
• RS2
• RS3
• RS4
• RS5
• BC1, BC2 and BC4
• MUMAX
Results for the phosphorus simulation are shown in Figure 7 and Table 11. Results for the nitrogen simulation are
shown in Figure 8 and Table 12. The model fit is generally acceptable.
L-22
-------
100
Total P
gnptojs
-Regression Loads
-Simulated Loads
Figure 7. Fit for monthly load of total phosphorus at USGS 152927000 Talkeetna River near Talkeetna,
AK.
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 152927000 Talkeetna River near Talkeetna, AK
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1985-2001)
83.2%
86%
8%
Validation period
(1972-1984)
82.18%
88%
8.2%
L-23
-------
Total N
o
-Averaging Loads
-Simulated Loads
Figure 8. Fit for monthly load of total nitrogen at USGS 152927000 Talkeetna River near Talkeetna, AK.
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 152927000 Talkeetna River near Talkeetna, AK
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1985-2001)
57.3%
59%
22.1%
Validation period
(1972-1984)
50.4%
51%
18.7%
Water Quality Results for Larger Watershed
As with hydrology, the SWAT model parameters used to calibrate at the Talkeetna River (USGS 152927000)
station for water quality were directly transferred to other portions of the watershed. Application of the SWAT
model without spatial adjustments resulted in relatively large errors in predicting loads and concentrations at some
stations. Summary statistics for the SWAT water quality calibration and validation at other stations in the
watershed are provided in Table 13 and Table 14.
L-24
-------
Table 13. Summary statistics for water quality at all stations - calibration period 1985-2001
Station
Relative Percent Error TSS Load
Relative Percent Error TP Load
Relative Percent Error TN Load
USGS 15266300 Kenai R at
Soldotna, AK
-14.3%
49.8%
34.4%
USGS 152927000 Talkeetna R nr
Talkeetna, AK
66.4%
83.2%
57.3%
Table 14. Summary statistics for water quality at all stations - validation period 1971-1984
Station
Relative Percent Error TSS Load
Relative Percent Error TP Load
Relative Percent Error TN Load
USGS 15266300 Kenai R at
Soldotna, AK
-14.7%
50.24%
28.9%
USGS 152927000 Talkeetna R nr
Talkeetna, AK
64.1%
82.18%
50.4%
L-25
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
Tetra Tech. 1999. Improving Point Source Loadings Data for Reporting National Water Quality Indicators. Final
Technical Report prepared for U.S. Environmental Protection Agency, Office of Waste water Management,
Washington, DC, by Tetra Tech, Inc., Fairfax, VA.
USEPA (United States Environmental Protection Agency). 2008. Using the BASINS Meteorological Database
(Version 2006). BASINS Technical Note 10. Office of Water, U.S. Environmental Protection Agency,
Washington, DC. http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf
(Accessed June, 2009).
L-26
-------
Appendix M
Model Configuration, Calibration and
Validation
Basin: Georgia-Florida Coastal Plain
(GaFI)
M-l
-------
Contents
Watershed Background M-4
Water Body Characteristics M-4
Soil Characteristics M-6
Land Use Representation M-6
Point Sources M-10
Meteorological Data M-11
Watershed Segmentation M-13
Calibration Data and Locations M-15
SWAT Modeling M-16
Assumptions M-16
Hydrology Calibration M-16
Hydrology Validation M-20
Hydrology Results for Larger Watershed M-23
Water Quality Calibration and Validation M-25
Water Quality Results for Larger Watershed M-28
References M-29
M-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes M-8
Table 2. Land use distribution for the Georgia-Florida Coastal Plain (2001 NLCD) (mi2) M-9
Table 3. Major point source discharges in the Georgia-Florida Coastal Plain model M-10
Table 4. Precipitation stations for the Georgia-Florida Coastal Plain model M-11
Table 5. Calibration and validation locations in the Georgia-Florida Coastal Plain M-15
Table 6. Summary statistics at USGS 02329000 Ochlockonee River at Havana, FL - calibration
period M-20
Table 7. Summary statistics at USGS 02329000 Ochlockonee River at Havana, FL - validation
period M-23
Table 8. Summary statistics (percent error): all stations - calibration period M-24
Table 9. Summary statistics (percent error): all stations - validation period M-25
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 02329000 Ochlockonee River at Havana, FL M-26
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 02329000 Ochlockonee River at Havana, FL M-27
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 02329000 Ochlockonee River at Havana, FL M-28
Table 13. Summary statistics for water quality at all stations - calibration period 1992-2002 M-28
Table 14. Summary statistics for water quality at all stations - validation period 1982-1992 M-28
Figures
Figure 1. Location of the Georgia-Florida Coastal Plain model M-5
Figure 2. Land use in the Georgia-Florida Coastal Plain model M-7
Figure 3. Model segmentation and USGS stations utilized for the Georgia-Florida Coastal Plain M-14
Figure 4. Mean monthly flow at USGS 02329000 Ochlockonee River at Havana, FL - calibration
period. M- 17
Figure 5. Seasonal regression and temporal aggregate at USGS 02329000 Ochlockonee River at
Havana, FL - calibration period M-18
Figure 6. Seasonal medians and ranges at USGS 02329000 Ochlockonee River at Havana, FL -
calibration period M-18
Figure 7. Flow exceedance at USGS 02329000 Ochlockonee River at Havana, FL - calibration
period M-19
Figure 8. Mean monthly flow at USGS 02329000 Ochlockonee River at Havana, FL - validation
period M-21
Figure 9. Seasonal regression and temporal aggregate at USGS 02329000 Ochlockonee River at
Havana, FL -validation period M-21
Figure 10. Seasonal medians and ranges at USGS 02329000 Ochlockonee River at Havana, FL -
validation period M-22
Figure 11. Flow exceedance at USGS 02329000 Ochlockonee River at Havana, FL - validation
period M-22
Figure 12. Fit for monthly load of TSS at USGS 02329000 Ochlockonee River at Havana, FL M-26
Figure 13. Fit for monthly load of total phosphorus at USGS 02329000 Ochlockonee River at
Havana, FL M-27
Figure 14. Fit for monthly load of total nitrogen at USGS 02329000 Ochlockonee River at Havana,
FL M-27
M-3
-------
The Georgia-Florida Coastal Plain drainages weres selected as one of the 15 non-pilot application watersheds for
the 20 Watershed study. Watershed modeling for the non-pilot areas is accomplished using the SWAT model
only, and model calibration and validation results are presented in abbreviated form.
Water Body Characteristics
9
The Georgia-Florida Coastal Plain model covers an area of about 17.500 mi in portions of Georgia and
Florida. The modeled area includes 15 HTJCSs in two groups, one group draining to Tampa Bay (HUC
0310) and the remainder in southern Georgia and northwest Florida (in HUC 0311 and 0312; Figure 1).
The watershed contains an EPA ORD Ecosystems Research Area (in the Tampa Bay drainage) and
Tampa Bay is part of EPA's National Estuary Program.
Climate in the watershed is humid subtropical and influenced by air masses from the Gulf of Mexico.
Average annual rainfall is around 45 to 53 inches per year, while the average annual temperatures is
around 70 - 72 °F. The majority of precipitation is associated with summer convective storms, and
tropical storms cross the area frequently. The study area has a climatic range from temperate in the north
to subtropical in the south and along the Gulf Coast.
The major land uses in the watershed include forest, agriculture (citrus and row crops), wetlands, urban,
and rangeland. Forested areas cover approximately 34 percent of the watershed. Much of the forest lands
are softwood pines used to manufacture paper products (facial tissue, toilet paper, hand towels, bags,
and boxes). Wetlands occupy about 26 percent of the watershed. Cultivated land covers approximately
11 percent, while developed land occupies over 10 percent of the area.
The populations of cities in the watershed increased from 10 to 30 percent between 1990 and 1999. The
largest city in the watershed is Tampa, FL. Most water used in the watershed is derived from
groundwater, primarily from the highly productive Floridan aquifer system.
M-4
-------
Alapana
03110252)
Hydrography
Water (Nat. Atlas Dataseti
US Census Populated Places
^^| Municipalities (pop ^ 50,000)
| County Boundanes
T Watershed with HUCBs
Withlacoochee
(03110203)
.Upper f~
Ochlockonee
b312000"2)
Upper Suwannee
(03110201)
Uacksonvi Me
Aucilla
(03110103)
J
Lower
Suwannee
(03.110205)
\
Santa Fe
(03110206)
I Lower* V
Ochlockonee\
(03120003)\\
achee
St. Marks
(03120001)
Gainesville,-
Pithlacrfascotee
(03100207)
Spring Hill
Tampa
Hillsbo'rough
" 0*3100205)
Alafia . f
(03100*204)
Clearwater
St. Petersburg
GULF OF
MEXICO
ittleManatee
^310,0203)
GCRP Model Areas - Georgia-Florida Coaslal Plain
Base Map
Figure 1. Location of the Georgia-Florida Coastal Plain model.
M-5
-------
Soil Characteristics
Soils in the watershed are described in STATSGO soil surveys. SWAT uses information drawn directly from the
soils data layer to populate the model.
Land Use Representation
Land use/cover in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage (Figure
2). NLCD land cover classes were aggregated according to the scheme shown in Table 1 for representation in the
20 Watershed model. SWAT uses the built-in hydrologic response unit (HRU) overlay mechanism in the
ArcSWAT interface. SWAT HRUs are formed from an intersection of land use and STATSGO major soils. The
distribution of land use in the watershed is summarized in Table 2.
M-6
-------
j _J County Boundaries
2001 NLCD Land Use
| Open water
] Developed, open space
] Developed, low intensity
| Developed, medium intensity
j^B Developed, high intensity
I | Barren land
| Deciduous forest
m Evergreen forest
I I Mixed forest
| | Scrub/shrub
^ Grassland/herbaceous
] Pasture/hay
] Cultivated crops
^] Woody wetlands
I | Emergent herbaceous wetlands
GULF OF
MEXICO
GCRP Model Areas - Georgia/Florida River Basins
Land Use Map
NAD_1983_Albers_metere - Map produced 12-23-2010 - P. Cada
20
40
80
• Kilometers
20
40
80
• Miles
TETRATECH
Figure 2. Land use in the Georgia-Florida Coastal Plain model.
M-7
-------
Table 1. Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area usually
accounted for as reach area
Deciduous
Evergreen
Mixed
Emergent & woody wetlands
Aquatic bed wetlands (not
emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
M-8
-------
Table 2. Land use distribution for the Georgia-Florida Coastal Plain (2001 NLCD) (mi )
HUC8
watershed
Little Manatee
03100203
Alafia
03100204
Hillsborough
03100205
Tampa Bay
03100206
Crystal-
Pithlachascotee
03100207
Aucilla
03110103
Upper
Suwannee
03110201
Alapaha
03110202
Withlacoochee
03110203
Little 031 10204
Lower
Suwannee
03110205
Santa Fe
03110206
Apalachee Bay-
St. Marks
03120001
Upper
Ochlockonee
03120002
Lower
Ochlockonee
03120003
Total
Open
water
3.28
14.54
6.95
14.74
13.46
2.56
8.51
11.15
7.96
6.22
8.63
19.51
4.47
5.53
18.55
121.29
Developed9
Open
space
12.89
57.09
109.61
85.28
171.27
37.36
92.49
83.39
82.00
38.31
85.64
70.92
91.70
42.28
70.17
950.82
Low
density
5.92
20.05
52.21
97.65
155.40
4.58
21.25
22.97
22.55
13.15
16.23
15.33
19.32
13.04
12.40
413.86
Medium
density
3.77
8.94
30.46
66.80
66.08
0.88
2.83
5.34
6.45
2.53
1.66
3.41
7.09
2.11
2.08
167.24
High
density
0.41
1.69
9.82
23.62
14.78
0.40
1.05
3.31
3.29
1.09
0.35
0.97
2.23
1.11
0.69
52.89
Barren
land
11.02
18.31
1.53
0.60
2.96
0.87
9.98
1.38
0.95
0.88
2.02
4.47
2.36
1.11
3.27
30.87
Forest
8.32
16.32
36.32
16.16
193.61
407.91
930.65
648.83
565.93
263.72
536.15
534.92
560.43
328.10
679.84
5,666.23
Shrub and
Grassland
9.61
63.67
30.35
17.35
49.35
80.61
304.63
141.50
133.57
57.54
295.65
268.37
90.09
67.64
91.14
1,597.45
Pasture/Hay
35.45
42.11
115.04
22.23
78.46
48.99
51.51
110.92
115.66
83.81
204.14
166.54
33.53
62.55
37.14
1,015.47
Cultivated
51.00
19.43
15.32
30.04
2.07
79.08
40.42
390.06
362.75
282.16
158.13
68.38
33.62
264.89
76.99
1,788.59
Wetland
64.58
154.34
238.80
115.03
304.20
301 .57
1,144.83
364.90
209.62
126.03
193.62
202.26
295.04
113.10
490.00
3,860.20
Total
206.25
416.49
646.41
489.49
1,051.64
964.81
2,608.16
1,783.76
1,510.74
875.42
1,502.22
1,355.07
1,139.88
901.45
1,482.26
15,664.90
aThe percent imperviousness applied to each of the developed land uses is as follows: open space (7.20%), low density (31.87%), medium density (60.14%), and high
density (87.47%).
M-9
-------
Point Sources
There are numerous point source discharges in the watershed. Only the major dischargers, generally defined as
those with a design flow greater than 1 MGD are included in the simulation (Table 3). The major dischargers are
represented at long-term average flows, without accounting for changes over time or seasonal variations.
Table 3. Major point source discharges in the Georgia-Florida Coastal Plain model
NPDES ID
FL0029033
FL0025518
GA0001279
GA0024082
GA0001678
FL0001465
GA0024911
GA0000124
FL0027880
GA0020222
GA0025852
FL0028126
FL0002518
FL0025526
FL0027839
FL0026557
FL0040983
FL0029653
FL0000523
FL0001589
FL0034657
FL0043869
FL0028061
FL0030406
FL0021326
FL0021857
FL0034789
FL0000159
Name
CITYOFQUINCYWWTP
ARVAH B. HOPKINS GENERATING
AFFINITY FOODS OF GA
THOMASVILLEWPCP
ENGELHARD CORPORATION
GOLDKIST INC - LIVE OAK PROCES
ADELWPCP
TIFTONALUMUNUMCO
JASPER-WWTP
VALDOSTA (MUD CREEK WPCP)
ASHBURN (WPCP)
STARKE-MUNICIPAL STP
ST. MARKS POWDER, INC.
SAM O. PURDOM GEN STATION
MONTICELLO-STP
PLANT CITY STP
HILLSBOROUGH CTY VALRICO WWTP
AOC, LLC
CF INDUSTRIES - BARTOW PHOS.
MOSAIC FERTILIZER, LLC - BARTO
CORONET INDUSTRIES INC
TAMPA ELEC-POLK POWER STATION
HILLSBOROUGH CO-SOUTHWEST WTP
TARPON SPRINGS STP
DUNEDIN-MAINLAND STP
CLEARWATER-MARSHALL ST STP
MID-COUNTY SERVICES, INC
PROGRESS ENERGY CRYSTAL RIVER
Design flow
(MGD)
1.5
1.9
6.5
1.5
2.5
1.2
3.2
1.2
1.7
0.8
1.0
8.0
6.0
4.0
4.0
6.0
10.0
0.9
0.7
Observed flow
(MGD)
(1991-2006 average)
1.0
0.3
0.5
3.7
1.1
1.3
1.3
0.3
0.7
2.1
0.9
1.7
21.4
21.0
10.7
3.6
5.2
0.1
4.0
1.7
62.7
2.7
1.6
3.2
5.8
6.7
1.1
0.0
M-10
-------
NPDES ID
FL0036366
FL0027821
FL0021865
FL0026603
FL0000264
FL0000809
FL0020940
FL0040614
FL0027651
FL0041670
Name
PROGRESS ENERGY CRYSTAL R 4&5
RIVER OAKS AWWTP
CLEARWATER-EAST WWTF
LARGO, CITY OF
IMC-AGRICO CO - PORT SUTTON
TAMPA ELEC COMPANY-FJ GANNON
HOWARD F CURREN AWTP
HILLSBORO CO - FALKENBURG RD A
CITY OF OLDSMAR
NORTHWEST REGIONAL WRF
Design flow
(MGD)
99.0
10.0
4.3
15.0
0.5
96.0
6.0
2.3
5.0
Observed flow
(MGD)
(1991-2006 average)
4.0
8.0
10.2
9.1
3.1
0.2
148.8
9.3
1.2
3.5
Most of these point sources have reasonably complete monitoring for total phosphorus and total suspended solids
(TSS). In the Georgia-Florida Coastal basin more dischargers also report total nitrogen (unlike other study areas)
due to concerns over nitrogen impacts on the coastal estuaries. The point sources were initially represented in the
model with the median of reported values for total phosphorus, total suspended solids and total nitrogen.
Meteorological Data
The required meteorological time series for the 20 Watershed SWAT simulations are precipitation and air
temperature. The 20 Watershed simulations do not include water temperature and uses a degree-day method for
snowmelt. SWAT estimates Penman-Monteith potential evapotranspiration using a statistical weather generator
for inputs other than temperature and precipitation. These meteorological time series are drawn from the
BASINS4 Meteorological Database (USEPA 2008), which provides a consistent, quality-assured set of
nationwide data with gaps filled and records disaggregated. Scenario application requires simulation over 30
years, so the available stations are those with a common 30-year period of record (or one that can be filled from
an approximately co-located station) that covers the year 2002. A total of 51 precipitation stations were identified
for use in the Georgia-Florida Coastal watershed model with a common period of record of 10/1/1971-9/30/2002
(Table 4). Temperature records are sparser; where these are absent temperature is taken from nearby stations with
an elevation correction.
Table 4. Precipitation stations for the Georgia-Florida Coastal Plain model
COOP ID
097276
098666
090140
098703
096087
080478
081046
084731
Name
QUITMAN 2 NW
THOMASVILLE 3 NE
ALBANY 3 SE
TIFTON
MOULTRIE 2 ESE
BARTOW
BROOKSVILLE CHIN HILL
LAKE CITY 2 E
Latitude
30.7836
30.8673
31.5339
31.4462
31.1769
27.8986
28.6164
30.1854
Longitude
-83.5691
-83.9318
-84.1488
-83.4766
-83.7492
-81.8432
-82.3657
-82.5942
Temperature
X
X
X
X
X
X
X
X
Elevation (m)
56
79
55
116
104
38
73
59
M-ll
-------
COOP ID
085275
087205
087851
099186
084273
084797
080975
082391
088758
088788
090586
093312
096879
082008
083956
084289
085879
087025
087886
088824
090010
091500
092266
092783
093386
093460
083986
086880
093465
087440
085539
085099
084394
098974
Name
MADISON
PLANT CITY
SAINT LEO
WAYCROSS 4 NE
INGLIS3E
LAKELAND
BRANFORD
DOWLING PARK 1 W
TALLAHASSEE WSO AP
TAMPA WSCMO AP
BAINBRIDGE INTL PAPER C
FARGO
PEARSON
CROSS CITY 2 WNW
HIGH SPRINGS
INVERNESS 3 SE
MONTICELLOWTP
PERRY
ST PETERSBURG
TARPON SPRINGS SWG PLNT
ABBEVILLE 4 S
CAMILLA 3 SE
CORDELE
DOUGLAS
FITZGERALD
FOLKSTON 3 SW
HILLSBOROUGH RVR ST PK
PARRISH
FOLKSTON 9 SW
RAIFORD STATE PRISON
MAYO
LIVE OAK
JASPER
VALDOSTA 2 S
Latitude
30.4517
28.0236
28.3379
31.2515
29.0254
28.0207
29.9625
30.2498
30.3932
27.9615
30.8229
30.6908
31.2928
29.6497
29.8287
28.8032
30.4923
30.0987
27.7632
28.1500
31.9381
31.1904
31.9848
31.4890
31.7108
30.7987
28.1429
27.6089
30.7400
30.0678
30.0565
30.2890
30.5229
30.8056
Longitude
-83.4119
-82.1422
-82.2600
-82.3127
-82.6157
-81.9218
-82.9107
-83.2593
-84.3533
-82.5403
-84.6175
-82.5632
-82.8422
-83.1663
-82.5972
-82.3124
-83.7832
-83.5742
-82.6272
-82.7500
-83.3078
-84.2035
-83.7758
-82.8205
-83.2516
-82.0181
-82.2269
-82.3478
-82.1277
-82.1928
-83.1818
-82.9650
-82.9446
-83.2736
Temperature
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Elevation (m)
37
37
58
44
9
44
9
16
17
6
58
35
62
13
20
12
30
14
2
2
73
53
94
71
113
9
16
18
37
37
20
37
45
81
M-12
-------
COOP ID
083153
089795
089120
094429
090406
080945
087429
089430
Name
FORT GREEN 12WSW
WOODRUFF DAM
USHER TOWER
HOMERVILLE 5 N
ASHBURN 3 ENE
BRADENTON 5 ESE
QUINCY3SSW
WEEKI WACHEE
Latitude
27.5706
30.7220
29.4084
31.0767
31.7003
27.4467
30.6001
28.5175
Longitude
-82.1377
-84.8742
-82.8186
-82.8002
-83.6230
-82.5014
-84.5499
-82.5755
Temperature
X
X
X
X
X
X
Elevation (m)
34
33
10
57
133
6
75
6
Watershed Segmentation
The Georgia-Florida Coastal basin was divided into 108 sub-watersheds for the purposes of modeling (Figure 3).
Ochlockonee River at USGS 02329000 was chosen for initial calibration. The model encompasses the complete
watershed and does not require specification of any upstream boundary conditions for application.
M-13
-------
.
USGS 02319000
*
USGS 02315500
USGS 02329000
68 \ £
USGS 02326900
USGS 02320500
USGS 02323500
^>
* * !» > -»
USGS 0230300
GULF OF
MEXICO
USGS Gages
Hydrography
Interstate
Water (Nat. Atlas Dataset)
US Census Populated Places
] Model Subbasins
GCRP Model Areas - Georgia/Florida River Basins
Model Segmentation
Figure 3. Model segmentation and USGS stations utilized for the Georgia-Florida Coastal Plain.
M-14
-------
Calibration Data and Locations
The specific site chosen for initial calibration was the Ochlockonee River at Havana, FL, a flow and water quality
monitoring location that approximately coincides with the mouth of an 8-digit HUC at its outflow to the
Ochlockonee River. The Ochlockonee River watershed was selected because there is a good set of flow and water
quality data available and the watershed lacks major point sources and impoundments. Additional calibration and
validation was pursued at multiple locations (Table 5). Parameters derived on the Ochlockonee River were not
fully transferable to other portions of the Georgia-Florida Coastal basin, and additional calibration was conducted
at multiple gage locations.
Table 5. Calibration and validation locations in the Georgia-Florida Coastal Plain
Station name
Alafia River at Lithia, FL
Hillsborough River near Zephyrhills, FL
Suwanee River at White Springs, FL
Withlacoochee River near Pinetta, FL
Suwanee River near Branford, FL
Suwanee River near Wilcox, FL
St. Marks River near Newport, FL
Ochlockonee River at Havana, FL
USGS ID
02301500
02303000
02315500
02319000
02320500
02323500
02326900
02329000
Drainage area
(mi2)
335
220
2430
2120
7880
9640
535
1140
Hydrology
calibration
X
X
X
X
X
X
X
X
Water quality
calibration
X
X
X
X
The model hydrology calibration period was set to Water Years 1992-2002 (within the 32-year period of record
for modeling). Hydrologic validation was then performed on Water Years 1982-1992. Water quality calibration
used calendar years 1992-2002, while validation used 1982-1992.
M-15
-------
SWAT Modeling
Assumptions
Hydrology Calibration
A spatial calibration approach was adopted for GCRP-SWAT modeling for the Georgia-Florida Coastal basin. A
systematic adjustment of parameters was adopted and some adjustments were applied throughout the basin. Most
of the calibration efforts were geared towards getting a closer match between simulated and observed flows at the
outlet of calibration focus area.
Land Use/Soil/Slope Definition
A 5/10/5 percent threshold was used for land use/soil/slope in the SWAT model while defining the HRUs. Urban
land use classes were exempted from the HRU overlay thresholds.
The calibration focus area (Ochlockonee River) includes nine subwatersheds and is generally representative of the
general land use characteristics of the overall watershed. The parameters were adjusted within the practical range
to obtain reasonable fit between the simulated and measured flows in terms of Nash-Sutcliffe modeling efficiency
and the high flow and low flow components as well as the seasonal flows.
The water balance of the upper portion of the Georgia-Florida Coastal basin predicted by the SWAT model over
the 32-year simulation period is as follows:
PRECIP = 1323.6 MM
SNOW FALL = 1.53 MM
SNOW MELT = 1.52 MM
SUBLIMATION = 0.01 MM
SURFACE RUNOFF Q = 167.16 MM
LATERAL SOIL Q = 18.39 MM
TILE Q = 0.00 MM
GROUNDWATER (SHAL AQ) Q = 223.44 MM
REVAP (SHAL AQ => SOIL/PLANTS) = 104.93 MM
DEEP AQ RECHARGE = 44.75 MM
TOTAL AQ RECHARGE = 374.05 MM
TOTAL WATER YLD = 397.68 MM
PERCOLATION OUT OF SOIL = 369.14 MM
ET = 766.5 MM
PET = 1576.8MM
TRANSMISSION LOSSES = 11.31 MM
The water balance of the lower portion of the Georgia-Florida Coastal basin predicted by the SWAT model over
the 32-year simulation period is as follows:
PRECIP = 1314.8 MM
SNOW FALL = 0.10 MM
SNOW MELT = 0.10 MM
SUBLIMATION = 0.00 MM
SURFACE RUNOFF Q = 169.83 MM
M-16
-------
LATERAL SOIL Q = 31.62 MM
TILE Q = 0.00 MM
GROUNDWATER (SHAL AQ) Q = 143.04 MM
REVAP (SHAL AQ => SOIL/PLANTS) = 100.62 MM
DEEP AQ RECHARGE = 162.45 MM
TOTAL AQ RECHARGE = 406.12 MM
TOTAL WATER YLD = 344.49 MM
PERCOLATION OUT OF SOIL = 409.81 MM
ET = 701.1 MM
PET = 1678.8MM
TRANSMISSION LOSSES = 0.00 MM
Hydrologic calibration adjustments focused on the following parameters:
• CN2 (initial SCS runoff curve number for moisture condition II)
• ESCO (soil evaporation compensation factor)
• SURLAG (surface runoff lag coefficient)
• SOL_AWC (available water capacity of the soil layer, mm water/mm of soil)
• ALPHA_BF (baseflow alpha factor, days)
• GW_DELAY (groundwater delay time, days)
• GWQMIN (threshold depth of water in the shallow aquifer required for return flow to occur, mm)
• GW_REVAP (groundwater "revap" coefficient)
• CH_N1 (Manning's "n" value for tributary channels)
• CH_N2 (Manning's "n" value for main channels)
Calibration results for the Ochlockonee River are summarized in Figure 4, Figure 5, Figure 6, Figure 7 and Table
6.
8000
6000
^000
2000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002 )
-Avg Modeled Flow (Same Period)
O-92
A-94
O-95
A-97 O-98
Month
A-00
O-01
Figure 4. Mean monthly flow at USGS 02329000 Ochlockonee River at Havana, FL - calibration period.
M-17
-------
• Avg Flow (10/1/1992 to 9/30/2002)
•Line of Equal Value
Best-Fit Line
2500
y|o.9707>j + 67.387
R2 = 0.8514
2500
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002)
Avg Modeled Flow (Same Period)
n ' lit! '
'an\Feb\Mar\Apr\May\Jun Jul Aug\Sep
500
1000 1500 2000 2500
Average Observed Flow (cfs)
10 11 12 1 234567
Month
Figure 5. Seasonal regression and temporal aggregate at USGS 02329000 Ochlockonee River at
Havana, FL - calibration period.
To Lower Bound Average Monthly Rainfall (in) -Median Observed Flow (10/1/1992 to 9/30/2002) Modeled (Median, 25th, 75th)
3500
Ocf Nov Dec Jan Feb
12
Figure 6. Seasonal medians and ranges at USGS 02329000 Ochlockonee River at Havana, FL -
calibration period.
M-18
-------
100000
•Observed Flow Duration (10/1/1992 to 9/30/2002 )
Modeled Flow Duration (10/1/1992 to 9/30/2002 )
10%
20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90%
100%
Figure 7. Flow exceedance at USGS 02329000 Ochlockonee River at Havana, FL - calibration period.
M-19
-------
Table 6. Summary statistics at USGS 02329000 Ochlockonee River at Havana, FL - calibration period
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 13
10-Year Analysis Period: 10/1/1992 - 9/30/2002
Flow/volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
11.57
6.08
0.66
2.06
3.14
5.10
1.27
3.36
0.58
Error Statistics
4.25
14.27
19.83
Observed Flow Gage
USGS 02329000 OCHLOCKONEE RIVER NR HAVANA, FLA.
Hydrologic Unit Code: 3120003
Latitude: 30.55408644
Longitude: -84.3840715
Drainage Area (sq-mi): 1140
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring Flow Volume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
15
30
33.10 » 30
1.60
-36.43
-4.28
10.50
0.711
0.539
0.793
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
11.10
5.32
0.93
1.72
2.36
5.01
2.00
3.51
0.53
Clear [
. ~^j_
Hydrology Validation
Hydrology validation for Ochlockonee River was performed for the period 10/1/1982 through 9/30/1992. The
validation achieves a moderately high coefficient of model fit efficiency, but is over on 10 percent highest flow
volume, and summer and fall seasonal volumes (Figure 8, Figure 9, Figure 10, Figure 11 and Table 7).
M-20
-------
10000
O-82
A-84
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1982 to 9/30/1992 )
•Avg Modeled Flow (Same Period)
O-85
A-87
O-88
Month
A-90
O-91
ro
OL
Figure 8. Mean monthly flow at USGS 02329000 Ochlockonee River at Havana, FL - validation period.
• Avg Flow (10/1/1982 to 9/30/1992)
Line of Equal Value
Best-Fit Line
4000
£
o
o
3000 -
% 2000 -•-
1000
1000 2000 3000
Average Observed Flow (cfs)
4000
4000
3000
•e
s"^*n-i
1
2000
1000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1982 to 9/30/1992)
Avg Modeled Flow (Same Period)
Figure 9. Seasonal regression and temporal aggregate at USGS 02329000 Ochlockonee River at
Havana, FL - validation period.
M-21
-------
To Lower Bound Average Monthly Rainfall (in) -Median Observed Flow (10/1/1982 to 9/30/1992) Modeled (Median, 25th, 75th)
6000
5000
10 11 12 1
Figure 10. Seasonal medians and ranges at USGS 02329000 Ochlockonee River at Havana, FL -
validation period.
•Observed Flow Duration (10/1/1982 to 9/30/1992 )
Modeled Flow Duration (10/1/1982 to 9/30/1992 )
100000
10%
20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90% 100%
Figure 11. Flow exceedance at USGS 02329000 Ochlockonee River at Havana, FL - validation period.
M-22
-------
Table 7. Summary statistics at USGS 02329000 Ochlockonee River at Havana, FL - validation period
REACH OUTFLOW FROM OUTLET 13
10-Year Analysis Period: 10/1/1982 - 9/30/1992
Flow/volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
13.85
7.22
0.51
1.95
1.60
8.22
2.07
4.01
0.53
Error Statistics
-5.54
-48.00
-1.84
-1.04
USGS 02329000 OCHLOCKONEE RIVER NR HAVANA, FLA.
Hydrologic Unit Code: 3120003
Latitude: 30.55408644
Longitude: -84.3840715
Drainage Area (sq-mi): 1140
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring Flow Volume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
11.57 » [ 30
-0.26
-31.19
-18.85
-12.44
0.799
0.649
0.895
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
14.66
7.36
0.99
1.97
1.44
8.24
3.00
4.94
0.61
Clear [
Hydrology Results for Larger Watershed
As described above, parameters determined for the gage at Ochlockonee River were initially transferred to other
gages in the watershed. However, changes to subwatershed level parameters were required to fit the model to the
observed flows. In all, calibration and validation was pursued at a total of eight gages throughout the watershed.
Results of the calibration and validation exercise are summarized in Table 8 and Table 9, respectively. Calibration
and validation results were acceptable at most gages.
M-23
-------
Table 8. Summary statistics (percent error): all stations - calibration period
Station
Error in total
volume:
Error in 50%
lowest flows:
Error in 10%
highest flows:
Seasonal volume
error - Summer:
Seasonal volume
error- Fall:
Seasonal volume
error - Winter:
Seasonal volume
error - Spring:
Error in storm
volumes:
Error in summer
storm volumes:
Daily Nash-
Sutcliffe
Coefficient of
Efficiency, E:
Monthly Nash-
Sutcliffe
Efficiency:
02301500
-4.77
32.79
-10.46
5.48
0.84
-26.89
-5.58
-28.50
-18.37
0.727
0.789
02303000
-0.21
8.10
-12.69
8.36
-2.28
-17.54
13.71
-9.30
0.77
0.675
0.736
02315500
-8.07
55.83
-14.74
28.40
14.11
-21.36
-21.16
-12.16
0.31
0.823
0.858
02319000
-1.91
9.25
-9.20
41.01
23.55
-15.83
-19.73
-34.73
-10.47
0.756
0.851
02320500
2.22
14.34
-3.57
10.89
19.48
-7.80
-2.98
19.91
48.43
0.821
0.865
02323500
3.45
11.46
1.83
20.53
18.31
-7.73
-6.27
-17.80
-12.89
0.802
0.838
02326900
4.32
15.97
0.32
-6.54
22.26
1.85
3.42
18.67
-12.42
0.623
0.654
02329000
4.25
-28.91
14.27
19.83
33.10
1.60
-36.43
-4.28
10.50
0.711
0.793
M-24
-------
Table 9. Summary statistics (percent error): all stations - validation period
Station
Error in total
volume:
Error in 50%
lowest flows:
Error in 10%
highest flows:
Seasonal volume
error - Summer:
Seasonal volume
error- Fall:
Seasonal volume
error - Winter:
Seasonal volume
error - Spring:
Error in storm
volumes:
Error in summer
storm volumes:
Daily Nash-
Sutcliffe
Coefficient of
Efficiency, E:
Monthly Nash-
Sutcliffe
Efficiency:
02301500
1.39
8.63
-3.36
3.79
54.39
-35.46
-7.98
-19.86
-22.13
0.416
0.393
02303000
3.17
-3.25
-4.80
9.87
7.84
-14.76
6.66
3.38
-3.79
0.720
0.760
02315500
-7.80
42.55
-19.13
6.07
97.49
-13.63
-30.57
-7.60
-12.02
0.722
0.755
02319000
-2.51
-36.60
-0.51
-14.33
29.83
1.90
-21.71
-35.83
-36.70
0.801
0.903
02320500
-11.68
-22.56
-6.19
-20.22
-3.66
-6.19
-17.18
26.96
-0.67
0.850
0.893
02323500
-17.18
-32.66
-4.00
-19.58
-17.71
-10.92
-22.74
-6.77
-23.61
0.788
0.820
02326900
-2.87
-4.93
2.96
-12.90
-3.76
1.01
3.05
11.11
-19.59
0.584
0.624
02329000
-5.54
-48.00
-1.84
-1.04
11.57
-0.26
-31.19
-18.85
-12.44
0.799
0.895
Water Quality Calibration and Validation
Initial calibration and validation of water quality was done on the Ochlockonee River (USGS 02329000), using
1992-2002 for calibration and 1982-1992 for validation. As with hydrology, water quality calibration was
performed on the later period as this better reflects the land use included in the model.
Calibration adjustments for sediment focused on the following parameters:
• SPCON (linear parameter for estimating maximum amount of sediment that can be re-entrained during
channel sediment routing)
• SPEXP (exponential parameter for estimating maximum amount of sediment that can be re-entrained
during channel sediment routing)
• CH_COV (channel cover factor)
• CH_EROD (channel erodibility factor)
• USLE_P (USLE support practice factor)
Simulated and estimated sediment loads at the Ochlockonee River station for both the calibration and validation
periods are shown in Figure 12 and statistics for the two periods are provided separately in .
Table 10. The key statistic in Table 10 is the relative percent error, which shows the error in the prediction of
monthly load normalized to the estimated load. Table 10 also shows the relative average absolute error, which is
the average of the relative magnitude of errors in individual monthly load predictions. This number is inflated by
outlier months in which the simulated and estimated loads differ by large amounts (which may be as easily due to
M-25
-------
uncertainty in the estimated load due to limited data as to problems with the model) and the third statistic, the
relative median absolute error, is likely more relevant and shows better agreement.
TSS
100,000
10,000
• Regression Loads
- Simulated Loads
Figure 12. Fit for monthly load of TSS at USGS 02329000 Ochlockonee River at Havana, FL.
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 02329000 Ochlockonee River at Havana, FL
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1992-1995)
9.5%
55.5%
36.7%
Validation period
(1982-1992)
-6.6%
59.4%
24.4%
Calibration adjustments for total phosphorus and total nitrogen focused on the following parameters:
• RHOQ (algal respiration rate at 20° C)
• PHOSKD (phosphorus soil partitioning coefficient)
• PSP (phosphorus availability index)
• RS2 (benthic source rate for dissolved P in the reach at 20° C)
• RS5 (organic P settling rate in the reach at 20° C)
• BC4 (rate constant for mineralization of organic P to dissolved P in the reach at 20° C)
• RS4 (rate coefficient for organic N settling in the reach at 20° C)
Results for the phosphorus simulation are shown in Figure 13 and .
Table 11. Results for the nitrogen simulation are shown in Figure 14 and Table 12. The model fit is generally
acceptable.
M-26
-------
Total P
1000
- Regression Loads
-Simulated Loads
Figure 13. Fit for monthly load of total phosphorus at USGS 02329000 Ochlockonee River at Havana, FL.
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 02329000 Ochlockonee River at Havana, FL
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1992-1995)
-7.4%
48.3%
22.8%
Validation period
(1982-1992)
-5.8%
52.4%
34.2%
Total N
1,600
-Averaging Loads
-Simulated Loads
rMcoco^-^-inintDtDi^i^ooooojojooT-T-rMOMcoco^-^-ir)
opopopopopopopopopopopopopopopq>q>q>q>q>q>q>q>q>q>q>
' ' ' ' ' ' ' ' ' ' ' ' '
'O Q.'O Q.'O Q.'O Q.'O Q.'O Q.' Q.' Q.' Q.' Q.' Q.' Q.' Q.
O
-------
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 02329000 Ochlockonee River at Havana, FL
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1992-1995)
-8.0%
49.2%
32.9%
Validation period
(1982-1992)
-5.0%
58.6%
21.0%
Water Quality Results for Larger Watershed
As with hydrology, a spatial calibration approach was adopted. Ochlockonee River watershed SWAT model
parameters for water quality were transferred to other portions of the watershed with necessary changes to
subbasin level parameters. Summary statistics for the SWAT water quality calibration and validation at other
stations in the watershed are provided in Table 13 and Table 14.
Table 13. Summary statistics for water quality at all stations - calibration period 1992-2002
Station
Relative Percent Error TSS Load
Relative Percent Error TP Load
Relative Percent Error TN Load
02301500
21.4%
16.5%
24.1%
02303000
10.0%
27.1%
-4.8%
02320500
-12.5%
6.6%
9.2%
02329000
9.5%
-7.4%
-8.0%
Table 14. Summary statistics for water quality at all stations - validation period 1982-1992
Station
Relative Percent Error TSS Load
Relative Percent Error TP Load
Relative Percent Error TN Load
02301500
-11.1%
-1.9%
-26.1%
02303000
-7.8%
4.2%
-20.2%
02320500
18.1%
10.9%
15.5%
02329000
-6.6%
-5.8%
-5.0%
M-28
-------
References
USEPA. 2008. Using the BASINS Meteorological Database (Version 2006). BASINS Technical Note 10.
Office of Water, U.S. Environmental Protection Agency, Washington, DC.
http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf (Accessed June,
2009).
M-29
-------
Appendix N
Model Configuration, Calibration and
Validation
Basin: Illinois River (Illin)
N-l
-------
Contents
Watershed Background N-4
Water Body Characteristics N-4
Soil Characteristics N-7
Land Use Representation N-7
Point Sources N-11
Meteorological Data N-14
Watershed Segmentation N-16
Calibration Data and Locations N-18
SWAT Modeling N-19
Assumptions N-19
Hydrology Calibration N-19
Hydrology Validation N-23
Hydrology Results for Larger Watershed N-26
Water Quality Calibration and Validation N-28
Water Quality Results for Larger Watershed N-31
References N-33
N-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes N-9
Table 2. Land use distribution for the Illinois River basin (2001 NLCD) (mi2) N-10
Table 3. Major point source discharges in the Illinois River basin N-ll
Table 4. Precipitation stations for the Illinois River basin model N-15
Table 5. Calibration and validation locations in the Illinois River basin N-18
Table 6. Summary statistics at USGS 05526000 Iroquois River near Chebanse, IL - calibration
period N-23
Table 7. Summary statistics at USGS 05526000 Iroquois River near Chebanse, IL - validation
period N-26
Table 8. Summary statistics (percent error): all stations - calibration period N-27
Table 9. Summary statistics: all stations - validation period N-28
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 05526000 Iroquois River near Chebanse, IL N-30
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 05526000 Iroquois River near Chebanse, IL N-31
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 05526000 Iroquois River near Chebanse, IL N-31
Table 13. Summary statistics for water quality at all stations - calibration period 1985-2001 (unless
otherwise noted) N-32
Table 14. Summary statistics for water quality at all stations - validation period 1978-1984 (unless
otherwise noted) N-32
Figures
Figure 1. Location of the Illinois River basin N-6
Figure 2. Land use in the Illinois River basin N-8
Figure 3. Model segmentation and USGS stations utilized for the Illinois River basin N-17
Figure 4. Mean monthly flow at USGS 05526000 Iroquois River near Chebanse, IL- calibration
period N-21
Figure 5. Seasonal regression and temporal aggregate at USGS 05526000 Iroquois River near
Chebanse, IL - calibration period N-21
Figure 6. Seasonal medians and ranges at USGS 05526000 Iroquois River near Chebanse, IL -
calibration period N-22
Figure 7. Flow exceedance at USGS 05526000 Iroquois River near Chebanse - calibration period N-22
Figure 8. Mean monthly flow at USGS 05526000 Iroquois River near Chebanse, IL - validation
period N-24
Figure 9. Seasonal regression and temporal aggregate at USGS 05526000 Iroquois River near
Chebanse, IL - validation period N-24
Figure 10. Seasonal medians and ranges at USGS 05526000 Iroquois River near Chebanse, IL -
validation period N-25
Figure 11. Flow exceedance at USGS 05526000 Iroquois River near Chebanse, IL - validation period... N-25
Figure 12. Fit for monthly load of TSS at USGS 05526000 Iroquois River near Chebanse, IL N-29
Figure 13. Fit for monthly load of total phosphorus at USGS 05526000 Iroquois River near Chebanse,
IL N-30
Figure 14. Fit for monthly load of total nitrogen at USGS 05526000 Iroquois River near Chebanse, IL.. N-31
N-3
-------
The Illinois River basin study area was selected as one of the 15 non-pilot application watersheds for the 20
Watershed study. Watershed modeling for the non-pilot areas is accomplished using the SWAT model only, and
model calibration and validation results are presented in abbreviated form. The majority of the Illinois River basin
lies in the state of Illinois except small portions extending into Wisconsin and Indiana. The Illinois River basin is
comprised of 12 HUC8 cataloging units.
Water Body Characteristics
The Illinois River is approximately 273 miles in length and is one of the major tributaries to the Mississippi River.
The Illinois River joins the Mississippi River near Grafton, IL, about 20 miles upstream from the confluence of
the Missouri and the Mississippi rivers. This study addresses the upper portion of the basin (Figure 1), which has
a drainage area of 17,004 mi2 (44,040 km2) and includes eleven HUCSs within HUC 0712 and HUC 0713
(Figure 1).
Within the upper portion of the basin (HUC 0712), over 80 percent of the land area is classified as part of the
Central Corn Belt Plains ecoregion. With the exception of the Chicago metropolitan area, land use in the Central
Corn Belt Plains is mostly corn and soybean cultivation for livestock feed crops and some livestock production.
The flat topography of the lower portion of the basin (HUC 0713) is in the Till Plains Section of the Central
Lowland physiographic province. The altitude of the land surface ranges from 600 to 800 ft above sea level. The
area of greatest topographic relief is along the river valley, where topographic relief can range from 200 to 400 ft.
The majority of the basin is extremely flat with less than 20 ft of relief.
Agriculture accounts for about 66 percent of the land use in the study area. Most of the recent urbanization is the
result of development of new suburban and residential areas. Urban areas account for about 18 percent of the land
use in the basin and are mainly concentrated in the metropolitan areas in and around Chicago. Forests cover about
10 percent of the study area and are concentrated along large-stream riparian areas.
Wetlands now make up a relatively small amount (1 percent) of land cover, but were once a major feature of the
basin. The majority of wetlands in the basin were drained prior to the 1850's for the development of farmland.
Remaining wetlands in the basin are mainly in riparian areas.
The climate of the Illinois River basin is classified as humid continental because of the cool, dry winters and
warm, humid summers. The average annual temperature for the Illinois River basin ranges from 46° F in the north
of the basin to 55° F in the south of the study area. Lake Michigan has a moderating effect on temperature near
the shoreline. Average annual precipitation, including snowfall, ranges from less than 32 inches in the northern
Wisconsin part of the basin to more than 38 inches near the southern and eastern Lake Michigan shoreline in the
Indiana part of the basin.
Streamflow in the study area consists of overland flow, groundwater discharge, agricultural drainage, and point-
source return flow. Local flooding generally is caused by isolated thunderstorms, whereas widespread flooding is
caused by more extensive thunderstorms that cover a wide area, by rapid snowmelt in the spring, or by a
combination of these factors. Flooding is common in the basin, in some years resulting in significant loss of life
and property.
The Illinois River connects Great Lakes at Chicago to the Mississippi River via the Illinois Waterway System. It
is also an important part of the Great Loop (the circumnavigation of Eastern North America by water). Originally
the Illinois and Michigan canal, opened in 1849, connected the Illinois River to the Chicago River. Later, the
Sanitary District of Chicago replaced the canal with the Chicago Sanitary and Shipping Canal and also reversed
N-4
-------
the flow of the Chicago River, originally flowing into Lake Michigan. Now, the Chicago Sanitary and Shipping
Canal is part of the Illinois Waterway. The Illinois Waterway System is a system of rivers, lakes, and canals that
provides shipping connection from the Great Lakes to the Gulf of Mexico via the Mississippi River. It consists of
336 miles (541 km) of water from the mouth of the Calumet River to the mouth of the Illinois River at Grafton,
IL. River traffic and flood control is managed by eight locks and dams operated by the Army Corps of Engineers.
The upper lock, T.J. O'Brien, is 7 miles from Lake Michigan on the Calumet River and the last lock is 90 miles
(140 km) upstream of the Mississippi River at the LaGrange lock and dam.
The watershed does not contain major reservoirs. However, there are a number of smaller lakes and reservoirs that
influence flow in this low gradient terrain. The river system in the Illinois River basin is highly manipulated by
human intervention including the reversal of the Chicago River and the massive Illinois Waterway. The river flow
and barge traffic is controlled by a series of lock and dams. Additional flow from Lake Michigan was not
considered in the model. Irrigation and groundwater pumping in the watershed are generally small and, therefore,
not included for the purposes of the 20 Watershed model.
N-5
-------
Hydrography
Water (Nat. Alias Dataset)
US Census Populated Places
^B Municipalities (pop » 50.000)
| | County Boundaries
~~1 Watershed with HUC8s
Grand Rapids
Michigan
Lake
Michigan
pper. Fox
(07120006
Lower Fox
(071200'07
Kankakee
(07,1-20'obl)
Davenport
Illinois
Iroqiiois
(07120002^
vermilion
(07130002)
Mackinaw
(07130004)
Bloomington
Lake-phautauqua
~^{07130003 '
m
Springfield
GCRP Model Areas - Ilinois River Basins
Base Map
Figure 1. Location of the Illinois River basin.
N-6
-------
Soil Characteristics
Soils in the watershed, as described in STATSGO soil surveys, fall primarily into hydrologic soil groups (HSGs)
B (moderately high infiltration capacity) and C (moderately high runoff potential). SWAT uses information drawn
directly from the soils data layer to populate the model.
Land Use Representation
Land use/cover in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage and is
predominantly row crop agriculture (Figure 2). NLCD land cover classes were aggregated according to the
scheme shown in Table 1 for representation in the 20 Watershed model. SWAT uses the built-in hydrologic
response unit (HRU) overlay mechanism in the ArcSWAT interface. SWAT HRUs are formed from an
intersection of land use and STATSGO soils. The distribution of land use in the watershed is summarized in
Table 2.
N-7
-------
Legend
Hydrography
= Interstate
I | County Boundaries
2001 NLCD Land Use
| Open water
~~| Developed, open space
| Developed, low intensity
^^| Developed, medium intensity
| Developed, high intensity
] Barren land
| Deciduous forest
^^| Evergreen forest
] Mixed forest
] Scrub/shrub
] Grassland/herbaceous
] Pasture/hay
] Cultivated crops
] Woody wetlands
Emergent herbaceous wetlands
GCRP Model Areas - Illnois River Basins
Land Use Map
NAD_1983_Albers_meters - Map produced 03-31-2011 - P. Cada
Figure 2. Land use in the Illinois River basin.
N-8
-------
Table 1. Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area usually
accounted for as reach area
Deciduous
Evergreen
Mixed
Emergent & woody wetlands
Aquatic bed wetlands (not
emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
N-9
-------
Table 2. Land use distribution for the Illinois River basin (2001 NLCD) (mi )
HUC 8 watershed
Kankakee
07120001
Iroquois
07120002
Chicago
07120003
Des Plaines
07120004
Upper Illinois
07120005
Upper Fox
07120006
Lower Fox
07120007
Lower Illinois-
Senachwine Lake
07130001
Vermilion
07130002
Lower Illinois-Lake
Chautauqua
07130003
Mackinaw
07130004
Total
Open
water
28.6
6.7
8.7
27.1
23.7
73.3
9.7
65.9
4.4
64.3
4.6
317.0
Developed3
Open
space
114.1
76.2
67.7
166.8
43.8
150.2
70.3
84.1
42.8
63.4
44.7
924.1
Low
density
133.0
60.1
238.1
414.4
42.5
187.8
103.9
76.3
47.9
70.9
41.5
1,416.3
Medium
density
21.7
4.9
175.2
174.0
6.7
49.1
30.2
20.8
6.2
21.5
7.5
517.9
High
density
7.7
1.4
87.3
81.2
2.5
14.9
8.9
7.9
1.5
4.5
1.8
219.5
Barren
land
1.1
0.8
1.2
2.0
1.1
4.3
1.7
2.4
1.6
0.7
0.2
17.1
Forest
329.9
86.9
60.9
145.2
66.3
285.5
69.9
228.2
29.5
376.2
66.4
1,745.1
Shrubland/
Grassland
95.8
7.8
29.4
76.4
27.6
63.9
30.2
15.5
6.6
4.7
2.2
360.1
Pasture/
Hay
108.3
42.6
4.1
26.0
14.2
135.4
38.5
70.3
20.4
97.8
47.7
605.2
Cultivated
2,013.6
1,837.2
32.2
258.3
900.3
536.0
739.1
1,353.6
1,171.0
866.0
929.5
10,636.9
Wetland
46.6
12.4
0.9
9.9
6.0
43.7
0.8
41.0
1.3
79.4
3.0
245.0
Total
2,900.4
2,136.9
705.9
1,381.3
1,134.8
1,544.1
1,103.1
1,966.0
1,333.1
1,649.4
1,149.1
17,004.1
aThe percent imperviousness applied to each of the developed land uses is as follows: open space (8.83%), low density (32.36%), medium density (61.24%), and high
density (88.70%).
N-10
-------
Point Sources
There are numerous point source discharges in the watershed. Only the major dischargers, with a design flow
greater than 1 MGD, are included in the simulation (Table 3). The major dischargers are represented at long-term
average flows, without accounting for changes overtime or seasonal variations.
Table 3. Major point source discharges in the Illinois River basin
NPDES ID
IL0025135
IL0001830
IL0027839
IL0001953
IL0001970
IL0002232
IL0002526
IL0034495
IL0001414
IL0002291
IL0021288
IL0028576
IL0042412
IL0046213
IL0001392
IL0002631
IL0029424
IL0030660
IL0031216
IL0030384
IL0023221
IL0022004
IL0030457
IL0021059
IL0048151
IL0048321
IL0002917
IL0021113
IL0002224
IL0021130
IL0021547
IL0023469
IL0026352
Name
BEARDSTOWN SD STP
CATERPILLAR INC.-MAPLETON
CANTON WEST STP
AVENTINE RENEWABLE ENERGY
AMEREN ENERGY RESOURCES-EDWARD
MIDWEST GENERATION-POWERTON
KEYSTONE STEEL AND WIRE
PEKINSTPtfl
CATERPILLAR INC-MOSSVILLE
CATERPILLAR INC.-EAST PEORIA
PEORIASDSTP
EAST PEORIA STP #1
WASHINGTON STP #2
EAST PEORIA STP #3
EMERALD PERFORMANCE MATERIALS
ARCELORMITTAL HENNEPIN INC
LASALLE STP
PERU STP #1
SPRING VALLEY STP
OTTAWA STP
MENDOTASTP
STREATORSTP
PONTIAC STP
MARSEILLES WWTP
EXELON GENERATION CO, LLC
EXELON GENERATION-BRAIDWOOD
EQUISTAR CHEMICALS, LP
MORRIS STP
EXELON GENERATION CO.,LLC
BLOOMINGDALE-REEVES WRF
GLENBARD WASTEWATER AUTH-MAIN
WEST CHICAGO REGIONAL STP
CARD L STREAM WRC
Design flow
(MGD)
1.13
14.15
3.43
34.09
357.90
530.10
10.12
4.50
1.44
3.10
37.00
4.22
1.50
1.20
0.95
7.25
3.33
3.00
1.10
4.00
2.40
3.30
3.50
1.23
35.02
22.85
6.47
2.50
481.10
3.45
16.02
7.64
6.50
Observed flow
(MGD)
(1991-2006
average)
1.31
0.89
2.5284
0.54
5.06
11.13
5.06
3.43
0.86
2.18
22.97
3.85
2.02
0.18
0.78
3.05
1.40
5.68
0.86
2.84
0.98
3.44
2.97
1.04
0.04
0.15
2.38
2.13
0.15
2.20
14.16
4.55
4.61
N-ll
-------
NPDES ID
IL0027618
IL0028967
IL0031739
IL0032735
IL0034061
IL0034479
IL0036137
IL0048721
IL0055913
IL0074373
IL0001619
IL0001643
IL0001732
IL0002216
IL0002453
IL0002861
IL0033553
IL0020532
IL0020559
IL0024201
IL0045403
IL0001589
IL0022519
IL0022586
IL0029611
IL0032760
IL0025089
IL0021784
IL0022179
IL0022161
IN0023621
IN0037176
IN0030651
IN0020991
IN0025577
IN0038172
IN0024520
IN0020427
IN0021466
Name
BARTLETT WWTP
GLENDALE HEIGHTS STP
WHEATONSDWWTF
BOLINGBROOKWRF#2
NAPERVILLE SPRINGBROOK WRC
HANOVER PARK STP #1
MWRDGC HANOVER PARK WRP
ROSELLE-J BOTTERMAN STP
MINOOKASTP
PLAINFIELD NORTH STP
INEOS NOVA LLC
BP AMOCO CHEMICAL-JOLIET
CATERPILLAR, INC.-JOLIET
MIDWEST GENERATION,LLC-JOLIET9
STEPAN COMPANY-ELWOOD
EXXONMOBIL OIL-JOLIET REFINERY
JOLIET WEST STP
FRANKFORT WEST WWTP
NEW LENOX STP #1
MOKENASTP
FRANKFORT NORTH STP
CITGO PETROLEUM CORPORATION
JOLIET EAST STP
FLAGG CREEK WRD MCELWAIN STP
LOCKPORT STP
IL AMERICAN WATER-SANTA FE
MANTENO WPCC
KANKAKEE RIVER METRO AGENCY
MOMENCESTP
WATSEKASTP
LOWELL MUNICIPAL STP
TWIN LAKES UTILITIES, INC WWTP
SOUTH HAVEN SEWER WORKS, INC.
PLYMOUTH MUNICIPAL STP
LAPORTE MUNICIPAL STP
ROLL CO ATE R, INC.
SOUTH BEND MUNICIPAL STP
BREMEN MUNICIPAL WWTP
NAPPANEE MUNICIPAL STP
Design flow
(MGD)
3.68
5.26
8.90
3.00
26.25
2.42
12.00
1.22
1.09
3.50
0.12
2.32
2.12
398.70
0.88
15.50
14.00
1.30
2.52
2.50
1.35
5.82
18.20
12.00
3.40
1.00
1.15
25.00
1.60
1.60
4.00
1.10
2.00
3.50
7.00
0.14
37.70
1.30
1.9
Observed flow
(MGD)
(1991-2006
average)
2.30
3.31
5.99
2.85
16.03
1.15
9.22
0.78
0.42
2.28
0.08
1.15
0.36
8.55
0.67
2.84
10.91
1.04
1.70
1.47
0.92
4.18
18.02
12.11
3.64
0.32
1.37
12.50
1.04
0.84
259.50
0.86
1.03
2.15
5.40
0.07
40.13
0.91
1.55
N-12
-------
NPDES ID
IL0030970
IL0036412
IL0062260
IL0020087
IL0022543
IL0022705
IL0035891
IL0020583
IL0021733
IL0023329
IL0027944
IL0028282
IL0028541
IL0028657
IL0028665
IL0001716
IL0020109
IL0020516
IL0021067
IL0031933
IL0053457
IL0066257
IL0031861
IL0034282
IL0020354
IL0020958
WI0022926
WI0031496
WI0028291
WI0028754
WI0038938
WI0020559
WI0023469
WI0029971
IL0020575
IL0020061
IL0026280
IL0027367
IL0030953
Name
SANDWICH STP
YORKVILLE-BRISTOL SD STP
ELBURN STP
GEN EVA STP
BATAVIA STP
ST. CHARLES EASTSIDE STP
FOX RIVER WRD WEST STP
FOX RIVER GROVE STP
LAKE IN THE HILLS SD STP
ALGONQUIN STP
CARPENTERSVILLE STP
CRYSTAL LAKE STP #2
EAST DUNDEE WWTP
FOX RIVER WRD SOUTH STP
FOX RIVER WRD NORTH STP
ROHM & HAAS CHEMICAL LLC
WAUCONDASTP
GARY STP
MCHENRY CENTRAL STP
NORTHERN MORAINE WW REC DIST
CRYSTAL LAKE STP #3
MCHENRY SOUTH WWTP
WOODSTOCK NORTH STP
WOODSTOCK SOUTH STP
ANTIOCH STP
FOX LAKE NW REGIONAL WRF
BURLINGTON WATER POLLUTION CON
SALEM UTILITY DISTRICT NO 2
UNION GROVE VILLAGE
WESTERN RACINE COUNTY SEWERAGE
TRENT TUBE DIV OF CRUCIBLE PLA
SUSSEX WASTEWATER TREATMENT FA
BROOKFIELD, CITY OF
WAUKESHACITY
PRINCETON STP
WOOD DALE NORTH STP
ITASCA STP
ADDISON SOUTH-A.J. LAROCCA STP
SALT CREEK SD STP
Design flow
(MGD)
1.50
3.62
1.27
5.00
4.20
9.00
5.00
1.25
4.20
3.00
4.50
5.80
2.30
25.00
7.75
2.00
1.40
2.80
3.00
2.00
1.70
1.50
3.50
1.75
1.60
9.00
2.50
1.57
1.00
0.92
0.00
1.00
10.00
16.00
2.15
1.97
2.60
3.20
3.30
Observed flow
(MGD)
(1991-2006
average)
0.64
1.26
0.55
3.18
2.98
4.41
1.20
0.80
2.21
2.31
2.50
4.01
0.37
15.62
5.10
1.62
1.48
1.58
2.06
0.93
0.56
0.65
2.20
1.02
1.46
6.32
3.01
0.89
0.91
0.95
0.20
1.42
7.52
6.49
1.82
1.83
2.01
1.95
3.01
N-13
-------
NPDES ID
IL0033812
IL0034274
IL0036340
IL0022055
IL0022071
IL0022501
IL0020796
IL0035092
IL0071366
IL0002178
IL0002186
IL0002208
IL0028347
IL0030171
IL0028088
IL0028061
IN0023060
IN0024457
IN0039331
IL0024473
IL0027723
Name
ADDISON NORTH STP
WOOD DALE SOUTH STP
MWRDGC EGAN WRP
LCDPW-DESPLAINESSTP
LCDPW-NEW CENTURY TOWN STP
MUNDELEINSTP
LINDENHURSTSDSTP
NSSDGURNEESTP
LAKE COUNTY DPW-MILL CREEK WRF
MIDWEST GENERATION,LLC-FISK
MIDWEST GENERATION,LLC-CRAWFRD
MIDWEST GENERATION,LLC-WILL CO
DEERFIELDWRF
NSSD CLAVEY ROAD STP
MWRDGC NORTHSIDE WRP
MWRDGC CALUMET WRP
HAMMOND MUNICIPAL STP
SCHERERVILLE MUNICIPAL STP
DYER MUNICIPAL WWTP
AQUA ILLINOIS-UNIV PARK
THORN CREEK BASIN SD STP
Design flow
(MGD)
5.30
1.13
30.00
16.00
6.00
4.95
2.00
23.60
1.00
241.20
356.80
715.70
3.50
17.80
333.00
354.00
37.80
8.75
2.60
2.17
15.94
Observed flow
(MGD)
(1991-2006
average)
3.58
0.68
24.76
10.32
2.48
4.26
1.11
15.32
0.59
0.75
1.07
0.93
3.02
14.11
262.75
266.94
39.92
4.57
1.36
2.05
15.83
Most of these point sources have reasonably good monitoring for total phosphorus and total suspended solids
(TSS), but not for total nitrogen. In many cases, only ammonia nitrogen is monitored. The point sources were
initially represented in the model with the median of reported values for the constituents (total phosphorus, total
nitrogen, and TSS) and an assumed total nitrogen concentration of 11.2 mg/L and assumed total phosphorus
concentration of 7.0 mg/L for secondary treatment facilities (Tetra Tech 1999). However, in cases where point
source contribution was deemed unusually high, average concentration of the available data was assumed for the
missing ones.
Meteorological Data
The required meteorological time series for the 20 Watershed SWAT simulations are precipitation and air
temperature. The 20 Watershed simulations do not include water temperature simulation and use a degree-day
method for snowmelt. SWAT estimates Penman-Monteith potential evapotranspiration using a statistical weather
generator for inputs other than temperature and precipitation. These meteorological time series are drawn from the
BASINS4 Meteorological Database (USEPA 2008), which provides a consistent, quality-assured set of
nationwide data with gaps filled and records disaggregated. Scenario application requires simulation over 30
years, so the available stations are those with a common 30-year period of record (or one that can be filled from
an approximately co-located station) that covers the year 2001. A total of 72 precipitation stations were identified
for use in the Illinois River basin model with a common period of record of 10/1/1971-9/30/2001 (Table 4).
Temperature records are sparser; where these are absent temperature is taken from nearby stations with an
elevation correction.
N-14
-------
Table 4. Precipitation stations for the Illinois River basin model
COOP
ID
471205
473058
474174
474457
475474
475479
478723
478937
479190
110203
110338
110442
110492
110583
110761
111250
111420
111475
111549
111577
111627
112011
112223
112736
112923
113413
113940
114013
114198
114530
114603
114710
114805
115272
115326
115334
115372
115413
115493
115712
Name
WI471205
WI473058
WI474174
WI474457
WI475474
WI475479
WI478723
WI478937
WI479190
IL110203
IL110338
111 10442
IL110492
IL110583
IL110761
IL111250
IL111420
IL111475
IL111549
IL111577
IL111627
IL112011
IL112223
IL112736
IL112923
IL113413
IL113940
IL114013
IL114198
IL114530
IL114603
IL114710
111 14805
IL115272
IL115326
IL115334
IL115372
IL115413
IL115493
IL115712
Latitude
42.6508
43.2390
42.5609
42.5937
43.0719
42.9550
42.6904
43.0064
42.8508
42.4811
41.7806
42.1153
40.0179
42.2551
40.4957
40.5447
41.3978
40.7407
41.9950
41.7373
40.9156
41.4492
41.9342
42.0629
40.7512
40.4732
40.3431
41.3017
40.4745
41.5034
41.1382
41.2484
41.0415
40.5526
42.2928
40.5019
41.3287
40.2018
42.3103
40.9126
Longitude
-88.2544
-88.1222
-87.8156
-88.4347
-88.0293
-87.9043
-88.0336
-88.2492
-88.7246
-88.0994
-88.3092
-88.1638
-90.4381
-88.8644
-89.0006
-90.0211
-88.2818
-88.7128
-87.9335
-87.7775
-89.5031
-87.6222
-88.7755
-88.2861
-88.4983
-88.3653
-90.0164
-89.3157
-87.6557
-88.1028
-87.8855
-89.8991
-89.4060
-89.3336
-88.6469
-90.3892
-88.7532
-89.6949
-88.2524
-89.0339
Temperature
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Elevation (ft)
751
850
600
846
725
669
732
830
876
751
659
876
449
738
774
650
505
709
659
620
535
663
873
764
689
751
459
459
709
545
640
781
459
709
814
640
489
574
741
751
N-15
-------
COOP
ID
116526
116616
116661
116711
116725
116753
116819
116910
117004
117150
117551
118353
118756
118870
118916
119021
119029
119221
119816
123418
124527
124782
125174
125535
127298
127482
128187
128999
129222
129240
129511
129670
Name
IL116526
IL116616
IL116661
IL116711
IL116725
IL116753
IL116819
IL116910
IL117004
IL117150
IL117551
IL118353
IL118756
IL118870
IL118916
IL119021
IL119029
IL119221
IL119816
IN123418
IN124527
IN124782
IN125174
IN125535
IN127298
IN127482
IN128187
IN128999
IN129222
IN129240
IN129511
IN129670
Latitude
41.3283
41.4933
41.7123
40.6675
41.3270
41.3503
40.7570
40.8854
40.9314
40.3131
40.1159
41.0909
41.3242
39.9451
41.5520
40.7928
42.3493
41.8129
40.7765
41.5575
40.7592
41.5269
41.2647
41.1590
40.9357
41.0659
41.7073
41.5115
41.4437
41.2390
41.1947
41.0265
Longitude
-88.9106
-87.6800
-88.9989
-89.6838
-87.7857
-89.1072
-88.1828
-88.6389
-89.7800
-88.1594
-90.5608
-88.8157
-88.9857
-90.2099
-89.5989
-87.7556
-87.8828
-88.0728
-90.0203
-85.8824
-87.4352
-86.2691
-87.4177
-86.9013
-87.1564
-86.2094
-86.3331
-87.0378
-86.9300
-85.8700
-87.0578
-86.5871
Temperature
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Elevation (ft)
525
709
951
650
719
620
669
650
100000
741
659
610
459
620
689
620
699
679
676
876
696
840
666
696
650
771
774
801
735
810
666
689
Watershed Segmentation
The Illinois River basin was divided into 100 subwatersheds for the purposes of modeling (Figure 3). The model
encompasses the complete watershed without any external area draining into it and, therefore, does not require
specification of any upstream boundary conditions for application. It should be noted, however, the Calumet
River (subbasin 98) discharging to the Chicago Sanitary and Shipping Canal (CSSC) and the CSSC itself
(subbasin 95) were disconnected from contributing to the downstream flow and the flow from CSSC was
simulated as a point source.
N-16
-------
Legend
A USGS gages
Hydrography
Interstate
Water (Nat. Atlas Dataset)
US Census Populated Places
County Boundaries
Model Subbasins
Lake
Michigan
Wisconsin
05532500
Chicago
USGS05543500
GCRP Model Areas - Illnois River Basins
Model Segmentation
NAD 1983 Albers meters-Map produced 03-31-2011 - P. Cada
Figure 3. Model segmentation and USGS stations utilized for the Illinois River basin.
N-17
-------
Calibration Data and Locations
The specific site chosen for initial calibration was the Iroquois River (HUC8: 07120002) near Chebanse, IL, a
flow and water quality monitoring location that approximately coincides with the mouth of an 8-digit HUC just
before Iroquois River joins Kankakee River. The Iroquois watershed was selected because there is a good set of
flow and water quality data available and the watershed has no major point sources or impoundments. Parameters
derived on the Iroquois River were not fully transferable to other portions of the Illinois River watershed and
additional calibration and validation was conducted at multiple gage locations (Table 5).
Table 5. Calibration and validation locations in the Illinois River basin
Station name
Iroquois River at Chebanse, IL
Kankakee River at Momence, IL
Des Plaines River at Riverside, IL
Fox River at Dayton, IL
Illinois river at Marseilles, IL
USGS ID
05526000
05520500
05532500
05552500
05543500
Drainage area
(mi2)
2,091
2,294
630
2,642
8,259
Hydrology
calibration
X
X
X
X
X
Water quality
calibration
X
X
X
X
X
The model hydrology calibration period was set to Water Years 1992-2001 (within the 30-year period of record
for modeling). Hydrologic validation was then performed on Water Years 1982-1992. Water quality calibration
used calendar years 1985-2001, while validation used 1978-1984. However, there was some variation to this time
period across the monitoring stations depending on the availability of monitored data.
N-18
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
McHenry Lock and Dam on the Fox River and Peoria Lock and Dam on the Illinois River were represented as
reservoirs in the model. Pertinent reservoir information including surface area and storage at principal (normal)
and emergency spillway levels for the reservoirs modeled were obtained from the National Inventory of Dams
(NID) database. The SWAT model provides four options to simulate reservoir outflow: measured daily outflow,
measured monthly outflow, average annual release rate for uncontrolled reservoir, and controlled outflow with
target release. Keeping in view the 20 Watershed climate change impact evaluation application, it was assumed
that the best representation of the reservoirs was to simulate them without supplying time series of outflow
records. Therefore, the target release approach was used in the GCRP-SWAT model.
Another important characteristic of the watershed is the widespread presence of subsurface tile drainage.
Installation of tile drainage has converted what were predominantly glacial plain outwash depressional wetlands
into productive farmland. The presence of tile drains, which include both surface and subsurface inlets, has
radically altered the natural hydrology of the area. Surface inlet tile drains, in particular, may also play a
significant role in the transport of sediment and pollutants from agricultural land to the river. It is not feasible to
simulate individual tile drain systems at the large watershed scale. Further, neither the location nor the total
density of tile drainage is known throughout the watershed. In most areas, only the public tile drains and ditches
are documented in spatial coverage, and the extent of private tile drains is known only for limited areas. The
SWAT model allows for some representation of tile drains in the form of three parameters: depth to the tile
drains, time to drain soil to field capacity, and tile drain lag time. Tile drains were applied on poorly drained soils
(identified from STATSGO data) under cultivation with slopes less than one percent.
Although, a spatial calibration approach was adopted for GCRP-SWAT modeling for the Illinois River basin,
adjustments to specific subwatesheds were kept as minimal as possible. However, a systematic adjustment of
parameters was been adopted and some adjustments were applied throughout the watershed. Most of the
calibration efforts were geared toward getting a closer match between simulated and observed flows at the outlet
of the calibration focus area.
A 5/10/5 percent threshold was used for land use/soil/slope in the SWAT model while defining the HRUs. The
cropland HRUs were split into corn and soybean in the ratio 1:1. Urban land use classes were exempted from the
HRU overlay thresholds.
The calibration focus area (Iroquois River) includes 10 subwatersheds and is representative of the general land use
characteristics of the overall Illinois River basin. The parameters were adjusted within the practical range to
obtain reasonable fit between the simulated and measured flows in terms of Nash-Sutcliffe modeling efficiency
and the high flow and low flow components as well as the seasonal flows.
The overall water balance of the whole Illinois River basin predicted by the SWAT 20 Watershed model over the
30-year simulation period is as follows:
PRECIP = 958.1 MM
SNOW FALL = 127.74 MM
N-19
-------
SNOW MELT = 122.18 MM
SUBLIMATION = 6.13 MM
SURFACE RUNOFF Q = 264.25 MM
LATERAL SOIL Q = 2.29 MM
TILE Q = 21.22 MM
GROUNDWATER (SHAL AQ) Q = 33.99 MM
REVAP (SHAL AQ => SOIL/PLANTS) = 78.22 MM
DEEP AQ RECHARGE = 4.71 MM
TOTAL AQ RECHARGE = 94.18 MM
TOTAL WATER YLD = 319.41 MM
PERCOLATION OUT OF SOIL = 91.80 MM
ET = 579.0 MM
PET = 1106.2MM
TRANSMISSION LOSSES = 2.35 MM
Hydrologic calibration adjustments focused on the following parameters:
• Curve numbers (all landuse except forest)
• CN (curve number) coefficient
• FFCB (fraction of field capacity)
• SURLAG (surface runoff lag coefficient)
• Baseflow factor
• GW_DELAY (groundwater delay time)
• GWQMN (threshold depth of water in the shallow aquifer required for return flow to occur)
• SHALLST (Initial depth of water in the shallow aquifer)
• CANMAX (maximum canopy storage)
• RevapMN (threshold depth of water in the shallow aquifer required for "revap" or percolation to the deep
aquifer to occur
• CH_K2 (channel hydraulic conductivity)
• CH_N2 (channel Mannings' coefficient)
• Sol_AWC (available water capacity of the soil layer, mm water/mm of soil)
• Depth to impervious surface
• Snow parameters SMTMP, SMFMX, SMFMN, TIMP
• Tile drain parameters (DDRAIN, TDRAIN and GDRAIN)
• Average wind speed in the weather database
Calibration results for the Iroquois River are summarized in Figures 4 through 7 and Table 6. In general, the
model captured the timing of the peaks well but tends to underestimate the high flows resulting in overall
underestimation by 17%.
N-20
-------
10000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2001 )
•Avg Modeled Flow (Same Period)
O-92 O-93 O-94 O-95
O-96 O-97
Month
O-98 O-99 O-OO
Figure 4. Mean monthly flow at USGS 05526000 Iroquois River near Chebanse, IL- calibration period.
Avg Flow (10/1/1992 to 9/30/2001)
• Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2001)
•Avg Modeled Flow (Same Period)
4000
•6
3000
T3
-------
Average Monthly Rainfall (in)
• Median Observed Flow (10/1/1992 to 9/30/2001)
[Observed (25th, 75th)
Modeled (Median, 25th, 75th)
6000
10 11 12 1
Figure 6. Seasonal medians and ranges at USGS 05526000 Iroquois River near Chebanse, IL -
calibration period.
•Observed Flow Duration (10/1/1992 to 9/30/2001 )
Modeled Flow Duration (10/1/1992 to 9/30/2001 )
100000
0.1
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 7. Flow exceedance at USGS 05526000 Iroquois River near Chebanse - calibration period.
N-22
-------
Table 6. Summary statistics at USGS 05526000 Iroquois River near Chebanse, IL - calibration period
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 52
9-Year Analysis F^riod: 10/1/1992 - 9/30/2001
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3):
Simulated Srjring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
11.89
4.85
1.03
2.68
1.63
3.49
4.09
4.06
0.85
Error Statistics
-16.99
-0.58
-27.20
29.82
Observed Flow Gage
USGS 05526000 IROQUOIS RIVER NEARCHEBANSE, IL
Hydrologic Unit Code: 7120002
Latitude: 41. 008921 5
Longitude: -87.8233719
Drainage Area (sq-rri): 2091
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
-18.41 » | 30
-23.84
-27.98
-35.59
-14.89
0.699
0.563
0.768
30
30 _
50
Model accuracy increases
as E or E' approaches 1.0
14.32
6.67
1.03
2.07
1.99
4.58
5.68
6.30
1.00
Clear [
Hydrology Validation
Hydrology validation for the Iroquois River was performed for the period 10/1/1982 through 9/30/1992. Results
are presented in Figures 8 through 11 and Table 7. The validation achieved a high Nash-Sutcliffe model
efficiency, but is under on 50 percent low volume and over on summer volume (Figure 8, Figure 9, Figure 10,
Figure 11, and Table 7). Although, the validation period is from 1982 to 1992 and the landuse data used in the
model was obtained from 2001 NLCD, the model performance was very much comparable for both calibration
and validation periods. This could be due to no major changes in the landuse/land management in the watershed
and a somewhat consistent weather pattern.
N-23
-------
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1982 to 9/30/1992 )
•Avg Modeled Flow (Same Period)
8000
f
ro
o
O-82
A-84
O-91
Figure 8. Mean monthly flow at USGS 05526000 Iroquois River near Chebanse, IL - validation period.
• Avg Flow (10/1 /1982 to 9/30/1992)
• • • • • Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1982 to 9/30/1992)
•Avg Modeled Flow (Same Period)
4000
4000
1000 2000 3000 4000
Average Observed Flow (cfs)
10 11 12 1 23456789
Month
Figure 9. Seasonal regression and temporal aggregate at USGS 05526000 Iroquois River near
Chebanse, IL - validation period.
N-24
-------
i Observed (25th, 75th)
• Median Observed Flow (10/1/1982 to 9/30/1992)
Average Monthly Rainfall (in)
Modeled (Median, 25th, 75th)
5000
Figure 10. Seasonal medians and ranges at USGS 05526000 Iroquois River near Chebanse, IL -
validation period.
•Observed Flow Duration (10/1/1982 to 9/30/1992 )
Modeled Flow Duration (10/1/1982 to 9/30/1992 )
100000
10000
I
LL 1000 -
D)
ro
Q
100 - =
10%
20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 11. Flow exceedance at USGS 05526000 Iroquois River near Chebanse, IL - validation period.
N-25
-------
Table 7. Summary statistics at USGS 05526000 Iroquois River near Chebanse, IL - validation period.
REACH OUTFLOW FROM OUTLET 52
10-Year Analysis F^riod: 10/1/1982 - 9/30/1992
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
_JpJal^f^inTulated_highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3):
Simulated Srjring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
12.15
5.02
0.95
1.57
3.58
4.14
2.86
4.28
0.54
Error Statistics
-2.98
-21.27
-11.00
36.69
USGS 05526000 IROQUOIS RIVER NEARCHEBANSE, IL
Hydrologic Unit Code: 7120002
Latitude: 41. 008921 5
Longitude: -87.8233719
Drainage Area (sq-rri): 2091
Total Observed In-stream Flow:
_j£t^l_of^b^ej^«djTighe^tjm%JlOT/s^_
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
21.34 » | 30
-10.30
-24.88
-17.48
-4.76
0.674
0.457
0.712
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
12.52
5.64
1.21
1.15
2.95
4.61
3.81
5.19
0.57
Clear [
Hydrology Results for Larger Watershed
Parameters determined for the Iroquois River monitoring station were not fully transferable to all other areas in
the watershed, especially the Kankakee and Fox gages. Kankakee River originally drained one of the largest
wetlands in North America and has since been significantly altered. The river now flows through primarily rural
lands of reclaimed croplands. During the calibration, Channel Mannings' N had to be reduced for the Kankakee
drainage area to reduce the peak flow rates in the channel. Higher upland erosion rates simulated by SWAT were
controlled by reducing the P-factor for cropland HRUs for both Kankakee and Fox drainage areas. In general,
calibration results were acceptable at most gages, as summarized in Table 8. Model performance was relatively
poor at the Kankakee and Fox stations.
Results of the validation exercise are summarized in Table 9. Summer season flows were overestimated both
during the calibration and validation periods at all stations except Kankakee. Unlike the calibration period, fall
season flows were overestimated during the validation period. Overall, hydrology was simulated well at all gages,
except the Kankakee station where low flows were underpredicted and high flows were over predicted, resulting
in an overprediction of total storm volumes.
N-26
-------
Table 8. Summary statistics (percent error): all stations - calibration period
Station
Error in total
volume:
Error in 50%
lowest flows:
Error in 10%
highest
flows:
Seasonal
volume error
- Summer:
Seasonal
volume error
-Fall:
Seasonal
volume error
-Winter:
Seasonal
volume error
- Spring:
Error in
storm
volumes:
Error in
summer
storm
volumes:
Daily Nash-
Sutcliffe
Coefficient of
Efficiency, E:
Monthly
Nash-
Sutcliffe
Coefficient of
Efficiency, E:
05526000
Iroquois River
at Chebanse, IL
-16.99
-0.58
-27.20
29.82
-18.41
-23.84
-27.98
-35.59
-14.89
0.699
0.768
05520500
Kankakee River
at Momence, IL
-16.74
-37.19
32.90
-5.31
-24.23
-15.45
-20.05
44.06
5.27
0.241
0.540
05532500
Des Plaines River
at Riverside, IL
-7.97
-6.98
-14.69
29.05
-7.60
-18.09
-16.18
-32.00
-21.41
0.561
0.622
05552500
Fox River
at Dayton, IL
-2.94
-5.58
1.14
47.39
-8.61
-24.48
-6.00
16.17
57.63
0.367
0.530
05543500
Illinois River
at Marseilles, IL
-6.75
-8.18
-5.35
11.96
-4.52
-12.03
-14.21
-5.03
3.30
0.787
0.883
N-27
-------
Table 9. Summary statistics: all stations - validation period
Station
Error in total
volume:
Error in 50%
lowest flows:
Error in 10%
highest
flows:
Seasonal
volume error
- Summer:
Seasonal
volume error
-Fall:
Seasonal
volume error
-Winter:
Seasonal
volume error
- Spring:
Error in
storm
volumes:
Error in
summer
storm
volumes:
Daily Nash-
Sutcliffe
Coefficient of
Efficiency, E:
Monthly
Nash-
Sutcliffe
Coefficient of
Efficiency, E:
05526000
Iroquois River
at Chebanse, IL
-2.98
-21.27
-11.00
36.69
21.34
-10.30
-24.88
-17.48
-4.76
0.674
0.712
05520500
Kankakee River
at Momence, IL
-9.24
-33.25
37.98
-5.79
9.77
-11.44
-23.42
68.38
13.56
0.036
0.473
05532500
Des Plaines River
at Riverside, IL
12.69
13.77
12.33
42.17
12.41
1.08
4.55
-13.72
-12.01
0.586
0.635
05552500
Fox River
at Dayton, IL
9.52
9.19
23.25
52.54
2.17
-5.13
10.02
37.25
67.69
0.418
0.702
05543500
Illinois River
at Marseilles, IL
5.78
-3.45
17.49
18.46
21.78
-2.24
-9.45
19.98
21.93
0.571
0.635
Water Quality Calibration and Validation
Initial calibration and validation of water quality was done on the Iroquois River (USGS 05526000), using time
period 1985-2001 for calibration and 1978-1984 for validation. As with hydrology, calibration was performed on
the later period as this better reflects the land use included in the model. The validation period at stations is
constrained by data availability.
Calibration adjustments for sediment focused on the following parameters:
N-28
-------
• SPCON (Linear parameters for estimating maximum amount of sediment that can be re-entrained during
channel sediment routing)
• PRF (Peak rate adjustment factor for sediment routing in the main channel)
• USLE-K (USLE erodibility factor)
• CH_COV (Channel cover factor)
• CH_EROD (Channel erodibility factor)
• USLE-P (USLE support practice factor)
• DIRTMX and curb length density in urban database
Simulated and estimated sediment loads at the Iroquois River station for both the calibration and validation
periods are shown in Figure 12 and statistics for the two periods are provided separately in Table 10. The key
statistic in Table 10 is the relative percent error, which shows the error in the prediction of monthly load
normalized to the estimated load. Table 10 also shows the relative average absolute error, which is the average of
the relative magnitude of errors in individual monthly load predictions. This number is inflated by outlier months
in which the simulated and estimated loads differ by large amounts (which may be as easily due to uncertainty in
the estimated load due to limited data as to problems with the model) and the third statistic, the relative median
absolute error, is likely more relevant and shows better agreement.
TSS
• Regression Loads
-Simulated Loads
O)O)O)O)O)O)O)O)O)O)O
Figure 12. Fit for monthly load of TSS at USGS 05526000 Iroquois River near Chebanse, IL.
N-29
-------
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 05526000 Iroquois River near Chebanse, IL
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1985-2001)
38%
56%
19.8%
Validation period
(1978-1984)
39%
51%
9.9%
Calibration adjustments for total phosphorus and total nitrogen focused on the following parameters:
• Initial soil organic N and P
• PPERCO (phosphorus percolation coefficient)
• NPERCO (nitrogen percolation coefficient)
• PHOSKD (phosphorus soil partitioning coefficient)
• SOL_CBN1 (Organic carbon in the first soil layer)
• MUMAX
• QUAL2E parameters such as algal, organic nitrogen, and organic phosphorus settling rate in the reach,
benthic source rate for dissolved phosphorus and NH4-N in the reach, fraction of algal biomass that is
nitrogen and phosphorus, Michaelis-Menton half-saturation constant for nitrogen and phosphorus
Results for the phosphorus simulation are shown in Figure 13 and Table 11. Results for the nitrogen simulation
are shown in Figure 14 and Table 12. The model fit is good for phosphorus and nitrogen was underpredicted.
Total P
1000
• Regression Loads
-Simulated Loads
COO>O^-(NCO^LOCOr^COO>O^-(NCO^LOCOr^COO>O^-
iY^a3a3a3a3a3a3a3a3a3a3O>q>q>q>q>q>q>q>q>q>oo
Figure 13. Fit for monthly load of total phosphorus at USGS 05526000 Iroquois River near Chebanse, IL.
N-30
-------
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 05526000 Iroquois River near Chebanse, IL
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1985-2001)
5%
49%
16.9%
Validation period
(1978-1984)
-1
33%
11.9%
10,000
Total N
-Averaging Loads
-Simulated Loads
ooo>o-<-rMco-^-Lr>cor^ooa>o-<-rMco-^-Lr>cor^ooa>o-<-
r-pf-pooopopopopopopopopopcpcpcpcpcpcpcpcpcpcpoo
Figure 14. Fit for monthly load of total nitrogen at USGS 05526000 Iroquois River near Chebanse, IL.
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 05526000 Iroquois River near Chebanse, IL
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1985-2001)
56%
64%
29.8%
Validation period
(1978-1984)
60%
64%
20.5%
Water Quality Results for Larger Watershed
Only the USLE P-factor was spatially adjusted to account for high upland erosion for the Kankakee and Fox
watersheds. Summary statistics for the SWAT water quality calibration and validation at other stations in the
watershed are provided in Table 13 and Table 14. There were unexplained high simulated sediment loads at the
Fox stations that were not reflected in the measured sediment data at this station.
N-31
-------
Table 13. Summary statistics for water quality at all stations - calibration period 1985-2001 (unless
otherwise noted)
Station
Relative Percent Error
TSS Load
Relative Percent Error
TP Load
Relative Percent Error
TN Load
05526000
Iroquois River
at Chebanse, IL
38
5
56
05520500
Kankakee River
at Momence, IL
-7
-71
-5
05532500
Des Plaines River
at Riverside, IL
-3
-54
-46
05552500
Fox River
at Dayton, IL
(1990-2001)
-234
-51
-3
05543500
Illinois River
at Marseilles, IL
-97
14
26
Table 14. Summary statistics for water quality at all stations - validation period 1978-1984 (unless
otherwise noted)
Station
Relative Percent Error
TSS Load
Relative Percent Error
TP Load
Relative Percent Error
TN Load
05526000
Iroquois River
at Chebanse, IL
39
-1
60
05520500
Kankakee River
at Momence, IL
-1
-100
-13
05532500
Des Plaines River
at Riverside, IL
-23
-68
-58
05552500
Fox River
at Dayton, IL
(1978-1989)
-267
-71
-14
05543500
Illinois River
at Marseilles, IL
(1974-1984)
-107
9
24
N-32
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
Tetra Tech. 1999. Improving Point Source Loadings Data for Reporting National Water Quality Indicators. Final
Technical Report prepared for U.S. Environmental Protection Agency, Office of Waste water Management,
Washington, DC, by Tetra Tech, Inc., Fairfax, VA.
USGS (United States Geological Survey). 1994. NAWQA Fact Sheet. Available at:
http://il.water.usgs.gov/proj/lirb/pubs/pdfs/fctsheet.pdf
USEPA (United States Environmental Protection Agency). 2008. Using the BASINS Meteorological Database
(Version 2006). BASINS Technical Note 10. Office of Water, U.S. Environmental Protection Agency,
Washington, DC. http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf
(Accessed June, 2009).
N-33
-------
Appendix O
Model Configuration, Calibration and
Validation
Basin: Lake Erie Drainages (Erie)
O-l
-------
Contents
Watershed Background O-4
Water Body Characteristics O-4
Soil Characteristics O-5
Land Use Representation O-5
Point Sources O-9
Meteorological Data O-10
Watershed Segmentation O-13
Calibration Data and Locations O-13
SWAT Modeling O-15
Assumptions O-15
Hydrology Calibration O-15
Hydrology Validation O-19
Hydrology Results for Larger Watershed O-22
Water Quality Calibration and Validation O-23
Water Quality Results for Larger Watershed O-26
References O-28
O-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes O-7
Table 2. Land use distribution for the Lake Erie drainages (2001 NLCD) (mi2) O-8
Table 3. Major point source discharges in the Lake Erie drainages O-9
Table 4. Precipitation stations for the Lake Erie drainages watershed model O-ll
Table 5. Calibration and validation locations in the Lake Erie drainages O-14
Table 6. Summary statistics at USGS 04208000 Cuyahoga River at Independence, OH - calibration
period O-19
Table 7. Summary statistics at USGS 04208000 Cuyahoga River at Independence, OH - validation
period O-22
Table 8. Summary statistics (percent error): all stations - calibration period O-23
Table 9. Summary statistics (percent error): all stations - validation period O-23
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 04208000 Cuyahoga River at Independence, OH O-24
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 04208000 Cuyahoga River at Independence, OH O-26
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 04208000 Cuyahoga River at Independence, OH O-26
Table 13. Summary statistics for water quality at all stations - calibration period 1990-2000 O-27
Table 14. Summary statistics for water quality at all stations - validation period 1980-1990 O-27
Figures
Figure 1. Location of the Lake Erie drainages O-5
Figure 2. Land use in the Lake Erie drainages O-6
Figure 3. Model segmentation and USGS stations utilized for the Lake Erie drainages O-13
Figure 4. Mean monthly flow at USGS 04208000 Cuyahoga River at Independence, OH - calibration
period O-16
Figure 5. Seasonal regression and temporal aggregate at USGS 04208000 Cuyahoga River at
Independence, OH- calibration period O-17
Figure 6. Seasonal medians and ranges at USGS 04208000 Cuyahoga River at Independence, OH -
calibration period O-17
Figure 7. Flow exceedance at USGS 04208000 Cuyahoga River at Independence, OH - calibration
period O-18
Figure 8. Mean monthly flow at USGS 04208000 Cuyahoga River at Independence, OH - validation
period O-20
Figure 9. Seasonal regression and temporal aggregate at USGS 04208000 Cuyahoga River at
Independence, OH - validation period O-20
Figure 10. Seasonal medians and ranges at USGS 04208000 Cuyahoga River at Independence, OH -
validation period O-21
Figure 11. Flow exceedance at USGS 04208000 Cuyahoga River at Independence, OH - validation
period O-21
Figure 12. Fit for monthly Load of TSS at USGS 04208000 Cuyahoga River at Independence, OH O-24
Figure 13. Fit for monthly load of total phosphorus at USGS 04208000 Cuyahoga River at
Independence, OH O-25
Figure 14. Fit for monthly load of total nitrogen at USGS 04208000 Cuyahoga River at Independence,
OH O-26
O-3
-------
The Lake Erie drainages were selected as one of the 15 non-pilot application watersheds for the 20 Watershed
study. Watershed modeling for the non-pilot areas is accomplished using the SWAT model only, and model
calibration and validation results are presented in abbreviated form.
Water Body Characteristics
Lake Erie is the eleventh largest freshwater lake in the world. About two-thirds of the contributing watershed is in
the United States, and includes portions of Michigan, Indiana, Ohio, Pennsylvania, and New York. The model
study area focuses on drainages to the southwestern portion of Lake Erie and encompasses nearly 11,700 mi2 in
12 HUCSs within HUC 0410 and HUC 0411 (Figure 1).
Situated in two major physiographic provinces, the Appalachian Plateaus and the Central Lowland, the watershed
includes varied topographic and geomorphic features that affect the hydrology. The watershed consists of multiple
independent drainages. The principal river in the study unit, the Maumee River, drains an area of 6,644 mi2, or
roughly one-third of the model study area. Other principal streams and their drainage areas in Ohio are the
Sandusky River (1,420 mi2, the Cuyahoga River (809 mi2), and the Grand River (705 mi2). The land surface is
gently rolling to nearly flat (Myers et al., 2000).
The majority of the land use in the model area is agriculture (67 percent). The remaining land uses are urban land
(15 percent), forest (13 percent), and open water or wetlands (4 percent). Corn, soybeans, and wheat are the
typical parts in the western part of the basin. Other agricultural land uses include pasture and forage crops, grown
predominantly in the eastern part of the basin. Forest and wetlands have been greatly reduced in the watershed
since the mid-1800s. Major urban areas in the model area include Cleveland, Toledo, and Akron, Ohio, along
with Fort Wayne, Indiana. These cities are important industrial and manufacturing centers. Major urban centers
rely on abundant supplies of water for shipping, electric power generation, industry, domestic consumption, and
waste assimilation.
Average annual precipitation across the model study area ranges from about 30 to 45 inches. Precipitation is
highest to the northeast because of lake effect. The lowest amounts of precipitation are in the northwestern part of
the basin near the Michigan border. The highest streamflows are typically in February, March, and April, as a
result of increased precipitation, cold temperatures and little vegetative growth. The lowest streamflows are in
August, September, and October. During low streamflow, groundwater typically contributes most of the flow.
Cooling during power generation accounts for 71 percent of the water use in the watershed. Public and domestic
supply account for 17 percent, and industry and mining account for 10 percent of the total water use. Normal
precipitation is generally adequate for agriculture, so irrigation accounts for less than 1 percent of water use. Most
of the major cities are near Lake Erie and derive their water from the lake.
Population density and growth in the Lake Erie basin are among the highest in the Great Lakes basin. About 40
percent of the total population of the Great Lakes basin lives in the Lake Erie basin in 17 urban areas having
populations of 50,000 or more. Water resources in the study unit are central to the economy and culture of the
region. The surficial deposits of this area consist primarily of ground moraine and end moraine of glacial origin;
valleys are filled with glacial outwash. The area is characterized by broad, low ridges with smooth, gentle slopes
separated by flat, gently undulating plains. The Eastern Lake Section and the Till Plains Section within the
province consist of wide expanses of flat land underlain by clayey till or lake deposits; this flat land is
interspersed with sandy ridges that are remnants of glacial-lake beaches. Because soils are fertile and the climate
is temperate, the primary land use in this part of the study unit is agricultural, ranging from orchards and
vineyards near the Lake Erie shoreline to cropland in corn and soybeans further inland.
O-4
-------
Legend
Hydrography
^B Water (Nat. Atlas Dataso
US Census Populated Places
•• Municipalities ipop? 50,000
| | County Bounda
I I Watershed with HUC
'Grand
(04110004
VI Tiffin
<04iqoooe)
Black Rrirki/
HiacK-htocKy
(04110001)
uron
Vermilion
(04100012)
St. Joseph
', (04100003)
Sandusky
(04100011)
Blanchard
04100008)
GCRP Model Areas - Lake Erie Drainages
Base Map
Figure 1. Location of the Lake Erie drainages.
Soil Characteristics
Soils in the watershed are described in STATSGO soil surveys. SWAT uses information drawn directly from the
soils data layer to populate the model.
Land Use Representation
Land use/cover in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage
(Figure 2). NLCD land cover classes were aggregated according to the scheme shown in Table 1 for
representation in the 20 Watershed model. SWAT uses the built-in hydrologic response unit (HRU) overlay
mechanism in the ArcSWAT interface. SWAT HRUs are formed from an intersection of land use and SSURGO
major soils. The distribution of land use in the watershed is summarized in Table 2.
O-5
-------
Hydrography
Interstate
I I County Boundaries
2001 N LCD Land Use
I | Op en water
^ Developed, open space
| Developed, low intensity
^^| Developed, medium intensity
^^| Developed, high intensity
I | Barren land
| Deciduous forest
| Evergreen forest
I 1 Mixed forest
I | Scrub/shrub
I | Grassland/herbaceous
^ Pasture/hay
| Cultivated crops
I I Woody wetlands
I | Emergent herbaceous wetlands
GCRP Model Areas - Erie/St. Clair River Basins - Land Use Map
TETRATECH
Figure 2. Land use in the Lake Erie drainages.
O-6
-------
Table 1. Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area usually
accounted for as reach area
Deciduous
Evergreen
Mixed
Emergent & woody wetlands
Aquatic bed wetlands (not
emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
O-7
-------
Table 2. Land use distribution for the Lake Erie drainages (2001 NLCD) (mi )
HUC8
watershe
d
St Joseph
04100003
St Marys
04100004
Upper
Maumee
04100005
Tiffin
04100006
Auglaize
04100007
Blanchard
04100008
Lower
Maumee
04100009
Sandusky
04100011
Huron-
Vermilion
04100012
Black-
Rocky
04110001
Cuyahoga
04110002
Grand
04110004
Total
Open
water
13.1
22.0
4.4
5.4
8.9
3.6
14.8
15.2
6.1
6.7
19.2
8.9
128.3
Developed9
Open
space
62.6
65.0
28.4
42.2
118.0
51.3
82.3
118.2
45.3
127.6
141.7
42.9
925.5
Low
density
33.2
30.7
18.9
15.6
43.2
19.4
47.8
50.0
18.9
111.4
144.4
26.5
560.0
Medium
density
7.2
10.7
4.4
3.3
9.2
5.1
17.1
14.9
4.3
28.8
56.7
2.9
164.6
High
density
3.3
4.8
2.5
1.9
5.6
2.4
8.0
7.1
2.0
6.4
20.6
0.7
65.4
Barren land
0.6
0.0
0.1
0.7
1.6
0.4
1.1
5.9
0.1
0.9
0.6
0.1
12.1
Forest
120.3
46.4
19.4
55.4
87.8
43.0
72.2
148.7
147.3
224.9
251.3
300.0
1,516.8
Shru bland/
Grassland
13.3
12.5
2.6
3.6
20.0
12.6
10.7
21.3
2.4
8.2
27.9
40.9
176.1
Pasture/
Hay
185.2
18.4
16.0
81.0
30.5
7.9
14.4
44.7
38.6
109.6
61.5
64.2
672.0
Cultivated
572.6
650.1
286.2
522.8
1,336.3
623.8
806.8
1,401.0
490.0
218.0
65.4
173.2
7,146.2
Wetland
82.2
5.4
4.4
45.6
5.3
2.6
5.5
32.9
9.3
55.0
22.1
45.3
315.5
Total
1,093.6
866.1
387.4
777.6
1,666.5
772.0
1,080.5
1,859.8
764.4
897.6
811.4
705.6
11,682.5
aThe percent imperviousness applied to each of the developed land uses is as follows: open space (7.30%), low density (32.53%), medium density (60.72%), and high
density (86.75%).
O-8
-------
Point Sources
There are numerous point source discharges in the watershed. Only the major dischargers, generally defined as
those with a design flow greater than 1 MGD are included in the simulation (Table 3). The major dischargers are
represented at long-term average flows, without accounting for changes over time or seasonal variations.
Table 3. Major point source discharges in the Lake Erie drainages
NPDES ID
OH0034223
OH0002666
OH0003298
OH0020893
OH0024899
OH0025771
OH0027910
OH0026921
OH0027952
OH0023841
OH0037338
OH0002615
OH0002623
OH0026069
OH0020851
OH0025135
OH0020532
OH0020796
IN0032191
IN0000388
IN0022462
IN0020672
OH0025291
OH0052949
OH0020001
OH0020664
OH0022659
OH0026948
OH0000957
Name
LUCAS CO COMMISSIONERS
GENERAL MOTORS CORPORATION
CAMPBELL SOUP COMPANY
CITY OF NAPOLEON
CITY OF DEFIANCE
VILLAGE OF HICKSVILLE
CITY OF VAN WERT
VILLAGE OF OTTAWA
CITY OF WAPAKONETA
ALLEN COUNTY COMMISSIONERS
ALLEN CO. COMMISSIONERS
PCS NITROGEN OHIO, LP
LIMA REFINING COMPANY
CITY OF LIMA
VILLAGE OF BLUFFTON
CITYOFFINDLAY
CITY OF BRYAN
VILLAGE OF ARCHBOLD
FORT WAYNE MUNICIPAL WWTP
DANASPICER MANUFACTURING INC.
BUTLER MUNICIPAL WWTP
AUBURN MUNICIPAL WPCP
CITY OF FREMONT
CITY OF TIFFIN
CITY OF UPPER SANDUSKY
CITY OF CRESTLINE
VILLAGE OF CHARDON
CITYOFPAINESVILLE
ISG CLEVELAND
Design flow
(MGD)
15.00
2.50
4.00
2.78
4.00
18.50
9.00
3.14
1.75
60.00
1.36
2.00
4.50
11.00
6.00
1.50
6.00
Observed flow
(MGD)
15.21
1.90
5.11
2.92
3.30
0.75
3.39
1.54
2.60
1.44
3.17
3.50
17.70
26.33
2.28
10.18
2.35
1.36
88.87
0.98
0.67
2.70
17.30
3.27
1.47
0.95
2.35
2.93
3.69
O-9
-------
NPDES ID
OH0024651
OH0024040
OH0024058
OH0027430
OH0027863
OH0098043
OH0025917
OH0064009
OH0023221
OH0001562
OH0023981
OH0026093
OH0020427
OH0025372
OH0024660
OH0026794
OH0030503
OH0045748
OH0043567
OH0020125
OH0028118
OH0021628
OH0020672
OH0024686
Name
NEORSD - SOUTHERLY WWTP
CITY OF BEDFORD
CITY OF BEDFORD HEIGHTS
SOLON CITY CENTRAL
CITYOFTWINSBURG
EARTH TECH
CITY OF KENT
SUMMIT COUNTY - FISHCREEK #25
CITY OF RAVENNA
REPUBLIC ENGINEERED PRODUCTS
AVON LAKE WASTEWATER PLANT
CITYOFLORAIN
OBERLIN WATER ENV. PROTECTION
VILLAGE OF GRAFTON
NEORSD - WESTERLY WWTP
CITY OF NORTH ROYALTON
CITY OF ROCKY RIVER
MEDINA COUNTY COMM SD 300
MEDINA COUNTY COMM SD 500
HURON BASIN STP
CITYOFWILLARD
CITYOFAMHERST
CITYOFBELLEVUE
CITY OF CLYDE
Design flow
(MGD)
175.00
3.20
7.50
3.60
3.40
1.40
5.00
4.00
2.80
6.50
15.00
1.50
1.00
50.00
1.50
22.00
2.00
10.00
1.36
2.00
1.20
1.90
Observed flow
(MGD)
115.67
2.38
2.49
4.00
2.92
2.72
2.59
3.92
2.20
87.30
9.67
13.09
1.27
3.74
28.50
1.55
11.13
1.66
9.04
0.87
1.80
1.88
1.09
1.95
Most of these point sources have reasonably complete monitoring for total suspended solids (TSS). Long term
average values of total phosphorus and total nitrogen were assumed based upon the type of point source
discharger. The point sources were initially represented in the model with the median of reported values for total
phosphorus, total suspended solids and total nitrogen.
Meteorological Data
The required meteorological time series for the 20 Watershed SWAT simulations are precipitation and air
temperature. The 20 Watershed simulations do not include water temperature and uses a degree-day method for
snowmelt. SWAT estimates Penmann-Monteith potential evapotranspiration using a statistical weather generator
for inputs other than temperature and precipitation. These meteorological time series are drawn from the
BASINS4 Meteorological Database (USEPA 2008), which provides a consistent, quality-assured set of
nationwide data with gaps filled and records disaggregated. Scenario application requires simulation over 30
years, so the available stations are those with a common 30-year period of record (or one that can be filled from
O-10
-------
an approximately co-located station) that covers the year 2000. A total of 57 precipitation stations were identified
for use in the Lake Erie drainages watershed model with a common period of record of 10/1/1969-9/30/2000
(Table 4). Temperature records are sparser; where these are absent temperature is taken from nearby stations with
an elevation correction.
Table 4. Precipitation stations for the Lake Erie drainages watershed model
COOP ID
IN1 20200
IN1 20676
IN1 22096
IN123037
IN1 23206
IN1 24497
MI200032
MI203823
OH330058
OH330059
OH330061
OH330107
OH330256
OH330862
OH331042
OH331072
OH331390
OH331458
OH331541
OH331657
OH332098
OH332251
OH332599
OH332786
OH332791
OH332974
OH333021
OH333421
OH333780
OH333874
Name
ANGOLA
BERNE
DECATUR 1 N
FORTWAYNEWSOAP
GARRETT
KENDALLVILLE
ADRIAN 2 NNE
HILLSDALE
AKRON CANTON WSO AP
AKRON WPCS
AKRON
ALLIANCE 3 NNW
ASHLAND 2 SW
BOWLING GREEN WWTP
BRYAN 2 SE
BUCYRUS
CELINA3NE
CHARDON
CHIPPEWALAKE
CLEVELAND WSFOAP
DEFIANCE
DORSET
ELYRIA 3 E
FINDLAYFAA AIRPORT
FINDLAYWPCC
FREMONT
GALION WATER WORKS
GROVER HILL
HIRAM
HOYTVILLE 2 NE
Latitude
41.6397
40.6684
40.8482
41.0062
41.3330
41.4428
41.9165
41.9353
40.9167
41.1500
41.0804
40.9550
40.8334
41.3831
41.4670
40.8129
40.5695
41.5834
41.0517
41.4051
41.2778
41.6834
41.3833
41.0136
41.0462
41.3334
40.7236
41.0184
41.3000
41.2168
Longitude
-84.9899
-84.9305
-84.9294
-85.2056
-85.1329
-85.2613
-84.0157
-84.6410
-81.4333
-81.5669
-81.5169
-81.1169
-82.3499
-83.6110
-84.5330
-82.9693
-84.5364
-81.1833
-81.9360
-81.8528
-84.3853
-80.6667
-82.0499
-83.6685
-83.6621
-83.1166
-82.7999
-84.4724
-81.1500
-83.7667
Temperature
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Elevation (m)
308
265
250
252
82
297
232
329
368
70
329
322
386
206
68
291
262
344
360
235
213
299
223
244
234
183
357
223
375
213
O-ll
-------
COOP ID
OH333915
OH334189
OH334551
OH334865
OH334874
OH334942
OH335438
OH335505
OH335669
OH336118
OH336196
OH336342
OH336389
OH336405
OH336465
OH336949
OH337383
OH337447
OH337698
OH338110
OH338313
OH338357
OH338534
OH338539
OH338609
OH338769
OH338822
Name
HUNTSVILLE 3 N
KENTON
LIMAWWTP
MANSFIELD WSOAP
MANSFIELD 5 W
MARION 2 N
MONTPELIER
MOSQUITO CREEK LAKE
NAPOLEON
NORWALK WWTP
OBERLIN
OTTAWA
PAINESVILLE 4 NW
PANDORA
PAULDING
RAVENNA 2 S
ST MARYS 3 W
SANDUSKY
SIDNEY HIGHWAY DEPT
STRYKER
TIFFIN
TOLEDO EXPRESS WSO
AP
UPPER SANDUSKY
UPPER SANDUSKY
WATER WK
VAN WERT 1 S
WARREN 3 S
WAUSEON WATER PLANT
Latitude
40.4803
40.6489
40.7247
40.8204
40.7668
40.6168
41.5804
41.3000
41.3940
41.2668
41.2668
41.0318
41.7500
40.9543
41.1245
41.1333
40.5447
41.4501
40.2983
41.5057
41.1168
41.5886
40.8334
40.8167
40.8495
41.2001
41.5184
Longitude
-83.8131
-83.6060
-84.1294
-82.5177
-82.6166
-83.1333
-84.6077
-80.7667
-84.1144
-82.6166
-82.2167
-84.0528
-81.2999
-83.9616
-84.5922
-81.2832
-84.4374
-82.7167
-84.1633
-84.4300
-83.1667
-83.8014
-83.2832
-83.2832
-84.5807
-80.8166
-84.1453
Temperature
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Elevation (m)
314
303
259
395
411
294
262
277
208
204
249
223
183
235
221
337
267
178
314
213
226
204
260
250
241
274
229
O-12
-------
Watershed Segmentation
The Lake Erie drainages were divided into 100 subwatersheds for the purposes of modeling (Figure 3). The model
encompasses the complete watershed and does not require specification of any upstream boundary conditions for
application.
USGS 04208000
USGS 04199000
USGS 04199500
USGS 04193500
USGS 04198000
Michigan '| JK?I 56
USGS 04206000
USGS'04186500
USGS gages
Hydrography
Interstate
^H Water (Nat. Atlas Dataset)
US Census Populated Places
I I County Boundaries
Model Subbasins
GCRP Model Areas - Ene/St. Clair River Basins
Model Segmentation
Figure 3. Model segmentation and USGS stations utilized for the Lake Erie drainages
Calibration Data and Locations
The specific site chosen for initial calibration was the Cuyahoga River at Independence (USGS 04208000), a flow
and water quality monitoring location that approximately coincides with the mouth of an 8-digit HUC. The
Cuyahoga River watershed was selected because there is a good set of flow and water quality data available and
the watershed lacks major point sources and impoundments. Additional calibration and validation was pursued at
multiple locations (Table 5). Parameters derived on the Cuyahoga River were not fully transferable to other
portions of the Lake Erie drainages, and additional calibration was conducted at multiple gage locations.
O-13
-------
Table 5. Calibration and validation locations in the Lake Erie drainages
Station name
Auglaize River near Fort Jennings OH
Maumee River at Waterville OH
Sandusky River near Fremont OH
Huron River at Milan OH
Black River at Elyria OH
Cuyahoga River at Old Portage OH
Cuyahoga River at Independence OH
USGS ID
04186500
04193500
04198000
04199000
04200500
04206000
04208000
Drainage area
(mi2)
332
6,330
1,251
371
396
404
707
Hydrology
calibration
X
X
X
X
X
X
X
Water quality
calibration
X
X
X
The model hydrology calibration period was set to Water Years 1990-2000 (within the 32-year period of record
for modeling). Hydrologic validation was then performed on Water Years 1980-1990. Water quality calibration
used calendar years 1990-2000, while validation used 1980-1990.
O-14
-------
SWAT Modeling
Assumptions
There were no significant impoundments and/or diversions that needed representation in the watershed model for
the Lake Erie drainages.
Hydrology Calibration
A spatial calibration approach was adopted for GCRP-SWAT modeling for the Lake Erie drainages. A systematic
adjustment of parameters has been adopted and some adjustments are applied throughout the basin. Most of the
calibration efforts were geared towards getting a closer match between simulated and observed flows at the outlet
of calibration focus area.
Land Use/Soil/Slope Definition
A 5/10/5 percent threshold was used for land use/soil/slope in the SWAT model while defining the HRUs. Urban
land use classes were exempted from the HRU overlay thresholds.
The calibration focus area (Cuyahoga River) includes six subwatersheds and is generally representative of the
general land use characteristics of the overall watershed with the exception of a higher percentage of cultivated
lands. The parameters were adjusted within the practical range to obtain reasonable fit between the simulated and
measured flows in terms of Nash-Sutcliffe modeling efficiency and the high flow and low flow components as
well as the seasonal flows.
The water balance of the whole Lake Erie drainages predicted by the SWAT model over the 32-year simulation
period is as follows:
PRECIP = 934.7 MM
SNOW FALL = 125.69 MM
SNOW MELT = 122.44 MM
SUBLIMATION = 1.95 MM
SURFACE RUNOFF Q = 233.68 MM
LATERAL SOIL Q = 1.40 MM
TILE Q = 0.00 MM
GROUNDWATER (SHAL AQ) Q = 92.63 MM
REVAP (SHAL AQ => SOIL/PLANTS) = 4.67 MM
DEEP AQ RECHARGE = 13.62 MM
TOTAL AQ RECHARGE = 110.90 MM
TOTAL WATER YLD = 327.70 MM
PERCOLATION OUT OF SOIL = 110.91 MM
ET = 583.7 MM
PET = 1152.4MM
TRANSMISSION LOSSES = 0.00 MM
Hydrologic calibration adjustments focused on the following parameters:
• CN2 (initial SCS runoff curve number for moisture condition II)
• ESCO (soil evaporation compensation factor)
• SURLAG (surface runoff lag coefficient)
• SOL_AWC (available water capacity of the soil layer, mm water/mm of soil)
O-15
-------
ALPHA_BF (baseflow alpha factor, days)
GW_DELAY (groundwater delay time, days)
GWQMIN (threshold depth of water in the shallow aquifer required for return flow to occur, mm)
GW_REVAP (groundwater "revap" coefficient)
CH_N1 (Manning's "n" value for tributary channels)
CH_N2 (Manning's "n" value for main channels)
CH_K1 (Effective hydraulic conductivity in tributary channel alluvium)
CH_K2 (Effective hydraulic conductivity in main channel alluvium)
SFTMP (Snowfall temperature)
SMTMP (Snowmelt base temperature)
SMFMX (Maximum melt rate for snow during the year)
SMFMN (Minimum melt rate for snow during the year)
The calibration achieves a moderately high coefficient of model fit efficiency, but is below on 50 percent lowest
flow volume. Calibration results for the Cuyahoga River are summarized in Figure 4, Figure 5, Figure 6, Figure 7
and Table 6.
3000
2500
,2000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1990 to 9/30/2000 )
-Avg Modeled Flow (Same Period)
•g
01
O-90
A-92
O-93
A-95
O-96
A-98
O-99
Month
Figure 4. Mean monthly flow at USGS 04208000 Cuyahoga River at Independence, OH - calibration
period.
O-16
-------
Avg Flow (10/1/1990 to 9/30/2000)
-Line of Equal Value
Best-Fit Line
2000
'150° +~
T3
•fiooo
T3
O
ro 500
' = 0.5872X + 36J3.63
R? = 0.8807
"- ..... \
^--
.-•'I
500 1000 1500
Average Observed Flow (cfs)
2000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1990 to 9/30/2000)
Avg Modeled Flow (Same Period)
2000
1500
•e
s"^*n-i
o
1000
500
Figure 5. Seasonal regression and temporal aggregate at USGS 04208000 Cuyahoga River at
Independence, OH - calibration period.
To Lower Bound Average Monthly Rainfall (in) -Median Observed Flow (10/1/1990 to 9/30/2000) Modeled (Median, 25th, 75th)
2500
2000
Ocr Nov Dec Jan Feb
10 11 12 1
Figure 6. Seasonal medians and ranges at USGS 04208000 Cuyahoga River at Independence, OH -
calibration period.
O-17
-------
100000
•Observed Flow Duration (10/1/1990 to 9/30/2000)
Modeled Flow Duration (10/1/1990 to 9/30/2000 )
10%
20%
30% 40% 50% 60% 70%
Percent of Time that Flow is Equaled or Exceeded
80%
90%
100%
Figure 7. Flow exceedance at USGS 04208000 Cuyahoga River at Independence, OH - calibration
period.
O-18
-------
Table 6.
period
Summary statistics at USGS 04208000 Cuyahoga River at Independence, OH - calibration
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 77
10-Year Analysis Period: 10/1/1990 - 9/30/2000
Flow/volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
17.82
6.24
2.68
3.27
4.28
5.64
4.64
6.32
1.40
Error Statistics
-3.32
-18.67
-6.08
25.11
Observed Flow Gage
USGS 04208000 Cuyahoga River at Independence OH
Hydrologic Unit Code: 4110002
Latitude: 41.39533087
Longitude: -81.6298478
Drainage Area (sq-mi): 707
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring Flow Volume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
1.39 » [ 30
-14.11
-7.94
-10.20
28.01
0.610
0.442
0.700
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
18.43
6.64
3.29
2.61
4.22
6.56
5.04
7.04
1.09
Clear [
Hydrology Validation
Hydrology validation for Cuyahoga River at Independence was performed for the period 10/1/1980 through
9/30/1990. The validation achieves a moderately high coefficient of model fit efficiency, but is below on total
flow, 50 percent lowest flow and 10 percent highest flow volumes. Validation results for the Cuyahoga River are
summarized in Figure 8, Figure 9, Figure 10, Figure 11 and Table 7.
O-19
-------
4000
O-80
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1980 to 9/30/1990 )
-Avg Modeled Flow (Same Period)
A-82
O-83
A-85
O-86
A-88
Month
O-89
Figure 8. Mean monthly flow at USGS 04208000 Cuyahoga River at Independence, OH - validation
period.
• Avg Flow (10/1/1980to 9/30/1990)
Line of Equal Value
Best-Fit Line
2000
500
1000
1500
2000
2000
1500
t
o
1000
500
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1980 to 9/30/1990)
Avg Modeled Flow (Same Period)
Average Observed Flow (cfs)
10 11 12 1 23456789
Month
Figure 9. Seasonal regression and temporal aggregate at USGS 04208000 Cuyahoga River at
Independence, OH -validation period.
O-20
-------
To Lower Bound Average Monthly Rainfall (in) -Median Observed Flow (10/1/1980 to 9/30/1990) Modeled (Median, 25th, 75th)
2500
2000
Ocr Nov Dec Jan
10 11 12 1
Figure 10. Seasonal medians and ranges at USGS 04208000 Cuyahoga River at Independence, OH
validation period.
•Observed Flow Duration (10/1/1980 to 9/30/1990)
Modeled Flow Duration (10/1/1980 to 9/30/1990 )
100000
10%
20% 30% 40% 50% 60% 70%
Percent of Time that Flow is Equaled or Exceeded
80%
90% 100%
Figure 11. Flow exceedance at USGS 04208000 Cuyahoga River at Independence, OH - validation period.
O-21
-------
Table 7. Summary statistics at USGS 04208000 Cuyahoga River at Independence, OH - validation
period
REACH OUTFLOW FROM OUTLET 77
10-Year Analysis Period: 10/1/1980 - 9/30/1990
Flow/volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
16.61
5.59
2.75
2.63
3.98
5.19
4.82
5.64
1.08
Error Statistics
-13.38
-25.70
-15.93
-0.83
Observed Flow Gage
USGS 04208000 Cuyahoga River a
Hydrologic Unit Code: 4110002
Latitude: 41.39533087
Longitude: -81.6298478
Drainage Area (sq-mi): 707
t Independence OH
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring Flow Volume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
-8.18 » [ 30
-21.31
-14.01
-18.14
-2.36
0.622
0.441
0.732
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
19.18
6.65
3.70
2.65
4.33
6.59
5.60
6.90
1.11
Clear [
Hydrology Results for Larger Watershed
As described above, parameters determined for the gage at Cuyahoga River at Independence were initially
transferred to other gages in the watershed. However, changes to subbasin level parameter were required to fit the
model to the observed flows. In all, calibration and validation was pursued at a total of seven gages throughout the
watershed. Results of the calibration and validation exercise are summarized in Table 8 and Table 9, respectively.
Calibration and validation results were acceptable at most gages.
O-22
-------
Table 8. Summary statistics (percent error): all stations - calibration period
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient of
Efficiency, E:
Monthly Nash-Sutcliffe Efficiency:
USGS
04186500
5.12
6.38
-13.25
1.15
2.86
0.49
14.92
-14.61
-22.24
0.44
0.832
USGS
04193500
2.84
56.11
-17.02
32.48
11.98
-5.40
-3.26
-13.55
7.55
0.65
0.900
USGS
04198000
-4.64
-22.60
-15.38
7.14
-4.26
-10.53
-1.41
-9.60
2.62
0.51
0.882
USGS
04199000
-0.99
0.52
-14.74
47.98
-27.69
-12.37
11.29
-14.98
27.73
0.29
0.597
USGS
04200500
4.21
-1.16
-13.91
61.88
-9.71
2.40
-2.52
-9.93
47.01
0.35
0.786
USGS
04206000
3.58
6.29
0.47
34.36
10.28
-8.25
-0.57
4.12
30.36
0.69
0.744
Table 9. Summary statistics (percent error): all stations - validation period
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient of
Efficiency, E:
Monthly Nash-Sutcliffe Efficiency:
USGS
04186500
-1.78
-35.07
-11.70
-1.85
5.85
-10.23
3.98
-18.28
-14.19
0.30
0.633
USGS
04193500
1.97
26.39
-11.39
41.13
1.71
-7.70
3.07
-8.68
19.02
0.71
0.903
USGS
04198000
1.22
-28.85
-11.72
23.91
10.39
-18.67
13.05
-3.71
21.09
0.43
0.764
USGS
04199000*
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
USGS
04200500
-2.34
-26.57
-17.25
59.12
-2.63
-13.18
-1.16
-12.42
43.95
0.42
0.798
USGS
04206000
6.41
4.79
0.84
33.99
12.35
-2.88
1.39
10.14
37.97
0.60
0.715
*No data (ND) were available for the validation period at USGS station 04199000.
Water Quality Calibration and Validation
Initial calibration and validation of water quality was done on the Cuyahoga River at Independence (USGS
04208000), using 1990-2000 for calibration and 1980-1990 for validation. As with hydrology, water quality
calibration was performed on the later period as this better reflects the land use included in the model.
Calibration adjustments for sediment focused on the following parameters:
O-23
-------
• SPCON (linear parameter for estimating maximum amount of sediment that can be re-entrained during
channel sediment routing)
• SPEXP (exponential parameter for estimating maximum amount of sediment that can be re-entrained
during channel sediment routing)
• CH_COV (channel cover factor)
• CH_EROD (channel erodibility factor)
• USLE_P (USLE support practice factor)
Simulated and estimated sediment loads at the Cuyahoga River station for both the calibration and validation
periods are shown in Figure 12 and statistics for the two periods are provided separately in Table 10. The key
statistic in Table 10 is the relative percent error, which shows the error in the prediction of monthly load
normalized to the estimated load. Table 10 also shows the relative average absolute error, which is the average of
the relative magnitude of errors in individual monthly load predictions. This number is inflated by outlier months
in which the simulated and estimated loads differ by large amounts (which may be as easily due to uncertainty in
the estimated load due to limited data as to problems with the model) and the third statistic, the relative median
absolute error, is likely more relevant and shows better agreement.
TSS
1,000,000
100,000
10,000
1,000
•Regression Loads
•Simulated Loads
Figure 12. Fit for monthly Load of TSS at USGS 04208000 Cuyahoga River at Independence, OH.
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 04208000 Cuyahoga River at Independence, OH
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1990-2000)
67.9.9%
74.5%
11.9%
Validation period
(1980-1990)
69.8%
75.8%
12.2%
Calibration adjustments for total phosphorus and total nitrogen focused on the following parameters:
• RHOQ (algal respiration rate at 20° C)
• PHOSKD (phosphorus soil partitioning coefficient)
O-24
-------
• PSP (phosphorus availability index)
• RSI (Local algal settlement rate in the reach at 20° C)
• AL1 (Fraction of algal biomass that is nitrogen)
• AL2 (Fraction of algal biomass that is phosphorus)
• MUMAX (Rate of oxygen uptake per unit NO2-N oxidation at 20° C)
• RHOQ (Algal respiration rate at 20° C)
• RS2 (benthic source rate for dissolved P in the reach at 20° C)
• RS3 (Benthic source rate for NFLpN in the reach at 20° C)
• RS5 (organic P settling rate in the reach at 20° C)
• BC4 (rate constant for mineralization of organic P to dissolved P in the reach at 20° C)
• RS4 (rate coefficient for organic N settling in the reach at 20° C)
• CH_ONCO (Channel organic nitrogen concentration)
• CH_OPCO (Channel organic phosphorus concentration)
• SDNCO (Denitrification threshold water content)
• CDN (Denitrification exponential rate constant)
Results for the phosphorus simulation are shown in Figure 13 and Table 11.. Results for the nitrogen simulation
are shown in Figure 14 and Table 12. The model fit is generally acceptable.
Total P
1000
o
E
"5
°
-Regression Loads
-Simulated Loads
. . .
d5" o° o° o° d* d* d5" o° o° o° d1 d* o° o° o° o° d1 c? o° o°
Figure 13. Fit for monthly load of total phosphorus at USGS 04208000 Cuyahoga River at Independence,
OH.
O-25
-------
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 04208000 Cuyahoga River at Independence, OH
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1990-2000)
23.9%
54.2%
27.4%
Validation period
(1980-1990)
-12.5%
66.9%
33.2%
Total N
10,000
1,000
o
•Averaging Loads
-Simulated Loads
o° d5" d*
o°
o°
o° d* o° o^ c^ o° d* of o°
Figure 14. Fit for monthly load of total nitrogen at USGS 04208000 Cuyahoga River at Independence, OH.
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 04208000 Cuyahoga River at Independence, OH
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1990-2000)
35.8%
46.4%
37.4%
Validation period
(1980-1990)
13.7%
53.6%
37.5%
Water Quality Results for Larger Watershed
As with hydrology, a spatial calibration approach was adopted. Cuyahoga River watershed SWAT model
parameters for water quality were transferred to other portions of the watershed with necessary changes to
subbasin level parameters. Summary statistics for the SWAT water quality calibration and validation at other
stations in the watershed are provided in Table 13 and Table 14.
O-26
-------
Table 13. Summary statistics for water quality at all stations - calibration period 1990-2000
Station
Relative Percent Error TSS Load
Relative Percent Error TP Load
Relative Percent Error TN Load
USGS
04193500
9.9%
33.5%
12.1%
USGS
04198000
17.0%
10.9%
16.8%
Table 14. Summary statistics for water quality at all stations - validation period 1980-1990
Station
Relative Percent Error TSS Load
Relative Percent Error TP Load
Relative Percent Error TN Load
USGS
04193500
11.2%
15.4%
-5.0%
USGS
04198000
8.1%
-24.4%
-19.6%
O-27
-------
References
USEPA. 2008. Using the BASINS Meteorological Database (Version 2006). BASINS Technical Note 10.
Office of Water, U.S. Environmental Protection Agency, Washington, DC.
http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf (Accessed June,
2009).
O-28
-------
Appendix P
Model Configuration, Calibration and
Validation
Basin: New England Coastal (NewEng)
-------
Contents
Watershed Background P-4
Water Body Characteristics P-4
Soil Characteristics P-6
Land Use Representation P-6
Point Sources P-10
Meteorological Data P-11
Watershed Segmentation P-13
Calibration Data and Locations P-15
SWAT Modeling P-16
Hydrology Calibration P-16
Hydrology Validation P-19
Hydrology Results for Larger Watershed P-22
Water Quality Calibration and Validation P-23
Water Quality Results for Larger Watershed P-26
References P-27
P-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes P-8
Table 2. Land use distribution for the New England Coastal basin (2001 NLCD) (mi2) P-9
Table 3. Major point source discharges in the New England Coastal basin P-10
Table 4. Precipitation stations for the New England Coastal watershed model P-11
Table 5. Calibration and validation locations in the New England Coastal basin P-15
Table 6. Summary statistics at USGS 01066000 Saco River at Cornish, Maine - calibration period P-19
Table 7. Summary statistics at USGS 01066000 Saco River at Cornish, Maine - validation period P-22
Table 8. Summary statistics (percent error): all Stations - calibration period P-23
Table 9. Summary statistics (percent error): all stations - validation period P-23
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 01066000 Saco River at Cornish, Maine P-24
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 01066000 Saco River at Cornish, Maine P-25
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 01066000 Saco River at Cornish, Maine P-26
Table 13. Summary statistics for water quality at all stations - calibration period 1993-2003 P-26
Figures
Figure 1. Location of the New England Coastal basin P-5
Figure 2. Land use in the New England Coastal basin P-7
Figure 3. Model segmentation and USGS stations utilized for the New England Coastal basin P-14
Figure 4. Mean monthly flow at USGS 01066000 Saco River at Cornish, Maine - calibration period....P-17
Figure 5. Seasonal regression and temporal aggregate at USGS 01066000 Saco River at Cornish, Maine
- calibration period P-17
Figure 6. Seasonal medians and ranges at USGS 01066000 Saco River at Cornish, Maine - calibration
period P-18
Figure 7. Flow exceedance at USGS 01066000 Saco River at Cornish, Maine - calibration period P-18
Figure 8. Mean monthly flow at USGS 01066000 Saco River at Cornish, Maine - validation period P-20
Figure 9. Seasonal regression and temporal aggregate at USGS 01066000 Saco River at Cornish, Maine
-validation period P-20
Figure 10. Seasonal medians and ranges at USGS 01066000 Saco River at Cornish, Maine - validation
period P-21
Figure 11. Flow exceedance at USGS 01066000 Saco River at Cornish, Maine - validation period P-21
Figure 12. Fit for monthly load of TSS at USGS 01066000 Saco River at Cornish, Maine P-24
Figure 13. Fit for monthly load of total phosphorus at USGS 01066000 Saco River at Cornish, Maine.... P-25
Figure 14. Fit for monthly load of total nitrogen at USGS 01066000 Saco River at Cornish, Maine P-26
Table 14. Summary statistics for water quality at all stations - validation period 1983-1993 P-27
P-3
-------
The New England Coastal basin was selected as one of the 15 non-pilot application watersheds for the 20
Watershed study. Watershed modeling for the non-pilot areas is accomplished using the SWAT model only, and
model calibration and validation results are presented in abbreviated form.
Water Body Characteristics
r\
The New England Coastal basins study area encompasses 11 HUCSs and 10,359 mi in Massachusetts,
Maine, and New Hampshire (Figure 1). The study area includes one of EPA's National Estuary Program
sites (Massachusetts Bays), which is also one of EPA's Climate Ready Estuaries sites. The entire model
area is in the New England Physiographic Province. Elevations in the watershed range from sea level
along the coast to greater than 6,000 ft in the White Mountains of New Hampshire.
Average annual precipitation in the watershed ranges from 40 to 50 inches, with higher amounts in the
mountainous regions - up to 100 inches per year at the summit of Mount Washington. About one-half of
this precipitation becomes surface runoff. Average annual air temperature varies from about 43° F in the
north to about 50° F in the south.
Most of the rivers in this watershed originate in mountainous forested areas with headwaters defined by
fast-flowing water with cobble and boulder-bottom streams. Flow in these rivers is generally regulated
by upstream lakes, reservoirs, flood-control dams, and power plants. The watershed also contains a large
number of natural lakes, many of which are enlarged and controlled by dams.
The land uses in the watershed are approximately 64 percent forested; 16 percent residential,
commercial, and industrial; and 6 percent agricultural. Cities include Boston, MA, Portland, ME,
Worcester, MA, and a variety of smaller cities near the Boston area. Major industries include light
manufacturing, pulp and paper production, silviculture, hydroelectric-power generation, tourism, and
seasonal recreation.
P-4
-------
Hydrography
Water (Nat. Atlas Dalaset)
US Census Populated Places
^B Municipalities (pop 5 50,000)
| County Boundaries
Watershed wilhHUCSs
New Hampshire
Presumpscot
(01060001)
Pemigewasset
(01070001)
Winmpesaukee
Mernmack
(01070006)
Contoocook
(01070003)
Charles
(01090001)
Lowell
Lynn
Boston
Nashua
(Oi070004)
/ (01070005)
I •; -a
Cape Cod
(01*0%0002)
Massachusetts
Rhode
Island
GCRP Model Areas - New England Coastal Basins
Base Map
Figure 1. Location of the New England Coastal basin.
P-5
-------
Soil Characteristics
Soils in the watershed, as described in STATSGO soil surveys, fall primarily into hydrologic soil groups (HSGs)
B (moderately high infiltration capacity) and C (moderate infiltration capacity). SWAT uses information drawn
directly from the soils data layer to populate the model.
Land Use Representation
Land use/cover in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage
(Figure 2). NLCD land cover classes were aggregated according to the scheme shown in Table 1 for
representation in the 20 Watershed model. SWAT uses the built-in hydrologic response unit (HRU) overlay
mechanism in the ArcSWAT interface. SWAT HRUs are formed from an intersection of land use and STATSGO
major soils. The distribution of land use in the watershed is summarized in Table 2.
P-6
-------
New Hampshire
Hydrography
Interstate
I I County Boundaries
2001 NLCD Land Use
| Open water
] Developed, open space
3 Developed, low intensity
| Developed, medium intensity
| Developed, high intensity
] Barren land
| Deciduous forest
|^^| Evergreen forest
] Mixed forest
I I Scrub/shrub
I Grassland/herbaceous
I I Pasture/hay
I Cultivated crops
^] Woody wetlands
n Emergent herbaceous wetlands
Lowell
Lynn
Boston
GCRP Model Areas - New England Coastal River Basins
Land Use Map
Figure 2. Land use in the New England Coastal basin.
P-7
-------
Table 1. Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area usually
accounted for as reach area
Deciduous
Evergreen
Mixed
Emergent & woody wetlands
Aquatic bed wetlands (not
emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
P-8
-------
Table 2. Land use distribution for the New England Coastal basin (2001 NLCD) (mi )
HUC8
watershed
Presumpscot
01060001
Saco
01060002
Piscataqua-
Salmon Falls
01060003
Pemigewasset
01070001
Merrimack
01070002
Contoocook
01070003
Nashua
01070004
Concord
01070005
Winnipesaukee
01070006
Charles
01090001
Cape Cod
01090002
Total
Open
water
84.7
50.4
43.3
26.1
92.0
18.9
17.5
12.5
52.4
29.3
8.3
435.5
Developed9
Open
space
76.8
62.0
103.4
21.8
25.3
29.4
30.3
43.0
119.5
100.0
28.0
639.5
Low
density
44.5
23.5
63.4
9.3
8.7
11.3
33.1
48.1
139.0
134.2
27.5
542.7
Medium
density
15.3
5.5
24.2
2.8
3.6
3.1
23.1
41.8
106.5
160.5
12.5
398.9
High
density
7.5
1.5
8.6
0.4
1.3
0.4
6.9
8.9
24.0
58.7
2.6
120.7
Barren
land
4.1
7.1
9.9
1.9
1.2
1.4
1.9
1.1
9.9
5.9
3.9
48.5
Forest
631.1
1,334.1
852.0
905.7
306.5
613.1
323.1
165.1
1 ,068.3
292.0
93.3
6,584.3
Shru bland/
Grassland
28.0
52.5
50.3
16.6
12.1
11.1
7.8
4.5
29.1
8.6
2.8
223.5
Pasture/Hay
65.7
37.5
91.5
10.6
11.4
29.2
42.0
25.9
108.4
38.7
5.3
466.2
Cultivated
14.6
22.0
15.8
11.7
4.4
5.1
5.3
3.8
21.3
3.5
5.3
112.9
Wetland
77.9
104.8
150.2
15.8
19.1
40.9
43.2
45.4
120.1
130.8
38.0
786.2
Total
1,050.2
1,700.8
1,412.6
1,022.7
485.8
764.1
534.2
400.3
1,798.6
962.2
227.4
10,358.9
aThe percent imperviousness applied to each of the developed land uses is as follows: open space (8.22%), low density (32.81%), medium density (60.90%), and high
density (87.25%).
P-9
-------
Point Sources
There are numerous point source discharges in the watershed. Only the major dischargers, generally defined as
those with a design flow greater than 1 MGD are included in the simulation (Table 3). The major dischargers are
represented at long-term average flows, without accounting for changes over time or seasonal variations.
Table 3. Major point source discharges in the New England Coastal basin
NPDES ID
ME0101117
ME01 00048
NH01 00668
NH01 00277
NH0100871
MA0101745
MA0101427
MA0101621
MA01 00447
MA01 00633
MA01 00668
MA0101711
NH01 00056
NH0100170
NH0100161
MA0100013
MA0004561
MA01 00986
MA01 00404
MA01 00579
NH0100471
NH01 00447
NH0100901
NH01 00960
NH0100005
NH01 00706
NH0000230
MA01 00498
Name
SACO WASTEWATER TREATMENT FACI
BIDDEFORDCITYOF
ROCHESTER WWTF
SOMERSWORTH WPCF
EXETER WWTF
AMESBURYWWTP
NEWBURYPORTWWP
HAVERHILLWPAF
GREATER LAWRENCE SD
LOWELL REGIONAL WW UTILITY
CONCORD WWTF
BILLERICAWWTP
DERRYWWTP
NASHUA WWTF
MERRIMACKWWTF
AYER WWTP
HOLLINGSWORTH & VOSE CO
EAST FITCHBURG WWTF
MWRA-CLINTONSTP
MILFORDWWTF
MILFORDWWTF
MANCHESTER WWTF
CONCORD-HALL STREET WWTF
WINNIPESAUKEE RIVER BASIN
ASHLAND WWTF
LINCOLN WWTP
MONADNOCK PAPER MILLS, INC.
MARLBOROUGH EASTERLY WWTP
Design flow
(MGD)
4.2
2.6
5.0
2.4
3.0
1.9
3.4
18.0
52.0
32.0
1.2
5.5
4.0
16.0
5.0
1.8
12.3
3.0
4.3
2.2
34.0
10.1
11.5
1.6
1.5
1.3
5.5
Observed flow
(MGD)
(1991-2006 average)
2.0
4.9
2.8
1.5
2.7
1.8
2.6
10.7
34.2
30.1
1.4
2.9
1.8
12.1
70.0
1.2
2.4
8.0
2.6
3.6
1.4
43.3
4.2
5.9
1.0
7.9
1.2
3.4
P-10
-------
NPDES ID
MA0101001
MA0101788
MA01 00480
MA0 1004 12
ME0002321
MA01 00978
MA01 02598
ME01 00633
ME0100617
NH01 00625
NH01 00234
MA01 02695
MA01 00587
ME0100102
Name
MAYNARD WWTF
HUDSON WWTF
MARLBOROUGH WESTERLY WWTP
WESTBOROUGH WWTP
S D WARREN COMPANY
MEDFIELDWWTP
CHARLES RIVER PCD
SOUTH PORTLAND CITY OF
SANFORD SEWER DISTRICT
HAMPTON WWTP
PORTSMOUTH-PIERCE ISLAND WWTP
SCITUATE_WWTP
PLYMOUTH WWTP
BRUNSWICK SEWER DISTRICT
Design flow
(MGD)
1.5
2.7
2.9
7.7
1.5
4.5
9.3
4.4
4.7
4.8
1.6
1.8
3.9
Observed flow
(MGD)
(1991-2006 average)
1.0
2.9
3.6
5.2
17.4
1.3
3.5
5.0
2.8
2.2
47.9
1.3
1.7
2.7
Most of these point sources have reasonably complete monitoring for total suspended solids (TSS). Assumptions
were made for total nitrogen and total phosphorus depending upon the type of facility. The point sources were
initially represented in the model with the median of reported values for total phosphorus, TSS and total nitrogen.
Meteorological Data
The required meteorological time series for the 20 Watershed SWAT simulations are precipitation and air
temperature. The 20 Watershed simulations do not include water temperature and uses a degree-day method for
snowmelt. SWAT estimates Penmann-Monteith potential evapotranspiration using a statistical weather generator
for inputs other than temperature and precipitation. These meteorological time series are drawn from the
BASINS4 Meteorological Database (USEPA 2008), which provides a consistent, quality-assured set of
nationwide data with gaps filled and records disaggregated. Scenario application requires simulation over 30
years, so the available stations are those with a common 30-year period of record (or one that can be filled from
an approximately co-located station) that covers the year 2003. A total of 52 precipitation stations were identified
for use in the New England Coastal watershed model with a common period of record of 10/1/1972-9/30/2003
(Table 4). Temperature records are sparser; where these are absent temperature is taken from nearby stations with
an elevation correction.
Table 4. Precipitation stations for the New England Coastal watershed model
COOP ID
274732
273530
170934
199316
Name
LINCOLN
GRAFTON
BRUNSWICK
WEST MEDWAY
Latitude
44.0500
43.5667
43.9000
42.1333
Longitude
-71.6667
-71.9500
-69.9333
-71.4333
Temperature
X
X
X
X
Elevation (m)
267
253
21
64
p-11
-------
COOP ID
278972
270998
273182
275780
278885
275013
276550
274218
275150
174566
190408
194105
176905
190736
190770
272174
271683
192451
190190
190860
192997
193505
193876
194313
194744
194760
195285
196486
199923
273024
274480
275211
275639
275712
Name
WEARE
BRISTOL
FRANKLIN FALLS DAM
NEW DURHAM 3 NNW
WARREN
MACDOWELL DAM
OTTER BROOK LAKE
FORT SCOTT
MARLOW
LEWISTON
BARRE FALLS DAM
LAWRENCE
PORTLAND WSFO AP
BLUE HILL
BOSTON WSFO AP
DURHAM
CONCORD WSO AIRPORT
EAST WAREHAM
ASHBURNHAM
BROCKTON
FRANKLIN
HAVERHILL
IPSWICH
LOWELL
MIDDLETON
MILFORD
NEWBURYPORT 3 WNW
PLYMOUTH-KINGSTON
WORCESTER WSO AP
FITZWILLIAM 2 W
LAKEPORT 2
MASSABESIC LAKE
MOUNT WASHINGTON
NASHUA 2 NNW
Latitude
43.0847
43.6001
43.4668
43.4833
43.9098
42.9000
42.9501
43.1830
43.1168
44.1001
42.4334
42.7001
43.6423
42.2123
42.3606
43.1500
43.1954
41.7656
42.6168
42.0500
42.0834
42.7592
42.6667
42.6500
42.6001
42.1667
42.8334
41.9833
42.2673
42.7833
43.5500
42.9833
44.2668
42.7833
Longitude
-71.7382
-71.7167
-71.6500
-71.1833
-71.8877
-71.9832
-72.2332
-71.7500
-72.2000
-70.2167
-72.0332
-71.1667
-70.3044
-71.1146
-71.0105
-70.9500
-71.5010
-70.6693
-71.8833
-71.0000
-71.4167
-71.0608
-70.8666
-71.3666
-71.0167
-71.5167
-70.9333
-70.7000
-71.8760
-72.1833
-71.4667
-71.4000
-71.2999
-71.4832
Temperature
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Elevation (m)
219
143
131
195
216
293
207
41
357
55
277
18
14
192
6
24
105
6
335
24
73
5
24
34
27
85
5
14
301
354
152
76
1909
40
P-12
-------
COOP ID
276818
276945
177479
179314
170844
190535
275629
196783
193624
272800
270681
172238
198757
Name
PINKHAM NOTCH
PLYMOUTH
SANFORD 2 NNW
WEST BUXTON 2 NNW
BRIDGTON 3 NW
BEDFORD
MOUNT SUNAPEE
READING
HINGHAM
EPPING
BENTON 5 SW
EAST HIRAM
WALPOLE 2
Latitude
44.2668
43.7833
43.4668
43.7001
44.0834
42.4833
43.3334
42.5168
42.2333
43.0333
44.0333
43.8833
42.1667
Longitude
-71.2500
-71.6500
-70.7832
-70.6166
-70.7332
-71.2832
-72.0832
-71.1333
-70.9167
-71.0832
-71.9333
-70.7500
-71.2500
Temperature
X
X
X
X
X
X
X
X
X
X
X
Elevation (m)
612
201
85
46
171
49
387
27
9
49
366
161
50
Watershed Segmentation
The New England Coastal basin was divided into 90 subwatersheds for the purposes of modeling (Figure 3). Saco
River at USGS 01066000 was chosen for initial calibration. The model encompasses the complete watershed and
does not require specification of any upstream boundary conditions for application.
P-13
-------
USGS 01100000
USGS 01099500
USGS 01096500
Massachusetts
USGS gages
Hydrography
Interstate
Water (Nat. Atlas Dataset)
US Census Populated PI?.?:
n Model Subbasins
GCRP Model Areas - New England Coastal River Basins
Model Segmentation
NAD_1983_Albers_metere - Map produced 12-23-2010 - P. Cad
Figure 3. Model segmentation and USGS stations utilized for the New England Coastal basin.
P-14
-------
Calibration Data and Locations
The specific site chosen for initial calibration was the Saco River at Cornish, Maine a flow and water quality
monitoring location that approximately coincides with the mouth of an 8-digit HUC at its outflow to the Saco
River. The Saco River watershed was selected because there is a good set of flow and water quality data available
and the watershed lacks major point sources and impoundments. Additional calibration and validation was
pursued at multiple locations (Table 5). Parameters derived on the Saco River were not fully transferable to other
portions of the New England Coastal basin, and additional calibration was conducted at multiple gage locations.
Table 5. Calibration and validation locations in the New England Coastal basin
Station name
Saco River at Cornish, Maine
Nashua River at Eat Pepperel, MA
Concord River at River Meadow Brook at
Lowell, MA
Merrimack River below Concord River at Lowell,
MA
USGS ID
01066000
01096500
01099500
01100000
Drainage area
(mi2)
1293
435
400
4635
Hydrology
calibration
X
X
X
X
Water quality
calibration
X
X
The model hydrology calibration period was set to Water Years 1993-2003 (within the 32-year period of record
for modeling). Hydrologic validation was then performed on Water Years 1983-1993. Water quality calibration
used calendar years 1993-2003, while validation used 1983-1993.
P-15
-------
SWAT Modeling
Hydrology Calibration
A spatial calibration approach was adopted for GCRP-SWAT modeling for the New England Coastal basin. A
systematic adjustment of parameters has been adopted and some adjustments are applied throughout the basin.
Most of the calibration efforts were geared toward getting a closer match between simulated and observed flows
at the outlet of calibration focus area.
Land Use/Soil/Slope Definition
A 5/10/5 percent threshold was used for land use/soil/slope in the SWAT model while defining the HRUs. Urban
land use classes were exempted from the HRU overlay thresholds.
The calibration focus area (Saco River) includes nine subwatersheds and is generally representative of the general
land use characteristics of the overall watershed. The parameters were adjusted within the practical range to
obtain reasonable fit between the simulated and measured flows in terms of Nash-Sutcliffe modeling efficiency
and the high flow and low flow components as well as the seasonal flows.
The water balance of whole New England Coastal basin predicted by the SWAT model over the 32-year
simulation period is as follows:
PRECIP = 1188.0 MM
SNOW FALL = 250.83 MM
SNOW MELT = 232.97 MM
SUBLIMATION = 16.71 MM
SURFACE RUNOFF Q = 236.32 MM
LATERAL SOIL Q = 144.44 MM
TILE Q = 0.00 MM
GROUNDWATER (SHAL AQ) Q = 214.89 MM
REVAP (SHAL AQ => SOIL/PLANTS) = 20.61 MM
DEEP AQ RECHARGE = 29.40 MM
TOTAL AQ RECHARGE = 265.08 MM
TOTAL WATER YLD = 571.26 MM
PERCOLATION OUT OF SOIL = 241.22 MM
ET = 555.8 MM
PET = 1041.6MM
TRANSMISSION LOSSES = 24.39 MM
Hydrologic calibration adjustments focused on the following parameters:
• CN2 (initial SCS runoff curve number for moisture condition II)
• ESCO (soil evaporation compensation factor)
• SURLAG (surface runoff lag coefficient)
• SOL_AWC (available water capacity of the soil layer, mm water/mm of soil)
• ALPHA_BF (baseflow alpha factor, days)
• GW_DELAY (groundwater delay time, days)
• GWQMIN (threshold depth of water in the shallow aquifer required for return flow to occur, mm)
• GW_REVAP (groundwater "revap" coefficient)
• CH_N1 (Manning's "n" value for tributary channels)
• CH_N2 (Manning's "n" value for main channels)
P-16
-------
Calibration results for the Saco River are summarized in Figure 4, Figure 5, Figure 6, Figure 7 and Table 6.
12000
10000
Avg Monthly Rainfall (in)
Avg Observed Flow (10/1/1993 to 9/30/2003)
Modeled Flow (Same Period)
O-93
A-95
O-96
A-98
O-99
A-01
O-02
Month
Figure 4. Mean monthly flow at USGS 01066000 Saco River at Cornish, Maine - calibration period.
Avg Flow (10/1/1993 to 9/30/2003)
•Line of Equal Value
-Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1993 to 9/30/2003)
•Avg Modeled Flow (Same Period)
10000
^000
_g
^000
JB
/.
•
•
3x - 344.
0.8392
S
/
27
.-*'
0 2000 4000 6000 8000 10000
Average Observed Flow (cfs)
10 11 12 1 23456789
Month
Figure 5. Seasonal regression and temporal aggregate at USGS 01066000 Saco River at Cornish, Maine
- calibration period.
P-17
-------
• Observed (25th, 75th) Average Monthly Rainfall (in) -Median Observed Flow (10/1/1993 to 9/30/2003) Modeled (Median, 25th, 75th)
12000
10000
2000
10 11
Month
Figure 6. Seasonal medians and ranges at USGS 01066000 Saco River at Cornish, Maine - calibration
period.
•Observed Flow Duration (10/1/1993 to 9/30/2003)
Modeled Flow Duration (10/1/1993 to 9/30/2003)
100000
100
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 7. Flow exceedance at USGS 01066000 Saco River at Cornish, Maine - calibration period.
P-18
-------
Table 6. Summary statistics at USGS 01066000 Saco River at Cornish, Maine - calibration period
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET(S) 4, 5
10-Year Analysis F^riod: 10/1/1993 - 9/30/2003
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Srjring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error -Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sjjt£liffeJDpj5fficjejTt^^
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
28.52
10.55
5.01
4.53
5.82
3.71
14.45
6.37
1.17
Error Statistics
1.08
-3.20
12.54
42.85
Observed Flow Gage
USGS 01066000 Saco River at Cornish, Maine
Hydrologic Unit Code: 1060002
Latitude: 43.80805556
Longitude: -70.781 6667
Drainage Area (sq-rri): 1293
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
-4.86 » | 30
-42.36
15.83
4.18
39.96
0.611
0.471
0.713
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
28.21
9.38
5.17
3.17
6.12
6.44
12.48
6.11
0.83
Clear [
Hydrology Validation
Hydrology validation for Saco River was performed for the period 10/1/1983 through 9/30/1993. Results are
presented in below. The validation achieves a moderately high coefficient of model fit efficiency, but is over on
summer flow volumes (Figure 8, Figure 9, Figure 10, Figure 11 and Table 7).
P-19
-------
15000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1983 to 9/30/1993)
•Avg Modeled Flow (Same Period)
c
ro
OL
O-83
A-85
O-86
A-88
O-89
A-91
O-92
Month
Figure 8. Mean monthly flow at USGS 01066000 Saco River at Cornish, Maine - validation period.
Avg Flow (10/1/1983 to 9/30/1993)
•Line of Equal Value
-Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1983 to 9/30/1993)
•Avg Modeled Flow (Same Period)
10000
^000
_g
^000
-------
• Observed (25th, 75th) Average Monthly Rainfall (in) -Median Observed Flow (10/1/1983 to 9/30/1993) Modeled (Median, 25th, 75th)
10000
9000
8000
Jan Feb Mar Apr May Jun JM! Aug Sep Oct Nov Dec
ro
CD
or
12
Figure 10. Seasonal medians and ranges at USGS 01066000 Saco River at Cornish, Maine -validation
period.
•Observed Flow Duration (10/1/1983 to 9/30/1993)
Modeled Flow Duration (10/1/1983 to 9/30/1993)
100000
10
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 11. Flow exceedance at USGS 01066000 Saco River at Cornish, Maine - validation period.
P-21
-------
Table 7. Summary statistics at USGS 01066000 Saco River at Cornish, Maine - validation period
REACH OUTFLOW FROM OUTLET(S) 4, 5
10-Year Analysis F^riod: 10/1/1983 - 9/30/1993
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Srjring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sjjt£liffeJDpj5fficjejTt^^
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
27.57
9.12
5.69
4.40
5.81
4.51
12.86
5.75
1.04
Error Statistics
0.67
6.45
-2.12
43.06
USGS 01066000 Saco River at Cornish, Maine
Hydrologic Unit Code: 1060002
Latitude: 43.80805556
Longitude: -70.781 6667
Drainage Area (sq-rri): 1293
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
-7.98 » | 30
-20.75
4.43
-9.91
31.54
0.764
0.555
0.844
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
27.39
9.32
5.35
3.07
6.31
5.69
12.31
6.39
0.79
Clear [
Hydrology Results for Larger Watershed
As described above, parameters determined for the gage at Saco River were initially transferred to other gages in
the watershed. However, changes to subbasin level parameter were required to fit the model to the observed
flows. In all, calibration and validation was pursued at a total of eight gages throughout the watershed. Results of
the calibration and validation exercise are summarized in Table 8 and Table 9, respectively. Calibration and
validation results were acceptable at most gages.
P-22
-------
Table 8. Summary statistics (percent error): all Stations - calibration period
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient of Efficiency, E:
Monthly Daily Nash-Sutcliffe Coefficient of Efficiency, E:
01066000
1.08
-3.20
12.54
42.85
-4.86
-42.36
15.83
4.18
39.96
0.611
0.713
01096500
-7.58
-26.88
-2.19
14.23
2.20
-20.66
-5.49
-18.32
6.63
0.627
0.760
01099500
-4.83
7.26
-6.16
58.91
25.28
-21.25
-19.03
2.38
44.46
0.715
0.741
01100000
1.83
4.19
-1.16
63.71
19.04
-22.06
-3.67
-1.23
48.94
0.714
0.806
Table 9. Summary statistics (percent error): all stations - validation period
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient of Efficiency, E:
Monthly Daily Nash-Sutcliffe Coefficient of Efficiency, E:
01066000
0.67
6.45
-2.12
43.06
-7.98
-20.75
4.43
-9.91
31.54
0.764
0.844
01096500
-6.51
-17.04
-7.27
29.39
0.24
-16.04
-13.04
-25.22
-0.50
0.700
0.856
01099500
-9.06
-8.56
-1.91
52.96
8.74
-28.69
-21.37
0.61
38.02
0.759
0.797
01100000
-8.01
-2.95
-15.94
63.08
3.27
-30.46
-20.07
-17.38
32.46
0.719
0.751
Water Quality Calibration and Validation
Initial calibration and validation of water quality was done on the Saco River (USGS 01066000), using 1993-2003
for calibration and 1983-1993 for validation. As with hydrology, calibration was performed on the later period as
this better reflects the land use included in the model.
Calibration adjustments for sediment focused on the following parameters:
• SPCON (linear parameter for estimating maximum amount of sediment that can be re-entrained during
channel sediment routing)
P-23
-------
• SPEXP (exponential parameter for estimating maximum amount of sediment that can be re-entrained
during channel sediment routing)
• CH_COV (channel cover factor)
• CH_EROD (channel erodibility factor)
• USLE_P (USLE support practice factor)
Simulated and estimated sediment loads at the Saco River station for both the calibration and validation periods
are shown in Figure 12 and statistics for the two periods are provided separately in Table 10. The key statistic in
Table 10 is the relative percent error, which shows the error in the prediction of monthly load normalized to the
estimated load. Table 10 also shows the relative average absolute error, which is the average of the relative
magnitude of errors in individual monthly load predictions. This number is inflated by outlier months in which the
simulated and estimated loads differ by large amounts (which may be as easily due to uncertainty in the estimated
load due to limited data as to problems with the model) and the third statistic, the relative median absolute error,
is likely more relevant and shows better agreement.
TSS
100,000
10,000
• Regression Loads
•Simulated Loads
Figure 12. Fit for monthly load of TSS at USGS 01066000 Saco River at Cornish, Maine.
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 01066000 Saco River at Cornish, Maine
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1993-1995)
-9.0%
39.9%
26.2%
Validation period
(1983-1993)
3.2%
45.9%
18.8%
P-24
-------
Calibration adjustments for total phosphorus and total nitrogen focused on the following parameters:
• RHOQ (algal respiration rate at 20° C)
• PHOSKD (phosphorus soil partitioning coefficient)
• PSP (phosphorus availability index)
• RS2 (benthic source rate for dissolved P in the reach at 20° C)
• RS5 (organic P settling rate in the reach at 20° C)
• BC4 (rate constant for mineralization of organic P to dissolved P in the reach at 20° C)
• RS4 (rate coefficient for organic N settling in the reach at 20° C)
Results for the phosphorus simulation are shown in Figure 13 and Table 11. Results for the nitrogen simulation
are shown in Figure 14 and Table 12. The model fit is generally acceptable.
Total P
100
o
E
7/5
o
- Regression Loads
-Simulated Loads
Figure 13. Fit for monthly load of total phosphorus at USGS 01066000 Saco River at Cornish, Maine.
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 01066000 Saco River at Cornish, Maine
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1993-1995)
9.6%
48.5%
62.0%
Validation period
(1983-1993)
-11.5%
33.5%
23.6%
P-25
-------
Total N
450
-Averaging Loads
-Simulated Loads
v <* , <* <-v
-------
Table 14. Summary statistics for water quality at all stations - validation period 1983-1993
Station
Relative Percent Error TSS Load
Relative Percent Error TP Load
Relative Percent Error TN Load
01066000
3.2%
-11.5%
26.3%
01100000
-
-
-
References
USEPA. 2008. Using the BASINS Meteorological Database (Version 2006). BASINS Technical Note 10.
Office of Water, U.S. Environmental Protection Agency, Washington, DC.
http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf (Accessed June,
2009).
P-27
-------
Appendix Q
Model Configuration, Calibration and
Validation
Basin: Rio Grande Valley (RioGra)
Q-i
-------
Contents
Watershed Background Q-4
Water Body Characteristics Q-4
Soil Characteristics Q-6
Land Use Representation Q-6
Point Sources Q-10
Meteorological Data Q-10
Watershed Segmentation Q-12
Calibration Data and Locations Q-14
SWAT Modeling Q-15
Assumptions Q-15
Hydrology Calibration Q-15
Hydrology Validation Q-19
Hydrology Results for Larger Watershed Q-22
Water Quality Calibration and Validation Q-24
Water Quality Results for Larger Watershed Q-27
References Q-28
Q-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes Q-8
Table 2. Land use distribution for the Rio Grande Valley basin (2001 NLCD; mi2) Q-9
Table 3. Major point source discharges in the Rio Grande Valley basin Q-10
Table 4. Precipitation stations for the Rio Grande Valley watershed model Q-10
Table 5. Calibration and validation locations in the Rio Grande Valley basin Q-14
Table 6. Summary statistics USGS 08227000 Saguache Creek near Saguache, CO - calibration
period Q-19
Table 7. Summary statistics at USGS 08227000 Saguache Creek near Saguache, CO - validation
period Q-22
Table 7. Summary statistics (percent error): all stations - calibration period Q-23
Table 8. Summary statistics: all stations - validation period Q-24
Table 9. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 08276500 Rio Grande near Taos, NM Q-25
Table 10. Model fit statistics (observed minus predicted) for monthly phosphorus loads using
stratified regression at USGS 08276500 Rio Grande near Taos, NM Q-26
Table 11. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 08276500 Rio Grande near Taos, NM Q-27
Table 12. Summary statistics for water quality at all stations - calibration period 1985-2003 Q-28
Table 13. Summary statistics for water quality at all stations - validation period 1973-1984 Q-28
Figures
Figure 1. Location of the Rio Grande Valley Q-5
Figure 2. Land use in the Rio Grande Valley basin Q-7
Figure 3. Model segmentation and USGS stations utilized for the Rio Grande Valley basin Q-13
Figure 4. Mean monthly flow at USGS 08227000 Saguache Creek near Saguache, CO - calibration
period Q-16
Figure 5. Seasonal regression and temporal aggregate at USGS 08227000 Saguache Creek near
Saguache, CO - calibration period Q-17
Figure 6. Seasonal medians and ranges at USGS 08227000 Saguache Creek near Saguache, CO -
calibration period Q-17
Figure 7. Flow exceedance at USGS 08227000 Saguache Creek near Saguache, CO - calibration
period Q-18
Figure 8. Mean monthly flow at USGS 08227000 Saguache Creek near Saguache, CO - validation
period Q-20
Figure 9. Seasonal regression and temporal aggregate at USGS 08227000 Saguache Creek near
Saguache, CO - validation period Q-20
Figure 10. Seasonal medians and ranges at USGS 08227000 Saguache Creek near Saguache, CO -
validation period Q-21
Figure 11. Flow exceedance at USGS 08227000 Saguache Creek near Saguache, CO - validation
period Q-21
Figure 12. Fit for monthly load of TSS at USGS 08276500 Rio Grande near Taos, NM Q-25
Figure 14. Fit for monthly load of total phosphorus at USGS 08276500 Rio Grande near Taos, NM Q-26
Figure 15. Fit for monthly load of total nitrogen at USGS 08276500 Rio Grande near Taos, NM Q-27
Q-3
-------
The Rio Grande Valley was selected as one of the 15 non-pilot application watersheds for the 20 Watershed
study. Watershed modeling for the non-pilot areas is accomplished using the SWAT model only, and model
calibration and validation results are presented in abbreviated form.
Water Body Characteristics
The Rio Grande flows from southwestern Colorado to the Gulf of Mexico. The model study area is the upstream
portion of the Rio Grande Valley, spanning parts of Colorado and New Mexico (Figure 1). This includes an area
of about 19,000 mi2 in ten HUCSs within HUCs 1301 and 1302.
The watershed is located in three physiographic provinces: Southern Rocky Mountains; Basin and Range
Provinces; and Colorado Plateaus Provinces. Extreme contrasts in precipitation, runoff, and temperature
characteristics exist between the Southern Rocky Mountains and the Basin and Range Provinces. These
characteristics strongly affect land and water use in the watershed (Levings et al., 1998; USGS, 2009a).
The headwaters of the Rio Grande originate in the mountains of southern Colorado at an altitude of over 13,000
ft. At the lower end of the watershed, just downstream of Albuquerque, NM, the altitude is approximately 3,700
ft. The climate in the high mountain headwater areas of the Rio Grande and its northern tributaries is alpine tundra
where average annual precipitation can exceed 50 inches, most in the form of snow. In contrast, near the lower
boundary of the model area, the Rio Grande flows through desert where average annual precipitation is less than 9
inches, most in the form of summer thunderstorms.
Rangeland is dominant in the Basin and Range Province, and forest is dominant in the Southern Rocky Mountains
and Colorado Plateaus Provinces; they occupy 54 percent and 35 percent of the model study area respectively.
The cities of Taos, Santa Fe, Albuquerque, and Las Cruces, NM are located in the watershed but developed land
constitutes less than 3 percent of the land area. Agricultural land use (5 percent) is limited primarily to areas
where surface water or shallow groundwater is available for irrigation. Almost all public and domestic water
supplies rely on groundwater, primarily from deeper aquifers. Surface water availability typically is necessary for
agriculture with the exception of a few areas where groundwater is available in sufficient quantities.
Historically, streamflow in the Rio Grande was caused by spring snowmelt and summer monsoon thunderstorms.
This natural streamflow pattern has been altered and regulated by the construction of reservoirs on the main stem
and tributaries that impound and store water for later use, primarily irrigation. Complex interactions occur
between groundwater and surface water in the Rio Grande flood plain. A system of canals distributes surface
water for agricultural irrigation and a system of drains intercepts shallow groundwater and returns it to the Rio
Grande. Surface water leaks from the Rio Grande and canals to recharge the shallow groundwater system. In
places, deeper groundwater flows upward to recharge the shallow groundwater system and/or to contribute flow
to the Rio Grande. In addition, excess applied irrigation water infiltrates and recharges the shallow groundwater
system.
Q-4
-------
Hydrography
Water (Nat. Atlas Dataset)
US Census Populated Places
^H Municipalities (pap i 50.000)
| | County Boundaries
~~| Watershed with HUC8s
Saguache
(13010004)
Rio Grande
Headwaters
13010001)
San Luis
13010003)
'lamosa-Trmchera
(13010002)--/
Upper Rio Grande
(13020101)
Rio Chama
(13020102)
Jemez
(13020202)
Albuquerque
Santa Fe
»
Glorieta
Rio Grande-
Santa Fe
(13020201)
Cedar Grove
Rio Grande-
Albuquerque
(13020203)
GCRP Model Areas - Rio Grande Valley
Base Map
Figure 1. Location of the Rio Grande Valley.
Q-5
-------
Soil Characteristics
Soils in the watershed, as described in STATSGO soil surveys, fall primarily into hydrologic soil groups (HSGs)
B (moderately high infiltration capacity) and D (low infiltration capacity). SWAT uses information drawn directly
from the soils data layer to populate the model.
Land Use Representation
Land use/cover in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage and is
predominantly rangeland in the south, and forest in the Southern Rocky Mountains and Colorado Plateaus
Provinces in the north (Figure 2). NLCD land cover classes were aggregated according to the scheme shown in
Table 1 for representation in the 20 Watershed model. SWAT uses the built-in hydrologic response unit (HRU)
overlay mechanism in the ArcSWAT interface. SWAT HRUs are formed from an intersection of land use and
SSURGO major soils. The distribution of land use in the watershed is summarized in Table 2.
Q-6
-------
Legend
Hydrography
—^ Interstate
I I County Boundaries
2001 NLCD Land Use
I I Open water
^^| Perennial Ice/Snow
^] Developed, open space
1H[ Developed, low intensity
HB Developed, medium intensity
|^| Developed, high intensity
^J Barren land
| Deciduous forest
|^| Evergreen Forest
I I Mixed forest
I I Scrub/shrub
I 1 Grasslanct/herbaceous
^] Pasture/hay
^j Cultivated crops
I I Woody wetlands
1 Emergent herbaceous wellands
New Mex ico *'
GCRP McxJel Areas - Rio Grande River Basin
Land Use Map
Figure 2. Land use in the Rio Grande Valley basin.
Q-7
-------
Table 1. Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area usually
accounted for as reach area
Deciduous
Evergreen
Mixed
Emergent & woody wetlands
Aquatic bed wetlands (not emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
Q-8
-------
Table 2. Land use distribution for the Rio Grande Valley basin (2001 NLCD; mi )
HUC8
watershed
13010001
13010002
13010003
13010004
13010005
13020101
13020102
13020201
13020202
13020203
Total
Open
water
4.8
7.1
4.0
0.8
1.5
3.7
22.7
4.3
1.9
8.2
59.0
Developed9
Open
space
4.7
27.0
9.6
8.6
6.5
29.9
18.0
26.6
5.0
81.9
217.8
Low
density
1.4
17.1
14.2
7.0
1.9
13.1
2.1
12.9
1.0
94.4
165.1
Medium
density
0.1
3.3
1.2
0.2
0.3
1.4
0.1
2.6
0.0
33.6
42.8
High
density
0.0
0.2
0.0
0.0
0.0
0.2
0.0
0.3
0.0
5.6
6.4
Barren
land
48.4
21.8
71.8
10.3
5.1
20.3
5.1
0.3
2.9
10.8
196.9
Forest
770.8
639.1
301.9
464.5
300.2
1 ,457.6
1 ,546.6
470.7
464.2
273.7
6,689.3
Shrub and
Grassland
481.6
1,307.0
985.4
712.8
359.2
1,660.1
1,487.4
1,337.7
549.6
1,400.8
10,281.6
Pasture/
Hay
8.8
272.5
256.6
89.7
42.1
20.8
22.8
1.2
3.7
56.3
774.6
Cultivated
0.7
11.8
4.3
0.3
0.4
33.5
12.0
13.1
4.4
46.6
127.0
Wetland
59.0
119.4
43.5
48.7
51.9
13.5
41.2
2.2
6.2
12.1
397.9
Total
1,380.6
2,426.6
1,692.7
1,343.0
769.2
3,254.2
3,158.0
1 ,871 .6
1 ,039.0
2,024.1
18,959.0
aThe percent imperviousness applied to each of the developed land uses is as follows: open space (8.76%), low density (32.36%), medium density (60.49%), and high
density (84.32%).
Q-9
-------
Point Sources
There are several point source discharges in the watershed. Only the major dischargers, with a design flow greater
than 1 MGD are included in the simulation (Table 3). The major dischargers are represented at long-term average
flows, without accounting for changes over time or seasonal variations.
Table 3. Major point source discharges in the Rio Grande Valley basin
ID
NM0020150
NM0022250
NM0027987
NM0022292
NM0028355
NM0029351
NM0024066
NM0022101
NM0022306
NM0024899
NM0020141
CO0044458
Name
Belen, City of
Albuquerque, City of (WWTP#2)
Rancho, City of
Santa Fe, City of (Airport Rd)
Los Alamos National Laboratory
Espanola, City of
Taos, Town of
Village of Taos Ski Valley
Molycorp Inc - Questa
Red River AWWT, Town of
Los Alamos County (Bayo Canyon)
Alamosa, City of
Design glow
(mgd)
1.2
60
2.4
6.5
1.01
1.25
0.13
2.5
1.37
2.6
Observed
flow (mgd)
0.79
58.57
3.02
5.79
0.62
0.91
0.98
0.04
0.49
0.47
12.22
1.53
The point sources were initially represented in the model with the median of reported values for TSS and an
assumed total nitrogen concentration of 11.2 mg/L and assumed total phosphorus concentration of 7.0 mg/L for
secondary treatment facilities (TetraTech 1999).
Meteorological Data
The required meteorological time series data for the 20 Watershed SWAT simulations are precipitation and air
temperature. The 20 Watershed simulations do not include water temperature simulation and use a degree-day
method for snowmelt. SWAT estimates Penman-Monteith potential evapotranspiration using a statistical weather
generator for inputs other than temperature and precipitation. These meteorological time series are drawn from the
BASINS4 Meteorological Database (USEPA 2008), which provides a consistent, quality-assured set of
nationwide data with gaps filled and records disaggregated. Scenario application requires simulation over 30
years, so the available stations are those with a common 30-year period of record (or one that can be filled from
an approximately co-located station) that covers the year 2001. A total of 54 precipitation stations were identified
for use in the Rio Grande model with a common period of record of 10/1/1972-9/30/2002 (Table 4). Temperature
records are sparser; where these are absent temperature is taken from nearby stations with an elevation correction.
Table 4. Precipitation stations for the Rio Grande Valley watershed model
ID
50130
50776
51458
51713
52184
53541
Name
CO050130
CO050776
CO051458
CO051713
CO052184
CO053541
Latitude
37.4389
37.4723
37.7067
38.4462
37.6742
37.7333
Longitude
-105.8610
-105.5040
-106.1440
-106.7610
-106.3240
-105.5110
Elevation
2296
2390
2339
2438
2397
2494
Temperature
Yes
Yes
Yes
Yes
Yes
Yes
Q-10
-------
ID
53951
54734
55322
55706
56203
57337
57428
57460
57656
58220
58931
290041
290234
290245
290915
291000
291180
291389
291630
291664
291982
292241
292608
292700
292837
293031
293060
293488
293511
293586
293592
294366
294369
294960
295084
295150
295965
296676
297323
298015
Name
CO053951
CO054734
CO055322
CO055706
CO056203
CO057337
CO057428
CO057460
CO057656
CO058220
CO058931
NM290041
NM290234
NM290245
NM290915
NM291000
NM291180
NM291389
NM291630
NM291664
NM291982
NM292241
NM292608
NM292700
NM292837
NM293031
NM293060
NM293488
NM293511
NM293586
NM293592
NM294366
NM294369
NM294960
NM295084
NM295150
NM295965
NM296676
NM297323
NM298015
Latitude
37.7718
38.0248
37.1742
37.5811
38.0207
38.0858
37.1954
38.4040
37.8193
37.0708
38.1312
36.2403
35.0357
36.0909
34.4220
36.3120
36.7444
36.4820
36.7409
36.9178
35.6414
36.0106
36.9359
36.5575
36.5928
35.9882
34.8242
35.8918
36.3336
35.5817
35.2656
35.3886
35.7784
36.3043
35.8645
34.7675
34.5209
35.5490
36.7059
35.2106
Longitude
-107.1090
-107.3140
-105.9390
-106.1870
-107.6680
-106.1440
-107.6650
-105.4660
-106.4270
-106.6210
-106.0560
-106.8320
-105.5950
-106.5780
-106.9680
-107.0000
-105.2620
-106.7300
-106.0810
-106.0340
-105.4480
-106.6870
-107.0540
-106.3210
-106.7610
-106.2600
-105.6880
-105.4030
-106.3650
-105.9750
-105.9430
-105.5860
-106.5530
-107.1810
-106.7460
-105.8610
-105.5040
-106.1440
-106.7610
-106.3240
Elevation
2758
2643
2344
2345
2390
2347
2421
2579
2828
2521
2396
1945
1618
1731
1443
2635
2440
2386
2332
2393
1695
2147
2071
2524
2054
1702
1871
2515
1981
2292
2042
1642
1909
2201
2263
1475
1987
2096
2644
2143
Temperature
Yes
Yes
Yes
Yes
Yes
Yes
No
No
Yes
No
Yes
Yes
Yes
Yes
Yes
No
No
No
yes
yes
No
yes
yes
yes
yes
yes
yes
yes
No
No
No
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Q-ll
-------
ID
298085
298518
298668
298845
299031
299085
299820
Name
NM298085
NM298518
NM298668
NM298845
NM299031
NM299085
NM299820
Latitude
35.6194
35.1767
36.3906
36.7664
35.7992
36.6511
35.9479
Longitude
-105.5110
-107.1090
-107.3140
-105.9390
-106.1870
-107.6680
-106.1440
Elevation
2059
1943
2123
2275
2042
2481
2505
Temperature
Yes
Yes
Yes
Yes
Yes
No
Yes
Watershed Segmentation
The Rio Grande basin was divided into 74 subwatersheds for the purposes of modeling (Figure 3). The model
encompasses the complete watershed and does not require specification of any upstream boundary conditions for
application.
Q-12
-------
Legend
USGS gages
Hydrography
Interstate
V\Ster (Nal. Atias Datasel)
US Census Populated Places
I I Ccunly Boundaries
I Model Subbasins
GCRP Model Areas - Rio Grande River Basin
Model Segmentation
Figure 3. Model segmentation and USGS stations utilized for the Rio Grande Valley basin.
Q-13
-------
Calibration Data and Locations
The specific site chosen for initial calibration was Saguache Creek near Saguache, CO (USGS 08227000), which
is the only gaging station in the basin without any reservoirs. Calibration and validation were pursued at multiple
locations (Table 5). Parameters derived at the Saguache Creek station were transferred to other portions of the Rio
Grande basin.
Table 5. Calibration and validation locations in the Rio Grande Valley basin
Station Name
Saguache Creek near Saguache, CO
Rio Grande near Lobatos, NM
RioGrande nearTaos, NM
RioGrande at Otowi Bridge, NM
RioGrande at Albuquerque
USGS ID
08227000
08251500
08276500
08313000
08330000
Drainage area
(mi2)
595
7700
9730
14300
17440
Hydrology
calibration
X
X
X
X
X
Water quality
calibration
X
X
X
X
X
The model hydrology calibration period was set to Water Years 1992-2001 (within the 30-year period of record
for modeling). Hydrologic validation was then performed on Water Years 1982-1991. Water quality calibration
used calendar years 1985-2003, while validation used 1973-1984. However, there was some variation to this time
period across the monitoring stations depending on the availability of monitored data.
Q-14
-------
SWAT Modeling
Assumptions
Ten major reservoirs occur in the Rio Grande basin. Pertinent reservoir information including surface area and
storage at principal (normal) and emergency spillway levels for the reservoirs modeled were obtained from the
National Inventory of Dams (NID) database. The SWAT model provides four options to simulate reservoir
outflow: measured daily outflow, measured monthly outflow, average annual release rate for uncontrolled
reservoir, and controlled outflow with target release. Keeping in view the 20 Watershed climate change impact
evaluation application, it was assumed that the best representation of the reservoirs was to simulate them without
supplying time series of outflow records. Therefore, the target release approach was used in the GCRP-SWAT
model.
Elevation bands were also created in the subwatersheds where elevation was above 3,000 m to account for the
impact of higher elevation. Moreover, since the northern and southern part of the Rio Grande basin are
geographically different, certain parameters have different values for the Colorado part of Rio Grande and for the
New Mexico part of the Rio Grande basin, respectively.
Hydrology Calibration
A spatial calibration approach was not adopted for GCRP-SWAT modeling for the Rio Grande basin; however, a
systematic adjustment of parameters was adopted and some adjustments were applied throughout the basin. Most
of the calibration efforts were geared toward getting a closer match between simulated and observed flows at one
of the USGS gaging stations in the basin.
Land Use/Soil/Slope Definition
A 5/10/5 percent threshold was used for land use/soil/slope in the SWAT model while defining the HRUs. Urban
land use classes were exempted from the HRU overlay thresholds.
The parameters were adjusted within the practical range at the calibration focus area to obtain reasonable fit
between the simulated and measured flows in terms of Nash-Sutcliffe modeling efficiency and the high flow and
low flow components as well as the seasonal flows.
The water balance of the whole Rio Grande basin predicted by the SWAT model over the 30-year simulation
period is as follows:
PRECIP = 307.4 MM
SNOW FALL = 38.57 MM
SNOW MELT = 34.39 MM
SUBLIMATION = 4.58 MM
SURFACE RUNOFF Q = 5.73 MM
LATERAL SOIL Q = 15.40 MM
TILE Q = 0.00 MM
GROUNDWATER (SHAL AQ) Q = 6.51 MM
REVAP (SHAL AQ => SOIL/PLANTS) = 7.47 MM
DEEP AQ RECHARGE = 1.22 MM
TOTAL AQ RECHARGE = 15.21 MM
TOTAL WATER YLD = 24.24 MM
PERCOLATION OUT OF SOIL = 12.18 MM
ET = 274.8 MM
Q-15
-------
PET = 1946.4MM
TRANSMISSION LOSSES =
3.41 MM
Hydrologic calibration adjustments focused on the following parameters:
• Snow parameters SMTMP, SMFMX, SMFMN, TIMP
• Baseflow factor
• GW_DELAY (groundwater delay time)
• GWQMN (threshold depth of water in the shallow aquifer required for return flow to occur)
• Rchrg_DP
• CH_K2 (channel hydraulic conductivity)
• NDTarg
• Curve Number
• Temperature Lapse Rate
• SURLAG (surface runoff lag time [days])
• ESCO (soil evaporation compensation factor)
• FFCB (fraction of field capacity)
Calibration results for the Rio Grande basin at Saguache Creek near Saguache, CO are summarized in Figures 4
through 7 and Table 6. In general, the model represents the observed flow adequately, both in terms of volume
and timing of the peaks (Figure 7 and Table 6).
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2001 )
•Avg Modeled Flow (Same Period)
400
200
92 O-93 O-94 O-95 O-96 O-97 O-98 O-99 O-OO
Figure 4. Mean monthly flow at USGS 08227000 Saguache Creek near Saguache, CO - calibration
period.
Q-16
-------
200
a
o
o
150
100
T3
O
<
• AvgFlowfl 0/1/1992 to 9/30/2001)
Line of Equal Value
Best-Fit Line
:
:
.,-'
/
y = 0.6085x+ 21.667
K\ = 0.7629 j
j j \£
0 50 100 150 200
Average Observed Flow (cfs)
200
150
_g
100
Avg Monthly Rainfall (in)
-Avg Observed Flow(1 0/1/1 992 to 9/30/2001)
Avg Modeled Flow (Same Period)
10 11 12 1
234567
Month
Figure 5. Seasonal regression and temporal aggregate at USGS 08227000 Saguache Creek near
Saguache, CO - calibration period
To Lower Bound Average Monthly Rainfall (in) -Median Observed Flow (10/1/1992 to 9/30/2001) Modeled (Median, 25th, 75th)
250
11
12
234
Month
4.0
Figure 6. Seasonal medians and ranges at USGS 08227000 Saguache Creek near Saguache, CO -
calibration period
Q-17
-------
i
_g
LL
0)
D)
ro
OJ
ro
Q
1000
•Observed Flow Duration (10/1/1992 to 9/30/2001
Modeled Flow Duration (10/1/1992 to 9/30/2001
0.1
0% 10% 20% 30% 40% 50% 60% 70% 80% 90%, 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 7. Flow exceedance at USGS 08227000 Saguache Creek near Saguache, CO - calibration period.
Q-18
-------
Table 6. Summary statistics USGS 08227000 Saguache Creek near Saguache, CO - calibration period
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 49
9-Year Analysis F^riod: 10/1/1992 - 9/30/2001
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
_JpJalj3fj3inTulatejdjT^^
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
_EjTOnnJ05yTigjTe£tJlows^__
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
1.38
0.41
0.33
0.46
0.28
0.16
0.48
0.08
0.03
Error Statistics
-4.92
-2.42
-14.20
14.91
Observed Flow Gage
USGS 08227000 SAGUACHE CREEK NEAR SAGUACHE, CO
Hydrologic Unit Code: 13010004
Latitude: 38.16333294
Longitude: -106.2838
Drainage Area (sq-rri): 595
Total Observed In-stream Flow:
_j£t^l_of^b^ej^«djTigjTe^tJ^%Jbws:_^
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow Volume (1-3):
Observed Spring Flow Volume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
39.93 » | 30
-8.39
-29.24
-60.14
-47.15
0.467
0.336
0.526
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
1.45
0.48
0.34
0.40
0.20
0.18
0.67
0.20
0.06
Clear
Hydrology Validation
Hydrology validation for the Rio Grande basin was performed for the period 10/1/1973 through 9/30/1982 at the
Saguache Creek USGS station due to unavailability of data for 1983-1992. The validation period for the other
stations was 1983-1992. Results are presented in Figure 8 through Figure 11 and Table 7.
Q-19
-------
Avg Monthly Rainfall (in)
- Avg Observed Flow (10/1/1973 to 9/30/1982)
Avg Modeled Flow (Same Period)
600
400
o
200
O-74 O-75
O-76 O-77 O-78 O-79
Month
O-80 O-81
Figure 8. Mean monthly flow at USGS 08227000 Saguache Creek near Saguache, CO - validation period.
250 ->
• Avg Flow (10/1/1973 to 9/30/1982)
Line of Equal Value
Best-Fit Line
y= 1.3461X-0.8676
~ R2 = 0.9263 j /
^5200 *• '*--
I
100 200
Average Observed Flow (cfs)
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1973 to
9/30/1982)
250
200
100
50
V '•
• 1
: 1
-\-rrn—i-
i :
ri
i : :
I i
I i
I i
-\-\-r
-i i—i i—i i i i i—i 1-
.. 2
2.5
10 11 12 1 23456789
Month
Figure 9. Seasonal regression and temporal aggregate at USGS 08227000 Saguache Creek near
Saguache, CO - validation period.
Q-20
-------
• Observed (25th, 75th) Average Monthly Rainfall (in) -Median Observed Flow (10/1/1973 to 9/30/1982) Modeled (Median, 25th, 75th)
300
5.0
10 11 12 1
234
Month
Figure 10. Seasonal medians and ranges at USGS 08227000 Saguache Creek near Saguache, CO -
validation period.
•Observed Flow Duration (10/1/1973 to 9/30/1982 )
Modeled Flow Duration (10/1/1973 to 9/30/1982 )
700
•e
i
-------
Table 7. Summary statistics at USGS 08227000 Saguache Creek near Saguache, CO - validation period
(percent error)
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 49
9-Year Analysis F^riod: 10/1/1973 - 9/30/1982
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9)
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
_Naj3hJ5uteliffe^pj5f^^
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
1.55
0.57
0.33
0.48
0.24
0.16
0.66
0.10
0.03
Error Statistics
32.99
25.16
38.45
46.11
Observed Flow Gage
USGS 08227000 SAGUACHE CREEK NEAR SAGUACHE, CO
Hydrologic Unit Code: 13010004
Latitude: 38.16333294
Longitude: -106.2838
Drainage Area (sq-rri): 595
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
63.28 » | 30
11.01
22.61
-34.14
-46.45
0.071
0.159
0.313
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
1.17
0.41
0.27
0.33
0.15
0.15
0.54
0.15
0.05
Clear [
Hydrology Results for Larger Watershed
As described above, parameters determined for the Saguache Creek gage were assumed transferable to other areas
of the watershed. In addition, calibration and validation was pursued at a total of 5 gages throughout the
watershed. Calibration results are fair to poor at most gages (Table 8). The flow statistics at most gages was
affected by the presence of major reservoirs on the main stem and also due to the complex interaction between
surface water and groundwater in this region. Results of the validation exercise are summarized in Table 9.
Q-22
-------
Table 8. Summary statistics (percent error): all stations - calibration period
Station
Error in total volume:
Error in 50% lowest
flows:
Error in 10% highest
flows:
Seasonal volume
error- Summer:
Seasonal volume
error- Fall:
Seasonal volume
error - Winter:
Seasonal volume
error- Spring:
Error in storm
volumes:
Error in summer storm
volumes:
Daily Nash-Sutcliffe
Coefficient of
Efficiency, E:
Baseline adjusted
coefficient (Garrick),
E1:
Monthly Nash-Sutcliffe
Coefficient of
Efficiency, E:
Saguache Creek
nearSaguache
CO 08227000
-4.92
-2.42
-14.20
14.91
39.93
-8.39
-29.24
-60.14
-47.15
0.47
0.34
0.53
Rio Grande
near Lobatos,
NM 08251500
26.41
121.00
-27.38
140.32
137.69
-20.45
-41.03
-76.95
-60.47
-0.29
-0.26
-0.34
Rio Grande
nearTaos, NM
08276500
0.58
36.50
-44.58
59.47
91.40
-20.57
-50.51
-80.60
-73.62
-0.08
-0.09
-0.12
Rio Grande
at Otowi
Bridge, NM
08313000
-32.08
-12.26
-61.45
-12.41
56.44
-18.20
-74.12
-85.08
-81.09
-0.29
-0.04
-0.34
Rio Grande
at Albuquerque
08330000
-6.06
38.00
-55.65
40.71
89.73
9.78
-62.19
-73.33
-60.36
-0.11
-0.07
-0.10
Q-23
-------
Table 9. Summary statistics: all stations - validation period
Station
Error in total volume:
Error in 50% lowest
flows:
Error in 10% highest
flows:
Seasonal volume
error- Summer:
Seasonal volume
error- Fall:
Seasonal volume
error - Winter:
Seasonal volume
error- Spring:
Error in storm
volumes:
Error in summer storm
volumes:
Daily Nash-Sutcliffe
Coefficient of
Efficiency, E:
Baseline adjusted
coefficient (Garrick),
E1:
Monthly Nash-Sutcliffe
Coefficient of
Efficiency, E:
Saguache Creek
near Saguache,
CO 08227000
32.99
25.16
38.45
46.11
63.28
11.01
22.61
-34.14
-46.45
0.07
0.16
0.31
Rio Grande
near Lobatos,
NM 08251500
20.30
160.26
-53.82
354.31
150.71
-11.36
-57.22
-80.42
-69.42
-0.17
-0.25
-0.18
Rio Grande
nearTaos, NM
08276500
1.51
62.77
-57.80
140.26
112.77
-13.42
-58.06
-80.47
-73.25
-0.12
-0.13
-0.13
Rio Grande
at Otowi
Bridge, NM
08313000
-30.04
7.42
-64.33
29.36
69.62
-27.66
-75.28
-85.28
-85.39
-0.29
0.00
-0.33
Rio Grande
at Albuquerque
08330000
-2.77
93.64
-47.39
50.54
132.95
-7.73
-61.01
-76.43
-76.27
-0.23
-0.09
-0.26
Water Quality Calibration and Validation
Initial calibration and validation of water quality was done at the Rio Grande near Taos, NM (USGS 08276500)
using 1985-2003 for calibration and 1973-1984 for validation. As with hydrology, calibration was performed on
the later period as this better reflects the land use included in the model. The start of the validation period is
constrained by data availability.
Calibration adjustments for sediment focused on the following parameters:
• SPCON (Linear parameters for estimating maximum amount of sediment that can be re-entrained during
channel sediment routing)
• CH_COV (Channel cover factor)
• CH_EROD (Channel erodibility factor)
• SPEXP (exponent parameter for calculating sediment re-entrained during channel sediment routing)
Q-24
-------
Simulated and estimated sediment loads at the Rio Grande near Taos (USGS 08276500) station for both the
calibration and validation periods are shown in Figure 12 and statistics for the two periods are provided separately
in Table 10. The key statistic in the table is the relative percent error, which shows the error in the prediction of
monthly load normalized to the estimated load. The table also shows the relative average absolute error, which is
the average of the relative magnitude of errors in individual monthly load predictions. This number is inflated by
outlier months in which the simulated and estimated loads differ by large amounts (which may be as easily due to
uncertainty in the estimated load due to limited data as to problems with the model) and the third statistic, the
relative median absolute error, is likely more relevant and shows better agreement.
TSS
1,000,000
-Regression Loads
-Simulated Loads
opopopopopopopcpcpcpcpcpcpcpcpcpcp
•5 -5
•5 -5
9 9
•5 -5
OOOOOOOOOOOOOOOOOOOO
Figure 12. Fit for monthly load of TSS at USGS 08276500 Rio Grande near Taos, NM.
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 08276500 Rio Grande near Taos, NM
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1985-2001)
57.3%
82%
22.1%
Validation period
(1972-1984)
41%
69%
19.7%
Calibration adjustments for total phosphorus and total nitrogen focused on the following parameters:
• PHOSKD (Phosphorus soil partitioning coefficient)
• RS2 (benthic source rate for dissolved phosphorus in the reach [mg P/m2*day])
• RS3 (benthic source rate for NH4-N in the reach [mg N/m2*day])
• RS4 (rate coefficient for organic N settling in the reach [day-1])
• RS5 (organic phosphorus settling rate in the reach [day-1])
• BC1 (rate constant for biological oxidation of NH4 to NO2 in the reach [day-1])
• BC2 (rate constant for biological oxidation of NO2 to NO3 in the reach [day-1])
• BC4 (rate constant for mineralization of organic P to dissolved P in the reach [day-1])
• MUMAX (maximum specific algal growth rate [day-1])
Q-25
-------
Results for the phosphorus simulation are shown in Figure 13 and Table 11. Results for the nitrogen simulation
are shown in Figure 14 and Table 12. The model fit is generally weak due to problems in simulating the details of
the hydrograph.
Total P
10000 -T
1000
o
In
c
s
100
-Regression Loads
-Simulated Loads
999
•4-J +-J +-J
6666666666666666000
Figure 13. Fit for monthly load of total phosphorus at USGS 08276500 Rio Grande near Taos, NM.
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 08276500 Rio Grande near Taos, NM
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1985-2003)
-46.9%
180%
32.1%
Validation period
(1973-1984)
-653.98%
773%
45%
Q-26
-------
Total N
10,000
•Averaging Loads
-Simulated Loads
ooooooooooooooa>a>a>a>a>a>a>a>a>a>
00000000000000000000
OOOOOOOOOOOOOOOOOOOO
Figure 14. Fit for monthly load of total nitrogen at USGS 08276500 Rio Grande near Taos, NM.
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 08276500 Rio Grande near Taos, NM
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1985-2003)
-28.3%
155%
46.6%
Validation period
(1973-1984)
-909.1%
996%
58.5%
Water Quality Results for Larger Watershed
As with hydrology, the SWAT model parameters used to calibrate at the USGS 08276500 Rio Grande near Taos,
NM for water quality were directly transferred to other portions of the watershed. Application of the SWAT
model without spatial adjustments resulted in relatively large errors in predicting loads and concentrations at some
stations. Summary statistics for the SWAT water quality calibration and validation at other stations in the
watershed are provided in Table 13 and Table 14.
Q-27
-------
Table 13. Summary statistics for water quality at all stations - calibration period 1985-2003
Station
Relative Percent Error
TSS Load
Relative Percent Error
TP Load
Relative Percent Error
TN Load
Saguache Creek
near Saguache,
CO
08227000
20.2%
93.0%
77.8%
Rio Grande
near Lobatos, NM
08251500
55.0%
-198.0%
-193.6%
Rio Grande
near Taos,
NM 08276500
57.3%
-46.9%
-28.3%
Rio Grande at
Otowi Bridge,
NM 08313000
98.1%
42.2%
30.4%
Rio Grande
at
Albuquerque
08330000
95.6%
-85.1%
-41.3%
Table 14. Summary statistics for water quality at all stations - validation period 1973-1984
Station
Relative Percent Error
TSS Load
Relative Percent Error
TP Load
Relative Percent Error
TN Load
Saguache Creek
near Saguache,
CO 08227000
-63.9%
86.80%
44.1%
RioGrande near
Lobatos, NM
08251500
55.6%
-708.19%
-1093.7%
RioGrande near
Taos, NM
08276500
41.0%
-653.98%
-909.1%
RioGrande at
Otowi
Bridge, NM
08313000
97.6%
-151.77%
-411.8%
RioGrande at
Albuquerque
08330000
94.1%
9.41%
-26.7%
References
Tetra Tech. 1999. Improving Point Source Loadings Data for Reporting National Water Quality Indicators. Final
Technical Report prepared for U.S. Environmental Protection Agency, Office of Waste water Management,
Washington, DC, by Tetra Tech, Inc., Fairfax, VA.
USEPA (United States Environmental Protection Agency). 2008. Using the BASINS Meteorological Database
(Version 2006). BASINS Technical Note 10. Office of Water, U.S. Environmental Protection Agency,
Washington, DC. http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf
(Accessed June, 2009).
Q-28
-------
Appendix R
Model Configuration, Calibration and
Validation
Basin: Sacramento River (Sac)
-------
Contents
Watershed Background R-4
Water Body Characteristics R-4
Soil Characteristics R-6
Land Use Representation R-6
Point Sources R-10
Meteorological Data R-10
Watershed Segmentation R-11
Calibration Data and Locations R-13
SWAT Modeling R-14
Assumptions R-14
Hydrology Calibration R-14
Hydrology Validation R-18
Hydrology Results for Larger Watershed R-21
Water Quality Calibration and Validation R-23
References R-26
R-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes R-8
Table 2. Land use distribution for the Sacramento River watershed (2001 NLCD) (mi2) R-9
Table 3. Major point source discharges in the Sacramento River watershed R-10
Table 4. Precipitation stations for the Sacramento River watershed model R-ll
Table 5. Calibration and validation locations in the Sacramento River watershed R-13
Table 6. Summary statistics at USGS 11377100 Sacramento River at Bend Bridge near Red Bluff, CA -
calibration period R-18
Table 7. Summary statistics at USGS 11377100 Sacramento River Bend Bridge near Red Bluff, CA -
validation period R-21
Table 8. Summary statistics (percent error): all stations - calibration period R-22
Table 9. Summary statistics: all stations -validation period R-23
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 11377100 Sacramento River at Bend Bridge near Red Bluff, CA R-24
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 11377100 Sacramento River at Bend Bridge near Red Bluff, CA R-25
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using averaging
estimator at USGS 11377100 Sacramento River at Bend Bridge near Red Bluff, CA R-26
Figures
Figure 1. Location ofthe Sacramento River watershed R-5
Figure 2. Land use in the Sacramento River watershed R-7
Figure 3. Model segmentation and USGS stations utilized for the Sacramento River watershed R-12
Figure 4. Mean monthly flow at USGS 11377100 Sacramento River at Bend Bridge near Red Bluff,
CA - calibration period R-15
Figure 5. Seasonal regression and temporal aggregate at USGS 11377100 Sacramento River at Bend
Bridge near Red Bluff, CA - calibration period R-16
Figure 6. Seasonal medians and ranges at USGS 05317000 USGS 11377100 Sacramento River at
Bend Bridge near Red Bluff, CA - calibration period R-16
Figure 7. Flow exceedance at USGS 11377100 Sacramento River at Bend Bridge near Red Bluff,
CA - calibration period R-17
Figure 8. Mean monthly flow at USGS 11377100 Sacramento River at Bend Bridge near Red Bluff,
CA - validation period R-19
Figure 9. Seasonal regression and temporal aggregate at USGS 11377100 Sacramento River at Bend
Bridge near Red Bluff, CA - validation period R-19
Figure 10. Seasonal medians and ranges at USGS 11377100 Sacramento River at Bend Bridge near Red
Bluff, CA - validation period R-20
Figure 11. Flow exceedance at USGS 11377100 Sacramento River at Bend Bridge near Red Bluff, CA -
validation period R-20
Figure 12. Fit for monthly load of TSS at USGS 11377100 Sacramento River at Bend Bridge near Red
Bluff, CA R-24
Figure 13. Fit for monthly load of total phosphorus at USGS 11377100 Sacramento River at Bend Bridge
near Red Bluff, CA R-25
Figure 14. Fit for monthly load of total nitrogen at USGS 11377100 Sacramento River at Bend Bridge
near Red Bluff, CA R-25
-------
The Sacramento River basin was selected as one of the 15 non-pilot application watersheds for the 20 Watershed
study. Watershed modeling for the non-pilot areas is accomplished using the SWAT model only, and model
calibration and validation results are presented in abbreviated form.
Water Body Characteristics
The Sacramento River is the largest river in California, originating from eastern slopes of the Klamath Mountains
and emptying into Suisun Bay (an arm of the San Francisco Bay) and eventually into the Pacific Ocean.
The Sacramento River in northern California is vital to the state's economy and for providing freshwater flow to
the San Francisco Bay. Lake Shasta impounds the mainstem and is subject to complex operational rules. This
study considers only the portion of the Sacramento River basin from Lake Shasta to just before the confluence
with the Feather River (Figure 1). Information was not available for this study to represent changes in reservoir
operations in response to climate change. Lake Shasta outflow time series were thus considered a fixed upstream
boundary condition. The resulting model area contains over 8,300 mi2 in 11 HUCSs, all within HUC 1802.
The average annual precipitation in the entire watershed ranges from 18 in/yr near Sacramento to about 75 in/yr at
the highest elevations, mostly occurring from November through March. Snow melt is the major source of flow
for the rivers of the watershed.
The Sacramento River is a major source of drinking water for residents of northern and southern California, and is
a principal source of irrigation water for Sacramento and San Joaquin Valley farmers. The land uses in the valley
portion of the Sacramento River basin model area are dominated by agriculture, which makes up 22 percent of the
model area. The Sacramento Valley supports a diverse agricultural economy, much of which depends on the
availability of irrigation water. Dairy products and crops including rice, fruits and nuts, tomatoes, sugar beets,
corn, alfalfa, and wheat are important agricultural commodities. The larger cities in the watershed, located in the
Sacramento Valley, include Chico and Redding, with developed land occupying a little over 4 percent of the
watershed. The remaining areas are primarily forest and range.
Agriculture is the largest consumer of water in the basin. Up to about 6 million acre-feet per year of water also is
exported from the basin, principally to areas in southern California. Part of the runoff from winter rains and spring
snowmelt is stored in reservoirs and released during the normally dry summer months. Most of the water supplies
are derived from these reservoirs. The water is mainly used to provide irrigation water to the Sacramento and San
Joaquin Valley agricultural communities, and to provide drinking water to Central Valley residents and residents
of southern California, and to protect water quality of the delta of the Sacramento and San Joaquin Rivers.
R-4
-------
Hydrography
Water (Nat. Atlas Dataset)
US Census Populated Places
^H Municipalities (pop £ 50.000)
J County Boundaries
H Model Subbasins
Sacramento-Lowe rCow-
Lower Clear
(18020101)
ottonwood
Headwaters
(18020113)
Upper Cow-Battle
(18020118)
Lower
Cottonwood
(18020102)
Upper Elder-
Upper Thorn es
(18020114)
Mill-Big Chico
(18020119)
Sacramento-
Lower Thomes
(18020103)
Upper Stony
(18020115)
Upper Butte
(18020120)s
\
Chico
Lower Butte
(18020105)
Sacramento-
Stone Corral
(18020104)
<~~
Citrus Heights
Folsom
GCRP Model Areas - Sacramento River Basin
Base Map
Figure 1. Location of the Sacramento River watershed.
R-5
-------
Soil Characteristics
Soils in the watershed, as described in STATSGO soil surveys, fall primarily into hydrologic soil groups (HSGs)
D (low infiltration capacity and high runoff potential) and B (moderately high infiltration capacity;
correspondingly, moderately low runoff potential). SWAT uses information drawn directly from the soils data
layer to populate the model.
Land Use Representation
Land use/cover in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage and is
predominantly range grass and shrubland, which together occupy 48 percent of the area. The other dominate uses
are forested land and agriculture, each of which covers about 22 percent of the watershed (Figure 2). NLCD land
cover classes were aggregated according to the scheme shown in Table 1 for representation in the 20 Watershed
model. SWAT uses the built-in hydrologic response unit (HRU) overlay mechanism in the ArcSWAT interface.
SWAT HRUs are formed from an intersection of land use and SSURGO major soils. The distribution of land use
in the watershed is summarized in Table 2.
R-6
-------
Legend
— Hydrography
^^= Interstate
I ^| County Boundaries
2001 NLCD Land Use
| Open water
^ Developed, open space
| Developed, low intensity
| Developed, medium intensity
| Developed, high intensity
^ Barren land
| Deciduous forest
^^| Evergreen forest
| | Mixed forest
I | Scrub/shrub
^ Grassland/herbaceous
I I Pasture/hay
J Cultivated crops
^ Woody wetlands
Emergent herbaceous wetlands
GCRP Model Areas - Sacramento River Basin
Land Use Map
NAD_1983_Albers_meters - Map produced 05-05-2011 - P. Cada
Figure 2. Land use in the Sacramento River watershed.
R-7
-------
Table 1. Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area usually
accounted for as reach area
Deciduous
Evergreen
Mixed
Emergent & woody wetlands
Aquatic bed wetlands (not
emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
R-8
-------
Table 2. Land use distribution for the Sacramento River watershed (2001 NLCD) (mi )
HUC 8 watershed
Sacramento-Stone
Corral
18020104
Upper Stony
18020115
Clear Creek-
Sacramento River
18020151
Cow Creek
18020152
Cottonwood Creek
18020153
Battle Creek-
Sacramento River
18020154
Paynes Creek-
Sacramento River
18020155
Thomes Creek-
Sacramento River
18020156
Big Chico Creek-
Sacramento River
18020157
Butte Creek
18020158
Cache Slough-
Sacramento River
18020163
Total
Open
water
11.9
10.0
0.5
0.8
0.4
8.9
2.8
2.8
2.9
1.6
0.8
43.6
Developed9
Open
space
54.2
22.3
10.1
24.5
4.1
42.7
13.7
26.0
23.6
27.2
1.1
249.6
Low
density
19.1
0.8
1.1
1.4
0.2
16.9
2.6
3.3
10.9
10.4
0.2
66.8
Medium
density
7.7
0.3
0.2
0.5
0.0
9.6
2.2
1.0
8.5
5.2
0.2
35.2
High
density
1.4
0.1
0.0
0.1
0.0
2.2
0.6
0.3
1.5
1.4
0.0
7.7
Barren
land
11.9
7.2
0.6
2.2
1.7
1.4
1.2
7.4
4.3
2.4
0.1
40.3
Forest
3.1
203.0
188.6
251.3
223.3
301.4
35.1
257.2
236.2
162.2
0.0
1,861.5
Shru bland/
Grassland
653.8
518.5
204.0
636.7
134.0
262.8
356.1
625.7
469.7
158.9
0.1
4,020.2
Pasture/
Hay
43.8
5.5
19.2
12.7
1.7
22.7
3.6
30.6
37.4
12.5
0.5
190.3
Cultivated
1,018.1
4.2
4.4
0.4
0.9
9.5
2.2
48.5
136.1
385.2
25.4
1,635.0
Wetland
67.0
3.3
0.9
12.4
2.9
4.6
3.6
5.9
11.1
52.3
1.2
165.2
Total
1,891.8
775.5
429.5
943.1
369.2
682.7
423.6
1,008.7
942.4
819.4
29.6
8,315.5
aThe percent imperviousness applied to each of the developed land uses is as follows: open space (5.95%), low density (30.02%), medium density (55.41%), and high
density (81.20%).
R-9
-------
Point Sources
There are numerous point source discharges in the watershed (Table 3). These are represented at long-term
average flows, without accounting for changes overtime or seasonal variations.
Table 3. Major point source discharges in the Sacramento River watershed
NPDES ID
CA0079081
CA0004821
CA0077704
CA0079731
CA0078034
Name
CHICO, CITY OF
PACTIV CORP
ANDERSON, CITY OF
REDDING, CITY OF
WILLOWS, CITY OF
Design flow
(MGD)
9.00
2.70
2.00
8.80
1.12
Observed flow
(MGD)
(1991-2006 average)
4.17
1.97
1.42
8.01
0.91
Most of these point sources have reasonably good monitoring for total suspended solids (TSS), but not for total
phosphorus and total nitrogen. The point sources were initially represented in the model with the median of
reported values for the constituents (total phosphorus, total nitrogen, and TSS) and an assumed total nitrogen
concentration of 11.2 mg/L and assumed total phosphorus concentration of 7.0 mg/L for secondary treatment
facilities (Tetra Tech 1999). However, in cases where point source contribution was deemed unusually high,
assumed values were substituted with the average concentration of the reported data.
Meteorological Data
The required meteorological time series for the 20 Watershed SWAT simulations are precipitation and air
temperature. The 20 Watershed simulations do not include water temperature simulation and use a degree-day
method for snowmelt. SWAT estimates Penman-Monteith potential evapotranspiration using a statistical weather
generator for inputs other than temperature and precipitation. These meteorological time series are drawn from the
BASINS4 Meteorological Database (USEPA 2008), which provides a consistent, quality-assured set of
nationwide data with gaps filled and records disaggregated. Scenario application requires simulation over 30
years, so the available stations are those with a common 30-year period of record (or one that can be filled from
an approximately co-located station) that covers the year 2001. A total of 28 precipitation stations were identified
for use in the Sacramento River watershed model with a common period of record of 10/1/1971-9/30/2001 (Table
4). Temperature records are sparser; where these are absent temperature is taken from nearby stations with an
elevation correction.
R-10
-------
Table 4. Precipitation stations for the Sacramento River watershed model
COOP ID
40546
41159
41700
41715
41806
41907
41948
42084
42402
42640
43791
45311
45385
45679
46194
46506
46521
46685
46726
47292
47581
48135
48580
48587
49390
49621
49699
49781
Name
CA040546
CA041159
CA041700
CA041715
CA041806
CA041907
CA041948
CA042084
CA042402
CA042640
CA043791
CA045311
CA045385
CA045679
CA046194
CA046506
CA046521
CA046685
CA046726
CA047292
CA047581
CA048135
CA048580
CA048587
CA049390
CA049621
CA049699
CA049781
Latitude
40.4000
39.9372
40.3034
39.6911
38.9240
40.4000
39.1806
39.8261
39.8739
39.3593
40.3636
40.5419
39.1459
40.3458
38.9261
39.7459
39.5179
39.7540
39.8876
40.1519
40.7957
40.7142
39.3754
39.5862
40.4569
40.6117
39.5231
38.6829
Longitude
-122.1500
-121.3140
-121.2420
-121.8210
-122.5670
-122.1430
-122.0290
-123.0840
-121.6170
-122.5170
-122.9650
-121.5760
-121.5850
-121.6090
-121.5440
-122.1990
-121.5530
-121.6240
-122.5530
-122.2530
-121.9350
-122.4160
-122.5460
-122.5340
-121.8650
-122.5280
-122.3050
-121.7940
Temperature
No
No
Yes
Yes
Yes
No
Yes
No
Yes
Yes
No
Yes
Yes
Yes
No
Yes
Yes
Yes
No
Yes
No
Yes
No
Yes
No
Yes
Yes
Yes
Elevation (ft)
128
1890
4531
184
1348
420
49
1512
2710
1204
2749
5751
56
4875
43
253
171
1749
755
354
2100
1076
1171
801
2221
1296
233
69
Watershed Segmentation
The Sacramento River basin was divided into 71 sub-watersheds for the purposes of modeling (Figure 3). The
model doesn't encompass the complete watershed. The Upper Sacramento and Pit Rivers join in Lake Shasta, a
huge reservoir formed by Shasta Dam. The watershed area considered for this 20 Watershed study is the drainage
between downstream of Shasta to the confluence of Feather River and Sacramento River. The outflow from Lake
Shasta was considered as a boundary condition. No specific site was considered as a calibration focus area.
R-ll
-------
USGS gages
Hydrography
Interstate
^H Water (Nat. Atlas Dataset)
US Census Populated Places
| | County Boundaries
H Model Subbasins
GCRP Model Areas - Sacramento River Basin
Model Segmentation
NAD_1983_Albers_meters - Map produced 05-05-2011 - P. Cada
Figure 3. Model segmentation and USGS stations utilized for the Sacramento River watershed.
R-12
-------
Calibration Data and Locations
There are three gages at which long term streamflow data were available (Table 5). All the three gages are on the
main stem of Sacramento River. As mentioned earlier, the watershed area considered for this study is the drainage
between downstream of Shasta to the confluence of Feather River and Sacramento River. Therefore, no specific
site was chosen for initial calibration. The results at the gaging site on Sacramento River above Bend Bridge near
Red Bluff are presented in detail.
Table 5. Calibration and validation locations in the Sacramento River watershed
Station name
Sacramento River above Bend Bridge near
Red Bluff, CA
Sacramento River at Colusa, CA
Sacramento River at Keswick, CA
USGS ID
11377100
11389500
11370500
Drainage area
(mi2)
8,900
12,090
6,468
Hydrology
calibration
X
X
X
Water quality
calibration
X
The model hydrology calibration period was set to Water Years 1992-2001 (within the 30-year period of record
for modeling). Hydrologic validation was then performed on Water Years 1983-1992. Unfortunately, the
Sacramento River watershed had no good water quality data. Water quality data available at the Keswick station
were used as a point source in the simulation because of the lack of these data from Shasta Dam, unlike the
outflow data availability. Due to very limited data at Colusa station, only Bend Bridge station was used for water
quality calibration. Although sediment data were available for a longer period, the nutrient data were again
limited. A period from 1997-2001 was used for calibration and 1973-1996 was used for validation.
R-13
-------
SWAT Modeling
Assumptions
The reservoirs simulated in this study include Black Butte, East Park, Stony Gorge, and Whiskeytown dams.
Pertinent reservoir information including surface area and storage at principal (normal) and emergency spillway
levels for the reservoirs modeled were obtained from the National Inventory of Dams (NID) database. The SWAT
model provides four options to simulate reservoir outflow: measured daily outflow, measured monthly outflow,
average annual release rate for uncontrolled reservoir, and controlled outflow with target release. Keeping in view
the 20 Watershed climate change impact evaluation application, it was assumed that the best representation of the
reservoirs was to simulate them without supplying time series of outflow records. Therefore, the target release
approach was used in the GCRP-SWAT model.
The monthly reservoir target storage was calculated based on daily reservoir storage data obtained from
Department of Water Resources-California Data Exchange Center (http://cdec.water.ca.gov/selectQuery.html) and
input in the reservoir input files in the model. A diversion was simulated for subwatershed 25 to represent
removal of water for irrigation from the Sacramento River near Red Bluff. Elevation bands were input for selected
subwatersheds to account for high altitudes.
Hydrology Calibration
A spatial calibration approach was not adopted for GCRP-SWAT modeling for the Sacramento River watershed.
As no specific calibration focus area was considered, simulated results were compared with the observed flow at
all three gaging stations, simultaneously.
Land Use/Soil/Slope Definition
A 5/10/5 percent threshold was used for land use/soil/slope in the SWAT model while defining the HRUs. Urban
land use classes were exempted from the HRU overlay thresholds.
The parameters were adjusted within the practical range to obtain reasonable fit between the simulated and
measured flows in terms of Nash-Sutcliffe modeling efficiency and the high flow and low flow components as
well as the seasonal flows.
The water balance of the whole Sacramento GRCP model predicted by the SWAT model over the 30-year
simulation period is as follows:
PRECIP = 864.7 MM
SNOW FALL = 152.06 MM
SNOW MELT = 132.35 MM
SUBLIMATION = 20.93 MM
SURFACE RUNOFF Q = 138.52 MM
LATERAL SOIL Q = 200.95 MM
TILE Q = 0.00 MM
GROUNDWATER (SHAL AQ) Q = 24.57 MM
REVAP (SHAL AQ => SOIL/PLANTS) = 74.30 MM
DEEP AQ RECHARGE = 100.46 MM
TOTAL AQ RECHARGE = 200.92 MM
TOTAL WATER YLD = 360.76 MM
PERCOLATION OUT OF SOIL = 198.01 MM
ET = 353.1 MM
PET = 1590.3MM
TRANSMISSION LOSSES = 2.50 MM
R-14
-------
Hydrologic calibration adjustments focused on the following parameters:
• Curve Number
• FFCB (initial soil water storage)
• SURLAG (surface runoff lag coefficient)
• CNCOEFF (plant ET curve number coefficient)
• Baseflow factor
• GW_DELAY (groundwater delay time)
• GWQMN (threshold depth of water in the shallow aquifer for return flow to occur [mmH2O])
• RevapMN (threshold depth of water in the shallow aquifer required for "revap" or percolation to the deep
aquifer to occur
• RCHRG_DP (deep aquifer percolation fraction)
• Sol_AWC (available water capacity of the soil layer, mm water/mm of soil)
• Snow parameters SFTMP, SMTMP, SMFMX and SMFMN, TIMP
• ESCO (soil evaporation compensation factor)
• CH_K2 (channel hydraulic conductivity)
• Elevation bands
• TLAPS (temperature lapse rate)
Calibration results for the Sacramento River above Bend Bridge near Red Bluff, CA (USGS 11377100) are
summarized in Figures 4 through 7 and Table 6. In general, model simulated streamflows, both in magnitude and
timing, compared very well with observed except some overestimation of total storm volumes, especially the
summer storm volumes (Figure 4, Figure 5, Figure 6, Figure 7, and Table 6).
80000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2001
•Avg Modeled Flow (Same Period)
O-92 O-93
O-94
O-95
O-96 O-97
Month
O-98
O-99
O-OO
Figure 4. Mean monthly flow at USGS 11377100 Sacramento River at Bend Bridge near Red Bluff, CA
calibration period.
R-15
-------
• Avg Flow (10/1/1992 to 9/30/2001)
• • • • • Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2001)
•Avg Modeled Flow (Same Period)
30000
y = 0.9183x +2667.9
R2 = 0.9856
I
30000
20000
10000
10000 20000 30000
Average Observed Flow (cfs)
10 11 12 1 23456
Month
Figure 5. Seasonal regression and temporal aggregate at USGS 11377100 Sacramento River at Bend
Bridge near Red Bluff, CA - calibration period.
Average Monthly Rainfall (in)
• Median Observed Flow (10/1/1992 to 9/30/2001)
I Observed (25th, 75th)
Modeled (Median, 25th, 75th)
50000
45000
40000
35000
'ro
a:
Figure 6. Seasonal medians and ranges at USGS 05317000 USGS 11377100 Sacramento River at Bend
Bridge near Red Bluff, CA - calibration period.
R-16
-------
•Observed Flow Duration (10/1/1992 to 9/30/2001 )
Modeled Flow Duration (10/1/1992 to 9/30/2001 )
o
D)
ro
Q
1000000
100000
10000
1000
100
10
1
0.1
0% 10% 20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90%
100%
Figure 7. Flow exceedance at USGS 11377100 Sacramento River at Bend Bridge near Red Bluff, CA
calibration period.
R-17
-------
Table 6. Summary statistics at USGS 11377100 Sacramento River at Bend Bridge near Red Bluff, CA -
calibration period
REACH OUTFLOW FROM OUTLET 27
9-Year Analysis F^riod: 10/1/1992 - 9/30/2001
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
_^fotalj3fj3inTulatejdJ^^
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
["Seasonal volume error - Summer:
f Seasonal volume error - Fall:
f Seasonal volume error - Winter:
[ Seasonal volume error - Sjpring:
[ Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
24.47
7.94
6.21
5.15
3.85
9.45
6.02
6.33
0.47
Error Statistics
10.23
9.28
6.22
9.59
USGS 11377100 SACRAMENTO RAB BEND BRIDGE NR RED BLUFF CA
Hydrologic Unit Code: 18020103
Latitude: 40.28848836
Longitude: -122.1866645
Drainage Area (sq-rri): 8900
Total Observed In-stream Flow:
Total of O^r\«djTigjTe^tJ^%Jlows:_
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
24.24 » | 30
2.11
16.99
43.97
78.74
0.746
0.625
0.944
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
22.20
7.47
5.68
4.70
3.10
9.25
5.14
4.39
0.26
Clear [
Hydrology Validation
Hydrology validation results for the station on the Sacramento River above Bend Bridge near Red Bluff, CA
(USGS 11377100), performed for the period 10/1/1982 through 9/30/1992 are presented in Figures 8 throiugh 11
and Table 7. Based on the model performance statistics, it can be noted that the timing and magnitude of
simulated flows, overall as well as seasonal flows were consistent with the pattern observed during the calibration
period (Figure 8, Figure 9, Figure 10, Figure 11, and Table 7).
R-18
-------
I
o
80000
60000
40000
20000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1982 to 9/30/1992 )
•Avg Modeled Flow (Same Period)
i
ro
01
o
O-82
A-84
O-91
Figure 8. Mean monthly flow at USGS 11377100 Sacramento River at Bend Bridge near Red Bluff, CA
validation period.
Avg Flow (10/1/1982 to 9/30/1992)
• Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1982 to 9/30/1992)
•Avg Modeled Flow (Same Period)
25000
25000
5000 10000 15000 20000 25000
Average Observed Flow (cfs)
10 11 12 1
23456
Month
Figure 9. Seasonal regression and temporal aggregate at USGS 11377100 Sacramento River at Bend
Bridge near Red Bluff, CA - validation period.
R-19
-------
Average Monthly Rainfall (in)
• Median Observed Flow (10/1/1982 to 9/30/1992)
[Observed (25th, 75th)
Modeled (Median, 25th, 75th)
25000
c
'co
>,
o
10 11 12 1
345
Month
Figure 10. Seasonal medians and ranges at USGS 11377100 Sacramento River at Bend Bridge near Red
Bluff, CA - validation period.
•Observed Flow Duration (10/1/1982 to 9/30/1992 )
Modeled Flow Duration (10/1/1982 to 9/30/1992 )
I
D)
'co
Q
1000000
100000
10000
1000
100
10
0.1
0% 10% 20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90% 100%
Figure 11. Flow exceedance at USGS 11377100 Sacramento River at Bend Bridge near Red Bluff, CA
validation period.
R-20
-------
Table 7. Summary statistics at USGS 11377100 Sacramento River Bend Bridge near Red Bluff, CA -
validation period
REACH OUTFLOW FROM OUTLET 27
10-Year Analysis Period: 10/1/1982 - 9/30/1992
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
_JpJalj3fj3inTulatejdjT^^
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
_EjTOnnJ05yTigjTe£tJlows^__
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
18.96
5.87
4.93
4.38
3.80
5.98
4.79
3.91
0.42
Error Statistics
10.06
5.52
3.98
9.90
Observed Flow Gage
USGS 11377100 SACRAMENTO F
Hydrologic Unit Code: 18020103
Latitude: 40.28848836
Longitude: -122.1866645
Drainage Area (sq-rri): 8900
? AB BEND BRIDGE NR RED BLUFF CA
Total Observed In-stream Flow:
_j£t^l_of^b^ej^«djTigjTe^tJ^%JlOT/s:_^
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
5.42 » | 30
6.62
19.20
26.05
81.28
0.571
0.403
0.923
30
_ 2P _
50
Model accuracy increases
as E or E' approaches 1.0
17.23
5.65
4.67
3.99
3.61
5.61
4.02
3.10
0.23
Clear [
:::::::
Hydrology Results for Larger Watershed
As mentioned above, no specific site was chosen for initial calibration and no spatial calibration was carried out
except some adjustments made to the diversion introduced in one subwatershed. Along with the Bend Bridge site,
two other sites on the main stem of the Sacramento River were chosen to compare model simulated streamflows
with observed flow. The calibration results for all three sites are summarized in Table 8. The Keswick site is right
downstream of Shasta Dam and the measured outflow from Shasta was used as a boundary condition. Lack of
accurate representation of the amounts and timings of withdrawal from diversions above the Colusa site, most
likely, resulted in poor Nash-Sutcliffe modeling efficiency at this site. Results of the validation exercise
summarized in Table 9 reflect the same successes and problems experienced during the calibration period.
R-21
-------
Table 8. Summary statistics (percent error): all stations - calibration period
Station
Error in total
volume:
Error in 50% lowest
flows:
Error in 10%
highest flows:
Seasonal volume
error- Summer:
Seasonal volume
error- Fall:
Seasonal volume
error- Winter:
Seasonal volume
error- Spring:
Error in storm
volumes:
Error in summer
storm volumes:
Daily Nash-Sutcliffe
Coefficient of
Efficiency, E:
Monthly Nash-
Sutcliffe Coefficient
of Efficiency, E:
11377100
Sacramento River above Bend
Bridge near Red Bluff, CA
10.23
9.28
6.22
9.59
24.24
2.11
16.99
43.97
78.74
0.746
0.944
11389500
Sacramento River at Colusa, CA
-0.13
-8.68
26.63
15.44
9.62
-13.18
5.10
55.14
-45.91
-0.484
0.632
11370500
Sacramento River at
Keswick, CA
-1.31
-3.21
5.35
-6.04
5.14
2.18
-4.76
36.67
69.25
0.952
0.979
R-22
-------
Table 9. Summary statistics: all stations - validation period
Station
Error in total volume:
Error in 50% lowest
flows:
Error in 10% highest
flows:
Seasonal volume error -
Summer:
Seasonal volume error -
Fall:
Seasonal volume error -
Winter:
Seasonal volume error -
Spring:
Error in storm volumes:
Error in summer storm
volumes:
Daily Nash-Sutcliffe
Coefficient of Efficiency,
E:
Monthly Nash-Sutcliffe
Coefficient of Efficiency,
E:
11377100
Sacramento River
above Bend Bridge
near Red Bluff, CA
10.06
5.52
3.98
9.90
5.42
6.62
19.20
26.05
81.28
0.571
0.923
11389500
Sacramento River at
Colusa, CA
-8.63
-27.84
10.85
15.74
-13.05
-17.70
-9.92
15.20
-41.12
-0.469
0.546
11370500
Sacramento River at
Keswick, CA
-5.38
-11.24
-8.59
-6.58
-10.64
-2.81
-2.38
-24.11
69.72
0.617
0.902
Water Quality Calibration and Validation
The model calibration and validation relied on the data from only one station (11377100; Sacramento River above
Bend Bridge near Red Bluff, CA) in the entire modeled watershed. A period from 1997-2001 was used for
calibration and 1973-1996 was used for validation.
Calibration adjustments for sediment focused on the following parameters:
• SPCON (Linear parameters for estimating maximum amount of sediment that can be re-entrained during
channel sediment routing)
• PRF (Peak rate adjustment factor for sediment routing in the main channel)
• USLE-P (USLE support practice factor)
• USLE-K (USLE erodibility factor)
• SLSUBBSN (average slope length)
• RSDCO (Residue decomposition coefficient)
Simulated and estimated sediment loads at the Bend Bridge station for both the calibration and validation periods
are shown in Figure 12 and statistics for the two periods are provided separately in Table 10. The key statistic in
R-23
-------
Table 10 is the relative percent error, which shows the error in the prediction of monthly load normalized to the
estimated load. Table 10 also shows the relative average absolute error, which is the average of the relative
magnitude of errors in individual monthly load predictions. This number is inflated by outlier months in which the
simulated and estimated loads differ by large amounts (which may be as easily due to uncertainty in the estimated
load due to limited data as to problems with the model) and the third statistic, the relative median absolute error,
is likely more relevant and shows better agreement.
TSS
10,000,000
1,000,000
100,000
o
I
o
-Regression Loads
- Simulated Loads
Figure 12. Fit for monthly load of TSS at USGS 11377100 Sacramento River at Bend Bridge near Red
Bluff, CA.
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 11377100 Sacramento River at Bend Bridge near Red Bluff, CA
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1997-2001)
-2%
59%
28.4%
Validation period
(1973-1996)
-55%
92%
18.2%
Calibration adjustments for total phosphorus and total nitrogen focused on the following parameters:
• CMN (rate factor of humus mineralization of active organic nutrients)
• NPERCO (nitrogen percolation coefficient)
• PHOSKD (phosphorus soil partitioning coefficient)
• SOL_CBN1 (organic carbon in the first soil layer)
• QUAL2E parameters such as organic nitrogen settling rate in the reach, fraction of algal biomass that is
nitrogen, benthic source rate for ammonia in the reach, rate coefficient for organic N settling in the reach,
rate constant for biological oxidation of nitrite to nitrate in the reach, rate constant for hydrolysis of
organic N to ammonia, Michaelis-Menton half-saturation constant for nitrogen, Maximum specific algal
growth rate
Results for the phosphorus simulation are shown in Figure 13 and Table 11. Results for the nitrogen simulation
are shown in Figure 14 and Table 12. The model fit is generally good for phosphorus. However, the model
overestimates nitrogen loads. The calibration efforts were limited because of the lack of long term monitored
nutrient data. Moreover, only one station had monitored data that were used in model calibration.
R-24
-------
Total P
1000
-Regression Loads
- Simulated Loads
ccccccccccccccccszcncccccccccc
Figure 13. Fit for monthly load of total phosphorus at USGS 11377100 Sacramento River at Bend Bridge
near Red Bluff, CA.
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 11377100 Sacramento River at Bend Bridge near Red Bluff, CA
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1997-2001)
-8%
29%
20.3%
Validation period
(1973-1996)
-33
52%
27.8%
Total N
10,000
1,000
o
100
10
-Averaging Loads
-Simulated Loads
Figure 14. Fit for monthly load of total nitrogen at USGS 11377100 Sacramento River at Bend Bridge near
Red Bluff, CA.
R-25
-------
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 11377100 Sacramento River at Bend Bridge near Red Bluff, CA
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1997-2001)
-135%
136%
111.5%
Validation period
(1973-1996)
-156%
159%
124.5%
References
Tetra Tech. 1999. Improving Point Source Loadings Data for Reporting National Water Quality Indicators. Final
Technical Report prepared for U.S. Environmental Protection Agency, Office of Waste water Management,
Washington, DC, by Tetra Tech, Inc., Fairfax, VA.
USEPA. 2008. Using the BASINS Meteorological Database (Version 2006). BASINS Technical Note 10.
Office of Water, U.S. Environmental Protection Agency, Washington, DC.
http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf (Accessed June,
2009).
R-26
-------
Appendix S
Model Configuration, Calibration and
Validation
Basin: Southern California Coastal
(SoCal)
S-l
-------
Contents
Watershed Background S-4
Water Body Characteristics S-4
Soil Characteristics S-6
Land Use Representation S-6
Point Sources S-9
Meteorological Data S-10
Watershed Segmentation S-13
Calibration Data and Locations S-14
SWAT Modeling S-16
Assumptions S-16
Hydrology Calibration S-16
Hydrology Validation S-20
Hydrology Results for Larger Watershed S-23
Water Quality Calibration and Validation S-25
Water Quality Results for Larger Watershed S-28
References S-29
S-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes S-7
Table 2. Land use distribution for the Southern California Coastal basin (2001 NLCD) (mi2) S-8
Table 3. Major point source discharges in the Southern California Coastal basin S-9
Table 4. Precipitation stations for the Southern California Coastal watershed model S-ll
Table 5. Calibration and validation locations in the Southern California Coastal basin S-15
Table 6. Summary statistics at USGS 11066460 Santa Ana River at MWD Crossing, CA - calibration
period S-20
Table 7. Summary statistics at USGS 11066460 Santa Ana River at MWD Crossing, CA - validation
period S-23
Table 8. Summary statistics (percent error): all stations - calibration period S-24
Table 9. Summary statistics (percent error): all stations - validation period S-24
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 11066460 Santa Ana River at MWD Crossing, CA S-26
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 11066460 Santa Ana River at MWD Crossing, CA S-27
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 11066460 Santa Ana River at MWD Crossing, CA S-28
Table 13. Summary statistics for water quality at all stations - calibration period 1991 -2001 S-29
Table 14. Summary statistics for water quality at all stations - validation period 1981-1991 S-29
Figures
Figure 1. Location of the Coastal Southern California basin S-5
Figure 2. Land use in the Southern California Coastal basin S-6
Figure 3. Model segmentation and USGS stations utilized for the Southern California Coastal basin S-14
Figure 4. Mean monthly flow at USGS 11066460 Santa Ana River at MWD Crossing, CA -
calibration period S-17
Figure 5. Seasonal regression and temporal aggregate at USGS 11066460 Santa Ana River at MWD
Crossing, CA - calibration period S-18
Figure 6. Seasonal medians and ranges at USGS 11066460 Santa Ana River at MWD Crossing, CA -
calibration period S-18
Figure 7. Flow exceedance at USGS 11066460 Santa Ana River at MWD Crossing, CA - calibration
period S-19
Figure 8. Mean Monthly Flow at USGS 11066460 Santa Ana River at MWD Crossing, CA -
Validation Period S-21
Figure 9. Seasonal regression and temporal aggregate at USGS 11066460 Santa Ana River at MWD
Crossing, CA - validation period S-21
Figure 10. Seasonal medians and ranges at USGS 11066460 Santa Ana River at MWD Crossing, CA -
validation period S-22
Figure 11. Flow exceedance at USGS 11066460 Santa Ana River at MWD Crossing, CA - validation
period S-22
Figure 12. Fit for monthly load of TSS at USGS 11066460 Santa Ana River at MWD Crossing, CA S-25
Figure 13. Fit for monthly load of total phosphorus at USGS 11066460 Santa Ana River at MWD
Crossing, CA S-27
Figure 14. Fit for monthly load of total nitrogen at USGS 11066460 Santa Ana River at MWD
Crossing, CA S-28
S-3
-------
The Southern California Coastal basin was selected as one of the 15 non-pilot application watersheds for the 20
Watershed study. Watershed modeling for the non-pilot areas is accomplished using the SWAT model only, and
model calibration and validation results are presented in abbreviated form.
Water Body Characteristics
Coastal Southern California basins
The Coastal Southern California basins encompass a land area of over 11,000 mi2 located along the southern
coast of California. The modeled area includes 12 HUCSs within HUC 1807. Major subbasins included in this
study are the Santa Clara River, Los Angeles River, San Gabriel River, Santa Ana River, San Juan River, and
Santa Margarita River (Figure 1). The Coastal Southern California watersheds are characterized by a mild semi-
arid climate with an average rainfall of 15 inches per year. The region is highly urbanized, with substantial
amounts of residential, commercial, and industrial developed land (36 percent) on flatter terrain at lower
elevations; the rugged mountains in the watershed are primarily in forest and rangeland, which together account
for 58 percent of the area.
The Santa Clara River is the largest river system in southern California that remains in a relatively natural state.
The watershed drains 1,634 mi2 from its headwaters in the San Gabriel Mountains to its mouth at the Pacific
Ocean. Ninety percent of the watershed consists of rugged mountains, ranging up to 8,800 feet high; the reminder
consists of valley floor and coastal plain. The climate in the watershed varies from moist, Mediterranean in
Ventura County near the Pacific coast to near desert at the extreme eastern boundary in Los Angeles County.
The Los Angeles and San Gabriel River watersheds are highly urbanized watersheds that encompass 835 mi2 and
640 mi2, respectively. The Los Angeles and San Gabriel Rivers both originate in mountainous areas including a
large portion of the Angeles National Forest. They flow from the mountains into the San Fernando and San
Gabriel Valleys. The rivers then continue on over the coastal plain of Los Angeles and eventually into the Pacific
Ocean. Both rivers have been highly modified with dams (51 in the Los Angeles River watershed and 26 in San
Gabriel River watershed). Virtually the entire Los Angeles River has been channelized and paved. The San
Gabriel River is also channelized and developed for much of its length. These modifications have resulted in a
loss of habitat and human access to the rivers. Diversion of water for use in groundwater recharge, significant
discharges of sewage treatment plant reclaimed waters, and urban runoff have dramatically changed the natural
hydrology of the rivers.
The Santa Ana River is the largest stream system in southern California and encompasses an area of about 2,700
mi2 in parts of Orange, San Bernardino, Riverside, and Los Angeles Counties. The headwaters are in the San
Bernardino Mountains, which reach altitudes over 10,000 feet. The river flows more than 100 miles to the Pacific
Ocean. The population of over 4 million people relies on water resources that originate within the watershed as
well as water imported from northern California and the Colorado River. The Santa Ana watershed is highly
urbanized with about 32 percent of the land use residential, commercial, or industrial. Agricultural land use
accounts for about 10 percent of the watershed. Under natural conditions, the Santa Ana River would be
intermittent with little or no flow in the summer months. Groundwater is the main source of water supply in the
watershed, providing about 66 percent of the consumptive water demand. Imported water from northern
California and the Colorado River account for 27 percent of the consumptive demand. Other sources of supply
include surface water derived from precipitation within the watershed (4 percent) and recycled water (3 percent).
The San Juan River watershed encompasses about 500 mi2. Watershed concerns include channelization, poor
surface water quality from discharge of nonpoint sources, loss of habitat in the floodplain, loss of riparian habitat,
-------
paving of the flood plain, decline of water supply and flows, biodiversity loss, invasive species, surface erosion,
and over use of existing resources. The majority of the watershed is urbanized.
The Santa Margarita River watershed encompasses 750 mi2. The headwaters are on Palomar Mountain and there
are 27 miles of free-flowing river. It is the least disturbed river system south of the Santa Ynez River in Santa
Barbara County. Unlike most of the rivers of the southern coast of California, the riparian habitat is of particularly
high quality, and is essential for the protection of waterfowl and a number of endangered plants and animals.
Groundwater is the main source of water supply in the watershed, providing about 66 percent of the consumptive
water demand. Imported water from northern California and the Colorado River is also an important source of
water supply, accounting for 27 percent of the consumptive demand. Other sources of supply include surface
water derived from precipitation within the basin (4 percent) and recycled water (3 percent).
Enhanced recharge of groundwater is an important component of the hydrologic cycle in the Santa Ana
watershed. The volume of water recharged is 37 percent of the volume pumped, with most of the enhanced
recharge consisting of surface water derived from precipitation within the basin. Discharge from wastewater
treatment facilities is also an important component of the hydrologic cycle, providing base flow in many parts of
the drainage network. These activities are among the many factors affecting water quality in the watershed.
Legend
— Hydrography
•?x~? Water (Nat Atlas Dataset)
US Census Populated Places
H Municipalities (pop i 50,000*
I | County Boundaries
1 I Watershed with HUCBs
Lancaster
Palmdale
Santa Clara
(18070102)
os Angeles
18070105)
Thousand Oaks
Los Angeles
•*•«. San Gabriel
(18070106)
San Bernardino
Santa'Moni
(18070104)
Santa Ana
. (18070203)
Glendale
Pasadena
Redlands
—
Riverside
Seal Bead
(18070201)
San Jacinto
(18070202)
Newport
(18070204)
Santa Margarit
(18070302)
GCRP Model Areas - Southern California Coastal Basins
Base Map
Figure 1. Location of the Coastal Southern California basin.
S-5
-------
Soil Characteristics
Soils in the watershed are described in STATSGO soil surveys. SWAT uses information drawn directly from the
soils data layer to populate the model.
Land Use Representation
Land use/cover in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage (Figure
2). NLCD land cover classes were aggregated according to the scheme shown in Table 1 for representation in the
20 Watershed model. SWAT uses the built-in hydrologic response unit (HRU) overlay mechanism in the
ArcSWAT interface. SWAT HRUs are formed from an intersection of land use and SSURGO major soils. The
distribution of land use in the watershed is summarized in Table 2.
Riverside
San Bernardino
Thousand Oaks
Los Angeles
Glendale
M^MNBMR
Pasadena
Hydrography
Interstate
I I County Boundaries
2001 NLCD Land Use
Open water
I I Developed, open space
I Developed, low intensity
j^B Developed, medium intensity
^^| Developed, high intensity
^ Barren land
| Deciduous forest
j^B Evergreen forest
I I Mixed forest
I I Scrub/shrub
I I Grassland/herbaceous
^ Pasture/hay
I I Cultivated crops
I I Woody wetlands
~n Emergent herbaceous wetlands
GCRP Model Areas - Coastal So. Cal. River Basins
Land Use Map
Figure 2. Land use in the Southern California Coastal basin.
S-6
-------
Table 1. Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area usually
accounted for as reach area
Deciduous
Evergreen
Mixed
Emergent & woody wetlands
Aquatic bed wetlands (not
emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
S-7
-------
Table 2. Land use distribution for the Southern California Coastal basin (2001 NLCD) (mi )
HUC8
watershed
Ventura
18070101
Santa Clara
18070102
Calleguas
18070103
Santa
Monica Bay
18070104
Los Angeles
18070105
San Gabriel
18070106
Seal Beach
18070201
San Jacinto
18070202
Santa Ana
18070203
Newport
Bay
18070204
Aliso-San
Onofre
18070301
Santa
Margarita
18070302
Total
Open
water
3.6
9.6
1.2
1.4
1.5
2.0
0.4
9.7
9.0
0.4
0.5
9.2
48.4
Developed9
Open
space
17.7
82.0
41.9
65.2
87.7
62.0
4.0
88.9
209.8
22.4
40.8
62.2
784.4
Low
density
4.9
33.6
47.8
60.0
135.5
87.5
11.3
40.1
233.1
32.9
40.0
21.6
748.3
Medium
density
2.2
26.7
42.8
137.0
221.7
183.3
46.3
30.9
200.4
48.0
35.5
16.3
991.3
High
density
0.1
1.7
1.9
68.7
70.7
51.1
13.6
0.3
16.1
13.5
3.9
1.0
242.5
Barren
land
0.3
17.0
1.7
1.0
2.0
3.4
0.2
3.5
13.1
0.7
2.2
1.9
47.1
Forest
89.5
325.3
4.0
13.8
43.3
85.3
0.1
62.7
212.9
0.8
10.7
31.5
879.9
Shru bland
121.4
1,090.8
155.6
214.0
273.9
240.1
1.7
446.9
741.0
35.2
356.7
556.7
4,234.1
Pasture/Hay
1.7
8.2
5.1
2.1
0.1
0.5
0.0
24.3
22.6
0.4
0.4
13.7
79.2
Cultivated
5.4
44.2
70.8
0.0
0.1
1.2
1.6
57.5
25.9
2.6
2.1
20.8
232.2
Wetland
1.0
5.8
2.2
1.1
1.8
1.4
0.4
0.3
10.2
0.5
3.7
6.2
34.6
Total
247.8
1,644.7
374.9
564.5
838.4
717.7
79.6
765.3
1,694.1
157.5
496.5
741.1
8,321.9
aThe percent imperviousness applied to each of the developed land uses is as follows: open space (7.75%), low density (35.39%), medium density (61.31%), and high
density (88.93%).
S-8
-------
Point Sources
There are numerous point source discharges in the watershed. Only the major dischargers, generally defined as
those with a design flow greater than 1 MGD are included in the simulation (Table 3). The major dischargers are
represented at long-term average flows, without accounting for changes over time or seasonal variations.
Table 3. Major point source discharges in the Southern California Coastal basin
NPDES ID
CA0053651
CA0054224
CA0054216
CA0054313
CA0053716
CA0053953
CA0055531
CA0001309
CA0056227
CA0053911
CA0054011
CA0054119
CA0053619
CA8000383
CA01 05279
CA8000073
CA8000402
CA0 105236
CA01 05295
CA01 05350
CA8000304
CA01 05376
CA0105619
CA01 05392
CA0053961
CA0053597
CA0055221
Name
SAN BUENAVENTURA, CITY OF
SANTA PAULA, CITY OF
LA CO SANITATION DISTRICTS
LA CO SANITATION DISTRICTS
LA CO SANITATION DISTRICTS
LOS ANGELES, CITY OF
BURBANK, CITY OF
BOEING COMPANY
LOS ANGELES, CITY OF
LA CO SANITATION DISTRICTS
LA CO SANITATION DISTRICTS
LA CO SANITATION DISTRICTS
LA CO SANITATION DISTRICTS
CORONA, CITY OF
INLAND EMPIRE UTILITIES AGENCY
INLAND EMPIRE UTILITIES AGENCY
INLAND EMPIRE UTILITIES AGENCY
COLTON, CITY OF
RIALTO, CITY OF
RIVERSIDE, CITY
COLTON/SAN BERNARDINO RTT&WRA
BEAUMONT, CITY OF
YUCAI PA VALLEY WATER DISTRICT
SAN BERNARDINO, CITY OF
OJAI VALLEY SANITARY DISTRICT
CAMARILLO SANITARY DISTRICT
SIMI VALLEY, CITY OF
Design flow
(MGD)
14.000
2.550
21.600
6.500
15.000
20.000
9.000
178.000
80.000
100.000
37.500
25.000
15.000
9.000
51.000
10.200
15.000
8.400
11.700
40.000
40.000
4.000
4.500
28.000
3.000
6.750
12.500
Observed flow
(MGD)
(1991-2006 average)
7.499
41.259
12.377
5.184
27.341
12.964
11.401
4.041
52.668
122.795
31.785
15.331
7.553
5.578
59.101
11.589
12.199
5.309
6.212
45.261
57.208
2.035
2.598
28.501
1.935
3.300
8.994
S-9
-------
NPDES ID
CA0056294
CA0055387
CA8000326
Name
THOUSAND OAKS, CITY OF
EXXONMOBIL OIL CORP
IRVINE RANCH WATER DISTRICT
Design flow
(MGD)
10.800
1.430
18.000
Observed flow
(MGD)
(1991-2006 average)
8.652
2.824
12.394
Most of these point sources have reasonably complete monitoring for total suspended solids (TSS). Long term
average values of total phosphorus and total nitrogen were assumed based upon the type of point source
discharger. The point sources were initially represented in the model with the median of reported values for total
phosphorus, total suspended solids and total nitrogen.
Meteorological Data
The required meteorological time series for the 20 Watershed SWAT simulations are precipitation and air
temperature. The 20 Watershed simulations do not include water temperature and uses a degree-day method for
snowmelt. SWAT estimates Penmann-Monteith potential evapotranspiration using a statistical weather generator
for inputs other than temperature and precipitation. These meteorological time series are drawn from the
BASINS4 Meteorological Database (USEPA 2008), which provides a consistent, quality-assured set of
nationwide data with gaps filled and records disaggregated. Scenario application requires simulation over 30
years, so the available stations are those with a common 30-year period of record (or one that can be filled from
an approximately co-located station) that covers the year 2001. A total of 85 precipitation stations were identified
for use in the Southern California Coastal watershed model with a common period of record of 10/1/1970-
9/30/2001 (Table 4). Temperature records are sparser; where these are absent temperature is taken from nearby
stations with an elevation correction.
S-10
-------
Table 4. Precipitation stations for the Southern California Coastal watershed model
COOP ID
045218
046162
046569
049087
047723
040235
047953
046940
047779
042198
042494
042805
045632
041369
047600
048992
046379
040606
042164
048844
041484
047306
047470
044647
046719
046175
042214
047785
041194
045114
047050
040014
042941
Name
LYTLE CREEK R S
NEWHALL S FC32CE
OXNARD
TUSTIN IRVINE RANCH
SAN BERNARDINO F S 226
ANZA
SANTA MONICA PIER
PIRU2ESE
SAN GABRIEL DAM FC425B
CRYSTAL LAKE FC238C
DOWNEY FIRE STN FC107C
ELSINORE
MILL CREEK INTAKE
CAMP ANGELUS
RUNNING SPRINGS 1 E
TRABUCO CANYON
OCEANSIDE PUMPING PLT
BEAUMONT
CRESTLINE
TEMECULA
CANOGA PARK PIERCE COLL
REDLANDS
RIVERSIDE FIRE STA 3
LACUNA BEACH
PASADENA
NEWPORT BEACH HARBOR
CULVER CITY
SAN GABRIEL FIRE DEPT
BURBANK VALLEY PUMP PLA
LOS ANGELES INTLAP
POMONA FAIRPLEX
ACTON ESCONDIDO FC261
FAIRMONT
Latitude
34.2384
34.3869
34.1981
33.7026
34.1344
33.5558
34.0081
34.4062
34.2054
34.3178
33.9297
33.6692
34.0915
34.1493
34.2067
33.6583
33.2170
33.9293
34.2330
33.5000
34.1819
34.0528
33.9511
33.5472
34.1483
33.6025
34.0051
34.1061
34.1868
33.9381
34.0811
34.4948
34.7043
Longitude
-117.4700
-118.5340
-119.1750
-117.7530
-117.2530
-116.6730
-118.4980
-118.7550
-117.8600
-117.8400
-118.1450
-117.3310
-116.9360
-116.9800
-117.0860
-117.5890
-117.3490
-116.9740
-117.2990
-117.1500
-118.5740
-117.1890
-117.3880
-117.7800
-118.1440
-117.8800
-118.4120
-118.0990
-118.3480
-118.3880
-117.7650
-118.2710
-118.4270
Temperature
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Elevation (m)
832
379
15
72
347
1193
4
223
451
1637
34
392
1507
1759
1818
296
3
796
455
95
241
402
256
11
263
3
17
137
200
30
317
905
933
S-ll
-------
COOP ID
043896
044211
044422
044671
044863
045115
046006
046399
046657
047473
047735
047776
047888
047957
048014
049152
049285
044650
048230
040144
040798
042090
043452
044628
046602
046663
047749
048967
048973
049660
041272
047926
047762
041682
Name
HEMET
IDYLLWILD FIRE DEPT
JUNCAL DAM
LAKE ARROWHEAD
LEBEC
LOS ANGELES DOWNTOWN
MT WILSON NO 2
OJAI
PALOMAR MOUNTAIN OBSERV
RIVERSIDE CITRUS EXP ST
SANDBERG
SAN GABRIEL CANYON P H
SANTA ANA FIRE STATION
SANTA PAULA
SAUGUS POWER PLANT 1
UCLA
VENTURA
LACUNA BEACH 2
SIGNAL HILL FC 415
ALTADENA
BIG TUJUNGA DAM FC46DE
COVINA CITY YRD FC387B
GLENDORA FC 287B
LACRESCENTAFC251C
PACOIMADAMFC33A-E
PALOS VERDES EST FC43D
SAN DIMAS FIRE FC95
TOPANGA PATROL STN FC6
TORRANCE
WHITTIER CITY YD FC106C
CAJON WEST SUMMIT
SANTA FE DAM
SAN FERNANDO PH 3
CHATSWORTH RESERVOIR
Latitude
33.7459
33.7572
34.4909
34.2467
34.8328
34.0511
34.2309
34.4479
33.3782
33.9670
34.7437
34.1553
33.7442
34.3120
34.5894
34.0697
34.2825
33.5567
33.7968
34.1819
34.2942
34.0920
34.1464
34.2223
34.3326
33.7998
34.1072
34.0843
33.8017
33.9762
34.3901
34.1119
34.3133
34.2264
Longitude
-116.9400
-116.7060
-119.5060
-117.1880
-118.8640
-118.2350
-118.0710
-119.2270
-116.8400
-117.3610
-118.7240
-117.9070
-117.8660
-119.1330
-118.4540
-118.4420
-119.2910
-117.8000
-118.1680
-118.1380
-118.1870
-117.8800
-117.8470
-118.2420
-118.3990
-118.3910
-117.8050
-118.5980
-118.3410
-118.0220
-117.5920
-117.9700
-118.4920
-118.6160
Temperature
X
X
X
X
X
X
X
X
X
X
X
X
Elevation (m)
504
1640
679
1586
1093
70
1740
227
1692
301
1375
227
41
72
642
131
32
64
30
344
706
178
280
477
457
66
291
227
34
128
1457
130
381
277
S-12
-------
COOP ID
048092
040742
041057
041754
043285
046473
047123
047813
048243
046377
048436
049378
045085
040741
044181
046910
041540
045417
Name
SEPULVEDA DAM
BIG BEAR LAKE DAM
BREA DAM
CHUCHUPATE RANGER STN
FULLERTON DAM
ORANGE COUNTY RESERVOIR
PRADO DAM
SAN J AC INTO R S
SILVERADO RANGER STN
OCEANSIDE MARINA
SPADRA LANTERMAN HOSP
VISTA2NNE
LONG BEACH AP
BIG BEAR LAKE
HURKEY CREEK PARK
PINE MOUNTAIN INN
CARPINTERIA RSVR
MATILIJA DAM
Latitude
34.1662
34.2414
33.8906
34.8079
33.8964
33.9379
33.8904
33.7870
33.7425
33.2097
34.0419
33.2294
33.8118
34.2442
33.6830
34.6001
34.4000
34.4830
Longitude
-118.4730
-116.9740
-117.9260
-119.0110
-117.8880
-117.8850
-117.6450
-116.9580
-117.6590
-117.3940
-117.8090
-117.2260
-118.1460
-116.9030
-116.6830
-119.3490
-119.4830
-119.2990
Temperature
X
X
X
X
Elevation (m)
207
2077
84
1603
104
201
171
475
334
3
206
155
9
2060
408
392
36
98
Watershed Segmentation
The Southern California Coastal basin was divided into 65 subwatersheds for the purposes of modeling (Figure
3). The model encompasses the complete watershed and does not require specification of any upstream boundary
conditions for application.
S-13
-------
/Santa Clarita
USGS 11109000
USGS 11074000
USGS.11066460
Thousand Oaks
Santa Monica
Glendale
Redlands
^^
Riverside
USGS Gages
Hydrography
Interstate
^H Water (Nat. Atlas Dataset)
US Census Populated Places
I I County Boundaries
Model Subbasins
1044000
Oceanside
GCRP Model Areas - Coastal So. Cal. River Basins
Model Segmentation
Figure 3. Model segmentation and USGS stations utilized for the Southern California Coastal basin.
Calibration Data and Locations
The specific site chosen for initial calibration was the Santa Ana River at MWD Crossing, a flow and water
quality monitoring location that approximately coincides with the mouth of an 8-digit HUC. The Santa Ana River
watershed was selected because there is a good set of flow and water quality data available and the watershed
lacks major point sources and impoundments. Additional calibration and validation was pursued at multiple
locations (Table 5). Parameters derived on the Santa Ana River were not fully transferable to other portions of the
Southern California Coastal basin, and additional calibration was conducted at multiple gage locations.
S-14
-------
Table 5. Calibration and validation locations in the Southern California Coastal basin
Station name
Santa Ana River at MWD Crossing, CA
Santa Margarita River near Temecula, CA
Santa Ana river below Prado Dam, CA
San Gabriel River above Whittier Narrows Dam,
CA
Santa Clara River near Piru, CA
USGS ID
11066460
11044000
11074000
11087020
11109000
Drainage
area (mi2)
852
588
2,258
2,692
2,183
Hydrology
calibration
X
X
X
X
X
Water quality
calibration
X
X
The model hydrology calibration period was set to Water Years 1991-2001 (within the 32-year period of record
for modeling). Hydrologic validation was then performed on Water Years 1981-1991. Water quality calibration
used calendar years 1991-2001, while validation used 1981-1991.
S-15
-------
SWAT Modeling
Assumptions
There are numerous diversions and impoundments in the Southern California Coastal basin. Since the objective of
the 20 Watershed modeling effort is to measure relative change, only major impoundments and/or diversions have
been represented in the model. Vail Lake and Prado Dam were the two impoundments represented in the Southern
California Coastal model. Vail Lake is located on the Temecula Creek, a tributary of Santa Margarita River. Prado
Dam is located on the Santa Ana River. Pertinent reservoir information including surface area and storage at
principal (normal) and emergency spillway levels for the reservoirs were obtained from the US Army Corps of
Engineers. The SWAT model provides four options to simulate reservoir outflow: measured daily outflow,
measured monthly outflow, average annual release rate for uncontrolled reservoir, and controlled outflow with
target release. Keeping in view the 20 Watershed climate change impact evaluation application to future climate
scenarios, it was assumed that the best representation of the reservoirs were to simulate them without supplying
time series of outflow records. Therefore, target release approach was used in the GCRP-SWAT model.
Hydrology Calibration
A spatial calibration approach was adopted for GCRP-SWAT modeling for the Southern California Coastal basin.
A systematic adjustment of parameters has been adopted and some adjustments are applied throughout the basin.
Most of the calibration efforts were geared towards getting a closer match between simulated and observed flows
at the outlet of calibration focus area.
Land Use/Soil/Slope Definition
A 5/10/5 percent threshold was used for land use/soil/slope in the SWAT model while defining the HRUs. Urban
land use classes were exempted from the HRU overlay thresholds.
The calibration focus area (Santa Ana River at MWD Crossing) includes six subwatersheds and is generally
representative of the general land use characteristics of the overall watershed with the exception of a higher
percentage of cultivated lands. The parameters were adjusted within the practical range to obtain reasonable fit
between the simulated and measured flows in terms of Nash-Sutcliffe modeling efficiency and the high flow and
low flow components as well as the seasonal flows.
The water balance of the whole Southern California Coastal basin predicted by the SWAT model over the 32-year
simulation period is as follows:
PRECIP = 494.8 MM
SNOW FALL = 24.25 MM
SNOW MELT = 23.76 MM
SUBLIMATION = 0.50 MM
SURFACE RUNOFF Q = 143.40MM
LATERAL SOIL Q = 85.39 MM
TILE Q = 0.00 MM
GROUNDWATER (SHAL AQ) Q = 94.83 MM
REVAP (SHAL AQ => SOIL/PLANTS) = 1.99 MM
DEEP AQ RECHARGE = 9.06 MM
TOTAL AQ RECHARGE = 105.88 MM
TOTAL WATER YLD = 277.58 MM
S-16
-------
PERCOLATION OUT OF SOIL = 61.28 MM
ET = 225.2 MM
PET = 1712.7MM
TRANSMISSION LOSSES = 46.04 MM
Hydrologic calibration adjustments focused on the following parameters:
• CN2 (initial SCS runoff curve number for moisture condition II)
• ESCO (soil evaporation compensation factor)
• SURLAG (surface runoff lag coefficient)
• SOL_AWC (available water capacity of the soil layer, mm water/mm of soil)
• ALPHA_BF (baseflow alpha factor, days)
• GW_DELAY (groundwater delay time, days)
• GWQMIN (threshold depth of water in the shallow aquifer required for return flow to occur, mm)
• GW_REVAP (groundwater "revap" coefficient)
• CH_N1 (Manning's "n" value for tributary channels)
• CH_N2 (Manning's "n" value for main channels)
• CH_K1 (Effective hydraulic conductivity in tributary channel alluvium)
• CH_K2 (Effective hydraulic conductivity in main channel alluvium)
• SFTMP (Snowfall temperature)
• SMTMP (Snowmelt base temperature)
• SMFMX (Maximum melt rate for snow during the year)
• SMFMN (Minimum melt rate for snow during the year)
The calibration achieves a moderately high coefficient of model fit efficiency. Calibration results for the Santa
Ana River at MWD Crossing are summarized in Figure 4, Figure 5, Figure 6, Figure 7 and Table 6.
Avg Monthly Rainfall (in)
—•— Avg Observed Flow (10/1/1991 to 9/30/2001 )
Avg Modeled Flow (Same Period)
3000
ro
o:
o
A-93
0-94
A-96
0-97
A-99
0-00
Month
Figure 4. Mean monthly flow at USGS 11066460 Santa Ana River at MWD Crossing, CA - calibration
period.
S-17
-------
• Avg Flow (10/1/1991 to 9/30/2001)
Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1991 to 9/30/2001)
Avg Modeled Flow (Same Period) „
800
.600
y =
0771X
-7.J
627
R2 = 0.9189
'T/T
800 n
10\ 11 \ 12\ 1 \ 2 | 3 j 4 \ 5 \ 6 | 7 j j 9
600 -----
f
o
: : : i : i
mit
: : i : i
: : : i : i
: : : i : i
400
200 - —
0
1
'c'
2 '<^
sl
CD
4 *
200 400 600
Average Observed Flow (cfs)
800
10 11 12 1 234567
Month
8 9
Figure 5. Seasonal regression and temporal aggregate at USGS 11066460 Santa Ana River at MWD
Crossing, CA - calibration period.
800
700
To Lower Bound Average Monthly Rainfall (in) -Median Observed Flow (10/1/1991 to 9/30/2001) Modeled (Median, 25th, 75th)
0
1
10 11 12 1
789
Figure 6. Seasonal medians and ranges at USGS 11066460 Santa Ana River at MWD Crossing, CA -
calibration period.
S-18
-------
•Observed Flow Duration (10/1/1991 to 9/30/2001 )
Modeled Flow Duration (10/1/1991 to 9/30/2001 )
27000
o
-------
Table 6. Summary statistics at USGS 11066460 Santa Ana River at MWD Crossing, CA - calibration
period
REACH OUTFLOW FROM OUTLET 34
10-Year Analysis Period: 10/1/1991 -9/30/2001
Flow/volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume ^months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
3.40
1.74
0.60
0.34
0.52
1.95
0.58
1.22
0.02
Error Statistics
3.71
7.28
-4.65
7.40
Observed Flow Gage
USGS 11066460 SANTA ANA R A MWD CROSSING CA
Hydrologic Unit Code: 18070203
Latitude: 33.96862566
Longitude: -117.4483806
Drainage Area (sq-mi): 852
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring FlowVolume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
19.91 » | 30
5.46
-13.35
-11.70
-46.22
0.625
0.438
0.747
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
3.28
1.82
0.56
0.32
0.44
1.85
0.67
1.38
0.03
Clear [
Hydrology Validation
Hydrology validation for Santa Ana River at MWD Crossing was performed for the period 10/1/1981 through
9/30/1991. The validation achieves an acceptable coefficient of model fit efficiency (Figure 8, Figure 9, Figure
10, Figure 11 and Table 7).
S-20
-------
2000
1500
t
o
1000
500
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1981 to 9/30/1991 )
Avg Modeled Flow (Same Period)
jj\ ^,, f
'•N^^V "-^i^ 's^/^VA
O-81
A-83
O-84
A-86
O-87
A-89
O-90
Month
20
18
16 ^
14 f
12 f
10 £
8 -|>
6 ^
4
2
0
o
Figure 8. Mean Monthly Flow at USGS 11066460 Santa Ana River at MWD Crossing, CA - Validation
Period
• Avg Flow (10/1/1981 to 9/30/1991)
Line of Equal Value
Best-Fit Line
T3
O
500
400
300
200
100
0
i
y= 1.1571X-23.162
R2 = 0.9274
s
0 200 400
Average Observed Flow (cfs)
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1981 to 9/30/1991)
Avg Modeled Flow (Same Period)
500 n
400 -•—
11 j 12\ 1 j 2 [ 3 j 4 ]
10 11 12 1 23456789
Month
Figure 9. Seasonal regression and temporal aggregate at USGS 11066460 Santa Ana River at MWD
Crossing, CA - validation period.
S-21
-------
To Lower Bound Average Monthly Rainfall (in) -Median Observed Flow (10/1/1981 to 9/30/1991) Modeled (Median, 25th
500
450
10 11
Figure 10. Seasonal medians and ranges at USGS 11066460 Santa Ana River at MWD Crossing, CA -
validation period.
Flow Duration (10/1/1981 to 9/30/1991 )
Modeled Flow Duration (10/1/1981 to 9/30/1991 )
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 11. Flow exceedance at USGS 11066460 Santa Ana River at MWD Crossing, CA - validation
period.
S-22
-------
Table 7. Summary statistics at USGS 11066460 Santa Ana River at MWD Crossing, CA - validation
period
REACH OUTFLOW FROM OUTLET 34
10-Year Analysis Period: 10/1/1981 -9/30/1991
Flow/volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
2.65
1.20
0.58
0.33
0.64
1.18
0.50
0.89
0.03
Error Statistics
1.61
1.21
-2.74
-2.97
Observed Flow Gage
USGS 11066460 SANTA ANA R A IV
Hydrologic Unit Code: 18070203
Latitude: 33.96862566
Longitude: -117.4483806
Drainage Area (sq-mi): 852
WD CROSSING CA
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring FlowVolume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
14.41 » [ 30
10.63
-21 .87
4.72
-49.23
0.587
0.368
0.678
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
2.61
1.24
0.57
0.34
0.56
1.06
0.65
0.85
0.06
Clear [
Hydrology Results for Larger Watershed
As described above, parameters determined for the gage at the Santa Ana River were initially transferred to other
gages in the watershed. However, changes to subwatershed level parameters were required to fit the model to the
observed flows. In all, calibration and validation was pursued at a total of five gages throughout the watershed.
Results of the calibration and validation exercise are summarized in Table 8 and Table 9, respectively. Calibration
and validation results were acceptable at most gages.
S-23
-------
Table 8. Summary statistics (percent error): all stations - calibration period
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error- Summer:
Seasonal volume error- Fall:
Seasonal volume error - Winter:
Seasonal volume error- Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient of Efficiency,
E:
Monthly Nash-Sutcliffe Efficiency:
USGS
11044000
-14.16
-34.24
-24.17
63.03
14.79
-20.76
20.89
-30.11
0.26
0.746
0.907
USGS
11074000
-0.63
-48.91
17.53
-39.86
-33.47
13.39
14.16
-33.40
-41.45
0.635
0.854
USGS
11087020
-2.03
-9.33
-5.28
-10.77
-27.72
4.18
-3.02
-18.43
-84.03
0.539
0.771
USGS 11 109000
(1996-2001)
-27.78
-58.29
-6.60
-52.09
-13.02
-12.67
-61.49
21.68
-74.54
0.290
0.875
Table 9. Summary statistics (percent error): all stations - validation period
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient of Efficiency, E:
Monthly Nash-Sutcliffe Efficiency:
USGS
11044000
29.64
-7.02
7.39
156.76
153.72
-4.59
122.83
2.38
339.73
0.456
0.721
USGS
11074000
9.60
-37.19
30.78
-22.67
-9.62
32.70
9.33
-17.19
-36.25
0.372
0.494
USGS
11087020
-7.95
37.06
-1.74
-25.74
-9.15
-1.41
-14.90
-29.85
-65.40
0.589
0.797
S-24
-------
Water Quality Calibration and Validation
Initial calibration and validation of water quality was done on the Santa River at MWD Crossing (USGS
11066460), using 1998-2000 for calibration due to limited water quality data availability. No water quality data
were available for the validation period. As with hydrology, calibration was performed on the later period as this
better reflects the land use included in the model.
Calibration adjustments for sediment focused on the following parameters:
• SPCON (linear parameter for estimating maximum amount of sediment that can be re-entrained during
channel sediment routing)
• SPEXP (exponential parameter for estimating maximum amount of sediment that can be re-entrained
during channel sediment routing)
• CH_COV (channel cover factor)
• CH_EROD (channel erodibility factor)
• USLE_P (USLE support practice factor)
Simulated and estimated sediment loads at the Santa Ana River station for both the calibration and validation
periods are shown in Figure 12 and statistics for the two periods are provided separately in Table 10. The key
statistic in Table 10 is the relative percent error, which shows the error in the prediction of monthly load
normalized to the estimated load. Table 10 also shows the relative average absolute error, which is the average of
the relative magnitude of errors in individual monthly load predictions. This number is inflated by outlier months
in which the simulated and estimated loads differ by large amounts (which may be as easily due to uncertainty in
the estimated load due to limited data as to problems with the model) and the third statistic, the relative median
absolute error, is likely more relevant and shows better agreement.
TSS
100,000
10,000 -
o 1,000
E
I
100 -
10 -
-Regression Loads
-Simulated Loads
Figure 12. Fit for monthly load of TSS at USGS 11066460 Santa Ana River at MWD Crossing, CA.
S-25
-------
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 11066460 Santa Ana River at MWD Crossing, CA
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1998-2000)
19.0%
62.5%
7.6%
Calibration adjustments for total phosphorus and total nitrogen focused on the following parameters:
• RHOQ (algal respiration rate at 20° C)
• PHOSKD (phosphorus soil partitioning coefficient)
• PSP (phosphorus availability index)
• RSI (Local algal settlement rate in the reach at 20° C)
• AL1 (Fraction of algal biomass that is nitrogen)
• AL2 (Fraction of algal biomass that is phosphorus)
• MUMAX (Rate of oxygen uptake per unit NO2-N oxidation at 20° C)
• RHOQ (Algal respiration rate at 20° C)
• RS2 (benthic source rate for dissolved P in the reach at 20° C)
• RS3 (Benthic source rate for NFLpN in the reach at 20° C)
• RS5 (organic P settling rate in the reach at 20° C)
• BC4 (rate constant for mineralization of organic P to dissolved P in the reach at 20° C)
• RS4 (rate coefficient for organic N settling in the reach at 20° C)
• CH_ONCO (Channel organic nitrogen concentration)
• CH_OPCO (Channel organic phosphorus concentration)
• SDNCO (Denitrification threshold water content)
• CDN (Denitrification exponential rate constant)
Results for the phosphorus simulation are shown in Figure 13 and Table 11. Results for the nitrogen simulation
are shown in Figure 14 and Table 12. The model fit is generally acceptable.
S-26
-------
Total P
100
o
E
«
I
10 -
-Regression Loads
-Simulated Loads
Figure 13. Fit for monthly load of total phosphorus at USGS 11066460 Santa Ana River at MWD Crossing,
CA.
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 11066460 Santa Ana River at MWD Crossing, CA
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1998-2000)
-14.7%
45.8%
23.7%
S-27
-------
Total N
180 -
160 -
140 -
120 -
100 -
80 -
60 -
40 -
20 -
0 -
-Averaging Loads
-Simulated Loads
Figure 14. Fit for monthly load of total nitrogen at USGS 11066460 Santa Ana River at MWD Crossing,
CA.
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 11066460 Santa Ana River at MWD Crossing, CA
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1998-2000)
-5.5%
24.6%
19.2%
Water Quality Results for Larger Watershed
As with hydrology, a spatial calibration approach was adopted. Santa Ana River watershed SWAT model
parameters for water quality were transferred to other portions of the watershed with necessary changes to
subbasin level parameters. Summary statistics for the SWAT water quality calibration and validation at other
stations in the watershed are provided in Table 13 and Table 14.
S-28
-------
Table 13. Summary statistics for water quality at all stations - calibration period 1991 -2001
Station
Relative Percent Error TSS Load
Relative Percent Error TP Load
Relative Percent Error TN Load
USGS 11 044000
98.0%
-17.7%
41.7%
Table 14. Summary statistics for water quality at all stations - validation period 1981-1991
Station
Relative Percent Error TSS Load
Relative Percent Error TP Load
Relative Percent Error TN Load
USGS 11 044000
(1987-1991)
97.5%
1 .6%
75.0%
References
USEPA (United States Environmental Protection Agency). 2008. Using the BASINS Meteorological Database
(Version 2006). BASINS Technical Note 10. Office of Water, U.S. Environmental Protection Agency,
Washington, DC. http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf
(Accessed June, 2009).
S-29
-------
Appendix T
Model Configuration, Calibration and
Validation
Basin: South Platte (SoPlat)
T-l
-------
Contents
Watershed Background T-4
Water Body Characteristics T-4
Soil Characteristics T-6
Land Use Representation T-6
Point Sources T-10
Meteorological Data T-11
Watershed Segmentation T-12
Calibration Data and Locations T-14
SWAT Modeling T-15
Assumptions T-15
Hydrology Calibration T-15
Hydrology Validation T-19
Hydrology Results for Larger Watershed T-22
Water Quality Calibration and Validation T-24
Water Quality Results for Larger Watershed T-27
References T-29
T-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes T-8
Table 2. Land use distribution for the South Platte watershed (2001 NLCD) (mi2) T-9
Table 3. Major point source discharges in the South Platte watershed T-10
Table 4. Precipitation stations for the South Platte watershed model T-ll
Table 5. Calibration and validation locations in the South Platte watershed T-14
Table 7. Summary statistics at USGS 06714000, South Platte River at Denver, CO - validation
period T-22
Table 8. Summary statistics (percent error): all stations - calibration period T-23
Table 9. Summary statistics: all stations - validation period T-24
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 06714000, South Platte River at Denver, CO T-25
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 06714000, South Platte River at Denver, CO T-26
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 06714000, South Platte River at Denver, CO T-27
Table 13. Summary statistics for water quality at all stations - calibration period 1993-2002 T-28
Table 14. Summary statistics for water quality at all stations - validation period 1986-1992 T-28
Figures
Figure 1. Location of the South Platte watershed T-5
Figure 2. Land use in the South Platte watershed T-7
Figure 3. Model segmentation and USGS stations utilized for the South Platte River watershed T-13
Figure 4. Mean monthly flow at USGS 06714000, South Platte River at Denver, CO - calibration
period. T- T-17
Figure 5. Seasonal regression and temporal aggregate at USGS 06714000, South Platte River at
Denver, CO - calibration period T-17
Figure 6. Seasonal medians and ranges at USGS 06714000, South Platte River at Denver, CO -
calibration period T-18
Figure 7. Flow exceedance at USGS 06714000, South Platte River at Denver, CO - calibration
period T-18
Table 6. Summary statistics at USGS 06714000, South Platte River at Denver, CO - calibration
period T-19
Figure 8. Mean monthly flow at USGS 06714000, South Platte River at Denver, CO - validation
period T-20
Figure 9. Seasonal regression and temporal aggregate at USGS 06714000, South Platte River at
Denver, CO - validation period T-20
Figure 10. Seasonal medians and ranges at USGS 06714000, South Platte River at Denver, CO -
validation period T-21
Figure 11. Flow exceedance at USGS 06714000, South Platte River at Denver, CO - validation
period T-21
Figure 12. Fit for monthly load of TSS at USGS 06714000, South Platte River at Denver, CO T-25
Figure 13. Fit for monthly load of total phosphorus at USGS 06714000, South Platte River at
Denver, CO T-26
Figure 14. Fit for monthly load of total nitrogen at USGS 06714000, South Platte River at Denver,
CO T-27
T-3
-------
The South Platte River basin study area was selected as one of the 15 non-pilot application watersheds for the 20
Watershed study. Watershed modeling for the non-pilot areas is accomplished using the SWAT model only, and
model calibration and validation results are presented in abbreviated form.
Water Body Characteristics
The South Platte River originates in the mountains of central Colorado at the Continental Divide and flows about
450 miles northeast across the Great Plains to its confluence with the North Platte River at North Platte, Nebraska.
The model study area is almost 15,000 mi2 in size and extends from the headwaters to the plains of central
Colorado, consisting of 11 HUCSs within HUC 1019 (Figure 1). Elevation in the model study area ranges from
14,286 ft at Mt. Lincoln on the Continental Divide to about 4,400 ft at the downstream end of the model area. The
basin includes two physiographic provinces, the Front Range Section of the Southern Rocky Mountain Province
and the Colorado Piedmont Section of the Great Plains Province (Dennehy et al., 1993, 1998; USGS, 2008).
The basin has a continental-type climate modified by topography, in which there are large temperature ranges and
irregular seasonal and annual precipitation. Mean temperatures increase from west to east and on the plains from
north to south. Areas along the Continental Divide average 30 inches or more of precipitation annually, which
includes snowfall in excess of 300 inches. In contrast, the annual precipitation on the plains east of Denver,
Colorado, and in the South Park area in the southwest part of the basin, ranges from 7 to 15 inches. Most of the
precipitation on the plains occurs as rain, which typically falls between April and September, while most of the
precipitation in the mountains occurs as snow, which typically falls between October and March.
Land use and land cover in the South Platte River basin is divided into rangeland (46 percent), agricultural land
(18 percent), forest land (24 percent), urban land (7 percent), and other land (5 percent). Rangeland is present
across all areas of the basin except over the high mountain forests. Agricultural land is somewhat more restricted
to the plains and the South Park area near Fairplay, Colorado. Forest land occurs in a north-south band in the
mountains. Urban land is present primarily in the Front Range urban corridor. Irrigated agriculture comprises only
8 percent of the basin but accounts for 71 percent of the water use. Urban lands comprise only 7 percent of the
basin but account for 12 percent of the water use (or 27 percent if power generation is considered an urban water
use).
To augment water supplies in the basin there are significant diversions of water into the South Platte tributaries
from tunnels that connect to the wetter, western side of the Continental Divide, most notably the Colorado-Big
Thompson Project (Adams Tunnel) which transports about 285,000 acre-feet per year of Colorado River water
through a 13-mile tunnel under the Continental Divide into the Big Thompson River. Overall there are 15 inter-
basin transfers into the basin and almost 1,000 reservoirs. Only the three largest mainstem reservoirs are explicitly
represented in the model. The limited data available on reservoirs and inter-basin transfers creates significant
challenges for hydrologic simulation in this watershed.
The population of the South Platte River basin is about 2.8 million people, over 95 percent of them in Colorado.
The basin contains the most concentrated population density in the Rocky Mountain region, located in the Denver
metropolitan area and along the Front Range urban corridor in Colorado where the mountains meet the plains.
Population densities outside the urban corridor are small and centered in small towns located along the principal
streams. The principal economy in the mountainous headwaters is based on tourism and recreation; the economy
in the urbanized south-central region mostly is related to manufacturing, service and trade industries, and
government services; and the economy of the basin downstream from Denver is based on agriculture and
livestock production.
T-4
-------
Lone TreeJ-Owl
(10190008)
Crow
(10190009)
Cache
La Poudre
(10190007)
Hydrography
Water (Nat. Atlas Dataset)
US Census Populated Places
Municipalities (pop £ 50,000)
County Boundaries
Watershed with HUCSs
Middle South Platte-
Cherry Creek
(10190003)
St. Vrain
(10190005)-*)
Boulder
Denver
Clear
(10190004)
(10190011)
Upper South
I Platte/
'(10190002)
x'
South Platte
Headwater
(10190001)
Kiowa
(10190010)
Colorado
Springs
GCRP Model Areas - South Platte River Basin
Base Map
Figure 1. Location of the South Platte watershed.
T-5
-------
Soil Characteristics
Soils in the watershed, as described in STATSGO soil surveys, fall primarily (56 percent) into hydrologic soil
group (HSG) B (moderately high infiltration capacity). SWAT uses information drawn directly from the soils data
layer to populate the model.
Land Use Representation
Land use/cover in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage and is
predominantly grassland in the high plains and forest in the mountains, with substantial urban development in the
Denver area (Figure 2). NLCD land cover classes were aggregated according to the scheme shown in Table 1 for
representation in the GCRP model. SWAT uses the built-in hydrologic response unit (HRU) overlay mechanism
in the ArcSWAT interface. SWAT HRUs are formed from an intersection of land use and SSURGO major soils.
The distribution of land use in the watershed is summarized in Table 2.
T-6
-------
Legend
Hydrography
= Interstate
I | County Boundaries
2001 NLCD Land Use
] Open water
|^| Perennial Ice/Snow
] Developed, open space
^^| Developed, low intensity
| Developed, medium intensity
| Developed, high intensity
^ Barren land
| Deciduous forest
j^B Evergreen forest
^ Mixed forest
] Scrub/shrub
] Grassland/herbaceous
] Pasture/hay
| Cultivated crops
] Woody wetlands
Emergent herbaceous wetlands
GCRP Model Areas - South Platte River Basin
Land Use Map
NAD_1983_Albers_meters - Map produced 03-24-2011 - P. Cada
0 15 30
60
• Kilometers
15
30
60
• Miles
TETRATECH
Figure 2. Land use in the South Platte watershed.
T-7
-------
Table 1. Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area usually
accounted for as reach area
Deciduous
Evergreen
Mixed
Emergent & woody wetlands
Aquatic bed wetlands (not
emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
T-8
-------
Table 2. Land use distribution for the South Platte watershed (2001 NLCD) (mi )
HUC8
watershed
South Platte
Headwater.
10190001
Upper
Southe
Platte
10190002
Middle
South
Platte-
Cherry
Creek
10190003
Clear
10190004
St. Vrain,
10190005
Big
Thompson.
10190006
Cache La
Poudre.
10190007
Lone Tree-
Owl..
10190008
Crow.
10190009
Kiowa.
10190010
Bijou
10190011
Total
Open
water
13.7
9.5
40.0
3.3
16.6
13.1
27.0
0.7
1.9
0.2
1.1
127.2
Snow/Ice
7.5
3.2
0.0
19.8
27.5
13.4
24.7
0.0
0.0
0.0
0.0
96.1
Developed9
Open
space
19.8
68.3
131.9
18.7
29.0
17.1
36.1
14.9
33.2
23.1
42.9
434.8
Low
density
1.2
94.6
129.8
40.7
43.5
17.7
41.8
2.2
12.8
1.5
2.3
388.1
Medium
density
0.0
34.3
71.2
13.0
15.7
5.9
14.1
0.5
7.0
0.5
0.5
162.7
High
density
0.0
11.5
27.6
3.9
3.9
1.0
3.3
0.0
1.8
0.1
0.0
53.2
Barren
land
37.9
16.7
4.2
21.2
25.6
21.6
18.3
1.3
5.6
0.6
0.6
153.5
Forest
526.3
1,125.5
45.4
280.9
397.4
379.8
630.0
1.9
30.9
31.0
27.9
3,477.0
Shru bland
921.1
440.9
1,301.2
153.1
174.1
196.6
721.8
414.9
1,126.2
414.6
938.9
6,803.3
Pasture/Hay
6.6
3.2
71.6
1.5
42.1
17.1
44.6
6.9
12.7
4.2
8.7
219.2
Cultivated
0.1
2.5
977.5
0.6
166.8
131.2
288.4
130.9
148.5
231.0
342.0
2,419.4
Wetland
70.0
39.7
78.3
9.2
37.1
17.4
40.3
3.6
9.1
10.1
18.8
333.8
Total
1 ,604.3
1 ,849.7
2,878.6
565.8
979.2
832.0
1,890.5
578.0
1,389.8
716.8
1 ,383.7
14,668.3
aThe percent imperviousness applied to each of the developed land uses is as follows: open space (6.41%), low density (33.46%), medium density (60.79%), and high
density (86.76%).
T-9
-------
Point Sources
There are numerous point source discharges in the watershed. Only the major dischargers, with a design flow
greater than 1 MGD are included in the simulation (Table 3). The major dischargers are represented at long-term
average flows, without accounting for changes overtime or seasonal variations.
Table 3. Major point source discharges in the South Platte watershed
NPDES ID
CO0001091
C00001104
CO0001139
C00001147
CO0001163
C00001511
CO0020290
C00020320
CO0020478
C00020508
CO0020737
C00021440
CO0021547
C00021580
CO0023078
C00023124
CO0024147
C00024171
CO0026409
C00026425
C0002661 1
C00026638
C00026662
CO0026671
C00026701
CO0027707
C00032999
CO0037966
C00040258
CO0040681
C00041700
CO0043010
WY0000442
WY0022381
Name
PUBLIC SERVICE CO. OF
COLORADO
PUBLIC SERVICE CO. OF COLO.
PUBLIC SERVICE COMPANY OF
COLO
SUNCOR ENERGY (USA) INC.
COORS BREWING COMPANY
LOCKHEED MARTIN SPACE
SYSTEMS
ESTES PARK SANITATION DISTRICT
WINDSOR, TOWN OF
BOXELDER SANITATION DISTRICT
EVANS, CITY OF
SOUTH FORT COLLINS SAN DIST
FORT LUPTON, CITY OF
BRIGHTON, CITY OF
ST. VRAIN SANITATION DISTRICT
LOUISVILLE, CITY OF
LAFAYETTE, CITY OF
BOULDER, CITY OF
WESTMINSTER, CITY OF
BROOMFIELD, CITY OF
FORT COLLINS, CITY OF
AURORA, CITY OF
METRO WASTEWATER RECLAM
DIST
SOUTH ADAMS COUNTY W&S DIST
LONGMONT, CITY OF
LOVELAND, CITY OF
SWIFT BEEF COMPANY
LITTLETON/ENGLEWOOD, CITIES
OF
CENTENNIAL WATER & SAN. DIST.
GREELEY, CITY OF
ARAPAHOE COUNTY W&WW
AUTHORITY
ST. VRAIN SANITATION DISTRICT
SUPERIOR METROPOLITAN DIST
NO1
Frontier Refining Inc
Cheyenne BOPU
Design flow
(MGD)
1.5
1.5
2.34
0.9
3
2.75
3
1.5
3.4
4.4
20.5
9.2
3.2
7
2.6
220
2.28
17
10
2.8
28
8.48
14.7
2.4
3
0
3.5
Observed flow
(MGD)
(1991-2006 average)
0.2798
1.8487
0.0985
1.7383
11.436
0.3062
0.5006
1.001
1.6335
0.8319
1.0112
1.0882
1.8255
0.5402
1.8457
1.6297
16.177
6.3946
3.7007
16.261
2.7601
159.81
2.8959
7.9692
5.6427
2.5889
24.182
3.5751
8.0844
0.842
1.0304
0.2913
0.8737
3.675
T-10
-------
Most of these point sources have relatively sparse water quality monitoring for nutrients. The point sources were
initially represented in the model with an assumed total phosphorus concentration of 7.2 mg/L and total nitrogen
concentration of 11.2 mg/L for secondary treatment facilities (Tetra Tech 1999).
Meteorological Data
The required meteorological time series for the GCRP SWAT simulations are precipitation and air temperature.
The GCRP simulations do not include water temperature simulation and use a degree-day method for snowmelt.
SWAT estimates Penman-Monteith potential evapotranspiration using a statistical weather generator for inputs
other than temperature and precipitation. These meteorological time series are drawn from the BASINS4
Meteorological Database (USEPA 2008), which provides a consistent, quality-assured set of nationwide data with
gaps filled and records disaggregated. Scenario application requires simulation over 30 years, so the available
stations are those with a common 30-year period of record (or one that can be filled from an approximately co-
located station) that covers the year 2000. A total of 33precipitation stations were identified for use in the
Minnesota River model with a common period of record of 10/1/1969-9/30/2000 (Table 4). Temperature records
are sparser; where these are absent temperature is taken from nearby stations with an elevation correction.
Table 4. Precipitation stations for the South Platte watershed model
COOP ID
CO050183
CO050263
CO050454
CO050843
CO050848
CO050945
CO051179
CO051186
CO051528
CO051547
CO052162
CO052220
CO052494
CO052790
CO053005
CO053038
CO053530
CO053553
CO053584
CO054155
CO054452
CO054762
CO055116
CO055121
CO056023
CO057510
CO057664
CO058839
Name
ALLENSPARK2SE
ANTERO RESERVOIR
BAILEY
BOULDER 2
BOULDER
BRIGGSDALE
BYERS 5 ENE
CABIN CREEK
CHEESMAN
CHERRY CREEK DAM
DEER TRAIL 3 NW
DENVER STAPELTON
EASTONVILLE 2 NNW
EVERGREEN
FORT COLLINS
FORT MORGAN
GRANT
GREELEY UNC
GREENLAND 6 NE
HOYT
KASSLER
LAKEWOOD
LONGMONT 2 ESE
LONGMONT 6 NW
NUNN
SEDALIA 4 SSE
SIMLA
WATERDALE
Latitude
40.1881
38.9933
39.4047
40.0339
39.9919
40.635
39.7403
39.6553
39.2203
39.6261
39.6419
39.7633
39.1092
39.6381
40.6147
40.2617
39.4608
40.4022
39.2167
39.9875
39.49
39.7489
40.1589
40.2467
40.7064
39.4036
39.1397
40.4256
Longitude
-105.502
-105.892
-105.477
-105.281
-105.267
-104.327
-104.128
-105.709
-105.278
-104.832
-104.078
-104.869
-104.6
-105.315
-105.131
-103.804
-105.679
-104.699
-104.738
-104.085
-105.095
-105.121
-105.074
-105.146
-104.783
-104.952
-104.088
-105.21
Temperature
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Elevation (m)
2504
2719
2356
1650
1672
1473
1554
3054
2097
1721
1554
1611
2198
2129
1525
1320
2644
1437
2103
1460
1703
1719
1509
1570
1584
1821
1828
1594
T-ll
-------
COOP ID
CO059210
WY481547
WY481675
WY484930
WY485420
Name
WOODLAND PARK 8 NNW
CARPENTER 3N
CHEYENNE WSFO AP
JELM 2S
LARAMIE 2 WSW
Latitude
39.1006
41.0844
41.1578
41.06
41.3042
Longitude
-105.094
-104.379
-104.807
-106.026
-105.641
Temperature
X
X
Elevation (m)
2365
1657
1864
2310
2187
Watershed Segmentation
The South Platte River watershed was divided into 75 subwatersheds for the purposes of modeling (Figure 3). The
initial calibration was done at gage 06714000, South Platte River at Denver. The model encompasses the
complete watershed; however, there are significant inter-basin transfers across the Continental Divide into the
headwaters of the system. These external sources are represented based on best available data and not changed for
climate scenarios. The scenarios thus represent the changes due only to potential weather changes within the
watershed.
T-12
-------
Legend
A. USGS gages
Hydrography
Interstate
Water (Nat. Atlas Dataset)
US Census Populated Places
Model Subbasins
Colorado
USGS 06752500
USGS 06731000
USGS 06720500
USGS 06714000
GCRP Model Areas - South Platte River Basin
Model Segmentation
NAD 1983 Albers meters - Map produced 03-24-2011 - P. Cada
Figure 3. Model segmentation and USGS stations utilized for the South Platte River watershed.
T-13
-------
Calibration Data and Locations
The specific site chosen for initial calibration was the South Platte River at Denver (USGS 06714000), a flow and
water quality monitoring location that approximately coincides with the mouth of an 8-digit HUC covering the
upstream portion of the watershed. This station was selected because there is a good set of flow and water quality
data available and the watershed lacks major point sources and impoundments. Calibration and validation were
pursued at multiple locations (Table 5). Parameters derived at the initial station were not fully transferable to
other portions of the watershed, and additional calibration was conducted at multiple gage locations.
Table 5. Calibration and validation locations in the South Platte watershed
Station name
South Platte River at Denver, CO
South Platte River at Henderson, CO
South Platte River near Kersey, CO
South Platte River near Weldona, CO
USGS ID
06714000
06720500
06754000
06758500
Drainage area
(mi2)
3,861
4,768
9,659
13,190
Hydrology
calibration
X
X
X
X
Water quality
calibration
X
X
X
X
The model hydrology calibration period was set to Water Years 1991-2000 (within the 30-year period of record
for modeling). Hydrologic validation was then performed on Water Years 1981-1990. Water quality calibration
used calendar years 1993-2000, while validation used 1988-1990, as only limited data were available for earlier
years.
T-14
-------
The South Platte River basin has primarily a semiarid climate and, as a result, a long history of water development
beginning about 1870 with live stream diversions from the Cache la Poudre River. Water shortages in the 1930s
prompted the creation of large scale inter-basin transports, most notably the Colorado-Big Thompson Project
(Adams Tunnel) which transports about 285,000 acre-feet per year of Colorado River water through a 13-mile
tunnel under the Continental Divide into the Big Thompson River. Overall there are 15 inter-basin transfers into
the basin and almost 1,000 reservoirs (Dennehy et al. 1993). There are also multiple water diversions within the
system.
Representing these managed features is a challenge for modeling. The three largest inter-basin transfers, each with
a quantity of greater than 50,000 acre feet per year (Adams, Moffat, and Roberts, to the Big Thompson River,
Boulder Creek, and Bear Creek respectively) were represented in the model based on a constant monthly pattern.
These account for about 95 percent of the inter-basin transfers (383,000 acre feet per year). Only a few of the
many reservoirs in the basin are modeled explicitly, with a focus on those that control flow in the mainstem rather
than providing sidestream storage. Three reservoirs were represented in the South Platte watershed model,
namely, Eleven Mile Canyon, Cheesman and Chatfield. The target storage method was adopted for these
reservoirs, yielding an approximation of actual behavior. Pertinent reservoir information was collected from the
Colorado Decision Support Systems.
Numerous other reservoirs and water transfers are not included in the model. This limits the ability of the
simulation to mimic observed flows. Conclusions should thus be drawn on the relative change in flows predicted
under future scenarios, rather than on quantitative estimates of flow.
A spatial calibration approach was adopted for GCRP-SWAT modeling for the South Platte basin. The initial
calibration at the edge of the Rockies was extended downstream with adjustment of parameters for the high plains
portion of the watershed. The calibration is strongly influenced by assumptions about water transfers,
withdrawals, and discharges in the basin.
The initial calibration focus area (South Platte River at Denver) includes 15 subwatersheds and is representative
the Rocky Mountain portion of the overall watershed. The parameters were adjusted within the practical range to
obtain reasonable fit between the simulated and measured flows in terms of Nash-Sutcliffe modeling efficiency
and the high flow and low flow components as well as the seasonal flows.
The water balance of the whole South Platte River basin predicted by the SWAT model over the 30-year
simulation period is as follows:
PRECIP = 384.4 MM
SNOW FALL = 108.78 MM
SNOW MELT = 99.46 MM
SUBLIMATION = 9.04 MM
SURFACE RUNOFF Q = 41.26 MM
LATERAL SOIL Q = 26.55 MM
TILE Q = 0.00 MM
GROUNDWATER (SHAL AQ) Q = 32.28 MM
REVAP (SHAL AQ => SOIL/PLANTS) = 2.96 MM
T-15
-------
DEEP AQ RECHARGE = 0.00 MM
TOTAL AQ RECHARGE = 35.25 MM
TOTAL WATER YLD = 79.88 MM
PERCOLATION OUT OF SOIL = 15.10 MM
ET = 346.2 MM
PET = 1383.3MM
TRANSMISSION LOSSES = 20.20 MM
Hydrologic calibration adjustments focused on the following parameters:
• CN2 (initial SCS runoff curve number for moisture condition II)
• ESCO (soil evaporation compensation factor)
• SURLAG (surface runoff lag coefficient)
• SOL_AWC (available water capacity of the soil layer, mm water/mm of soil)
• ALPHA_BF (baseflow alpha factor, days)
• GW_DELAY (groundwater delay time, days)
• GWQMIN (threshold depth of water in the shallow aquifer required for return flow to occur, mm)
• GW_REVAP (groundwater "revap" coefficient)
• CH_N1 (Manning's "n" value for tributary channels)
• CH_N2 (Manning's "n" value for main channels)
• CH_K1 (Effective hydraulic conductivity in tributary channel alluvium)
• CH_K2 (Effective hydraulic conductivity in main channel alluvium)
• SFTMP (Snowfall temperature)
• SMTMP (Snowmelt base temperature)
• SMFMX (Maximum melt rate for snow during the year)
• SMFMN (Minimum melt rate for snow during the year)
Calibration results for the South Platte River at Denver are summarized in Figure 4, Figure 5, Figure 6, Figure 7,
and Table 6. The quality of the fit is fair, with a tendency to underpredict flows in the winter and overpredict
flows in the summer and fall. Much of this discrepancy is believed to be due to the complex series of water
imports, storage, and withdrawals in the watershed.
T-16
-------
3000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1990 to 9/30/2000
Avg Modeled Flow (Same Period)
2500 -
2000
-------
To Lower Bound Average Monthly Rainfall (in) -Median Observed Flow (10/1/1990 to 9/30/2000) BModeled (Median, 25th, 75th)
1200
10 11 12 1 2 3
1000 -
10 11 12 1 2 3 4 5 6 7 8 9
Figure 6. Seasonal medians and ranges at USGS 06714000, South Platte River at Denver, CO -
calibration period.
•Observed Flow Duration (10/1/1990 to 9/30/2000 )
•Modeled Flow Duration (10/1/1990 to 9/30/2000)
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 7. Flow exceedance at USGS 06714000, South Platte River at Denver, CO - calibration period.
T-18
-------
Table 6. Summary statistics at USGS 06714000, South Platte River at Denver, CO - calibration period
REACH OUTFLOW FROM OUTLET(S) 31, 32
10-Year Analysis Period: 10/1/1990 - 9/30/2000
Flow/volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
1.05
0.40
0.18
0.40
0.14
0.07
0.44
0.20
0.07
Error Statistics
9.82
1.05
-3.29
32.89
Observed Flow Gage
USGS 06714000 SOUTH PLATTE R
Hydrologic Unit Code: 10190003
Latitude: 39.759722222
Longitude: -105. 166666666
Drainage Area (sq-mi): 3861
VER AT DENVER, CO
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring FlowVolume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
26.06 » 30
-35.47 30
0.17 30
-29.43 20
-35.56
0.738
0.439
0.857
50
Model accuracy increases
as E or E' approaches 1.0
0.95
0.41
0.18
0.30
0.11
0.10
0.44
0.29
0.10
Clear [
Hydrology Validation
Hydrology validation for the South Platte River at Denver was performed for the period 10/1/1980 through
9/30/1990. Like the calibration, the validation fails to achieve all desired seasonal criteria and attains only a
mediocre value for model fit efficiency - likely due in large part to water imports and withdrawals. Results are
summarized in Figure 8, Figure 9, Figure 10, Figure 11, and Table 7.
T-19
-------
4000
3000 -
:2000 -
1000 -
O-80
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1980 to 9/30/1990 )
Avg Modeled Flow (Same Period)
A-82
O-83
A-85
O-86
A-88
Month
O-89
ro
or
Figure 8. Mean monthly flow at USGS 06714000, South Platte River at Denver, CO - validation period.
Avg Flow (10/1/1980 to 9/30/1990)
-Line of Equal Value
Best-Fit Line
1500
500
1000
1500
Average Observed Flow (cfs)
1500
1000
•e
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1980to 9/30/1990)
Avg Modeled Flow (Same Period)
10 i 11 i 12 i 1 i 2 3 i 4 i 5 i 6
500 - —
10 11 12 1 23456789
Month
Figure 9. Seasonal regression and temporal aggregate at USGS 06714000, South Platte River at Denver,
CO - validation period.
T-20
-------
To Lower Bound Average Monthly Rainfall (in) -Median Observed Flow (10/1/1980 to 9/30/1990) Modeled (Median, 25th, 75th)
2500
2000
500
Jan Feb Mar Apr May
10 11 12 1
Figure 10. Seasonal medians and ranges at USGS 06714000, South Platte River at Denver, CO -
validation period.
•Observed Flow Duration (10/1/1980 to 9/30/1990)
Modeled Flow Duration (10/1/1980 to 9/30/1990)
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 11. Flow exceedance at USGS 06714000, South Platte River at Denver, CO - validation period.
T-21
-------
Table 7. Summary statistics at USGS 06714000, South Platte River at Denver, CO - validation period
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET(S) 31, 32
10-Year Analysis Period: 10/1/1980 - 9/30/1990
Flow/volumes are (inches/year) for upstream drainage area
Observed Flow Gage
USGS 06714000 SOUTH PLATTE RIVER AT DENVER, CO
Hydrologic Unit Code: 10190003
Latitude: 39.759722222
Longitude: -105.166666666
Drainage Area (sq-mi): 3861
Total Simulated In-stream Flow:
1.07
Total Observed In-stream Flow:
1.28
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
0.39
0.20
_Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
0.58
0.20
^Simulated Summer Flow_Volurne ^months 7-9):
^irnulajed Fall Flow yoJujTTeJmonthsJ[p:12)^__
Simulated_Winter Flqw_yolurne (months 1_-3_):_
_Sirnulated Spring_FlowVplume_(months_4-_6)^
0.39_
0.18
Observed Summer Flow Volume (7-9):
0.37
0.16
ObservedFall
0.10
0.40"
Jpbseryed Winter_Flp\/vVolume_(1_-3):
Obse_rved_Spring Flow Volume
0.13
0.62"
Total Simulated Storm Volume:
0.25
Total Observed Storm Volume:
Simulated Summer Storm Volume (7-9):
0.08
Observed Summer Storm Volume (7-9):
0.42
0.14
Errors (Simulated-Observed)
ErrpMn tptal_volume:_
Error in 50% lowest flows:
Error Statistics
ErrpMn 1_0% jiighestjlows:_
Seasonal _yojume_ej;rorj Summer:
Seasonal volume error - Fall:
Seasonal _yoJume_ejTorj Winter^
JSeasqnal yojume errpr_-jJprjng: _
Jrrqrjn_stprm_ volumes: _
Error in summer storm volumes:
-35.95
-39.52"
-39.71
_Nash-Sutcliffe Cqefficjent of_Efficie_ncy,_&
Baseline adjusted coefficient (Garrick), E':
0.523
0.423
Model accuracy increases
as E or E' approaches 1.0
Monthly NSE
0.627
Hydrology Results for Larger Watershed
Minor adjustments were made to the parameters determined for the initial calibration gage to improve the fit
downstream. Calibration and validation was pursued at a total of four gages in the watershed. Calibration results
were fair at most gages, as summarized in Table 8, with significant seasonal volume errors. This is primarily the
result of the simplified representation of inter-basin transfers and reservoir storage in the watershed, and so is
deemed acceptable.
Results of the validation exercise are summarized in Table 9. Problems similar to those experienced in the
calibration period were seen at all the gages and total flows tended to be underpredicted, likely due to an increase
in storage and withdrawals since the 1980s.
T-22
-------
Table 8. Summary statistics (percent error): all stations - calibration period
Station
Error in total volume:
Error in 50% lowest
flows:
Error in 10% highest
flows:
Seasonal volume error
- Summer:
Seasonal volume error
-Fall:
Seasonal volume error
-Winter:
Seasonal volume error
- Spring:
Error in storm volumes:
Error in summer storm
volumes:
Daily Nash-Sutcliffe
Coefficient of
Efficiency, E:
Monthly Nash-Sutcliffe
Coefficient:
06714000
South Platte
River
at Denver, CO
9.82
1.05
-3.29
32.89
26.06
-35.47
0.17
-29.43
-35.56
0.738
0.857
06720500
South Platte
River
at Henderson, CO
3.89
1.79
-3.37
14.68
-5.77
-27.37
10.68
-0.10
-0.31
0.610
0.811
06754000
South Platte River
near Kersey, CO
10.69
-1.85
-1.04
54.26
-25.53
-23.18
20.03
-32.95
-24.85
0.597
0.689
06758500
South Platte
River
nearWeldona,
CO
-0.38
-9.55
8.26
33.55
-28.93
-45.57
11.49
-41.56
-25.86
0.628
0.734
T-23
-------
Table 9. Summary statistics: all stations - validation period
Station
Error in total volume:
Error in 50% lowest
flows:
Error in 10% highest
flows:
Seasonal volume
error- Summer:
Seasonal volume
error- Fall:
Seasonal volume
error- Winter:
Seasonal volume
error- Spring:
Error in storm
volumes:
Error in summer
storm volumes:
Daily Nash-Sutcliffe
Coefficient of
Efficiency, E:
Monthly Nash-
Sutcliffe Coefficient:
06714000
South Platte
River
at Denver, CO
-16.28
-0.62
-32.91
5.88
13.67
-21.65
-35.95
-39.52
-39.71
0.523
0.627
06720500
South Platte River
at Henderson, CO
-15.23
-3.83
-26.23
3.72
-6.26
-28.43
-25.81
-7.64
6.38
0.521
0.665
06754000
South Platte River
near
Kersey, CO
-18.38
-3.73
-36.86
16.98
-32.91
-39.92
-18.82
-42.20
-42.38
0.572
0.612
06758500
South Platte River
near Weldona, CO
-34.38
-15.24
-37.64
-3.46
-49.87
-57.46
-32.82
-52.40
-37.33
0.568
0.632
Water Quality Calibration and Validation
Initial calibration of water quality was done for the South Platte River at Denver (USGS 06714000), using
available data for 1993-2000. Insufficient earlier data were available to allow a separate validation period at this
station; however, validation for brief earlier periods was performed at downstream sites. As with hydrology, water
quality calibration was performed on the later period as this better reflects the land use included in the model. The
start of the validation period is constrained by data availability.
Calibration adjustments for sediment focused on the following parameters:
• SPCON (linear parameter for estimating maximum amount of sediment that can be re-entrained during
channel sediment routing)
• SPEXP (exponential parameter for estimating maximum amount of sediment that can be re-entrained
during channel sediment routing)
• CH_COV (channel cover factor)
• CH_EROD (channel erodibility factor)
• USLE_P (USLE support practice factor)
T-24
-------
Simulated and estimated sediment loads for the South Platte River at Denver station are shown in Figure 12 and
statistics are provided in Table 10. The key statistic in Table 10 is the relative percent error, which shows the error
in the prediction of monthly load normalized to the estimated load. Several large loading events were
underpredicted by the model, likely due to uncertainty in the simulation of reservoir trapping. Table 10 also shows
the relative average absolute error, which is the average of the relative magnitude of errors in individual monthly
load predictions. This number is inflated by outlier months in which the simulated and estimated loads differ by
large amounts (which may be as easily due to uncertainty in the estimated load due to limited data as to problems
with the model) and the third statistic, the relative median absolute error, is likely more relevant and shows better
agreement.
TSS
-Regression Loads
-Simulated Loads
Figure 12. Fit for monthly load of TSS at USGS 06714000, South Platte River at Denver, CO.
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 06714000, South Platte River at Denver, CO
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1993-2000)
86.6%
77%
4.2%
Calibration adjustments for total phosphorus and total nitrogen focused on the following parameters:
• RHOQ (algal respiration rate at 20° C)
• PHOSKD (phosphorus soil partitioning coefficient)
• PSP (phosphorus availability index)
• RSI (Local algal settlement rate in the reach at 20° C)
• AL1 (Fraction of algal biomass that is nitrogen)
• AL2 (Fraction of algal biomass that is phosphorus)
• MUMAX (Rate of oxygen uptake per unit NO2-N oxidation at 20° C)
• RHOQ (Algal respiration rate at 20° C)
T-25
-------
• RS2 (benthic source rate for dissolved P in the reach at 20° C)
• RS3 (Benthic source rate for NFLpN in the reach at 20° C)
• RS5 (organic P settling rate in the reach at 20° C)
• BC4 (rate constant for mineralization of organic P to dissolved P in the reach at 20° C)
• RS4 (rate coefficient for organic N settling in the reach at 20° C)
• CH_ONCO (Channel organic nitrogen concentration)
• CH_OPCO (Channel organic phosphorus concentration)
• SDNCO (Denitrification threshold water content)
• CDN (Denitrification exponential rate constant)
Results for the phosphorus simulation are shown in Figure 13 and Table 11. Results for the nitrogen simulation
are shown in Figure 14 and Table 12. The model fit is generally good for the nutrients.
Total P
1000
100
o
Ifl
I
-Regression Loads
-Simulated Loads
Figure 13. Fit for monthly load of total phosphorus at USGS 06714000, South Platte River at Denver, CO.
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 06714000, South Platte River at Denver, CO
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1993-2000)
-14.0%
34%
13.4%
T-26
-------
Total N
1,000
100
o
«
•Averaging Loads
-Simulated Loads
Figure 14. Fit for monthly load of total nitrogen at USGS 06714000, South Platte River at Denver, CO.
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 06714000, South Platte River at Denver, CO
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1993-2000)
6.1%
46%
29.4%
Water Quality Results for Larger Watershed
The SWAT model parameters for water quality determined for South Platte River at Denver were directly
transferred to other portions of the watershed. Application of the SWAT model without spatial adjustments
resulted in relatively large errors in predicting loads and concentrations at some stations. Sediment loads appear to
be under-predicted throughout, reflecting the underprediction at the upstream station, but nutrients are fit fairly
well. Summary statistics for the SWAT water quality calibration and validation at other stations in the watershed
are provided in Table 13 and Table 14.
T-27
-------
Table 13. Summary statistics for water quality at all stations - calibration period 1993-2002
Station
Relative Percent
Error TSS Load
Relative Percent
Error TP Load
Relative Percent
Error TN Load
06714000
South Platte River
at Denver, CO
86.6%
-14.0%
6.1%
06720500
South Platte River
at Henderson, CO
75.2%
-6.8%
11.9%
06754000
South Platte River
near Kersey, CO
73.3%
0.5%
9.1%
06758500
South Platte
River
nearWeldona,
CO
ND
36.9%
-36.0%
Table 14. Summary statistics for water quality at all stations - validation period 1986-1992
Station
Relative Percent
Error TSS Load
Relative Percent
Error TP Load
Relative Percent
Error TN Load
06714000
South Platte River
at Denver, CO
ND
ND
ND
06720500
South Platte River
at Henderson, CO
14.7%
-0.8%
2.7%
06754000
South Platte River
near Kersey, CO
ND
ND
ND
06758500
South Platte
River near
Weldona, CO
ND
5.0%
7.7%
T-28
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
Dennehy, K.F., D.W. Litke, C.M. Tate, and J.S. Heiny. 1993. South Platte River Basin - Colorado, Nebraska,
and Wyoming. Water Resources Bulletin, 29(4):647-683.
Dennehy, K.F., D.W. Litke, C.M. Tate, S.L. Qi, P.B. McMahon, B.W. Bruce, R.A. Kimbrough, and J.S. Heiny.
1998. Water Quality in the South Platte River Basin, Colorado, Nebraska, and Wyoming, 1992-95. U.S.
Geological Survey Circular 1167.
Tetra Tech. 1999. Improving Point Source Loadings Data for Reporting National Water Quality Indicators. Final
Technical Report prepared for U.S. Environmental Protection Agency, Office of Waste water Management,
Washington, DC, by Tetra Tech, Inc., Fairfax, VA.
USEPA (United States Environmental Protection Agency). 2008. Using the BASINS Meteorological Database
(Version 2006). BASINS Technical Note 10. Office of Water, U.S. Environmental Protection Agency,
Washington, DC. http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf
(Accessed June, 2009).
USGS (United States Geological Survey). 2008. National Water Quality Assessment (NAWQA) Program: South
Platte River Basin, http://co.water.usgs.gov/nawqa/splt/ (Accessed June, 2009).
T-29
-------
Appendix U
Model Configuration, Calibration and
Validation
Basin: Trinity River (Trin)
U-l
-------
Contents
Watershed Background U-4
Water Body Characteristics U-4
Soil Characteristics U-6
Land Use Representation U-6
Point Sources U-10
Meteorological Data U-11
Watershed Segmentation U-13
Calibration Data and Locations U-15
SWAT Modeling U-16
Assumptions U-16
Hydrology Calibration U-16
Hydrology Validation U-20
Hydrology Results for Larger Watershed U-23
Water Quality Calibration and Validation U-25
Water Quality Results for Larger Watershed U-28
References U-30
U-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes U-8
Table 2. Land use distribution for the Trinity River basin (2001 NLCD) (mi2) U-9
Table 3. Major point source discharges in the Trinity River basin U-10
Table 4. Precipitation stations for the Trinity River watershed model U-ll
Table 5. Calibration and validation locations in the Trinity River basin U-15
Table 6. Summary statistics at USGS 08066500 Trinity River at Romayor, TX - calibration period.... U-20
Table 7. Summary statistics at USGS 08066500 Trinity River at Romayor, TX - validation period U-23
Table 8. Summary statistics (percent error): all Stations - calibration period U-24
Table 9. Summary statistics: all stations - validation period U-25
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 08066500 Trinity River at Romayor, TX U-26
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 08066500 Trinity River at Romayor, TX U-27
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 08066500 Trinity River at Romayor, TX U-28
Table 13. Summary statistics for water quality at all stations - calibration period 1985-2001 U-29
Table 14. Summary statistics for water quality at all stations - validation period 1972-1984 U-29
Figures
Figure 1. Location ofthe Trinity River basin U-5
Figure 2. Land use in the Trinity River basin U-7
Figure 3. Model segmentation and USGS stations utilized for the trinity river watershed U-14
Figure 4. Mean monthly flow at USGS 08066500 Trinity River at Romayor, TX - calibration period.. U-17
Figure 5. Seasonal regression and temporal aggregate at USGS 08066500 Trinity River at Romayor,
TX - calibration period U-18
Figure 6. Seasonal medians and ranges at USGS 08066500 Trinity River at Romayor, TX - calibration
period U-18
Figure 7. Flow exceedance at USGS 08066500 Trinity River at Romayor, TX - calibration period U-19
Figure 8. Mean monthly flow at USGS 08066500 Trinity River at Romayor, TX- validation period U-21
Figure 9. Seasonal regression and temporal aggregate at USGS 08066500 Trinity River at Romayor,
TX - validation period U-21
Figure 10. Seasonal medians and ranges at USGS 08066500 Trinity River at Romayor, TX - validation
period U-22
Figure 11. Flow exceedance at USGS 08066500 Trinity River at Romayor, TX - validation period U-22
Figure 12. Fit for monthly load of TSS at USGS 08066500 Trinity River at Romayor, TX U-26
Figure 13. Fit for monthly load of total phosphorus at USGS 08066500 Trinity River at Romayor, TX.. U-27
Figure 14. Fit for monthly load of total nitrogen at USGS 08066500 Trinity River at Romayor, TX U-28
U-3
-------
The Trinity River basin was selected as one of the 15 non-pilot application watersheds for the 20 Watershed
study. Watershed modeling for the non-pilot areas is accomplished using the SWAT model only, and model
calibration and validation results are presented in abbreviated form.
Water Body Characteristics
The Trinity River basin is located in east central Texas. It extends on a southeast diagonal, from immediately
south of the Oklahoma-Texas border to the Trinity Bay at the Gulf of Mexico. The model study area encompasses
almost 18,000 mi2 in 12 HUCSs in HUC 1203 (Figure 1). The watershed is dissected by alternate bands of rolling,
treeless prairies, smooth to slightly rolling prairies, rolling timbered hills, and a relatively flat coastal plain. The
watershed slopes gradually from about 1,200 ft above sea level in the northwest, to about 600 ft mid-basin, and on
to sea level in the southeastern section of the area, at Trinity Bay (Land et al., 1998; Ulery et al., 1993).
Past and current human activities, including construction of reservoirs, urbanization, farming, ranching, and oil
and gas production, have greatly altered the natural environment in the Trinity River basin. Approximately 37
percent of the watershed is cropland or pasture. Major crops include corn, cotton, peanuts, sorghum, soybeans,
rice, and wheat. Wheat and cotton are dry cropland crops, while rice is an irrigated crop. Forest and wetlands
represent about 33 percent of the watershed and developed land makes up about 19 percent of the watershed. The
population in the watershed is mainly clustered in the Dallas-Fort Worth metropolitan area, with a few secondary
population clusters (Denton, McKinney, Corsicana, and Waxahachie).
The climate of the basin is described as modified-marine, subtropical-humid, having warm summers and a
predominant onshore flow of tropical maritime air from the Gulf of Mexico. Precipitation varies considerably
across the watershed. Average annual precipitation ranges from less than 27 inches in the northwest part of the
watershed to greater than 52 inches in the southeast. On average, the watershed experiences a winter surplus and a
summer deficiency of precipitation. Average annual temperature is fairly uniform throughout the basin, ranging
from about 69° F in the southeastern area of the watershed to about 65° F in the northwest.
There are 22 large reservoirs in the Trinity River basin and hundreds of smaller reservoirs, mostly flood control
structures. Reservoirs have been built to retain runoff on all major tributaries and the mainstem of the Trinity
River. Diversions move water within the basin and to and from adjacent river basins. The largest interbasin
diversion is out of the basin, from the Trinity River below Livingston Reservoir to the Houston metropolitan area.
There are numerous other inter- and intrabasin diversions.
The largest consumptive use in the watershed is domestic with the majority being used in Dallas and Tarrant
counties because of their large populations. Surface water, almost entirely from reservoirs, supplies more than 90
percent of the water used in the basin. Groundwater is used for municipal and domestic supply in some of the
smaller towns and in rural areas. Transfers of water, from the adjoining basins and from reservoirs below Dallas
and Fort Worth, are required to meet the needs of the Dallas-Fort Worth area. Relatively little water is used for
irrigating crops.
U-4
-------
Denton
(12030104}
Elm Fork
Trinity
Upper West
Fork Trinity
(120301:01}
ast-Fork
(12030103) | / Trinity
12030106)
Cedar
(12030'107)
rjH12030105)
Chambers
2030109)
Richland
(12030108)
Lower'Tj;inity-
Tehuacana
(12030201^-,
Hydrography
Waler (Nat. Alias Dataset)
US Census Populated Places
^B Municipalities (pop i SO.000)
I | County Boundaries
I I Watershed with HUCBs
• Lower Trinity
Kickapoo
(12030202)
College
Station
Lower
Trinity^
(12030203)
GCRP Model Areas - Trinity River Basin
Base Map
Figure 1. Location of the Trinity River basin.
U-5
-------
Soil Characteristics
Soils in the watershed, as described in STATSGO soil surveys, fall primarily into hydrologic soil groups (HSGs)
C (moderately low infiltration capacity) and D (low infiltration capacity). Soils range from course textured loamy
sands to fine textured montmorillonitic clays. Soil depths vary from very shallow to deep. SWAT uses
information drawn directly from the soils data layer to populate the model.
Land Use Representation
Land use/cover in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage and is
predominantly rangeland (Figure 2). NLCD land cover classes were aggregated according to the scheme shown in
Table 1 for representation in the 20 Waterhsed model. SWAT uses the built-in hydrologic response unit (HRU)
overlay mechanism in the ArcSWAT interface. SWAT HRUs are formed from an intersection of land use and
SSURGO major soils. The distribution of land use in the watershed is summarized in Table 2.
U-6
-------
Oklahoma
• Hydrography
= Interstate
I | County Boundaries
2001 NLCD Land Use
HI Open water
I Developed open space
| Developed, low intensity
^H Developed, medium intensity
|H| Developed, high intensity
^ Barren land
• Deciduous forest
^H Evergreen forest
1 Mixed forest
] Scrub/shrub
I I Grassland/herbaceous
~~1 Pasture/hay
l] Cultivated crops
^ VWbody wetlands
n Emergent herbaceous wetlands
GCRP Model Areas - Trinity River Basin
Land Use Map
Figure 2. Land use in the Trinity River basin.
U-7
-------
Table 1. Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area usually
accounted for as reach area
Deciduous
Evergreen
Mixed
Emergent & woody wetlands
Aquatic bed wetlands (not
emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
U-8
-------
Table 2. Land use distribution for the Trinity River basin (2001 NLCD) (mi )
HUC8
watershed
12030101
12030102
12030103
12030104
12030105
12030106
12030107
12030108
12030109
12030201
12030202
12030203
Total
Open
water
45.0
40.4
103.6
15.0
25.2
74.2
61.6
56.8
39.6
31.6
140.5
35.2
668.6
Developed9
Open space
116.5
159.9
116.8
37.6
112.4
94.3
40.0
42.7
59.3
74.2
123.5
40.9
1,018.2
Low
density
31.3
232.1
98.2
25.9
121.2
110.6
46.9
4.9
21.9
64.3
75.4
18.0
850.6
Medium
density
7.1
107.0
71.1
13.7
67.4
82.4
5.8
1.1
6.7
6.1
6.9
2.6
378.1
High
density
2.0
62.0
35.7
5.2
49.0
25.0
2.6
0.3
3.6
2.4
2.2
1.2
191.2
Barren
land
8.0
2.8
1.9
0.9
2.2
0.7
0.7
0.4
1.6
32.2
4.3
4.4
59.9
Forest
321.0
183.9
188.2
100.5
182.1
152.3
136.8
100.3
124.5
515.8
880.3
58.7
2,944.2
Shru bland
1 ,283.3
557.2
746.1
389.5
300.6
414.6
185.1
354.0
416.7
315.1
479.3
53.1
5,494.6
Pasture/Hay
88.1
108.4
256.0
83.1
299.2
160.5
479.1
200.7
182.3
705.5
956.9
169.0
3,688.9
Cultivated
52.9
49.4
228.9
45.8
152.2
167.7
55.5
145.2
208.0
63.8
41.3
45.8
1 ,256.5
Wetland
1.4
10.9
11.5
2.3
58.3
21.1
51.7
10.4
10.9
295.1
547.1
377.5
1 ,398.2
Total
1 ,956.7
1,514.1
1,858.1
719.4
1 ,369.7
1 ,303.2
1 ,065.8
916.7
1 ,075.0
2,106.0
3,257.7
806.6
17,949.1
aThe percent imperviousness applied to each of the developed land uses is as follows: open space (7.74%), low density (31.65%), medium density (60.78%), and high
density (89.15%).
U-9
-------
Point Sources
There are numerous point source discharges in the watershed. Only the major dischargers, with a design flow
greater than 1 MGD are included in the simulation (Table 3). The major dischargers are represented at long-term
average flows, without accounting for changes overtime or seasonal variations.
Table 3. Major point source discharges in the Trinity River basin
ID
TX0001007
TX0001023
TX0001198
TX0020354
TX002071 1
TX0022241
TX0022357
TX0022527
TX0022802
TX0023116
TX0023931
TX0024163
TX0024678
TX0024686
TX002491 1
TX002501 1
TX0025364
TX0025372
TX0025453
TX0025950
TX0030180
TX0031577
TX0032018
TX0047180
TX0047261
TX0047295
TX0047431
TX0047724
TX0047830
TX0047848
TX004791 1
TX0052892
TX0052990
TX0053112
Name
EXTEX LAPORTE LIMITED PARTNERS
LUMINANT GENEATION COMPANY LLC
EXTEX LAPORTE LIMITED PARTNERS
UPPER TRINITY REGIONAL WATER D
FLOWER MOUND, TOWN OF
NORTH TEXAS MWD
GAINESVILLE, CITY OF
TERRELL, CITY OF - KINGS CREEK
TRINITY RIVER AUTHORITY OF TEX
AZLE, CITY OF
NORTH TEXAS MWD
LIVINGSTON, CITY OF
GARLAND, CITY OF (DUCK CREEK)
GARLAND, CITY OF (ROWLETT CREE
DECATUR, CITY OF
TRINITY RIVER AUTHORITY OF TEX
ATHENS, CITY OF
ATHENS, CITY OF
PALESTINE CITY OF-TOWN CREEK
NORTH TEXAS MWD
BIG BROWN POWER COMPANY LLC
TEXAS DEPARTMENT OF CRIMINAL J
GRAPEVINE, CITY OF
DENTON, CITY OF (PECAN CREEK)
ENNIS, CITY OF
FORT WORTH, CITY OF
NORTH TEXAS MWD
WEATHERFORD, CITY OF
DALLAS, CITY OF (CENTRAL)
DALLAS, CITY OF (SOUTHSIDE)
NORTH TEXAS MWD
LEWISVILLE, CITY OF
MEXIA, CITY OF
THE COLONY, CITY OF
Design flow
(MGD)
927
870
1280
5
10
1.2
4.14
3
162
0.941
4.75
2.25
30
24
1.2
0.9
1.367
1.027
2.05
2
1015
2.85
5.75
12
3.1
166
25
4.5
150
110
16
12
2
3.39
Observed
flow (MGD)
1.24
0.58
10.67
2.15
3.49
0.72
1.74
2.45
133.25
0.69
2.025
1.029
21.15
14.67
0.92
11.54
0.81
54.76
2.88
3.00
1.15
1.77
2.51
10.28
1.64
106.75
12.17
2.046
130.29
67.85
18.08
7.63
0.55
1.68
U-10
-------
ID
TX0055735
TX0056731
TX0062189
TX0070831
TX0072974
TX0074284
TX0075388
TX0078565
TX0079391
TX0088633
TX0092789
TX0100170
TXO 103501
TXO 104345
TXO 104957
Name
TROPHY CLUB MUD NO. 1
CORSICANA, CITY OF
BRAZOS ELECTRIC POWER COOPERAT
CROCKETT, CITY OF
HUNTSVILLE, CITY OF
LIBERTY,CITY OF
TEXAS DEPARTMENT OF CRIMINAL J
NORTH TEXAS MUNICIPAL WATER Dl
KAUFMAN, CITY OF
NORTH TEXAS MWD
TEXAS DEPARTMENT OF CRIMINAL J
DAYTON, CITY OF
NORTH TEXAS MUNICIPAL WATER Dl
TRINITY RIVER AUTHORITY OF TEX
TRINITY RIVER AUTHORITY OF TEX
Design flow
(MGD)
1.4
4.95
85
2
4.15
2.5
1.44
2.25
1.2
24
1.5
2
5
3.5
5
Observed
flow (MGD)
0.58
3.05
1.78
0.68
2.97
1.38
0.73
0.90
0.57
26.1
0.76
1.47
3.98
1.95
1.31
Most of these point sources have reasonably good monitoring for total suspended solids (TSS), but not for total
nitrogen and total phosphorus. The point sources were initially represented in the model with the median of
reported values for TSS and an assumed total nitrogen concentration of 11.2 mg/L and assumed total phosphorus
concentration of 7.0 mg/L for secondary treatment facilities (Tetra Tech 1999).
Meteorological Data
The required meteorological time series for the 20 Waterhsed SWAT simulations are precipitation and air
temperature. The 20 Waterhsed simulations do not include water temperature simulation and use a degree-day
method for snowmelt. SWAT estimates Penman-Monteith potential evapotranspiration using a statistical weather
generator for inputs other than temperature and precipitation. These meteorological time series are drawn from the
BASINS4 Meteorological Database (USEPA 2008), which provides a consistent, quality-assured set of
nationwide data with gaps filled and records disaggregated. Scenario application requires simulation over 30
years, so the available stations are those with a common 30-year period of record (or one that can be filled from
an approximately co-located station) that covers the year 2001. A total of 64 precipitation stations were identified
for use in the Trinity River model with a common period of record of 10/1/1972-9/30/2002 (Table 4).
Temperature records are sparser; where these are absent, temperature is taken from nearby stations with an
elevation correction.
Table 4. Precipitation stations for the Trinity River watershed model
ID
410129
410206
410235
410271
410337
410440
410518
Name
TX410129
TX410206
TX410235
TX410271
TX410337
TX410440
TX410518
Latitude
32.6444
33.3867
29.7879
33.4407
32.7395
32.2067
32.2636
Longitude
-97.5617
-97.7163
-94.6342
-98.3708
-97.1277
-96.7957
-96.6375
Elevation
241
308
7
317
200
162
141
Temperature
No
No
Yes
No
No
No
Yes
U-ll
-------
ID
410691
410984
411063
411596
411800
411810
411870
412019
412096
412114
412244
412404
412772
413047
413080
413133
413284
413285
413370
413415
413642
413668
413691
414182
414315
414382
414517
414679
414705
414972
415094
415192
415196
415271
415477
415766
415869
416130
416210
416331
Name
TX410691
TX410984
TX411063
TX411596
TX411800
TX411810
TX411870
TX412019
TX412096
TX412114
TX412244
TX412404
TX412772
TX413047
TX413080
TX413133
TX413284
TX413285
TX413370
TX413415
TX413642
TX413668
TX413691
TX414182
TX414315
TX414382
TX414517
TX414679
TX414705
TX414972
TX415094
TX415192
TX415196
TX415271
TX415477
TX415766
TX415869
TX416130
TX416210
TX416331
Latitude
32.6476
33.5511
33.2067
31.2581
32.3139
30.3637
30.5334
32.1078
32.5562
31.3073
32.8525
33.1990
32.3657
31.7322
33.1397
32.5340
32.8193
32.8339
33.1519
33.6359
33.7970
33.1025
32.9507
32.0162
29.7284
30.7064
33.2384
33.0798
32.5590
33.2251
33.0353
33.0689
30.0593
30.7394
30.9392
33.2365
31.6833
33.6536
31.9611
33.4561
Longitude
-97.4438
-97.8472
-97.7716
-95.9744
-97.4064
-95.0838
-95.1500
-96.4746
-97.6697
-95.4508
-96.8555
-97.1050
-95.6085
-96.2078
-96.3974
-96.6607
-97.3614
-97.2974
-96.8122
-97.1444
-96.8568
-98.5849
-97.0553
-97.1093
-95.1306
-95.5421
-98.1453
-97.2967
-96.2724
-97.8316
-96.4860
-97.0100
-94.7950
-94.9256
-95.9202
-96.6419
-96.4832
-97.3752
-96.6881
-98.0253
Elevation
241
329
227
98
239
60
108
126
346
106
134
192
155
132
179
143
209
196
226
238
221
320
178
168
11
151
314
195
128
265
155
169
11
54
77
190
163
306
138
323
Temperature
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
No
No
Yes
No
Yes
No
Yes
No
No
No
No
No
Yes
No
Yes
No
Yes
Yes
No
Yes
No
Yes
No
Yes
Yes
Yes
Yes
Yes
No
Yes
No
U-12
-------
ID
416636
416641
416757
417028
417556
417586
417588
417659
417707
417773
418274
418929
419125
419286
419522
419532
419800
Name
TX416636
TX416641
TX416757
TX417028
TX417556
TX417586
TX417588
TX417659
TX417707
TX417773
TX418274
TX418929
TX419125
TX419286
TX419522
TX419532
TX419800
Latitude
33.3737
33.4372
31.7832
33.3659
32.9537
30.5382
32.9522
33.0068
32.9334
32.4612
33.7033
32.7668
33.4254
33.4869
32.4287
32.7484
32.7018
Longitude
-98.7657
-98.7806
-95.6038
-97.0122
-97.5738
-95.8457
-96.7664
-97.2246
-96.4667
-96.4493
-96.6419
-96.2831
-96.3393
-97.1572
-96.8432
-97.7699
-96.0150
Elevation
364
361
142
210
235
96
190
190
166
111
232
157
232
221
192
291
158
Temperature
Yes
No
Yes
Yes
No
No
No
No
No
No
Yes
No
No
No
Yes
Yes
Yes
Watershed Segmentation
The Trinity River basin was divided into 73 subwatersheds for the purposes of modeling (Figure 3). The model
encompasses the complete watershed and does not require specification of any upstream boundary conditions for
application.
U-13
-------
Oklahoma
TX418378
TX410206
TX 410984""
TX416331
TX410271
TX416641
TX416636 TX414517
TX415766
TX413370-I—
TX413080
TX411490
TX415094,
TX412244
TX412404
A
TX417028
TX413668
&
TX411063
TX417556
TX418563
TX417659
TX412096
TX410129
TX412019
TX416757
TX413691
TX412242
TX411800
s
TX410337
TX415897
TX419522
TX414182
S
TX410440
TX410518
TX41S196-
L—i
Legend
Hydrography
^^= Interstate
'_ _" Water (Nat. Atlas Datasel)
US Census Populated Places
| I County Boundaries
I I Watershed with HUCBs
GCRP Model Areas - Trinity River Basin
Weather Stations
Figure 3. Model segmentation and USGS stations utilized for the trinity river watershed
U-14
-------
Calibration Data and Locations
The specific site chosen for initial calibration was the Trinity River at Romayor, which is the most downstream
gaging station in the basin. Calibration and validation were pursued at multiple locations (Table 5). Parameters
derived on the Trinity River at Romayor were transferred to other portions of the Trinity River basin.
Table 5. Calibration and validation locations in the Trinity River basin
Station name
East Fork Trinity River at Grand
Prairie, TX
Clear Creek at Sanger, TX
East Fork Trinity near Crandall, TX
Trinity River at Rosser, TX
Trinity River at Trinidad, TX
Trinity River near Crockett, Tx
Trinity River at Romayor, Tx
USGS ID
08062000
08051500
08062000
08062500
08062700
08065350
08066500
Drainage area
(mi2)
629
1,300
851
2,410
1,110
14,900
16,200
Hydrology
calibration
X
X
X
X
X
X
X
Water quality
calibration
X
X
X
X
X
X
X
The model hydrology calibration period was set to Water Years 1992-2001 (within the 30-year period of record
for modeling). Hydrologic validation was then performed on Water Years 1982-1991. Water quality calibration
used calendar years 1985-2001, while validation used 1972-1984. However, there was some variation to this time
period across the monitoring stations depending on the availability of monitored data.
U-15
-------
SWAT Modeling
Assumptions
Eighteen major reservoirs occur in the upper portion of the Trinity River basin. Pertinent reservoir information
including surface area and storage at principal (normal) and emergency spillway levels for the reservoirs modeled
were obtained from the National Inventory of Dams (NID) database. The SWAT model provides four options to
simulate reservoir outflow: measured daily outflow, measured monthly outflow, average annual release rate for
uncontrolled reservoir, and controlled outflow with target release. Keeping in view the 20 Waterhsed climate
change impact evaluation application, it was assumed that the best representation of the reservoirs was to simulate
them without supplying time series of outflow records. Therefore, the target release approach was used in the
GCRP-SWAT model.
Hydrology Calibration
A spatial calibration approach was not adopted for GCRP-SWAT modeling for Trinity River basin; however, a
systematic adjustment of parameters was adopted and some adjustments were applied throughout the basin. Most
of the calibration efforts were geared toward getting a closer match between simulated and observed flows at the
outlet closest to the most downstream USGS gaging station of the basin.
Land Use/Soil/Slope Definition
A 5/10/5 percent threshold was used for land use/soil/slope in the SWAT model while defining the HRUs. Urban
land use classes were exempted from the HRU overlay thresholds.
The parameters were adjusted within the practical range at the calibration focus area to obtain reasonable fit
between the simulated and measured flows in terms of Nash-Sutcliffe modeling efficiency and the high flow and
low flow components as well as the seasonal flows.
The water balance of the whole Trinity River basin predicted by the SWAT model over the 30-year simulation
period is as follows:
PRECIP = 1046.9 MM
SNOW FALL = 16.34 MM
SNOW MELT = 16.21 MM
SUBLIMATION = 0.13 MM
SURFACE RUNOFF Q = 167.49 MM
LATERAL SOIL Q = 10.66 MM
TILE Q = 0.00 MM
GROUNDWATER (SHAL AQ) Q = 16.31 MM
REVAP (SHAL AQ => SOIL/PLANTS) = 151.18 MM
DEEP AQ RECHARGE = 8.86 MM
TOTAL AQ RECHARGE = 177.28 MM
TOTAL WATER YLD = 192.80 MM
PERCOLATION OUT OF SOIL = 175.92 MM
ET = 703.4 MM
PET = 1937.8MM
TRANSMISSION LOSSES = 1.66 MM
U-16
-------
Hydrologic calibration adjustments focused on the following parameters:
• SURLAG (surface runoff lag coefficient)
• CNCOEFF (plant ET curve number coefficient)
• Baseflow factor
• GWQMn (threshold depth of water in the shallow aquifer for return flow to occur [mmH2O])
• NDTarg (number of days needed to reach target storage from current pond storage)
• CN2 (initial SCS runoff curve number for moisture condition II)
• ESCO (soil evaporation compensation factor)
• Revap coeff
Calibration results for the Trinity River at Romayor are summarized in Figures 4 through 7 and Table 6. The
calibration results show a good match (both in volume and timing) between the observed and the simulated flows
(Figure 4, Figure 5, Figure 6, Figure 7, and Table 6).
8000
6000
4000 -
2000
O-92
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002 )
Avg Modeled Flow (Same Period)
O-01
Figure 4. Mean monthly flow at USGS 08066500 Trinity River at Romayor, TX - calibration period.
U-17
-------
20000
• Avg Flow (1/1/1993 to
12/31/2000)
Line of Equal Value
y = 0.5418x+3766.3
0 5000 10000 15000 20000
Average Observed Flow (cfs)
fi
o
Avg Monthly Rainfall (in)
-Avg Observed Flow (1/1/1993 to 12/31/2000)
Avg Modeled Flow (Same Period)
20000
15000
10000
5000 -
]Feb\Mai\Apt\Ma\\Jun\ Ju/UugjSepj Ocfj/VovJD
9 10 11 12
Month
Figure 5. Seasonal regression and temporal aggregate at USGS 08066500 Trinity River at Romayor, TX
calibration period.
To Lower Bound
Average Monthly Rainfall (in)
-Median Observed Flow (1/1/1993 to 12/31/2000)
11
12
"E
'ro
or
Figure 6. Seasonal medians and ranges at USGS 08066500 Trinity River at Romayor, TX - calibration
period.
U-18
-------
i
1
LL
0
0)
TO
0
ro
Q
•Observed Flow Duration (1/1/1993 to 12/31/2000)
Modeled Flow Duration (1/1/1993 to 12/31/2000)
1000000
100000
10000 -
1000
1 - =
0.1
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 7. Flow exceedance at USGS 08066500 Trinity River at Romayor, TX - calibration period.
U-19
-------
Table 6. Summary statistics at USGS 08066500 Trinity River at Romayor, TX - calibration period
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 3
8-Year Analysis F^riod: 1/1/1993 - 12/31/2000
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3):
Simulated Srjring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
7.10
2.80
0.67
0.83
2.04
2.18
2.04
2.85
0.39
Error Statistics
-6.88
-2.33
-9.48
63.58
Observed Flow Gage
USGS 08066500 Trinity Rv at Romayor, TX
Hydrologic Unit Code: 12030202
Latitude: 30.4252067
Longitude: -94.8507622
Drainage Area (sq-rri): 17186
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
9.93 » | 30
-24.04
-14.28
18.68
210.14
0.623
0.482
0.740
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
7.62
3.09
0.69
0.51
1.85
2.88
2.38
2.41
0.13
Clear \"
Hydrology Validation
Hydrology validation for the Trinity River was performed for the period 10/1/1983 through 9/30/1992. The results
are presented in Figures 8 through 11 and Table 7. The validation achieves a reasonable coefficient of model fit
efficiency, but is under on 50 percent low volume and over on seasonal volumes for summer and fall (Figure 8
through Figure 11 and Table 7).
U-20
-------
o
80000
60000
40000
20000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1982 to 9/30/1991 )
Avg Modeled Flow (Same Period)
O-82
O-83
O-84
O-85
O-86
O-87
O-88
O-89
O-90
Month
Figure 8. Mean monthly flow at USGS 08066500 Trinity River at Romayor, TX- validation period.
Avg Flow (10/1/1982 to
9/30/1991)
- Line of Equal Value
20000
15000
T3
I 10000
T3
O
ro 5000
= (j).7721x+il934.3
I R2 = 0.8388
L^......l
o
5000 10000 15000 20000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1982 to 9/30/1991)
Avg Modeled Flow (Same Period)
t
o
20000
15000
10000 -
5000 -
Average Observed Flow (cfs)
Figure 9. Seasonal regression and temporal aggregate at USGS 08066500 Trinity River at Romayor, TX •
validation period.
U-21
-------
30000
To Lower Bound
Average Monthly Rainfall (in)
-Median Observed Flow (10/1/1982 to 9/30/1991)
c
'CD
or
Figure 10. Seasonal medians and ranges at USGS 08066500 Trinity River at Romayor, TX - validation
period.
i
I
D)
5
0)
03
Q
•Observed Flow Duration (10/1/1982 to 9/30/1991 )
Modeled Flow Duration (10/1/1982 to 9/30/1991)
1000000
100000
10000
1000
10%
20% 30% 40% 50% 60%, 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90%
100%,
Figure 11. Flow exceedance at USGS 08066500 Trinity River at Romayor, TX - validation period.
U-22
-------
Table 7. Summary statistics at USGS 08066500 Trinity River at Romayor, TX - validation period
REACH OUTFLOW FROM OUTLET 3
9-Year Analysis F^riod: 10/1/1982 - 9/30/1991
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
_JpJal^f^inTulated_highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3):
Simulated Srjring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
6.54
2.76
0.74
0.92
1.18
1.64
2.79
2.48
0.41
Error Statistics
0.70
11.67
-0.63
21.78
USGS 08066500 Trinity Rv at Romayor, TX
Hydrologic Unit Code: 12030202
Latitude: 30.4252067
Longitude: -94.8507622
Drainage Area (sq-rri): 17186
Total Observed In-stream Flow:
_j£t^l_of^b^ej^«djTighe^tjm%JlOT/s^_
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume_(4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
12.35 » | 30
-16.62
2.88
3.66
30
30
20
77.98 | 50
0.471
0.431
0.760
Model accuracy increases
as E or E' approaches 1.0
6.49
2.78
0.67
0.76
1.05
1.97
2.71
2.39
0.23
Clear [
Hydrology Results for Larger Watershed
As described above, parameters determined for the Romayor gage were fully transferable to other gages in the
watershed. In addition, calibration and validation was pursued at a total of seven gages throughout the watershed.
Calibration results were acceptable at most gages (Table 8).
Results of the validation exercise are summarized in Table 9. Problems similar to those experienced on the
Romayor gage were seen at most of the tributary gages, with over-prediction of seasonal flows in summer and
under-prediction in winter and spring. However, as noted above, this is likely due to the use of land use and model
parameters that are more reflective of current conditions and is not believed to present a bar to application of the
model.
U-23
-------
Table 8. Summary statistics (percent error): all Stations - calibration period
Station
Error in total
volume:
Error in 50%
lowest flows:
Error in 10%
highest flows:
Seasonal
volume error -
Summer:
Seasonal
volume error -
Fall:
Seasonal
volume error -
Winter:
Seasonal
volume error -
Spring:
Error in storm
volumes:
Error in
summer storm
volumes:
Daily Nash-
Sutcliffe
Coefficient of
Efficiency, E:
Baseline
adjusted
coefficient
(Garrick), E':
Monthly Nash-
Sutcliffe
Coefficient of
Efficiency, E:
08049500
West Fork
Trinity River
at Grand
Prairie
-21.5
-22.96
-22.09
27.87
-2.90
-28.82
-42.02
-32.74
55.10
0.556
0.424
0.541
08051500
Clear
Creek
near
Sanger
-1.45
-68.43
4.2
169.54
20.8
-15.79
-16.05
3.55
192.89
0.3
0.380
0.584
08062000
East Fork
Trinity
near
Crandall
-4.33
-12.98
-2.26
110.48
28.55
-30.22
-6.16
-6.33
159.69
0.258
0.323
0.595
08062500
Trinity River
near Rosser
-36.72
-22.65
-30.65
44.47
-22.01
-45.18
-61.25
-27.19
116.13
0.388
0.358
0.388
08062700
Trinity River
near
Trinidad
-28.22
-15.64
-25.85
58.55
-15.66
-38.74
-46.16
-21.96
149.46
0.496
0.427
0.605
08065350
Trinity
River
near
Crockett
-13.76
-20.30
1.63
47.56
2.84
-30.16
-17.33
29.75
155.46
0.265
0.368
0.651
08066500
Trinity
River
at Romayor
-6.88
-2.33
-9.48
63.58
9.93
-24.04
-14.28
18.68
210.14
0.623
0.482
0.740
U-24
-------
Table 9. Summary statistics: all stations - validation period
Station
Error in total
volume:
Error in 50%
lowest flows:
Error in 10%
highest flows:
Seasonal
volume error -
Summer:
Seasonal
volume error -
Fall:
Seasonal
volume error -
Winter:
Seasonal
volume error -
Spring:
Error in storm
volumes:
Error in
summer storm
volumes:
Daily Nash-
Sutcliffe
Coefficient of
Efficiency, E:
Baseline
adjusted
coefficient
(Garrick), E':
Monthly Nash-
Sutcliffe
Coefficient of
Efficiency, E:
08049500
West Fork
Trinity
River
at Grand
Prairie
6.38
10.4
-4.43
104.35
29.01
-29.5
-2.74
-19.03
97.08
0.820
0.550
0.932
08051500
Clear Creek
near
Sanger
-2.4
-86.16
1.86
8.88
41.28
-35.41
5.30
1.84
42.3
0.605
0.540
0.864
08062000
East Fork
Trinity
near Crandall
12.95
22.86
13.16
55.60
82.61
-36.32
15.48
-2.03
160.45
0.367
0.268
0.732
08062500
Trinity River
near Rosser
-25.22
-8.38
-20.01
15.41
-11.38
-47.22
-30.13
-22.33
103.72
0.705
0.462
0.807
08062700
Trinity
River
near
Trinidad
-12.94
1.87
-9.75
29.18
-1.31
-41.84
-12.50
-11.06
142.79
0.626
0.455
0.833
08065350
Trinity
River
near
Crockett
-11.10
-17.16
4.38
11.0
-1.24
-41.97
0.66
18.61
59.70
0.128
0.32
0.730
08066500
Trinity
River
at Romayor
0.7
11.67
-0.63
21.78
12.35
-16.62
2.88
3.66
81.05
0.471
0.431
0.760
Water Quality Calibration and Validation
Initial calibration and validation of water quality was done on Trinity River at Romayor (USGS 08066500), using
1985-2001 for calibration and 1972-1984 for validation. As with hydrology, water quality calibration was
performed on the later period as this better reflects the land use included in the model. The start of the validation
period is constrained by data availability.
Calibration adjustments for sediment focused on the following parameters:
U-25
-------
• SPCON (linear parameters for estimating maximum amount of sediment that can be re-entrained during
channel sediment routing)
• CH_COV (channel cover factor)
• CH_EROD (channel erodibility factor)
Simulated and estimated sediment loads at the Romayor station for both the calibration and validation periods are
shown in Figure 12 and statistics for the two periods are provided separately in Table 10. The key statistic in
Table 10 is the relative percent error, which shows the error in the prediction of monthly load normalized to the
estimated load. Table 10 also shows the relative average absolute error, which is the average of the relative
magnitude of errors in individual monthly load predictions. This number is inflated by outlier months in which the
simulated and estimated loads differ by large amounts (which may be as easily due to uncertainty in the estimated
load due to limited data as to problems with the model) and the third statistic, the relative median absolute error,
is likely more relevant and shows better agreement.
TSS
-Regression Loads
•Simulated Loads
Figure 12. Fit for monthly load of TSS at USGS 08066500 Trinity River at Romayor, TX.
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 08066500 Trinity River at Romayor, TX
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
(1985-2001)
9.2%
129%
58.8%
Validation period
(1972-1984)
-17.4%
137%
56.4%
Calibration adjustments for total phosphorus and total nitrogen focused on the following parameters:
• PHOSKD (Phosphorus soil partitioning coefficient)
• RS2(benthic source rate for dissolved phosphorus in the reach [mg P/m2*day])
• RS3 (benthic source rate for NH4-N in the reach [mg N/m2*day])
• RS4 (rate coefficient for organic N settling in the reach [day-1])
U-26
-------
• RS5 (organic phosphorus settling rate in the reach [day-1])
• BC1 (rate constant for biological oxidation of NH4 to NO2 in the reach [day-1])
• BC2 (rate constant for biological oxidation of NO2 to NO3 in the reach [day-1])
• BC4 (rate constant for mineralization of organic P to dissolved P in the reach [day-1])
• MUMAX (maximum specific algal growth rate [day-1])
Results for the phosphorus simulation are shown in Figure 13 and Table 11. Results for the nitrogen simulation
are shown in Figure 14 and Table 12. The model fit is generally good.
Total P
1000
• Regression Loads
•Simulated Loads
Figure 13. Fit for monthly load of total phosphorus at USGS 08066500 Trinity River at Romayor, TX.
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 08066500 Trinity River at Romayor, TX
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1985-2001)
3.0%
108%
75.7%
Validation period
(1972-1984)
-21.58
110%
68.6%
U-27
-------
Total N
•Averaging Loads
-Simulated Loads
Figure 14. Fit for monthly load of total nitrogen at USGS 08066500 Trinity River at Romayor, TX.
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 08066500 Trinity River at Romayor, TX
Statistic
Relative Percent Error
Average Absolute Error
Median Absolute Error
Calibration period
(1985-2001)
-3.8%
107%
78.4%
Validation period
(1972-1984)
-31.9%
113%
66.7%
Water Quality Results for Larger Watershed
As with hydrology, the SWAT model parameters used to calibrate at the USGS 08066500 Trinity River at
Romayor, TX station for water quality were directly transferred to other portions of the watershed. Application of
the SWAT model without spatial adjustments resulted in relatively large errors in predicting loads and
concentrations at some stations. Summary statistics for the SWAT water quality calibration and validation at other
stations in the watershed are provided in Table 13 and Table 14.
U-28
-------
Table 13. Summary statistics for water quality at all stations - calibration period 1985-2001
Station
Relative
Percent Error
TSS Load
Relative
Percent Error
TP Load
Relative
Percent Error
TN Load
08049500
West Fork
Trinity
River at
Grand
Prairie
62.9%
38.9%
77.3%
08051500
Clear Creek
near
Sanger
98.3%
77%
83.9%
08062000
East Fork
Trinity near
Crandall
-26.4%
-186.9%
41.2%
08062500
Trinity River
near Rosser
44.9%
12.4%
63.0%
08062700
Trinity
River near
Trinidad
58.1%
9%
60.7%
08065350
Trinity
River near
Crockett
53.4%
15.8%
50.5%
08066500
Trinity
River at
Romayor
9.2%
3.0%
-3.8%
Table 14. Summary statistics for water quality at all stations - validation period 1972-1984
Station
Relative
Percent Error
TSS Load
Relative
Percent Error
TP Load
Relative
Percent Error
TN Load
08049500
West Fork
Trinity
River at
Grand
Prairie
58.1%
36.58%
60.0%
08051500
Clear Creek
near
Sanger
97.4%
50.06%
64.2%
08062000
East Fork
Trinity near
Crandall
-43.3%
-192.45%
18.8%
08062500
Trinity River
near Rosser
36.4%
14.42%
45.7%
08062700
Trinity
River near
Trinidad
55.8%
9.48%
42.4%
08065350
Trinity
River near
Crockett
54.0%
17.04%
36.9%
08066500
Trinity
River at
Romayor
-17.4%
-21.58%
-31.9%
U-29
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
Land, L.F., J.B. Moring, P.C. Van Metre, D.C. Reutter, B.J. Mahler, A.A. Shipp, and R.L. Ulery. 1998. Water
Quality in the Trinity River Basin, Texas, 1992-95 U.S. Geological Survey Circular 1171.
Tetra Tech. 1999. Improving Point Source Loadings Data for Reporting National Water Quality Indicators. Final
Technical Report prepared for U.S. Environmental Protection Agency, Office of Waste water Management,
Washington, DC, by Tetra Tech, Inc., Fairfax, VA.
Ulery, R.L., P.C. Van Metre, and A.S. Crossfield. 1993. Trinity River Basin, Texas: Water Resources Bulletin,
29(4): 685-711.
USEPA (United States Environmental Protection Agency). 2008. Using the BASINS Meteorological Database
(Version 2006). BASINS Technical Note 10. Office of Water, U.S. Environmental Protection Agency,
Washington, DC. http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf
(Accessed June, 2009).
U-30
-------
Appendix V
Model Configuration, Calibration and
Validation
Basin: Upper Colorado River (UppCol)
V-l
-------
Contents
Watershed Background V-4
Water Body Characteristics V-4
Soil Characteristics V-5
Land Use Representation V-5
Point Sources V-9
Meteorological Data V-9
Watershed Segmentation V-11
Calibration Data and Locations V-12
SWAT Modeling V-14
Assumptions V-14
Hydrology Calibration V-14
Hydrology Validation V-18
Hydrology Results for Larger Watershed V-21
Water Quality Calibration and Validation V-24
Water Quality Results for Larger Watershed V-27
References V-28
V-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes V-7
Table 2. Land use distribution for the Upper Colorado River basin (2001 NLCD) (mi2) V-8
Table 3. Major point source discharges in the Upper Colorado River basin V-9
Table 4. Precipitation stations for the Upper Colorado River watershed model V-10
Table 5. Calibration and validation locations in the Upper Colorado River basin V-13
Table 6. Summary statistics at USGS 09070500 Colorado River near Dotsero, Colorado - calibration
period V-18
Table 7. Summary statistics at USGS 09070500 Colorado River near Dotsero, Colorado - validation
period V-21
Table 8. Summary statistics (percent error): all stations - calibration period V-22
Table 9. Summary statistics (percent error): all stations - validation period V-23
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 09070500 Colorado River near Dotsero, Colorado V-25
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 09070500 Colorado River near Dotsero, Colorado V-26
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 09070500 Colorado River near Dotsero, Colorado V-27
Table 13. Summary statistics for water quality at all stations - calibration period 1992-2002 V-27
Table 14. Summary statistics for water quality at all stations - validation period 1982-1992 V-27
Figures
Figure 1. Location ofthe Upper Colorado River basin V-5
Figure 2. Land use in the Upper Colorado River basin V-6
Figure 3. Model segmentation and USGS stations utilized for the Upper Colorado River basin V-12
Figure 4. Mean monthly flow at USGS 09070500 Colorado River near Dotsero, Colorado - calibration
period V-15
Figure 5. Seasonal regression and temporal aggregate at USGS 09070500 Colorado River near Dotsero,
Colorado - calibration period V-16
Figure 6. Seasonal medians and ranges at USGS 09070500 Colorado River near Dotsero, Colorado -
calibration period V-16
Figure 7. Flow exceedance at USGS 09070500 Colorado River near Dotsero, Colorado - calibration
period V-17
Figure 8. Mean monthly flow at USGS 09070500 Colorado River near Dotsero, Colorado - validation
period V-19
Figure 9. Seasonal regression and temporal aggregate at USGS 09070500 Colorado River near Dotsero,
Colorado -validation period V-19
Figure 10. Seasonal medians and ranges at USGS 09070500 Colorado River near Dotsero, Colorado -
validation period V-20
Figure 11. Flow exceedance at USGS 09070500 Colorado River near Dotsero, Colorado - validation
period V-20
Figure 12. Fit for monthly load of TSS at USGS 09070500 Colorado River near Dotsero, Colorado V-24
Figure 13. Fit for monthly load of total phosphorus at USGS 09070500 Colorado River near Dotsero,
Colorado V-26
Figure 14. Fit for monthly load of total nitrogen at USGS 09070500 Colorado River near Dotsero,
Colorado V-26
V-3
-------
The Upper Colorado River basin was selected as one of the 15 non-pilot application watersheds for the 20
Watershed study. Watershed modeling for the non-pilot areas is accomplished using the SWAT model only, and
model calibration and validation results are presented in abbreviated form.
Water Body Characteristics
The Upper Colorado River basin model area has a drainage area of about 17,800 mi2 and contains 12 HUCSs
within HUC 1401 and 1402. All except 100 mi2 of this area is in Colorado (Figure 1).
The Colorado River and its tributaries originate in the mountains of central Colorado and flow about southwest
into Utah. The Continental Divide marks the eastern and southern boundary of the basin, with altitudes over
14,000 ft. Topography in the western part of the basin generally consists of high plateaus bordered by steep cliffs
along the valleys, and the lowest altitude (4,300 ft) is near the Colorado-Utah border. The basin is divided almost
equally into two physiographic provinces: the Southern Rocky Mountains in the east and the Colorado Plateau in
the west (USGS, 2006; Apodaca et al., 1996).
Because of large changes in altitude, the climate in the basin varies from alpine conditions in the east to semiarid
in the west. Mean annual temperatures range from as low as 32.8° F in Gunnison County near the Continental
Divide to as high as 54.1° F near Grand Junction, Colorado. Precipitation in the basin ranges from more than 40
inches per year in the eastern mountainous regions to less than 10 inches per year in the lower altitude western
regions. Mountain areas receive most of their precipitation during the winter months when accumulation of snow
can exceed an annual average of 100 inches.
The Upper Colorado River basin is largely rural. Rangeland and forest occupy about 88 percent of the basin.
Livestock (sheep and cattle) use large areas of rangeland for foraging. Forest land that includes most of the
mountain and plateau areas is used for some commercial lumber production. Large parts of the watershed are set
aside for recreational use, including all or parts of 4 National Park Service areas, 5 National Forests and numerous
wilderness areas, 11 state parks, numerous State Wildlife Management areas, and 17 ski areas. Mining activities
are also an important land use and have included the extraction of metals and energy fuels.
Less than 2 percent of the land area is developed. The largest population center is Grand Junction (population less
than 60,000 in 2010), which is located at the confluence of the Colorado and Gunnison Rivers. The larger cities in
the basin are located predominantly near agricultural lands or in mountain recreational communities. Agricultural
activities (about 4 percent of the area) include production of crops such as alfalfa, fruits, grains, hay, and
vegetables. Little crop production is possible without irrigation because of the semiarid climate. Irrigated lands
are predominantly in river valleys or low-altitude regions where the water is supplied by an extensive system of
canals and ditches.
The natural hydrology of the Upper Colorado River basin has been considerably altered by water development,
which includes numerous reservoirs and diversions. In the watershed, there are 9 major interbasin water transfers,
7 major water diversions, 9 major reservoirs, and 10 major municipal discharges. The interbasin water transfers
provide supplementary irrigation and municipal water supplies to the South Platte, Arkansas, and Rio Grande
drainages. About 25 percent of the interbasin water transfers are to the South Platte watershed for the municipal
water supply for the Denver metropolitan area. Most of the water used in the watershed comes from surface water
sources. Groundwater sources account for less than 1 percent of the water used. Irrigation accounts for about 97
percent of off-stream water use. Besides off-stream water uses, there are in-stream water uses such as
hydroelectric power generation.
V-4
-------
Hydrography
•B Water (Nat. Atlas Cataso
US Census Populated Places
^H Municipalities ;pop > 50,000)
| | County Boundaries
I I Watershed with HUCBs
Colorado Headwaters
(14010001) . V
Parachute-
Roan
(14010006)
Colorado Headwaters
Plateau
(14010005)
Highlands
Ranch
Roaring Fork
(14010004)
Lower Gunnison
(14020(305)
7V
North Fork
Gunnison
(14020004)
East-
Taylor
(14020001)
Upper Gunnison
, (14020002)
Uncompahgre
(14020006)
GCRP Model Areas - Upper Colorado River Basin
Base Map
Figure 1. Location of the Upper Colorado River basin.
Soil Characteristics
Soils in the watershed, as described in STATSGO soil surveys, fall primarily into hydrologic soil groups (HSGs)
B (moderately high infiltration capacity) and C (moderate infiltration capacity). SWAT uses information drawn
directly from the soils data layer to populate the model.
Land Use Representation
Land use/cover in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage (Figure
2). NLCD land cover classes were aggregated according to the scheme shown in Table 1 for representation in the
GCRP model. SWAT uses the built-in hydrologic response unit (HRU) overlay mechanism in the ArcSWAT
interface. SWAT HRUs are formed from an intersection of land use and STATSGO major soils. The distribution
of land use in the watershed is summarized in Table 2.
V-5
-------
Legend
Hydrography
= Interstate
I, | County Boundaries
2001 NLCD Land Use
I | Op en water
| Developed, open space
| Developed, low intensity
j^B Developed, medium intensity
^^| Developed, high intensity
I | Barren land
^^| Deciduous forest
^H Evergreen forest
I I Mixed forest
| | Scrub/shrub
^ Grassland/herbaceous
I | Pasture/hay
I | Cultivated crops
I I Woody wetlands
^ Emergent herbaceous wetlands
_
GCRP Model Areas - Upper Colorado River Basin
Land Use Map
TETRATECH
Figure 2. Land use in the Upper Colorado River basin.
V-6
-------
Table 1. Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area usually
accounted for as reach area
Deciduous
Evergreen
Mixed
Emergent & woody wetlands
Aquatic bed wetlands (not
emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
V-7
-------
Table 2. Land use distribution for the Upper Colorado River basin (2001 NLCD) (mi )
HUC8
watershed
Colorado
Headwaters
14010001
Blue
14010002
Eagle
14010003
Roaring Fork
14010004
Colorado
Headwaters-
Plateau
14010005
Parachute-
Roan
14010006
East-Taylor
14020001
Upper
Gunnison
14020002
Tomichi
14020003
North Fork
Gunnison
14020004
Lower
Gunnison
14020005
Uncompahgre
14020006
Total
Open
water
26.4
10.0
1.7
3.6
12.7
0.0
3.6
17.1
0.5
1.2
5.9
2.3
85.0
Snow/Ice
82.2
26.5
6.2
3.8
0.0
0.0
0.9
0.0
0.1
0.0
0.0
0.0
119.7
Developed9
Open
space
12.7
8.1
8.3
9.5
27.4
0.8
3.3
10.0
5.4
5.6
8.0
12.0
111.1
Low
density
5.8
6.2
8.5
6.8
42.9
0.9
0.9
4.2
1.2
2.6
11.2
13.2
104.4
Medium
density
0.8
2.2
2.6
2.3
13.5
0.1
0.2
0.8
0.4
0.3
2.7
4.1
30.0
High
density
0.1
0.2
0.2
0.2
3.1
0.0
0.0
0.0
0.1
0.1
0.4
1.0
5.3
Barren
land
64.7
66.7
48.8
126.0
46.4
44.7
61.8
96.4
17.4
26.2
23.8
52.9
675.8
Forest
1,597.3
334.9
563.4
876.2
1,544.8
353.4
441.5
1,312.2
544.5
655.2
821.2
576.8
9,621.5
Shru bland
986.8
205.7
306.6
361.2
1,167.6
282.1
215.4
888.0
488.5
207.9
664.2
282.4
6,056.4
Pasture/Hay
54.5
5.1
10.2
30.8
119.8
22.6
5.6
57.0
30.8
57.2
61.9
109.1
564.6
Cultivated
0.3
0.2
0.3
0.4
94.4
0.2
0.0
0.5
0.0
1.6
44.7
54.5
197.1
Wetland
70.2
17.2
15.6
33.9
46.0
2.5
33.7
25.1
13.9
11.1
18.7
6.3
294.3
Total
2,901.9
683.1
972.5
1,454.6
3,118.7
707.3
766.9
2,411.3
1,102.7
969.0
1,662.7
1,114.6
17,865.2
aThe percent imperviousness applied to each of the developed land uses is as follows:
density (87.41%).
open space (9.78%), low density (31.89%), medium density (60.48%), and high
V-8
-------
Point Sources
There are numerous point source discharges in the watershed. Only the major dischargers, generally defined as
those with a design flow greater than 1 MGD are included in the simulation (Table 3). The major dischargers are
represented at long-term average flows, without accounting for changes over time or seasonal variations.
Table 3. Major point source discharges in the Upper Colorado River basin
NPDES ID
CO0040053
CO0039641
CO0039624
CO0035394
CO0020516
CO0023086
CO0026387
CO0020451
CO0029955
CO0045420
CO0000230
CO0037681
CO0040142
CO0024431
CO0037311
CO0021369
CO0042480
Name
MESA CO./GRAND JUNCTION - CITY
DELTA, CITY OF
MONTROSE, CITY OF
U.S. MOLYCORP.
GLENWOOD SPRINGS, CITY OF
SNOWMASS WATER & SAN DISTRICT
ASPEN CONSOLIDATED SAN DISTRCT
FRISCO SANITATION DISTRICT
SUMMIT CO BOARD OF COMMISS
IOWA HILL WATER RECLAMATION
CLIMAX MOLYBDENUM COMPANY
THREE LAKES WATER & SAN DIST
FRASER SANITATION DISTRICT
EAGLE RIVER WATER & SAN. DIST.
EAGLE RIVER WATER & SAN. DIST.
EAGLE RIVER WATER & SAN. DIST.
CBS OPERATIONS, INC.
Design flow
(MGD)
12.500
3.800
3.200
0.000
2.300
1.600
1.870
1.700
2.600
1.500
0.000
2.000
2.000
4.300
12.500
3.800
3.200
Observed flow
(MGD)
(1991-2006 average)
7.484
1.060
1.709
0.347
0.875
0.792
1.683
0.577
0.631
0.683
1.385
0.419
0.729
2.195
7.484
1.060
1.709
Most of these point sources have reasonably complete monitoring for total suspended solids (TSS). Assumptions
were made for total nitrogen and total phosphorus depending upon the type of facility. The point sources were
initially represented in the model with the median of reported values for total phosphorus, total suspended solids
and total nitrogen.
Meteorological Data
The required meteorological time series data for the GCRP SWAT simulations are precipitation and air
temperature. The GCRP simulations do not include water temperature and uses a degree-day method for
snowmelt. SWAT estimates Penmann-Monteith potential evapotranspiration using a statistical weather generator
for inputs other than temperature and precipitation. These meteorological time series are drawn from the
BASINS4 Meteorological Database (USEPA 2008), which provides a consistent, quality-assured set of
nationwide data with gaps filled and records disaggregated. Scenario application requires simulation over 30
years, so the available stations are those with a common 30-year period of record (or one that can be filled from
an approximately co-located station) that covers the year 2002. A total of 47 precipitation stations were identified
for use in the Upper Colorado River watershed model with a common period of record of 10/1/1972-9/30/2003
V-9
-------
(Table 4). Temperature records are sparser; where these are absent temperature is taken from nearby stations with
an elevation correction.
Table 4. Precipitation stations for the Upper Colorado River watershed model
COOP ID
050183
050214
050797
050843
050909
051071
051186
051609
051660
051713
051772
051959
052281
053146
053246
053359
053488
053489
053496
053500
053530
053592
053662
053951
054664
054734
055507
055722
056012
056203
056266
Name
ALLENSPARK 2 NNW
ALTENBERN
BLUE MESA LAKE
BOULDER 2
BRECKENRIDGE
BUENAVISTA2S
CABIN CREEK
CIMARRON
CLIMAX
COCHETOPA CREEK
COLORADO NATL MONUMENT
CRESTED BUTTE
DILLON 1 E
FRUITA 1 W
GATEWAY
GLENWOOD SPGS #2
GRAND JUNCTION WALKER
GRAND JUNCTION 6 ESE
GRAND LAKE 1 NW
GRAND LAKE 6 SSW
GRANT
GREEN MT DAM
GUNNISON 3 SW
HERMIT 7 ESE
KREMMLING
LAKE CITY
MEREDITH
MONTROSENO2
NORWOOD
OURAY
PALISADE
Latitude
40.1881
39.5008
38.4668
40.0340
39.4862
38.8247
39.6553
38.4443
39.3672
38.4462
39.1014
38.8743
39.6262
39.1645
38.6825
39.5181
39.1342
39.0423
40.2669
40.1851
39.4608
39.8790
38.5250
37.7718
40.0575
38.0248
39.3619
38.4858
38.1318
38.0207
39.1136
Longitude
-105.5010
-108.3790
-107.1670
-105.2810
-106.0430
-106.1270
-105.7080
-107.5590
-106.1890
-106.7610
-108.7330
-106.9760
-106.0350
-108.7340
-108.9720
-107.3170
-108.5400
-108.4660
-105.8320
-105.8660
-105.6780
-106.3330
-106.9680
-107.1090
-106.3680
-107.3140
-106.7420
-107.8790
-108.2860
-107.6680
-108.3500
Temperature
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Elevation (m)
2504
1731
2316
1650
2920
2422
3054
2102
3442
2438
1762
2698
2763
1373
1387
1792
1481
1451
2658
2526
2644
2359
2329
2758
2274
2643
2385
1763
2140
2390
1466
V-10
-------
COOP ID
056306
056524
057031
057337
057460
057618
057656
057848
057936
058064
058184
058204
058501
058560
059175
059265
Name
PAONIA 1 SW
PLACERVILLE
RIFLE
SAGUACHE
SARGENTS
SHOSHONE
SILVERTON
SPICER
STEAMBOAT SPRINGS
SUGARLOAF RESERVOIR
TAYLOR PARK
TELLURIDE4WNW
TWIN LAKES RES
URAVAN
WINTER PARK
YAM PA
Latitude
38.8523
37.9944
39.5329
38.0858
38.4040
39.5717
37.8193
40.4725
40.4884
39.2495
38.8184
37.9492
39.0937
38.3762
39.8904
40.1562
Longitude
-107.6230
-108.0210
-107.7920
-106.1440
-106.4230
-107.2260
-107.6650
-106.4470
-106.8230
-106.3710
-106.6080
-107.8730
-106.3510
-108.7420
-105.7610
-106.9090
Temperature
X
X
X
X
X
X
X
X
X
X
X
X
Elevation (m)
1701
2301
1661
2347
2579
1807
2828
2556
2094
2968
2806
2643
2806
1527
2775
2405
Watershed Segmentation
The Upper Colorado River basin was divided into 89 subwatersheds for the purposes of modeling (Figure 3).
Colorado River near Dotsero at USGS 09070500 was chosen for initial calibration. The model encompasses the
complete watershed and does not require specification of any upstream boundary conditions for application.
V-ll
-------
09070000
09070500
09085100
09085000
09163500
09152500
USGS gages
Hydrography
Interstate
^H Water (Nat. Atlas Dataset)
US Census Populated Places
I I County Boundaries
Model Subbasins
GCRP Model Areas - Upper Colorado River Basin
Model Segmentation
Figure 3. Model segmentation and USGS stations utilized for the Upper Colorado River basin.
Calibration Data and Locations
The specific site chosen for initial calibration was the Colorado River near Dotsero, CO (USGS 09070500) a flow
and water quality monitoring location. The USGS gage located at Colorado River near Dotsero was selected
because there is a good set of flow and water quality data available and the watershed lacks major point sources
and impoundments. Additional calibration and validation was pursued at multiple locations (Table 5). Parameters
derived from the initial calibration were not fully transferable to other portions of the Upper Colorado River
basin, and additional calibration was conducted at multiple gage locations.
V-12
-------
Table 5. Calibration and validation locations in the Upper Colorado River basin
Station name
Colorado River near Kremmling, CO
Eagle River below Gypsum, CO
Colorado River near Dotsero, CO
Colorado river below Glenwood Springs, CO
Roaring fork River at Glenwood Springs, CO
Colorado River near Cameo, CO
Gunnison River near Gunnison, CO
Tomichi Creek at Gunnison, CO
Gunnison River near Grand Junction, CO
Colorado River near Utah State Line, CO
USGS ID
09058000
09070000
09070500
09085100
09085000
09095500
09114500
09119000
09152500
09163500
Drainage area
(mi2)
2,379
945
4,390
6,014
1,451
7,986
1,011
1,061
7,928
17, 843
Hydrology
calibration
X
X
X
X
X
X
X
X
X
X
Water quality
calibration
X
X
X
X
X
X
X
X
The model hydrology calibration period was set to Water Years 1992-2002 (within the 32-year period of record
for modeling). Hydrologic validation was then performed on Water Years 1982-1992. Water quality calibration
used calendar years 1992-2002, while validation used 1982-1992.
V-13
-------
SWAT Modeling
Assumptions
The Upper Colorado River basin is comprised of the areas drained by the Colorado River and Gunnison River.
There are a number of reservoirs and diversion structures on Colorado and Gunnison rivers. Only major reservoirs
in the basin, namely, Green Mountain, Blue Mesa and Morrow Point, were represented in the model. The Green
Mountain reservoir is located on the Colorado River, while the Blue Mesa and Morrow Point reservoirs are
located on the Gunnison River. Pertinent reservoir information including surface area and storage at principal
(normal) and emergency spillway levels for the reservoir were obtained from the Colorado Bureau of
Reclamation. The SWAT model provides four options to simulate reservoir outflow: measured daily outflow,
measured monthly outflow, average annual release rate for uncontrolled reservoir, and controlled outflow with
target release. Keeping in view, the GCRP climate change impact evaluation application to future climate
scenarios, it was assumed that the best representation of the reservoir was to simulate it without supplying time
series of outflow records. The target release approach was used for the Green Mountain reservoir. Due to lack of
detailed data annual average release approach was used fort the Blue Mesa and Morrow Point reservoirs.
Hydrology Calibration
A spatial calibration approach was adopted for GCRP-SWAT modeling for the Upper Colorado River basin. A
systematic adjustment of parameters has been adopted and some adjustments are applied throughout the basin.
Most of the calibration efforts were geared towards getting a closer match between simulated and observed flows
at the outlet of calibration focus area.
Land Use/Soil/Slope Definition
A 5/10/5 percent threshold was used for land use/soil/slope in the SWAT model while defining the HRUs. Urban
land use classes were exempted from the HRU overlay thresholds.
The calibration focus area includes twenty-eight subwatersheds and is generally representative of the general land
use characteristics of the overall watershed. The parameters were adjusted within the practical range to obtain
reasonable fit between the simulated and measured flows in terms of Nash-Sutcliffe modeling efficiency and the
high flow and low flow components as well as the seasonal flows.
The water balance of whole Upper Colorado River basin predicted by the SWAT model over the 32-year
simulation period is as follows:
PRECIP = 418.2 MM
SNOW FALL = 177.21 MM
SNOW MELT = 136.22 MM
SUBLIMATION = 38.94 MM
SURFACE RUNOFF Q = 15.54 MM
LATERAL SOIL Q = 80.11 MM
TILE Q = 0.00 MM
GROUNDWATER (SHAL AQ) Q = 44.64 MM
REVAP (SHAL AQ => SOIL/PLANTS) = 0.00 MM
DEEP AQ RECHARGE = 5.87 MM
TOTAL AQ RECHARGE = 50.51 MM
TOTAL WATER YLD = 127.56 MM
V-14
-------
PERCOLATION OUT OF SOIL = 38.88 MM
ET = 332.7 MM
PET = 1075.9MM
TRANSMISSION LOSSES = 12.72 MM
Hydrologic calibration adjustments focused on the following parameters:
• CN2 (initial SCS runoff curve number for moisture condition II)
• ESCO (soil evaporation compensation factor)
• SURLAG (surface runoff lag coefficient)
• SOL_AWC (available water capacity of the soil layer, mm water/mm of soil)
• ALPHA_BF (baseflow alpha factor, days)
• GW_DELAY (groundwater delay time, days)
• GWQMIN (threshold depth of water in the shallow aquifer required for return flow to occur,
mm)
• GW_REVAP (groundwater "revap" coefficient)
• CH_N1 (Manning's "n" value for tributary channels)
• CH_N2 (Manning's "n" value for main channels)
• CH_K1 (Effective hydraulic conductivity in tributary channel alluvium)
• CH_K2 (Effective hydraulic conductivity in main channel alluvium)
• SFTMP (Snowfall temperature)
• SMTMP (Snowmelt base temperature)
• SMFMX (Maximum melt rate for snow during the year)
• SMFMN (Minimum melt rate for snow during the year)
Calibration results for the Colorado River near Dotsero are summarized in Figure 4, Figure 5, Figure 6,
Figure 7 and Table 6.
15000
,40000
I
o
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002 )
•Avg Modeled Flow (Same Period)
2.5 -
"5000
0.5
O-92
A-94
O-95
A-97
O-98
A-00
O-01
Month
Figure 4. Mean monthly flow at USGS 09070500 Colorado River near Dotsero, Colorado - calibration
period.
V-15
-------
8000
Avg Flow (10/1/1992 to 9/30/2002)
•Line of Equal Value
•Best-Fit Line
y = 1.0101x +148.34
R2 = 0.9315
8000
6000
4000
2000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1992 to 9/30/2002)
-Avg Modeled Flow (Same Period)
2000 4000 6000 8000
Average Observed Flow (cfs)
Figure 5. Seasonal regression and temporal aggregate at USGS 09070500 Colorado River near Dotsero,
Colorado - calibration period.
• Observed (25th, 75th) Average Monthly Rainfall (in) -Median Observed Flow (10/1/1992 to 9/30/2002) •Modeled (Median, 25th
9000
10 11 12 1 23456789
Figure 6. Seasonal medians and ranges at USGS 09070500 Colorado River near Dotsero, Colorado -
calibration period.
V-16
-------
100000
10000
D)
>, 1000
ro
Q
100
•Observed Flow Duration (10/1/1992 to 9/30/2002 )
Modeled Flow Duration (10/1/1992 to 9/30/2002 )
10%
20% 30% 40% 50% 60% 70% 80%
Percent of Time that Flow is Equaled or Exceeded
90%
100%
Figure 7. Flow exceedance at USGS 09070500 Colorado River near Dotsero, Colorado - calibration
period.
V-17
-------
Table 6. Summary statistics at USGS 09070500 Colorado River near Dotsero, Colorado - calibration
period
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET(S) 58, 59
10-Year Analysis Period: 10/1/1992 - 9/30/2002
Flow/volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9):
Simulated Fall Flow Volume (months 10-12):
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
6.97
2.27
1.49
2.18
1.09
0.65
3.04
0.69
0.23
Error Statistics
8.18
1.31
-7.51
29.83
Observed Flow Gage
USGS 09070500 COLORADO RIVER NEAR DOTSERO, CO
Hydrologic Unit Code: 14010001
Latitude: 39.6446111
Longitude: -107.0780139
Drainage Area (sq-mi): 4394
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow Volume (10-12):
Observed Winter Flow Volume (1-3):
Observed Spring FlowVolume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
27.91 » [ 30
-12.66
-3.78
-29.75
0.27
0.829
0.580
0.864
30
30 _
50
Model accuracy increases
as E or E' approaches 1.0
6.44
2.46
1.47
1.68
0.85
0.74
3.16
0.98
0.23
Clear
Hydrology Validation
Hydrology validation for Saco River was performed for the period 10/1/1982 through 9/30/1992. The validation
achieves a moderately high coefficient of model fit efficiency, but is over on summer flow volumes (Table 8,
Table 9, Table 10, Table 1 land Table 7).
V-18
-------
15000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1982 to 9/30/1992 )
-Avg Modeled Flow (Same Period)
O-82
Figure 8. Mean monthly flow at USGS 09070500 Colorado River near Dotsero, Colorado - validation
period.
• Avg Flow (10/1 /1982 to 9/30/1992)
• Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1982 to 9/30/1992)
•Avg Modeled Flow (Same Period)
8000
6000
o
T3
O
foo
0
y = q.971x + 85.342
R2 = 0.9308
<
• .
,A
> ^r
•
,/•
*•''
8000
6000
4000
2000
2000 4000 6000
Average Observed Flow (cfs)
8000
10 11 12 1 2
3456
Month
Figure 9. Seasonal regression and temporal aggregate at USGS 09070500 Colorado River near Dotsero,
Colorado - validation period.
V-19
-------
• Observed (25th, 75th) Average Monthly Rainfall (in) -Median Observed Flow (10/1/1992 to 9/30/2002) Modeled (Median, 25th, 75th)
9000
10 11 12 1
Figure 10. Seasonal medians and ranges at USGS 09070500 Colorado River near Dotsero, Colorado -
validation period.
•Observed Flow Duration (10/1/1982 to 9/30/1992 )
Modeled Flow Duration (10/1/1982 to 9/30/1992 )
100000
100
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 11. Flow exceedance at USGS 09070500 Colorado River near Dotsero, Colorado - validation
period.
V-20
-------
Table 7. Summary statistics at USGS 09070500 Colorado River near Dotsero, Colorado - validation
period.
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET(S) 58, 59
10-Year Analysis Period: 10/1/1982 - 9/30/1992
Flow/volumes are (inches/year) for upstream drainage area
Observed Flow Gage
USGS 09070500 COLORADO RIVER NEAR DOTSERO, CO
Hydrologic Unit Code: 14010001
Latitude: 39.6446111
Longitude: -107.0780139
Drainage Area (sq-mi): 4394
Total Simulated In-stream Flow:
7.03
Total Observed In-stream Flow:
6.97
Total ofsimulated highest_1p% flows_
Total of Simulated lowest 50% flows:
2.19
1.48
_Total of Observed^highestJ0_%_ flows:
Total of Observed Lowest 50% flows:
2.63
1.66
^Simulated Summer Flow_Volurne ^months 7-9) :
Observed Summer Flow Volume (7-9):
1.84
1.08
Simulated_Winter Flqw_yolurne (months 1_-3):_
_Sirnulated Spri ng_Flow_ Vplume_(months_4-_6):_
0.64
3.10"
Jpbseryed Winter_Flp\/vVolume_(1_-3):
Obse_rved_Spring Flow Volume
0.95
0.82
3.36
Total Simulated Storm Volume:
0.69
Total Observed Storm Volume:
0.98
Simulated Summer Storm Volume (7-9):
0.20
Observed Summer Storm Volume (7-9):
0.24
Errors (Simulated-Observed)
Error Statistics
Recommended Criteria
Error in total volume:
.Error jn_50°/<^ lowest Jlqws:_
ErrpMn 1_0% jiighestjlows:_
Seasonal ^ojume_ejrorj Summer:
_Seasona^ volume error_ -_Fall:_
Seasonal volume error - Winter:
_Seasonal volume error_-_Sprini
Jrrqr]n_stprm_ volumes:
Error in summer storm volumes:
50
_Nash-Suteliffe Cqefficjent of_Efficie_ncy,_&
Baseline adjusted coefficient (Garrick), E':
0.780
0.545
Model accuracy increases
as E or E' approaches 1.0
Monthly NSE
0.819
Hydrology Results for Larger Watershed
As described above, parameters determined for the gage at Colorado River at Dotsero were initially transferred to
other gages in the watershed. However, changes to subwatershed level parameters were required to fit the model
to the observed flows. In all, calibration and validation was pursued at a total often gages throughout the
watershed. Results of the calibration and validation exercise are summarized in Table 8 and Table 9, respectively.
Calibration and validation results were acceptable at most gages.
V-21
-------
Table 8. Summary statistics (percent error): all stations - calibration period
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient of
Efficiency, E:
Monthly Nash-Sutcliffe Coefficient of
Efficiency, E:
09058000
1.25
-10.98
4.73
2.01
8.41
-21 .75
5.65
-7.91
2.53
0.484
0.636
09070000
8.77
-19.88
3.06
33.81
13.61
-28.40
2.88
-3.45
42.75
0.812
0.904
09085100
-1.09
-9.84
-6.45
15.28
-1.55
-11.66
-8.73
8.00
28.46
0.847
0.912
09085000
8.13
-2.04
-2.62
29.89
18.07
-12.04
-2.31
-18.68
16.84
0.877
0.922
09095500
1.99
-3.28
-14.74
28.02
21.42
-18.43
-12.04
-32.12
0.28
0.858
0.892
09114500
-0.20
6.02
3.27
8.52
29.01
5.64
-12.12
17.03
6.25
0.638
0.790
09119000
3.22
-23.16
13.39
34.72
4.60
-48.21
2.18
-27.67
-11.96
0.678
0.762
09152500
3.82
3.70
-9.13
30.15
-2.68
-18.15
1.02
6.01
43.58
0.629
0.694
09163500
4.43
7.11
-13.89
39.13
13.70
-19.72
-8.47
-29.66
-7.11
0.817
0.843
V-22
-------
Table 9. Summary statistics (percent error): all stations - validation period
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient of
Efficiency, E:
Monthly Nash-Sutcliffe Coefficient of
Efficiency, E:
09058000
-6.24
-20.25
-6.85
-4.31
0.85
-27.64
-3.48
-8.73
-11.69
0.504
0.586
09070000
1.02
-16.89
-4.81
9.24
1.21
-27.33
1.09
-7.96
5.33
0.795
0.882
09085100
-3.72
-10.53
-9.36
8.19
-9.61
-13.61
-6.83
2.88
17.51
0.847
0.887
09085000
1.29
-8.69
-12.98
20.69
8.36
-17.52
-7.04
-24.74
-2.57
0.853
0.914
09095500
-3.41
-3.35
-24.88
20.15
16.70
-19.76
-16.89
-35.18
-20.98
0.796
0.834
09114500
-9.88
-9.66
-5.91
2.80
10.92
-16.02
-20.52
-6.67
-22.07
0.440
0.639
09119000
0.51
-15.95
-8.04
35.99
3.17
-29.15
-8.20
-37.92
-25.04
0.480
0.525
09152500
1.76
9.77
-19.66
39.80
3.70
-27.50
-4.02
1.83
41.16
0.554
0.618
09163500
-3.29
4.74
-27.78
30.79
10.27
-25.03
-16.84
-33.40
-13.83
0.701
0.739
V-23
-------
Water Quality Calibration and Validation
Initial calibration and validation of water quality was done at USGS 09070500, Colorado River near Dotsero from
water years 1995 to 2002, due to limited water quality data. Subject to the availability of water quality data for the
other gages, 1992-2002 was adopted as the calibration period and 1982-1992 was adopted as the validation
period. As with hydrology, calibration was performed on the later period as this better reflects the land use
included in the model.
Calibration adjustments for sediment focused on the following parameters:
• SPCON (linear parameter for estimating maximum amount of sediment that can be re-entrained during
channel sediment routing)
• SPEXP (exponential parameter for estimating maximum amount of sediment that can be re-entrained
during channel sediment routing)
• CH_COV (channel cover factor)
• CH_EROD (channel erodibility factor)
• USLE_P (USLE support practice factor)
Simulated and estimated sediment loads at the Colorado River station near Dotsero for both the calibration and
validation periods are shown in Figure 12 and statistics are provided separately in Table 10. The key statistic in
Table 10 is the relative percent error, which shows the error in the prediction of monthly load normalized to the
estimated load. Table 10 also shows the relative average absolute error, which is the average of the relative
magnitude of errors in individual monthly load predictions. This number is inflated by outlier months in which the
simulated and estimated loads differ by large amounts (which may be as easily due to uncertainty in the estimated
load due to limited data as to problems with the model) and the third statistic, the relative median absolute error,
is likely more relevant and shows better agreement.
TSS
1,000,000
100,000
10,000
1,000
-Regression Loads
•Simulated Loads
100
* X/ <' X/ <^ X/
o* ^s o* V^ 0°
Figure 12. Fit for monthly load of TSS at USGS 09070500 Colorado River near Dotsero, Colorado.
V-24
-------
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 09070500 Colorado River near Dotsero, Colorado
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration/validation period
(1995-2002)
0.4%
53.7%
21.6%
Calibration adjustments for total phosphorus and total nitrogen focused on the following parameters:
• RHOQ (algal respiration rate at 20° C)
• PHOSKD (phosphorus soil partitioning coefficient)
• PSP (phosphorus availability index)
• RSI (Local algal settlement rate in the reach at 20° C)
• AL1 (Fraction of algal biomass that is nitrogen)
• AL2 (Fraction of algal biomass that is phosphorus)
• MUMAX (Rate of oxygen uptake per unit NO2-N oxidation at 20° C)
• RHOQ (Algal respiration rate at 20° C)
• RS2 (benthic source rate for dissolved P in the reach at 20° C)
• RS3 (Benthic source rate for NFLpN in the reach at 20° C)
• RS5 (organic P settling rate in the reach at 20° C)
• BC4 (rate constant for mineralization of organic P to dissolved P in the reach at 20° C)
• RS4 (rate coefficient for organic N settling in the reach at 20° C)
• CH_ONCO (Channel organic nitrogen concentration)
• CH_OPCO (Channel organic phosphorus concentration)
• SDNCO (Denitrification threshold water content)
• CDN (Denitrification exponential rate constant)
Results for the phosphorus simulation are shown in Figure 13 and Table 11. Results for the nitrogen simulation
are shown in Figure 14 and Table 12. The model fit is generally acceptable.
V-25
-------
Total P
1000
100
-Regression Loads
-Simulated Loads
.
cP
.
o°
< . , . , <
cP v8 o° ^ o° ^ cP ^ o° ^
Figure 13. Fit for monthly load of total phosphorus at USGS 09070500 Colorado River near Dotsero,
Colorado.
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 09070500 Colorado River near Dotsero, Colorado
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration/validation period
(1995-2002)
47.4%
75.9%
23.8%
Total N
-Averaging Loads
-Simulated Loads
o°
< < *
d5" ^ o° ^ d3" ^ o°
< v
d5"
Figure 14. Fit for monthly load of total nitrogen at USGS 09070500 Colorado River near Dotsero,
Colorado.
V-26
-------
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 09070500 Colorado River near Dotsero, Colorado
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration/Validation period
(1995-2002)
15.1%
52.2%
32.4%
Water Quality Results for Larger Watershed
As with hydrology, a spatial calibration approach was adopted. SWAT model parameters for water quality
derived from calibrations performed at the USGS gage at Colorado River near Dotsero were transferred to other
portions of the watershed with necessary changes to subbasin level parameters. Summary statistics for the SWAT
water quality calibration and validation at other stations in the watershed are provided in Table 13 and Table 14.
Table 13. Summary statistics for water quality at all stations - calibration period 1992-2002
Station
Relative Percent Error TSS
Load
Relative Percent Error TP
Load
Relative Percent Error TN
Load
09058000
14.1
29.7
09085000
(1996-
2002)
13.1
-25.9
09095500
33.3
80.1
-22.0
09114500
(1995-
2002)
4.9
-27.5
-37.4
09119000
(1995-
2002)
19.3
28.7
17.9
09152500
25.1
-9.7
-42.9
09163500
43.9
60.6
-60.9
Table 14. Summary statistics for water quality at all stations - validation period 1982-1992
Station
Relative Percent Error TSS Load
Relative Percent Error TP Load
Relative Percent Error TN Load
09058000
(1989-1992)
-26.4
7.7
09085000
09095500
47.9
84.8
-11.5
09114500
09119000
09152500
36.9
-1.2
-38.6
09163500
60.2
66.8
-48.9
V-27
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
Apodaca, L.E., N.E. Driver, V.C. Stephens, and N.E. Spahr. 1996. Environmental Setting and Implications on
Water Quality, Upper Colorado River Basin, Colorado and Utah. U.S. Geological Survey. Water-Resources
Investigations Report 95-4263. Denver, Colorado.
USEPA. 2008. Using the BASINS Meteorological Database (Version 2006). BASINS Technical Note 10.
Office of Water, U.S. Environmental Protection Agency, Washington, DC.
http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf (Accessed June,
2009).
USGS (United States Geological Survey). 2006. National Water Quality Assessment (NAWQA) Program: Upper
Colorado River Basin Study Unit, http://co.water.usgs.gov/nawqa/ucol/ (Accessed June, 2009).
V-28
-------
Appendix W
Model Configuration, Calibration and
Validation
Basin: Powder and Tongue Rivers
(PowTon)
W-l
-------
Contents
Watershed Background W-4
Water Body Characteristics W-4
Soil Characteristics W-6
Land Use Representation W-6
Point Sources W-10
Meteorological Data W-10
Watershed Segmentation W-11
Calibration Data and Locations W-13
SWAT Modeling W-14
Assumptions W-14
Hydrology Calibration W-14
Hydrology Validation W-18
Hydrology Results for Larger Watershed W-21
Water Quality Calibration and Validation W-23
Water Quality Results for Larger Watershed W-26
References W-27
W-2
-------
Tables
Table 1. Aggregation of NLCD land cover classes W-8
Table 2. Land use distribution for the Powder and Tongue River basin (2001 NLCD) (mi2) W-9
Table 3. Major point source discharges in the Powder and Tongue River basin W-10
Table 4. Precipitation stations for the Powder and Tongue River watershed model W-10
Table 5. Calibration and validation locations in the Powder and Tongue River basin W-13
Table 6. Summary statistics at USGS 06306300 Tongue River at State Line near Decker, MT -
calibration period W-18
Table 7. Summary statistics at USGS 06306300 Tongue River at State Line near Decker, MT -
validation period W-21
Table 8. Summary statistics (percent error): all stations - calibration period W-22
Table 9. Summary statistics (percent error): all stations -validation period W-22
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 06306300 Tongue River at State Line near Decker, MT W-24
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 06306300 Tongue River at State Line near Decker, MT W-25
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 06306300 Tongue River at State Line near Decker, MT W-26
Table 13. Summary statistics for water Quality at all stations - calibration period 1993-2003 W-26
Table 14. Summary statistics for water quality at all stations - validation period 1983-1993 W-27
Figures
Figure 1. Location of the Powder and Tongue River basin W-5
Figure 2. Land use in the Powder and Tongue River basin W-7
Figure 3. Model segmentation and USGS stations utilized for the Powder and Tongue River basin W-12
Figure 4. Mean monthly flow at USGS 06306300 Tongue River at State Line near Decker, MT -
calibration period W-16
Figure 5. Seasonal regression and temporal aggregate at USGS 06306300 Tongue River at State Line
near Decker, MT- calibration period W-16
Figure 6. Seasonal medians and ranges at USGS 06306300 Tongue River at State Line near Decker,
MT- calibration period W-17
Figure 7. Flow exceedance at USGS 06306300 Tongue River at State Line near Decker, MT -
calibration period W-17
Figure 8. Mean monthly flow at USGS 06306300 Tongue River at State Line near Decker, MT -
validation period W-19
Figure 9. Seasonal regression and temporal aggregate at USGS 06306300 Tongue River at State
Line near Decker, MT - validation period W-19
Figure 10. Seasonal medians and ranges at USGS 06306300 Tongue River at State Line near Decker,
MT - validation period W-20
Figure 11. Flow exceedance at USGS 06306300 Tongue River at State Line near Decker, MT -
validation period W-20
Figure 12. Fit for monthly load of TSS at USGS 06306300 Tongue River at State Line near Decker,
MT W-23
Figure 13. Fit for monthly load of total phosphorus at USGS 06306300 Tongue River at State Line near
Decker, MT W-25
Figure 14. Fit for monthly load of total nitrogen at USGS 06306300 Tongue River at State Line near
Decker, MT W-26
W-3
-------
The Powder/Tongue River basin was selected as one of the 15 non-pilot application watersheds for the 20
Watershed study. This basin was selected as representative of conditions in the northern plains. Watershed
modeling for the non-pilot areas is accomplished using the SWAT model only, and model calibration and
validation results are presented in abbreviated form.
Water Body Characteristics
The Powder River and Tongue River are major tributaries to the Yellowstone River, which in turn is part of the
Missouri River system on the east side of the Rocky Mountains. The model study area consists of almost 19,000
mi2 in Montana and Wyoming and consists of 12 HUCSs in HUC 1009 (Figure 1).
The watershed lies in parts of the Great Plains, Middle Rocky Mountains, Wyoming Basin, and Northern Rocky
Mountains physiographic provinces (Zelt et al., 1999). Elevation ranges from over 13,000 ft on the crest of the
Big Horn Range to less than 3,000 ft at the confluence of the Powder and Yellowstone Rivers. This large
elevation range has important impacts on climate in the watershed, which ranges from cold and moist in the
mountainous areas to temperate and semiarid in the plains areas. Mean annual temperatures range from less than
32° F at the highest elevations to about 50° F along the river valleys in Montana. Annual temperature extremes
range from about -40° F during the winter to hotter than 100° F during the summer. Mean annual precipitation
ranges from about 12 inches in the plains to more than 35 inches at high elevations. Snowfall composes a
substantial part of annual precipitation in most years.
Streams in the mountainous areas of the basin generally are perennial and derived primarily from snowmelt
runoff. Most streams originating in the plains areas of the basin are ephemeral, flowing only as a result of local
snowmelt or intense rainstorms (Peterson et al. 2004). In some subbasins, where irrigated agriculture is a major
land use, most of the streamflow results from agricultural return flow and sustained base flows.
Rangeland is the dominant land cover (85 percent of the watershed). Cropland and pasture compose less than 2
percent of the watershed. Silviculture is another important land use activity and forests cover about 10 percent of
the model study area. The watershed is sparsely populated and developed land accounts for only 0.5 percent of the
watershed.
In addition to agriculture, silviculture, and urban uses, other important land uses in the watershed include metals
and coal mining and hydrocarbon production. One of the nation's largest natural gas fields lies in the watershed
and production from the low-sulfur coal beds in the Powder River basin is increasing rapidly in response to the
demand for low-sulfur steam coal by electric utility consumers. All of the active coal mines in the watershed are
surface (strip) mines.
There are no major storage reservoirs in the watershed, although the Tongue River is impounded near the state
line. However, hundreds of small impoundments for water supply, recreation, power, and flood control have been
constructed in the watershed.
The plains streams tend to have relatively high concentrations of nitrogen, mostly as organic nitrogen and from
natural sources. Phosphorus concentrations are also relatively high and due to natural sources in marine
sedimentary rocks. The sparse vegetative cover and erodible soils in the plains areas contribute to large suspended
sediment concentrations, and the Powder River is estimated to produce an annual suspended sediment yield of
about 275 tons per square mile (Peterson et al. 2004).
W-4
-------
Hydrography
Water (Nat. Alias Dataset)
US Census Populated Places
j^B Municipalities (pop £ 50.000)
| | Caunty Boundaries
~~| \Atelershed wilh HUCBs
Mizpan
Lbwer TongUe—f(-1°°9°2J£):
(10090102)
Lower
Powder
(10090209)
ladle
Powder
(10090207)
Upper Tongue
(10090101)
Little
Powder
(10090208)
Clear
(10090206)
Crazy
Woman
(10090205)J Upper Powder
(10090202)
iddle
Fork
Powder
(10090201)
South Fork
Powder
(10090203
Antelope Hills
GCRP Model Areas - Powder and Tongue River Basins
Base Map
Figure 1. Location of the Powder and Tongue River basin.
W-5
-------
Soil Characteristics
Soils in the watershed, as described in STATSGO soil surveys, fall primarily into hydrologic soil groups (HSGs)
B (moderately high infiltration capacity) and C (moderate infiltration capacity). SWAT uses information drawn
directly from the soils data layer to populate the model.
Land Use Representation
Land use/cover in the watershed is based on the 2001 National Land Cover Database (NLCD) coverage (Figure
2). NLCD land cover classes were aggregated according to the scheme shown in Table 1 for representation in the
20 Watershed model. SWAT uses the built-in hydrologic response unit (HRU) overlay mechanism in the
ArcSWAT interface. SWAT HRUs are formed from an intersection of land use and STATSGO major soils. The
distribution of land use in the watershed is summarized in Table 2.
W-6
-------
Legend
Hydrography
Interstate
I I County Boundaries
2001 NLCD Land Use
I I Open water
I I Developed, open space
| Developed, low intensity
IB Developed, medium intensity
|^B Developed, high intensity
I I Barren land
| Deciduous forest
^^| Evergreen forest
] Mixed forest
I I Scrub/shrub
I Grassland/herbaceous
] Pasture/hay
I I Cultivated crops
I I Wfoody wetlands
| Emergent herbaceous wetlands
GCRP Model Areas -Yellowstone River Basins
Land Use Map
Figure 2. Land use in the Powder and Tongue River basin.
W-7
-------
Table 1. Aggregation of NLCD land cover classes
NLCD Class
1 1 Water
12 Perennial ice/snow
21 Developed open space
22 Dev. Low Intensity
23 Dev. Med. Intensity
24 Dev. High Intensity
31 Barren Land
41 Forest
42 Forest
43 Forest
51-52 Shrubland
71-74 Herbaceous Upland
81 Pasture/Hay
82 Cultivated
91 -97 Wetland
98-99 Wetland
Comments
Water surface area usually
accounted for as reach area
Deciduous
Evergreen
Mixed
Emergent & woody wetlands
Aquatic bed wetlands (not
emergent)
SWAT class
WATR
WATR
URLD
URMD
URHD
UIDU
SWRN
FRSD
FRSE
FRST
RNGB
RNGE
HAY
AGRR
WETF, WETL,
WETN
WATR
W-8
-------
Table 2. Land use distribution for the Powder and Tongue River basin (2001 NLCD) (mi )
HUC8
watershed
Lower
Powder
10090209
Lower
Tongue
10090102
Mizpah
10090210
Upper
Tongue
10090101
Middle
Powder
10090207
Little Powder
10090208
Clear
10090206
Upper
Powder
10090202
Crazy
Woman
10090205
Middle Fork
Powder
10090201
South Fork
Powder
10090203
Salt
10090204
Total
Open
water
0.48
0.96
0.18
3.35
1.04
0.27
6.95
0.15
0.43
0.10
0.19
0.08
14.19
Snow/Ice
0.00
0.00
0.00
0.00
0.00
0.00
0.36
0.00
0.00
0.00
0.00
0.00
0.36
Developed9
Open
space
4.71
9.19
3.59
16.81
2.22
6.06
7.31
5.10
3.27
1.52
2.54
2.26
64.58
Low
density
1.36
1.91
0.62
5.92
0.68
0.73
2.34
1.02
1.33
0.59
0.63
2.44
19.57
Medium
density
0.08
0.54
0.00
2.03
0.11
0.40
0.73
0.01
0.05
0.04
0.01
0.21
4.21
High
density
0.00
0.06
0.00
0.51
0.01
0.09
0.10
0.00
0.00
0.00
0.00
0.00
0.77
Barren
land
0.73
2.85
0.35
13.14
3.85
11.99
1.32
9.19
2.09
8.34
48.47
22.02
124.32
Forest
61.75
525.93
20.03
493.15
111.08
77.75
229.32
18.02
143.47
170.87
20.03
8.40
1,879.79
Shru bland
1,732.77
2,209.00
728.90
1,838.36
913.59
1,867.11
848.49
2,449.96
767.88
781.65
1,115.46
759.54
16,012.71
Pasture/Hay
2.01
12.89
0.49
52.52
3.36
2.52
17.76
2.77
6.80
4.96
1.96
0.02
108.05
Cultivated
38.36
31.45
35.57
23.30
10.16
23.95
10.55
7.98
1.52
0.18
0.01
0.00
183.04
Wetland
35.26
62.91
7.58
79.80
16.88
23.82
24.20
29.52
11.95
19.42
3.48
2.49
317.31
Total
1,877.51
2,857.68
797.31
2,528.87
1,062.99
2,014.69
1,149.40
2,523.71
938.79
987.66
1,192.80
797.47
18,728.90
aThe percent imperviousness applied to each of the developed land uses is as follows:
density (85.99%).
open space (7.42%), low density (31.64%), medium density (59.16%), and high
W-9
-------
Modeling of BMP Implementation Options in Support of
TMDL Compliance Plans and Strategic Watershed Planning
Point Sources
There are numerous point source discharges in the watershed. Only the major dischargers, generally defined as
those with a design flow greater than 1 MGD are included in the simulation (Table 3). The major dischargers are
represented at long-term average flows, without accounting for changes over time or seasonal variations.
Table 3. Major point source discharges in the Powder and Tongue River basin
NPDES ID
MT0000892
MT0020001
MT0024210
WY0020010
Name
DECKER COAL CO (WEST MINE)
MILES CITY- CITY OF
DECKER COAL CO (EAST MINE)
SHERIDAN, CITY OF
Design flow
(MGD)
1.980
Observed flow
(MGD)
(1991-2006 average)
0.861
1.0638
0.884
2.489
Most of these point sources have reasonably complete monitoring for total suspended solids (TSS). Assumptions
were made for total nitrogen and total phosphorus depending upon the type of facility. The point sources were
initially represented in the model with the median of reported values for total phosphorus, total suspended solids
and total nitrogen.
Meteorological Data
The required meteorological time series for the 20 Watershed SWAT simulations are precipitation and air
temperature. The 20 Watershed simulations do not include water temperature and uses a degree-day method for
snowmelt. SWAT estimates Penmann-Monteith potential evapotranspiration using a statistical weather generator
for inputs other than temperature and precipitation. These meteorological time series are drawn from the
BASINS4 Meteorological Database (USEPA 2008), which provides a consistent, quality-assured set of
nationwide data with gaps filled and records disaggregated. Scenario application requires simulation over 30
years, so the available stations are those with a common 30-year period of record (or one that can be filled from
an approximately co-located station) that covers the year 2003. A total of 37 precipitation stations were identified
for use in the Powder and Tongue River watershed model with a common period of record of 10/1/1972-
9/30/2003 (Table 4). Temperature records are sparser; where these are absent temperature is taken from nearby
stations with an elevation correction.
Table 4. Precipitation stations for the Powder and Tongue River watershed model
COOP ID
241084
241127
241297
241905
242266
242689
244442
245303
245690
Name
BRANDENBERG
BROADUS
BUSBY
COLSTRIP
DECKER 4 NNE
EKALAKA
ISMAY
MAC KENZIE
MILES CITY AP
Latitude
45.8161
45.4443
45.5398
45.8944
45.0117
45.8904
46.4997
46.1423
46.4267
Longitude
-106.2310
-105.4070
-106.9590
-106.6330
-106.8630
-104.5460
-104.7990
-104.7350
-105.8820
Temperature
X
X
X
X
X
X
X
Elevation (m)
844
924
1045
981
1073
1044
762
856
800
-------
COOP ID
245754
245870
246691
247034
247740
248165
248607
249175
480740
481165
481220
481570
482881
483801
483855
485055
485506
486195
486395
487375
487376
487545
488155
488160
488626
488852
488858
489580
Name
MIZPAH 4 NNW
MOORHEAD 9 NE
POWDERVILLE 8 NNE
RIDGEWAY 1 S
SONNETTE 2 WNW
TERRY
VOLBORG
WYOLA 1 SW
BILLY CREEK
BUFFALO
BURGESS JUNCTION
CASPER WSCMO
ECHETA2NW
GAS HILLS 4 E
GILLETTE 6 SE
KAYCEE
LEITER 9 N
MIDWEST
MOORCROFT 3 S
POWDER RIVER SCHOOL
POWDER RIVER NO 2
RECLUSE
SHERIDAN AP
SHERIDAN FIELD STN
STORY
TEN SLEEP 4 NE
TEN SLEEP 16SSE
WESTON 1 E
Latitude
46.2859
45.1759
45.8525
45.5023
45.4184
46.7940
45.8437
45.1217
44.1243
44.3450
44.7743
42.8976
44.4828
42.8394
44.2645
43.7144
44.8501
43.4132
44.2170
43.0359
43.0350
44.7409
44.7694
44.8407
44.5772
44.0657
43.8112
44.6406
Longitude
-105.2910
-105.7510
-105.0340
-104.4470
-105.8680
-105.3020
-105.6800
-107.4060
-106.7310
-106.7200
-107.5210
-106.4630
-105.8990
-107.5130
-105.4910
-106.6370
-106.2880
-106.2770
-104.9290
-106.9880
-106.9880
-105.7260
-106.9680
-106.8380
-106.8960
-107.3800
-107.3640
-105.3050
Temperature
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Elevation (m)
756
981
853
1011
1189
685
908
1137
1516
1423
2457
1627
1219
1972
1414
1420
1268
1481
1318
1736
1737
1265
1202
1143
1549
1469
1426
1074
Watershed Segmentation
The Powder and Tongue River basin was divided into 77 subwatersheds for the purposes of modeling (Figure 3).
Tongue River at State Line near Decker at USGS 06306300 was chosen for initial calibration. The model
encompasses the complete watershed and does not require specification of any upstream boundary conditions for
application.
W-ll
-------
Hydrography
Interstate
Water (Nat. Atlas Dataset)
US Census Populated Places
I | County Boundaries
T Model Subbasin
Wyoming
0 15 30 60
Kilometers
GCRP Model Areas - Yellowstone River Basins
Model Segmentation
NAD 1983 Albers meters- Map produced 12-23-2010- P. Cada
Figure 3. Model segmentation and USGS stations utilized for the Powder and Tongue River basin.
W-12
-------
Calibration Data and Locations
The specific site chosen for initial calibration was the Tongue River at State Line near Decker, MT a flow and
water quality monitoring location. The USGS gage located at Tongue River at State Line near Decker was
selected because there is a good set of flow and water quality data available and the watershed lacks major point
sources and impoundments. Additional calibration and validation was pursued at multiple locations (Table 5).
Parameters derived from the initial calibration were not fully transferable to other portions of the Powder and
Tongue River basin, and additional calibration was conducted at multiple gage locations.
Table 5. Calibration and validation locations in the Powder and Tongue River basin
Station name
Tongue River at Tongue R Dam nr Decker MT
Tongue River at State Line nr Decker MT
Tongue River at Birney Day School Br nr Birney MT
Tongue River at Miles City MT
Powder River at Moorhead MT
Powder River near Locate MT
USGS ID
06307500
06306300
06307616
06308500
06324500
06326500
Drainage area
(mi2)
1,770
1,453
2,621
5,397
8,086
13,068
Hydrology
calibration
X
X
X
X
X
X
Water quality
calibration
X
X
The model hydrology calibration period was set to Water Years 1993-2003 (within the 32-year period of record
for modeling). Hydrologic validation was then performed on Water Years 1983-1993. Water quality calibration
used calendar years 1993-2003, while validation used 1983-1993.
W-13
-------
a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
The Powder and Tongue River basin is comprised of the areas drained by the Tongue River and Powder River.
Tongue River reservoir is the only major impoundment that is represented in the model. Pertinent reservoir
information including surface area and storage at principal (normal) and emergency spillway levels for the
reservoir were obtained from the State Water Resources Bureau. The SWAT model provides four options to
simulate reservoir outflow: measured daily outflow, measured monthly outflow, average annual release rate for
uncontrolled reservoir, and controlled outflow with target release. Keeping in view the 20 Watershed climate
change impact evaluation application to future climate scenarios, it was assumed that the best representation of
the reservoir was to simulate it without supplying time series of outflow records. Hence, the target release
approach was used for the Tongue River reservoir.
A spatial calibration approach was adopted for GCRP-SWAT modeling for the Powder and Tongue River basin.
A systematic adjustment of parameters has been adopted and some adjustments are applied throughout the basin.
Most of the calibration efforts were geared towards getting a closer match between simulated and observed flows
at the outlet of calibration focus area.
A 5/10/5 percent threshold was used for land use/soil/slope in the SWAT model while defining the HRUs. Urban
land use classes were exempted from the HRU overlay thresholds.
The calibration focus area includes three subwatersheds and is generally representative of the general land use
characteristics of the overall watershed. The parameters were adjusted within the practical range to obtain
reasonable fit between the simulated and measured flows in terms of Nash-Sutcliffe modeling efficiency and the
high flow and low flow components as well as the seasonal flows.
The Tongue River and Powder River basins were modeled separately. The water balance of the Tongue River
basin predicted by the SWAT model over the 32-year simulation period is as follows:
PRECIP = 374.4 MM
SNOW FALL = 116.02 MM
SNOW MELT = 98.58 MM
SUBLIMATION = 16.87 MM
SURFACE RUNOFF Q = 26.86 MM
LATERAL SOIL Q = 22.46 MM
TILE Q = 0.00 MM
GROUNDWATER (SHAL AQ) Q = 17.25 MM
REVAP (SHAL AQ => SOIL/PLANTS) = 0.09 MM
DEEP AQ RECHARGE = 10.78 MM
TOTAL AQ RECHARGE = 28.11 MM
TOTAL WATER YLD = 66.57 MM
PERCOLATION OUT OF SOIL = 28.46 MM
ET = 388.6 MM
PET = 1268.1MM
TRANSMISSION LOSSES = 0.00 MM
W-14
-------
The water balance of the Powder River basin predicted by the SWAT model over the 32-year simulation period is
as follows:
PRECIP = 363.5 MM
SNOW FALL = 106.55 MM
SNOW MELT = 91.10 MM
SUBLIMATION = 15.05 MM
SURFACE RUNOFF Q = 25.65 MM
LATERAL SOIL Q = 17.88 MM
TILE Q = 0.00 MM
GROUNDWATER (SHAL AQ) Q = 14.03 MM
REVAP (SHAL AQ => SOIL/PLANTS) = 0.04 MM
DEEP AQ RECHARGE = 0.00 MM
TOTAL AQ RECHARGE = 14.06 MM
TOTAL WATER YLD = 55.33 MM
PERCOLATION OUT OF SOIL = 12.00 MM
ET = 441.3 MM
PET = 1427.9MM
TRANSMISSION LOSSES = 2.23 MM
Hydrologic calibration adjustments focused on the following parameters:
• CN2 (initial SCS runoff curve number for moisture condition II)
• ESCO (soil evaporation compensation factor)
• SURLAG (surface runoff lag coefficient)
• SOL_AWC (available water capacity of the soil layer, mm water/mm of soil)
• ALPHA_BF (baseflow alpha factor, days)
• GW_DELAY (groundwater delay time, days)
• GWQMIN (threshold depth of water in the shallow aquifer required for return flow to occur, mm)
• GW_REVAP (groundwater "revap" coefficient)
• CH_N1 (Manning's "n" value for tributary channels)
• CH_N2 (Manning's "n" value for main channels)
• CH_K1 (Effective hydraulic conductivity in tributary channel alluvium)
• CH_K2 (Effective hydraulic conductivity in main channel alluvium)
• SFTMP (Snowfall temperature)
• SMTMP (Snowmelt base temperature)
• SMFMX (Maximum melt rate for snow during the year)
• SMFMN (Minimum melt rate for snow during the year)
The calibration achieves a moderately high coefficient of model fit efficiency, but is above on summer flow
volumes. Calibration results for the Tongue River at State Line near Decker are summarized in Figure 4, Figure 5,
Figure 6, Figure 7 and Table 6).
W-15
-------
4000
3000
2000 -
1000
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1993 to 9/30/2003)
Avg Modeled Flow (Same Period)
O-93
A-95
O-96
A-98
O-99
A-01
O-02
Month
Figure 4. Mean monthly flow at USGS 06306300 Tongue River at State Line near Decker, MT -
calibration period.
• Avg Flow (10/1 /1993 to 9/30/2003)
• Line of Equal Value
•Best-Fit Line
Avg Monthly Rainfall (in)
- Avg Observed Fl ow (10/1 /1993 to 9/30/2 003)
•Avg Modeled Flow (Same Period)
1500
^1000
-------
To Lower Bound Average Monthly Rainfall (in) -Median Observed Flow (10/1/1993 to 9/30/2003) Modeled (Median, 25th, 75th)
2500
2000
w)500
j^1000
500
10 11 12 1 2
10 11 12 1
234
Month
8 9
Figure 6. Seasonal medians and ranges at USGS 06306300 Tongue River at State Line near Decker, MT
- calibration period.
•Observed Flow Duration (10/1/1993 to 9/30/2003 )
Modeled Flow Duration (10/1/1993 to 9/30/2003)
10000
•6
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 7. Flow exceedance at USGS 06306300 Tongue River at State Line near Decker, MT - calibration
period.
W-17
-------
Table 6. Summary statistics at USGS 06306300 Tongue River at State Line near Decker, MT -
calibration period
SWAT Simulated Flow
REACH OUTFLOW FROM OUTLET 16
10-Year Analysis F^riod: 10/1/1993 - 9/30/2003
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9)
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
Seasonal volume error - Fall:
Seasonal volume error - Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
3.92
1.58
0.69
0.77
0.47
0.39
2.30
0.83
0.13
Error Statistics
9.26
4.58
-2.92
38.16
Observed Flow Gage
USGS 06306300 Tongue River at State Line nr Decker MT
Hydrologic Unit Code: 10090101
Latitude: 45.0088632
Longitude: -106.8361 78
Drainage Area (sq-rri): 1453
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
-3.70 » | 30
-18.41
10.99
13.40
27.25
0.719
0.494
0.832
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
3.59
1.63
0.66
0.55
0.48
0.48
2.07
0.73
0.10
Clear [
Hydrology Validation
Hydrology validation for Tongue River was performed forthe period 10/1/1983 through 9/30/1993. The
validation achieves a moderately high coefficient of model fit efficiency, but is below on 50% low flow and
winter flow volumes (Figure 8, Figure 9, Figure 10, Figure 11 and Table 7).
W-18
-------
3000
2500
Avg Monthly Rainfall (in)
-Avg Observed Flow (10/1/1983 to 9/30/1993 )
-Avg Modeled Flow (Same Period)
'ro
01
O-83
A-85
O-86
A-88
O-89
A-91
O-92
Month
Figure 8. Mean monthly flow at USGS 06306300 Tongue River at State Line near Decker, MT - validation
period.
• Avg Flow (10/1 /1983 to 9/30/1993)
• - - - • Line of Equal Value
Best-Fit Line
Avg Monthly Rainfall (in)
- Avg Observed Fl ow (10/1 /1983 to 9/30/1993)
-Avg Modeled Flow (Same Period)
1500
^1000
y = 0.8258X + 28.336
R2 = 0.9023
-------
To Lower Bound Average Monthly Rainfall (in) -Median Observed Flow (10/1/1983 to 9/30/1993) Modeled (Median, 25th, 75th)
2000
10 11 12 1
Figure 10. Seasonal medians and ranges at USGS 06306300 Tongue River at State Line near Decker, MT
- validation period.
•Observed Flow Duration (10/1/1983 to 9/30/1993 )
Modeled Flow Duration (10/1/1983 to 9/30/1993)
10000
•6
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 11. Flow exceedance at USGS 06306300 Tongue River at State Line near Decker, MT - validation
period.
W-20
-------
Table 7. Summary statistics at USGS 06306300 Tongue River at State Line near Decker, MT - validation
period
REACH OUTFLOW FROM OUTLET 16
10-Year Analysis F^riod: 10/1/1983 - 9/30/1993
Flow volumes are (inches/year) for upstream drainage area
Total Simulated In-stream Flow:
Total of simulated highest 10% flows:
Total of Simulated lowest 50% flows:
Simulated Summer Flow Volume (months 7-9)
Simulated Fall Flow Volume (months 10-12^
Simulated Winter Flow Volume (months 1-3):
Simulated Spring Flow Volume (months 4-6):
Total Simulated Storm Volume:
Simulated Summer Storm Volume (7-9):
Errors (Simulated-Observed)
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error - Summer:
(' Seasonal volume error - Fall:
Seasonal volume error- Winter:
Seasonal volume error - Sjpring:
Error in storm volumes:
Error in summer storm volumes:
Nash-Sutcliffe Coefficient of Efficiency, E:
Baseline adjusted coefficient (Garrick), E':
Monthly NSE
3.25
1.36
0.53
0.63
0.34
0.32
1.96
0.73
0.10
Error Statistics
-9.95
-14.65
2.57
Observed Flow Gage
USGS 06306300 Tongue River a
Hydrologic Unit Code: 10090101
Latitude: 45.0088632
Longitude: -106.8361 78
Drainage Area (sq-rri): 1453
: State Line nr Decker MT
Total Observed In-stream Flow:
Total of Observed highest 10% flows:
Total of Observed Lowest 50% flows:
Observed Summer Flow Volume (7-9):
Observed Fall Flow VolumeJ10-12):
Observed Winter Flow VolumeJ1-3):
Observed Spring Flow Volume (4-6):
Total Observed Storm Volume:
Observed Summer Storm Volume (7-9):
Recommended Criteria
10
10
15
30
-24.24 » | 30
-33.45
-5.12
-0.73
-3.56
0.703
0.494
0.818
30
30
20
50
Model accuracy increases
as E or E' approaches 1.0
_ Cle
3.60
1.59
0.71
0.61
0.45
0.48
2.07
0.74
0.11
d: : : : : :
Hydrology Results for Larger Watershed
As described above, parameters determined for the gage at Tongue River at State Line near Decker were initially
transferred to other gages in the watershed. However, changes to subwatershed level parameters were required to
fit the model to the observed flows. In all, calibration and validation was pursued at a total of six gages
throughout the watershed. Results of the calibration and validation exercise are summarized in Table 8 and Table
9, respectively. Calibration and validation results were acceptable at most gages.
W-21
-------
Table 8. Summary statistics (percent error): all stations - calibration period
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error- Summer:
Seasonal volume error- Fall:
Seasonal volume error - Winter:
Seasonal volume error- Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient of
Efficiency, E:
Monthly Nash-Sutcliffe Efficiency:
06307500
0.01
1.04
-6.35
-18.08
2.93
-3.32
10.83
-18.89
22.94
0.68
0.800
06306300
9.26
4.58
-2.92
38.16
-3.70
-18.41
10.99
13.40
27.25
0.72
0.832
06307616
1.10
-6.88
-3.80
-21.91
0.48
16.34
10.69
-19.82
5.64
0.67
0.822
06308500
-5.78
-18.16
0.83
28.90
-33.44
-31.74
3.58
-40.42
1.54
0.36
0.718
06324500
7.55
17.10
-7.16
35.93
16.00
3.45
-0.63
13.11
16.28
0.49
0.631
06326500
-1.83
16.25
3.35
10.13
54.07
1.59
-22.23
-7.26
-24.77
0.28
0.535
Table 9. Summary statistics (percent error): all stations - validation period
Station
Error in total volume:
Error in 50% lowest flows:
Error in 10% highest flows:
Seasonal volume error- Summer:
Seasonal volume error- Fall:
Seasonal volume error - Winter:
Seasonal volume error- Spring:
Error in storm volumes:
Error in summer storm volumes:
Daily Nash-Sutcliffe Coefficient of
Efficiency, E:
Monthly Nash-Sutcliffe Efficiency:
06307500
-15.51
-11.29
-13.96
-30.86
-14.00
-17.72
-5.72
-10.41
68.91
0.65
0.760
06306300
-9.95
-25.59
-14.65
2.57
-24.24
-33.45
-5.12
-0.73
-3.56
0.70
0.818
06307616
-13.85
-29.38
-8.00
-24.08
-18.99
-5.64
-8.23
-7.07
76.33
0.53
0.631
06308500
-0.30
-41.64
35.98
62.05
-22.35
-42.94
-6.63
-14.47
38.90
-0.55
-0.532
06324500
-14.83
-21.34
-22.75
2.81
-29.02
-7.25
-19.24
-0.77
-7.71
0.47
0.727
06326500
-10.20
-33.18
21.75
59.27
-22.23
-12.83
-29.50
-16.65
-2.84
-0.43
-0.367
W-22
-------
Water Quality Calibration and Validation
Initial calibration and validation of water quality was done at USGS 06306300, Tongue River at State Line near
Decker from water years 1983 to 2003. Subject to the availability of water quality data for the other gages, 1993-
2003 was adopted as the calibration period and 1982-1992 was adopted as the validation period. As with
hydrology, calibration was performed on the later period as this better reflects the land use included in the model.
Calibration adjustments for sediment focused on the following parameters:
• SPCON (linear parameter for estimating maximum amount of sediment that can be re-entrained during
channel sediment routing)
• SPEXP (exponential parameter for estimating maximum amount of sediment that can be re-entrained
during channel sediment routing)
• CH_COV (channel cover factor)
• CH_EROD (channel erodibility factor)
• USLE_P (USLE support practice factor)
Simulated and estimated sediment loads at the Tongue River station at State Line near Decker for both the
calibration and validation periods are shown in Figure 12 and statistics are provided separately in Table 10.
The key statistic in Table 10 is the relative percent error, which shows the error in the prediction of monthly load
normalized to the estimated load. Table 10 also shows the relative average absolute error, which is the average of
the relative magnitude of errors in individual monthly load predictions. This number is inflated by outlier months
in which the simulated and estimated loads differ by large amounts (which may be as easily due to uncertainty in
the estimated load due to limited data as to problems with the model) and the third statistic, the relative median
absolute error, is likely more relevant and shows better agreement.
TSS
o
E
"5
I
100,000
10,000
1,000
100 -•-
10 -•-
• Regression Loads
-Simulated Loads
co^-mcoi^ooa>o-!-(Nco^-mcoi^
opopopopopopopq>q>q>q>q>q>q>q>
8°8
oooooooooooooooooooo
oooooooooooooooooooo
Figure 12. Fit for monthly load of TSS at USGS 06306300 Tongue River at State Line near Decker, MT.
W-23
-------
Table 10. Model fit statistics (observed minus predicted) for monthly sediment loads using stratified
regression at USGS 06306300 Tongue River at State Line near Decker, MT
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
-21.8%
128.5%
43.2%
Validation period
-3.4%
109.7%
38.2%
Calibration adjustments for total phosphorus and total nitrogen focused on the following parameters:
• RHOQ (algal respiration rate at 20° C)
• PHOSKD (phosphorus soil partitioning coefficient)
• PSP (phosphorus availability index)
• RSI (Local algal settlement rate in the reach at 20° C)
• AL1 (Fraction of algal biomass that is nitrogen)
• AL2 (Fraction of algal biomass that is phosphorus)
• MUMAX (Rate of oxygen uptake per unit NO2-N oxidation at 20° C)
• RHOQ (Algal respiration rate at 20° C)
• RS2 (benthic source rate for dissolved P in the reach at 20° C)
• RS3 (Benthic source rate for NFLpN in the reach at 20° C)
• RS5 (organic P settling rate in the reach at 20° C)
• BC4 (rate constant for mineralization of organic P to dissolved P in the reach at 20° C)
• RS4 (rate coefficient for organic N settling in the reach at 20° C)
• CH_ONCO (Channel organic nitrogen concentration)
• CH_OPCO (Channel organic phosphorus concentration)
• SDNCO (Denitrification threshold water content)
• CDN (Denitrification exponential rate constant)
Results for the phosphorus simulation are shown in Figure 13 and Table 11. Results for the nitrogen simulation
are shown in Figure 14 and Table 12. The model fit is generally acceptable.
W-24
-------
Total P
1000
- Regression Loads
-Simulated Loads
0.01
opopopopopopopq>q>q>q>q>q>q>q>q>q>ooo
§§§§§§§§§§§§§§§§§§§§
Figure 13. Fit for monthly load of total phosphorus at USGS 06306300 Tongue River at State Line near
Decker, MT.
Table 11. Model fit statistics (observed minus predicted) for monthly phosphorus loads using stratified
regression at USGS 06306300 Tongue River at State Line near Decker, MT
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
8.8%
94.1%
27.6%
Validation period
35.1%
76.0%
25.7%
W-25
-------
/(JO
oUU
o
« 40°
c
300 •
0 1
i
(
Total N
i
V
LlJUJuJ^JL^jJjJkJ^
?-*Lr>cor^ooa>o-<-rMco-^-Lr>cor^oo
OOOOOOOOOOOOO)O)O)O)O)O)O)O)O)
oooooooooooooooo
DOOOOOOOOOOOOOOO
|
t I
Uu*jt
g 8 5 8
o o o o
O O O O
• Averaging Loads
—•—Simulated Loads
Figure 14. Fit for monthly load of total nitrogen at USGS 06306300 Tongue River at State Line near
Decker, MT.
Table 12. Model fit statistics (observed minus predicted) for monthly total nitrogen loads using
averaging estimator at USGS 06306300 Tongue River at State Line near Decker, MT
Statistic
Relative Percent Error
Relative Average Absolute Error
Relative Median Absolute Error
Calibration period
3.9%
93.3%
33.6%
Validation period
31.5%
74.5%
37.1%
Water Quality Results for Larger Watershed
As with hydrology, a spatial calibration approach was adopted. SWAT model parameters for water quality
derived from calibrations performed at the USGS gage at Tongue River at State Line near Decker were
transferred to other portions of the watershed with necessary changes to subbasin level parameters. Summary
statistics for the SWAT water quality calibration and validation at other stations in the watershed are provided in
Table 13 and Table 14.
Table 13. Summary statistics for water Quality at all stations - calibration period 1993-2003
Station
Relative Percent Error TSS Load
Relative Percent Error TP Load
Relative Percent Error TN Load
06308500
35.6
12.5
3.8
W-26
-------
Table 14. Summary statistics for water quality at all stations - validation period 1983-1993
Station
Relative Percent Error TSS Load
Relative Percent Error TP Load
Relative Percent Error TN Load
06308500
-14.1
-45.5
-52.9
References
Peterson, D.A., K.A. Miller, T.T. Bartos, M.L. Clark, S.D. Porter, and T.L. Quinn. 2004. Water Quality in the
Yellowstone River Basin, Wyoming, Montana, and North Dakota, 1999-2001. Circular 1234. U.S. Geological
Survey, Reston, VA.
USEPA (United States Environmental Protection Agency). 2008. Using the BASINS Meteorological Database
(Version 2006). BASINS Technical Note 10. Office of Water, U.S. Environmental Protection Agency,
Washington, DC. http://water.epa.gov/scitech/datait/models/basins/upload/2009_04_13_BASINSs_tecnotel0.pdf
(Accessed June, 2009).
Zelt, R.B., G. Boughton, K.A. Miller, J.P. Mason, and L.M. Gianakos. 1999. Environmental Setting of the
Yellowstone River Basin, Montana, North Dakota, and Wyoming Water-Resources Investigations Report 98-
4269. Cheyenne, Wyoming.
W-27
-------
Appendix X
Scenario Results for the Five Pilot
Watersheds
X-l
-------
Contents
Apalachicola-Chattahoochee-Flint Basin, HSPF Model X-3
Apalachicola-Chattahoochee-Flint Basin, SWAT Model X-21
Verde-Salt-San Pedro Basins, HSPF Model X-38
Verde-Salt-San Pedro Basins, SWAT Model X-59
Minnesota River (Upper Mississippi Basin), HSPF Model X-75
Minnesota River (Upper Mississippi Basin), SWAT Model X-91
Susquehanna River Basin, HSPF Model X-109
Susquehanna River Basin, SWAT Model X-121
Willamette River Basin, HSPF Model X-140
Willamette River Basin, SWAT Model X-154
X-2
-------
Apalachicola-Chattahoochee-Flint Basin, HSPF Model
Results at Downstream Stationk
900
800
700
*i/r
E 600
u
| 500
LL.
"™ 400
c
< 300
c
ra
1 200
-inn
n
Apalachicola Mouth
-+-
f
ft
¥
A
ft
0
*
ft
o
lAi
H
ft ft
BASE
G(
ICLUS
:M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 1. Mean Annual Flow (cms), Apalachicola River at Mouth (HSPF)
X-3
-------
Apalachicola Mouth
1 6 000
14 nnn
"uT
§0
u
lH
* *r
2k zfe
u
1 i
EP zn
g Q
*
$-A,
*
A A
BASE ICLUS
GCM
BASE ICLUS
NARCCAP
BASE ICLUS
BCSD
Figure 2. 100-yr Flow Peak (Log-Pearson III, cms), Apalachicola River at Mouth (HSPF)
| 250
u
1
u. 200
5
| 150
"TO
c -inn -
Average An
8 I
*
—
Apalachicola Mouth
a
$
+
+
A
*
*
I 1 S *
BASE
G(
ICLUS
:M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 3. Average Annual 7-day Low Flow (cms), Apalachicola River at Mouth (HSPF)
X-4
-------
n -lo -
ker Flashiness Index
DO C
D 0 P -
n oo -* h
ra "•""
CO
1/1
ro 0.04
^
u
5
0.02
Apalachicola Mouth
J
i
• 9 i
© ©
BASE
G(
ICLUS
;M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 4. Richards-Baker Flashiness Index, Average Annual 7-day Low Flow (cms), Apalachicola
River at Mouth (HSPF)
S 160
m
oj
"° mn
g 10°
"c
0 80
- 60
U_ DU
0
& 4U
s
20
Apalachicola Mouth
on
^ * A A * *
yfv * i 1 ^ nk
" " S S
BASE ICLUS BASE ICLUS BASE ICLUS
GCM NARCCAP BCSD
Figure 5. Days to Flow Centroid (Water Year Basis), Apalachicola River at Mouth (HSPF)
X-5
-------
2 000 000
"Z"
„>- -i cnn nnn
§.
•o
ra
•3 1 000 000 -
LO
£
n
8
A
HS
BASE
G(
Apalachicola Mouth
4, 4>
*
O
A
33
ICLUS
:M
A
&
BASE
NAR(
A
*
ICLUS
;CAP
A
L — *
BASE
BC
A
ICLUS
SD
Figure 6. TSS Load (MT/yr), Apalachicola River at Mouth (HSPF)
Apalachicola Mouth
on nnn
TO
o
_j ft nnn
Z
6 000
4 nnn
2 000
, i
A A
<£
f f * *
(/^^^^ ^^^^^^
n O
^ Bt
+ +
& ^
BASE ICLUS
GCM
BASE ICLUS
NARCCAP
BASE ICLUS
BCSD
Figure 7. TN Load (MT/yr), Apalachicola River at Mouth (HSPF)
X-6
-------
Apalachicola Mouth
4 000
"Z"
>" -a nnn
i
"D
TO
•3 2 000
Q.
1 000
500
n
A
ft
* (A)
(A) £
*-^^^™ ^^^^™
A
A *
,
Jf ^
^5
T « *
BASE ICLUS
GCM
BASE ICLUS
NARCCAP
BASE ICLUS
BCSD
Figure 8. TP Load (MT/yr), Apalachicola River at Mouth (HSPF)
X-7
-------
Results at Multiple Stations
Table 1. Summary of Range of Change Relative to Existing Conditions for NARCCAP Dynamically
Downscaled Scenarios, Apalachicola-Chattahoochee-Flint Basin HSPF Model
Results without LU Change
Min
Median
Mean
Max
Results with LU Change
Min
Median
Mean
Max
Flint River near Montezuma (gage 02349605)
Flow
TSS
TN
TP
-18.62%
-13.15%
-23.92%
-15.02%
8.70%
40.66%
17.08%
21.84%
6.49%
41.77%
14.35%
18.58%
20.02%
94.31%
36.43%
41.91%
-18.44%
-15.22%
-23.37%
-14.72%
8.67%
32.89%
16.27%
19.58%
6.47%
33.16%
13.57%
16.26%
19.92%
76.35%
34.78%
37.40%
Chattahoochee River near Cornelia (gage 02331600)
Flow
TSS
TN
TP
-5.84%
20.99%
-3.45%
3.50%
5.18%
51.30%
11.31%
19.70%
4.56%
56.56%
11.24%
18.22%
11.13%
99.56%
21.95%
32.21%
-5.82%
20.06%
-3.52%
2.87%
5.19%
47.02%
11.00%
18.66%
4.56%
51.73%
10.91%
17.19%
11.11%
90.90%
21.35%
30.49%
Peachtree Creek (gage 02336300)
Flow
TSS
TN
TP
-11.06%
-6.08%
-8.98%
-9.95%
6.31%
17.47%
6.82%
12.15%
5.58%
14.88%
5.84%
9.31%
14.40%
27.63%
13.47%
18.03%
-10.62%
-6.21%
-8.76%
-9.98%
6.32%
16.92%
6.48%
11.86%
5.59%
14.10%
5.39%
9.02%
14.24%
25.95%
12.48%
17.51%
Chattahoochee at Atlanta (gage 02336000)
Flow
TSS
TN
TP
-13.84%
-9.66%
-8.27%
-9.55%
0.51%
22.22%
3.57%
10.03%
0.07%
21.98%
3.35%
7.95%
8.41%
47.03%
10.66%
19.82%
-13.36%
-10.37%
-8.55%
-10.73%
0.89%
19.29%
3.37%
8.98%
0.41%
17.43%
3.06%
6.77%
8.65%
36.75%
9.91%
17.76%
Ichawaynochaway Creek (gage 02353500)
Flow
TSS
TN
TP
-18.49%
-0.63%
-25.45%
-15.16%
9.21%
138.28%
20.94%
38.74%
5.64%
140.22%
14.52%
33.16%
23.80%
282.06%
45.86%
77.57%
-18.48%
-0.64%
-25.45%
-15.16%
9.21%
137.95%
20.93%
38.72%
5.64%
139.86%
14.51%
33.14%
23.80%
281.35%
45.85%
77.51%
Chattahoochee River at West Point (gage 02339500)
XQ
-O
-------
Flow
TSS
TN
TP
Results without LU Change
Min
-14.64%
-11.13%
-14.38%
-10.14%
Median
2.82%
32.90%
6.47%
10.65%
Mean
1.71%
31.91%
5.20%
8.03%
Max
11.46%
72.52%
18.35%
20.33%
Results with LU Change
Min
-14.26%
-13.41%
-14.34%
-11.18%
Median
3.00%
26.00%
6.24%
10.21%
Mean
1.90%
25.48%
4.96%
7.43%
Max
11.54%
58.86%
17.66%
19.53%
Chattahoochee River near Columbia (gage 02343801)
Flow
TSS
TN
TP
-17.57%
-7.19%
-25.53%
-12.29%
5.35%
55.37%
13.71%
11.02%
3.73%
57.84%
11.00%
8.54%
16.78%
127.37%
36.63%
23.25%
-17.30%
-6.08%
-25.29%
-12.62%
5.40%
53.44%
13.47%
10.86%
3.79%
55.81%
10.77%
8.29%
16.72%
122.70%
35.95%
22.77%
Flint River at Newton (gage 02353000)
Flow
TSS
TN
TP
-17.76%
-1.54%
-22.22%
-12.51%
9.62%
61.44%
19.43%
24.78%
6.69%
59.67%
15.16%
20.68%
21.63%
131.16%
40.01%
46.91%
-17.66%
-3.84%
-22.16%
-12.75%
9.60%
56.79%
19.16%
23.48%
6.68%
54.98%
14.91%
19.40%
21.56%
121.31%
39.49%
44.40%
Apalachicola River at Seminole (gage 02358000)
Flow
TSS
TN
TP
-18.57%
-8.60%
-26.90%
-13.60%
7.23%
68.64%
18.07%
16.94%
4.89%
70.10%
14.00%
13.63%
19.75%
150.58%
43.52%
34.81%
-18.38%
-7.25%
-26.77%
-13.77%
7.24%
65.74%
17.86%
16.36%
4.91%
67.07%
13.81%
13.03%
19.68%
144.64%
43.01%
33.52%
Apalachicola Mouth
Flow
TSS
TN
TP
-20.76%
-17.46%
-37.64%
-16.43%
6.58%
58.76%
14.70%
16.96%
4.66%
63.45%
13.18%
13.90%
21.68%
147.71%
54.13%
38.06%
-20.57%
-16.56%
-37.39%
-16.39%
6.60%
57.62%
14.64%
16.47%
4.68%
62.15%
13.09%
13.37%
21.59%
144.58%
53.65%
36.72%
X-9
-------
Flint near Montezuma
250
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 9. Monthly Average Flows, Flint River near Montezuma (HSPF)
Flint near Montezuma
10000
1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 10. Flow Duration, Flint River near Montezuma (HSPF)
X-10
-------
Chatt near Cornelia
11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 11. Monthly Average Flows, Chattahoochee River near Cornelia (HSPF)
Chatt near Cornelia
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 12. Flow Duration, Chattahoochee River near Cornelia (HSPF)
X-ll
-------
Peachtree
6 7
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 13. Monthly Average Flows, Peachtree Creek (HSPF)
Peachtree
1000
o.
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 14. Flow Duration, Peachtree Creek (HSPF)
X-12
-------
Chatt at Atlanta
120
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 15. Monthly Average Flows, Chattahoochee River at Atlanta (HSPF)
Chatt at Atlanta
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 16. Flow Duration, Chattahoochee River at Atlanta (HSPF)
X-13
-------
Ichawaynochaway
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 17. Monthly Average Flows, Ichawaynochaway Creek (HSPF)
Ichawaynochaway
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 18. Flow Duration, Ichawaynochaway Creek (HSPF)
X-14
-------
Chatt at West Point
350
10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 19. Monthly Average Flows, Chattahoochee River at West Point (HSPF)
Chatt at West Point
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 20. Flow Duration, Chattahoochee River at West Point (HSPF)
X-15
-------
Chatt near Columbia
6 7
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 21. Monthly Average Flows, Chattahoochee River near Columbia (HSPF)
Chatt near Columbia
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 22. Flow Duration, Chattahoochee River near Columbia (HSPF)
X-16
-------
Flint at Newton
6 7
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 23. Monthly Average Flows, Flint River at Newton (HSPF)
Flint at Newton
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 24. Flow Duration, Flint River at Newton (HSPF)
X-17
-------
Apalachicola at Seminole
1600
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 25. Monthly Average Flows, Apalachicola River at Seminole (HSPF)
Apalachicola at Seminole
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 26. Flow Duration, Apalachicola River at Seminole (HSPF)
X-18
-------
Apalachicola Mouth
567
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 27. Monthly Average Flows, Apalachicola River at Mouth (HSPF)
Apalachicola Mouth
100000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 28. Flow Duration, Apalachicola River at Mouth (HSPF)
X-19
-------
12%
-4%
Month
Figure 29. Average of Median Percent Change in Flow; NARCCAP Scenarios W1-W6 at all
Stations, Apalachicola-Chattahoochee-Flint Basin (HSPF)
X-20
-------
Apalachicola-Chattahoochee-Flint Basin, SWAT Model
Results at Downstream Station
800
tfl
E 600
o 50°
U.
"ro 400
c
< 300
C
ro
flJ 200
100
Apalachicola
,
•li /|\
•
^^^
/^^ ^T.^^ ^^J
0
0
O
0
^
^
+ """ # #
BASE
G(
ICLUS
;M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 30. Mean Annual Flow, Apalachicola River at Mouth (SWAT)
3,000
V)
£ 2,500
u
g 2,000
V)
ro
Ol
°r 1,500
5
"TO 1,000
£
2 500
U.
Q
^
6—
+
Apalachicola
*
A
+
A
4
g
A
4
&
1
— 1
BASE
GC
ICLUS
;M
BASE
NAR
ICLUS
DCAP
BASE
BC
ICLUS
SD
Figure 31. 100-yr flow Peak (Log-Pearson III), Apalachicola River at Mouth (SWAT)
X-21
-------
600 -i
1" 500
u
O
Average Annual 7-day Low F
-* M OO J:
o 8 8 8 E
Apalachicola
I i
f f T"T IV
A A
? ? ° ° * *
* *
BASE ICLUS BASE ICLUS BASE ICLUS
GCM NARCCAP BCSD
Figure 32. Average Annual 7-day Low Flow, Apalachicola River at Mouth (SWAT)
ker Flashiness Index
DO O C
D 0 P 0 C
f^ W.W-"
CO
ro 0.004
^
^o
5
0.002
n
Q
+
Apalachicola
A 4
®
^j*^^
t
t
/->
8
ft
^Ts^^
ft
^T^i
BASE
G(
ICLUS
;M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 33. Richards-Baker Flashiness Index, Apalachicola River at Mouth (SWAT)
X-22
-------
250 -i
"t/T
8 200
m
HI
ra 150
:lowCentroid
I
0
g, 50
0
Apalachicola
W
BASE
ICLUS
GCM
BASE
ICLUS
NARCCAP
y*v
^
BASE
^^^v
1?
ICLUS
BCSD
Figure 34. Days to Flow Centroid, Apalachicola River at Mouth (SWAT)
Apalachicola
30,000 -i
25,000
20 000 -
1
""*" 1 5 000
I
5,000
n
i~£
^
"M
A
A
*• — *
t
JL
' — *
t
c~
+
O
+
0
O
fe>
* *
BASE
ICLUS
GCM
BASE
ICLUS
NARCCAP
BASE
ICLUS
BCSD
Figure 35. TSS Load, Apalachicola River at Mouth (SWAT)
X-23
-------
2,500
^
"D
TO
5
Q.
1— -] 000
500
n
Apalachicola
i j.
A
¥
A
¥
K
2
S
O
4
1
BASE
G(
ICLUS
:M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 36. TP Load, Apalachicola River at Mouth (SWAT)
9^ nnn
20 000
s.
•n
S
z
5 000
n
A
Apalachicola
* 7k
"K
*
•
0
.u
*
~" iV *
BASE
G(
ICLUS
;M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 37. TN Load, Apalachicola River at Mouth (SWAT)
X-24
-------
Results at Multiple Stations
Table 2. Summary of Range of Change Relative to Existing Conditions for NARCCAP
Dynamically Downscaled Scenarios, Apalachicola-Chattahoochee-Flint Basin SWAT
Model
Results without LU Change
Min
Median
Mean
Max
Results with LU Change
Min
Median
Mean
Max
Chattahoochee at Atlanta HUC 03130001
Flow
TSS
TN
TP
-4.97%
5.73%
3.07%
19.25%
15.45%
40.72%
11.73%
35.55%
14.11%
34.45%
10.27%
33.73%
23.72%
49.87%
12.43%
38.87%
-4.52%
7.02%
5.05%
18.55%
15.17%
35.31%
14.50%
37.32%
13.98%
31.78%
13.06%
35.00%
23.28%
46.32%
15.71%
41.04%
Middle Chattahoochee - Lake Harding HUC 03130002
Flow
TSS
TN
TP
-14.55%
-16.80%
-2.12%
8.33%
14.78%
49.05%
5.91%
26.99%
11.98%
44.01%
5.11%
24.78%
24.84%
77.45%
7.86%
31.90%
-14.15%
-15.78%
-1.84%
7.72%
14.49%
48.57%
6.30%
27.42%
11.81%
43.61%
5.49%
25.08%
24.58%
76.55%
8.28%
32.49%
Middle Chattahoochee - WF George HUC 03130003
Flow
TSS
TN
TP
-18.68%
-14.99%
-1.24%
8.01%
12.52%
59.60%
9.81%
29.42%
9.37%
52.46%
8.61%
26.82%
24.24%
92.84%
12.95%
36.55%
-18.37%
-14.31%
-0.98%
7.66%
12.34%
59.50%
10.20%
29.68%
9.28%
52.32%
8.99%
27.02%
23.94%
92.49%
13.35%
36.92%
Lower Chattahoochee HUC 03130004
Flow
TSS
TN
TP
-19.63%
-30.48%
-1.48%
9.76%
11.89%
26.43%
10.33%
29.85%
8.52%
21.16%
9.07%
27.41%
24.74%
55.50%
14.40%
37.97%
-19.35%
-29.74%
-1.22%
9.47%
11.74%
27.07%
10.68%
30.05%
8.44%
21.83%
9.41%
27.57%
24.48%
56.08%
14.77%
38.25%
Peachtree Creek (gage 02336300)
Flow
TSS
TN
TP
-7.54%
-5.41%
20.49%
32.59%
17.56%
31.74%
32.31%
45.86%
16.16%
27.21%
33.30%
48.09%
27.69%
43.52%
49.14%
64.21%
-7.00%
-2.24%
25.31%
35.39%
16.48%
30.58%
38.46%
51.13%
15.25%
26.46%
40.42%
53.11%
26.39%
41.69%
59.42%
72.85%
X-25
-------
Results without LU Change
Min
Median
Mean
Max
Results with LU Change
Min
Median
Mean
Max
Upper Flint HUC 03130005
Flow
TSS
TN
TP
-21.50%
-20.29%
8.28%
19.70%
11.78%
33.66%
38.60%
62.54%
8.12%
27.97%
32.62%
54.68%
22.24%
50.19%
41.83%
66.95%
-21.32%
-20.18%
8.72%
18.54%
11.44%
33.00%
39.65%
61.28%
7.86%
27.52%
33.73%
53.50%
21.87%
49.49%
42.82%
66.48%
Middle Flint HUC 03130006
Flow
TSS
TN
TP
-23.41%
-24.04%
1.80%
8.52%
11.34%
29.06%
39.97%
50.28%
6.51%
21.37%
31.73%
42.57%
23.31%
50.70%
45.52%
59.55%
-23.28%
-23.84%
2.33%
8.42%
11.20%
28.81%
41.64%
50.47%
6.41%
21.26%
32.75%
42.62%
23.10%
50.79%
46.13%
59.46%
Kinchafonee-Muckalee HUC 03130007
Flow
TSS
TN
TP
-23.90%
-27.37%
-2.98%
0.10%
10.90%
16.78%
37.39%
39.40%
5.58%
8.52%
31.24%
34.00%
24.80%
29.04%
48.55%
51.33%
-23.75%
-27.09%
-1.72%
0.79%
10.99%
16.63%
40.15%
41.44%
5.69%
8.42%
33.23%
35.46%
24.88%
28.94%
50.51%
53.06%
Lower Flint HUC 03130008
Flow
TSS
TN
TP
-25.22%
-26.70%
-1.07%
7.81%
10.59%
23.36%
35.94%
44.98%
4.96%
15.83%
28.43%
38.40%
24.78%
46.79%
45.00%
56.09%
-25.12%
-26.59%
-0.82%
7.71%
10.50%
23.20%
36.87%
45.09%
4.90%
15.78%
28.99%
38.44%
24.64%
46.57%
45.39%
56.16%
Ichawaynochaway HUC 03130009
Flow
TSS
TN
TP
-30.02%
-28.90%
-3.74%
8.37%
6.79%
12.70%
31.14%
39.07%
0.18%
5.06%
26.85%
34.88%
22.33%
31.45%
52.16%
55.22%
-30.02%
-28.89%
-3.86%
8.28%
6.79%
12.70%
31.19%
39.13%
0.19%
5.07%
26.84%
34.89%
22.33%
31.51%
52.19%
55.27%
Spring HUC 03130010
Flow
-31.06%
3.87%
-1.58%
22.04%
-31.05%
3.88%
-1.57%
22.05%
X-26
-------
TSS
TN
TP
Results without LU Change
Min
-31.24%
-11.68%
2.54%
Median
8.99%
27.23%
35.05%
Mean
3.96%
19.99%
29.00%
Max
33.14%
45.46%
49.26%
Results with LU Change
Min
-31.26%
-11.61%
2.56%
Median
8.98%
27.33%
35.06%
Mean
3.95%
20.12%
29.07%
Max
33.11%
45.70%
49.43%
Apalachicola HUC 03130011
Flow
TSS
TN
TP
-27.10%
-47.39%
-4.82%
5.56%
7.37%
26.84%
15.84%
36.17%
3.66%
15.07%
13.38%
32.51%
23.75%
46.35%
25.28%
51.52%
-26.92%
-47.25%
-4.56%
5.41%
7.28%
26.16%
16.26%
36.25%
3.62%
14.63%
13.74%
32.58%
23.58%
46.09%
25.62%
51.62%
Chipola HUC 03130012
Flow
TSS
TN
TP
-45.73%
-46.43%
-20.99%
-6.27%
-8.86%
1.58%
13.68%
28.47%
-8.70%
7.65%
16.33%
29.53%
22.72%
62.14%
63.05%
72.78%
-45.71%
-46.52%
-21.17%
-6.62%
-8.84%
1.58%
13.55%
28.23%
-8.69%
7.63%
16.34%
29.46%
22.73%
62.12%
63.38%
73.13%
Chattahoochee at Atlanta
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
6 7
Month
10 11 12
Figure 38. Monthly Average Flows, Chattahoochee River at Atlanta (SWAT)
X-27
-------
Chattahoochee at Atlanta
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 39. Flow Duration, Chattahoochee River at Atlanta (SWAT)
Middle Chatt - Harding
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 40. Monthly Average Flows, Middle Chattahoochee River at Harding (SWAT)
X-28
-------
Middle Chatt - Harding
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 41. Flow Duration, Middle Chattahoochee River at Harding (SWAT)
Middle Chatt - WFG
700
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 42. Monthly Average Flows, Middle Chattahoochee River at WF Geroge (SWAT)
X-29
-------
Middle Chatt - WFG
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 43. Flow Duration, Middle Chattahoochee River at WF George (SWAT)
Lower Chattahoochee
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
10 11 12
Figure 44. Monthly Average Flows, Lower Chattahoochee River (SWAT)
X-30
-------
Lower Chattahoochee
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 45. Flow Duration, Lower Chattahoochee River (SWAT)
Peachtree Creek
34567
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 46. Monthly Average Flows, Peachtree Creek (SWAT)
X-31
-------
Peachtree Creek
1000
ro
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.001
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 47. Flow Duration, Peachtree Creek (SWAT)
Upper Flint
6 7
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 48. Monthly Average Flows, Upper Flint River (SWAT)
X-32
-------
Upper Flint
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 49. Flow Duration, Upper Flint River (SWAT)
Middle Flint
6 7
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 50. Monthly Average Flows, Middle Flint River (SWAT)
X-33
-------
Middle Flint
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 51. Flow Duration, Middle Flint River (SWAT)
Kinchafonee
6 7
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 52. Monthly Average Flows, Kinchafonee River (SWAT)
X-34
-------
Kinchafonee
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 53. Flow Duration, Kinchafonee River (SWAT)
Lower Flint
700
10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 54. Monthly Average Flows, Lower Flint River (SWAT)
X-35
-------
Lower Flint
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 55. Flow Duration, Lower Flint River (SWAT)
Ichawaynochaway
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
LOW1
LOW2
LOWS
LOW4
LOWS
LOW6
Figure 56. Monthly Average Flows, Ichawaynochaway Creek (SWAT)
X-36
-------
Ichawaynochaway
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 57. Flow Duration, Ichawaynochaway Creek (SWAT)
14%
12%
-2%
-4%
Month
Figure 58. Average of Median Percent Change in Flow; NARCCAP Scenarios W1-W6 at all
Stations, Apalachicola-Chattahoochee-Flint Basin (SWAT)
X-37
-------
Verde-Salt-San Pedro Basins, HSPF Model
Results at Downstream Station
OC -
Mean Annual Flow (cms)
_i. _i hj r-
D Ul O Ul O C
Verde R Tangle
Creek
$ * * • A A
-t
Z5
-&
A
§
A
*-*
BASE
G(
ICLUS
;M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 59. Mean Annual Flow, Verde River below Tangle Creek (HSPF)
Verde R Tangle Creek
8,000
_ 7'000
£
1 6,000
0 5,000
S.
n
a! 4,000
0
d. 3,000
S. 2,000
o
iZ 1,000
k.
1 0
o o
A A
+ +
A A A
i i4- LJ
d) (D
f
ww yy vy
W " «
CURRENT ICLUS CURRENT
GCM NARI
A
A
/\
^
ICLUS
;CAP
CURRENT ICLUS
BCSD
Figure 60. 100-yr Flow Peak, Verde River below Tangle Creek (HSPF)
X-38
-------
2.5 -i
1 2
_O
1 ,=
ra
1 1
c
c
01
9 0.5
1
0
Verde R Tangle
Creek
1 | • * * d
CURRENT
ICLUS
GCM
CURRENT
ICLUS
NARCCAP
CURRENT
ICLUS
BCSD
Climate and Land-Use Scenario
Figure 61. Average Annual 7-day Low Flow, Verde River below Tangle Creek (HSPF)
Verde R Tangle Creek
n A
9
•n 0.35
8 0.3
ra 0.25
LL.
i_
Jj Q2
ra
CO
w n 1*=:
i_
ra
o n 1
DC
0 05
n
* ^
_i_ rh CD
+ + A A w
A A A
9 A 5?
A,
A ^ A J?
CURRENT ICLUS CURRENT ICLUS CURRENT ICLUS
GCM NARCCAP BCSD
Figure 62. Richards-Baker Flashiness Index, Verde River below Tangle Creek (HSPF)
X-39
-------
w Centroid (Water Year Basis)
•^ -* M M I
3 8 8 8 I
u.
0
1 50
n
Verde R Tangle
Creek
*
r\
w
+
y^~
/ ^
£
d
w
+
/\
L — ^
1
^^
0
±
II
O
^^^^
CURRENT
G(
ICLUS
-M
CURRENT
NAR(
ICLUS
;CAP
CURRENT
BC
ICLUS
SD
Figure 63. Days to flow Centroid, Verde River below Tangle Creek (HSPF)
7 nnn nnn
6 000 000
^* 4 000 000
•o
O 3 000 000
i
9 nnn nnn
n
vv
Verde R Tangle Creek
O
BASE
G(
O
A
A
ICLUS
:M
A
A
$
$
j | * ft
^^
O
BASE
NAR(
^^
O
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 64. TSS Load, Verde River below Tangle Creek (HSPF)
X-40
-------
"Z"
"O
TO
5
Q.
l~ 400
n
Verde R Tangle Creek
* *
o
+
A
BASE
G(
o
¥ ¥
+ ! ! $ d
A
ICLUS
:M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 65. TP Load, Verde River below Tangle Creek (HSPF)
Verde R Tangle Creek
Qcn
300
250
h^1 200
1
•n
0 150
•z.
\-
50 -
n
1
A
n
J
A
«
^
XT\
©
•
^
_i_
._
><
+
A
&
o
A
it
BASE
ICLUS
GCM
BASE
ICLUS
NARCCAP
BASE
ICLUS
BCSD
Figure 66. TN Load, Verde River below Tangle Creek (HSPF)
X-41
-------
Results at Multiple Stations
Table 3. Summary of Range of Change Relative to Existing Conditions for NARCCAP
Dynamically Downscaled Scenarios, Verde-Salt-San Pedro Basins (HSPF)
Results without LU Change
Min
Median
Mean
Max
Results with LU Change
Min
Median
Mean
Max
Verde River nr Paulden (gage 09503700)
Flow
TSS
TN
TP
-19.22%
-46.64%
-24.46%
-35.96%
13.20%
279.87%
17.02%
215.57%
4.35%
325.08%
6.57%
248.32%
21.91%
921.15%
24.87%
702.29%
-19.77%
-46.06%
-26.72%
-35.51%
11.36%
264.17%
15.54%
210.93%
2.87%
308.86%
6.06%
244.42%
20.01%
890.16%
23.92%
702.04%
Verde River nr Clarkdale (gage 09504000)
Flow
TSS
TN
TP
-28.25%
-54.11%
-31.46%
-50.03%
-3.86%
117.39%
2.79%
82.97%
-6.28%
154.81%
-1.53%
115.72%
16.58%
443.83%
26.70%
349.54%
-28.44%
-53.86%
-32.01%
-49.86%
-3.93%
113.80%
2.92%
81.08%
-6.39%
152.75%
-1.43%
115.16%
16.14%
439.42%
26.45%
348.35%
Oak Creek at Sedona (gage 09504430)
Flow
TSS
TN
TP
-38.15%
-61.24%
-35.08%
-56.30%
-19.65%
-22.59%
-11.82%
-21.87%
-18.62%
11.14%
-10.35%
8.85%
14.90%
164.74%
33.10%
146.41%
-37.84%
-61.06%
-35.14%
-56.16%
-19.53%
-21.83%
-11.77%
-21.22%
-18.48%
11.97%
-10.23%
9.58%
14.93%
167.07%
33.35%
148.54%
Oak Creek at Cornville (gage 09504500)
Flow
TSS
TN
TP
-37.33%
-61.08%
-28.75%
-52.97%
-18.41%
-23.51%
-9.49%
-22.59%
-17.40%
11.98%
-8.30%
7.87%
15.51%
164.82%
26.81%
135.13%
-36.86%
-60.86%
-28.89%
-52.89%
-18.22%
-22.75%
-9.43%
-21.97%
-17.19%
12.66%
-8.15%
8.55%
15.51%
166.85%
27.18%
137.46%
West Clear Creek at Camp Verde (gage 09505800)
Flow
TSS
TN
TP
-43.20%
-44.58%
-43.49%
-41.20%
-18.41%
73.76%
-16.25%
58.84%
-16.72%
63.12%
-13.93%
51.02%
10.66%
161.47%
18.10%
135.70%
-43.12%
-44.63%
-43.41%
-41.22%
-18.37%
73.64%
-16.20%
58.66%
-16.70%
62.98%
-13.90%
50.85%
10.65%
161.31%
18.13%
135.47%
X-42
-------
Results without LU Change
Min
Median
Mean
Max
Results with LU Change
Min
Median
Mean
Max
Verde River at Camp Verde (gage 09506000)
Flow
TSS
TN
TP
-39.47%
-37.08%
-33.23%
-34.75%
-10.71%
97.58%
-3.95%
79.98%
-12.65%
107.27%
-6.35%
87.03%
17.18%
280.92%
26.09%
222.59%
-38.81%
-36.84%
-32.66%
-34.52%
-10.51%
96.97%
-3.65%
79.70%
-12.50%
105.74%
-6.14%
86.04%
16.91%
275.63%
26.13%
218.97%
East Verde River nr Childs (gage 09507980)
Flow
TSS
TN
TP
-34.22%
60.12%
-19.93%
54.92%
-9.91%
202.10%
-7.00%
189.14%
-9.06%
219.40%
-5.28%
205.77%
11.98%
478.93%
16.29%
451.05%
-34.20%
60.11%
-19.91%
54.92%
-9.91%
202.07%
-7.00%
189.12%
-9.06%
219.38%
-5.27%
205.75%
11.98%
478.93%
16.29%
451.05%
Verde River below Tangle Creek (gage 09508500)
Flow
TSS
TN
TP
-36.24%
-8.29%
-28.54%
-6.45%
-7.44%
127.73%
-1.20%
116.26%
-9.71%
120.35%
-4.04%
106.51%
17.07%
217.62%
24.81%
178.54%
-35.83%
-8.01%
-28.17%
-6.17%
-7.36%
127.01%
-1.01%
115.84%
-9.65%
119.53%
-3.91%
106.05%
16.89%
214.23%
24.85%
176.27%
Salt River nr Roosevelt (gage 09498500)
Flow
TSS
TN
TP
-39.83%
-0.96%
-33.90%
4.11%
-27.83%
142.09%
-20.75%
145.45%
-20.06%
930.13%
-10.78%
962.01%
25.58%
4839.81
%
35.77%
5023.14
%
-39.80%
-1.01%
-33.89%
4.05%
-27.83%
141.98%
-20.79%
145.33%
-20.06%
929.57%
-10.82%
961.41%
25.51%
4836.98
%
35.70%
5020.12
%
Salt River Outlet
Flow
TSS
TN
TP
-38.23%
14.29%
-34.80%
15.63%
-22.22%
218.77%
-18.63%
225.65%
-16.28%
865.40%
-10.07%
899.44%
24.37%
4239.97
%
31.43%
4422.64
%
-38.21%
14.27%
-34.79%
15.61%
-22.23%
218.68%
-18.66%
225.54%
-16.29%
865.08%
-10.09%
899.09%
24.33%
4238.43
%
31.39%
4420.97
%
San Pedro River nr Redington (gage 09472000)
X-43
-------
Flow
TSS
TN
TP
Results without LU Change
Min
-33.37%
-63.44%
-53.16%
-28.80%
Median
-1.82%
447.10%
295.85%
-13.76%
Mean
19.76%
647.37%
429.27%
5.12%
Max
87.15%
1888.65
%
1268.83
%
56.11%
Results with LU Change
Min
-33.37%
-63.43%
-53.14%
-28.79%
Median
-1.83%
447.01%
295.76%
-13.77%
Mean
19.76%
647.19%
429.10%
5.12%
Max
87.13%
1888.12
%
1268.31
%
56.10%
Aravaipa Crk nr Mammoth (gage 09473000)
Flow
TSS
TN
TP
-1.83%
-0.17%
-0.15%
-0.30%
5.17%
6.59%
20.28%
-0.06%
12.10%
69.35%
171.24%
0.87%
36.57%
338.59%
814.29%
3.28%
-1.99%
-0.22%
-0.42%
-0.35%
4.52%
6.70%
20.85%
-0.09%
11.57%
69.41%
170.75%
0.84%
36.08%
338.36%
809.21%
3.29%
San Pedro River Outlet
Flow
TSS
TN
TP
-15.97%
-9.43%
-19.88%
-2.32%
2.69%
72.40%
129.84%
-0.88%
18.78%
176.29%
312.06%
1.66%
70.25%
672.54%
1179.94
%
8.51%
-16.15%
-9.46%
-19.86%
-2.41%
2.32%
72.38%
129.61%
-0.96%
18.36%
176.10%
310.88%
1.62%
69.40%
671.55%
1174.21
%
8.51%
X-44
-------
Verde R Paulden
567
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 67. Monthly Average Flows, Verde River near Paulden (HSPF)
Verde R Paulden
1000
0.1
• LO
•LO
•LO
•LO
•LO
•LO
•LO
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 68. Flow Duration, Verde River near Paulden (HSPF)
X-45
-------
Verde R Clarkdale
56789
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 69. Monthly Average Flows, Verde River near Clarkdale (HSPF)
Verde R Clarkdale
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 70. Flow Duration, Verde River near Clarkdale (HSPF)
X-46
-------
Oak C Sedona
567
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 71. Monthly Average Flows, Oak Creek at Sedona (HSPF)
Oak C Sedona
1000
0.1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 72. Flow Duration, Oak Creek at Sedona (HSPF)
X-47
-------
Oak C Cornville
567
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 73. Monthly Average Flows, Oak Creek at Cornville (HSPF)
Oak C Cornville
1000
0.1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 74. Flow Duration, Oak Creek at Cornville (HSPF)
X-48
-------
W Clear C Camp Verde
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 75. Monthly Average Flows, West Clear Creek at Camp Verde (HSPF)
W Clear C Camp Verde
1000
• LOW
•LOW
•LOW
•LOW
•LOW
•LOW
LOW
0.1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 76. Flow Duration, West Clear Creek at Camp Verde (HSPF)
X-49
-------
Verde R Camp Verde
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 77. Monthly Average Flows, Verde River at Camp Verde (HSPF)
Verde R Camp Verde
10000
to
0.1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 78. Flow Durations, Verde River at Camp Verde (HSPF)
X-50
-------
E Verde R Childs
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 79. Monthly Average Flows, East Verde River at Childs (HSPF)
E Verde R Childs
1000
0.1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 80. Flow Duration, East Verde River at Childs (HSPF)
X-51
-------
Verde R Tangle Creek
567
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 81. Monthly Average Flows, Verde River at Tangle Creek (HSPF)
Verde R Tangle Creek
10000
1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 82. Flow Duration, Verde River at Tangle Creek (HSPF)
X-52
-------
30%
-40%
Month
Figure 83. Average of Median Percent Change in Flow; NARCCAP Scenarios W1-W6 at all
Stations, Verde-Basin (HSPF)
Salt R Roosevelt
LOWO
• LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
6 7
Month
10 11 12
Figure 84. Monthly Average Flows, Salt River nr Roosevelt (HSPF)
X-53
-------
Salt R Roosevelt
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 85. Flow Duration, Salt River nr Roosevelt (HSPF)
Salt R Outlet
6 7
Month
10 11 12
LOWO
• LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 86. Monthly Average Flows, Salt River at Outlet (HSPF)
X-54
-------
Salt R Outlet
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 87. Flow Duration, Salt River at Outlet (HSPF)
San Pedro R Redington
567
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 88. Monthly Average Flows, San Pedro River at Redington (HSPF)
X-55
-------
San Pedro R Redington
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 89. Flow Duration, San Pedro River at Redington (HSPF)
Aravaipa C Mammoth
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
• LOWS
•LOW4
LOWS
LOW6
Figure 90. Monthly Average Flows, Aravaipa Creek at Mammoth (HSPF)
X-56
-------
Aravaipa C Mammoth
to
I
CD
c? 0.001
L
CD
>0.0001
<
^•.00001
CD
6?oooooi
0.0000001
1E-08
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Percent of Time that Flow is Equaled or Exceeded
Figure 91. Flow Duration, Aravaipa Creek at Mammoth (HSPF)
San Pedro R Outlet
34567
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 92. Monthly Average Flows, San Pedro River at Outlet (HSPF)
X-57
-------
San Pedro R Outlet
1000
to
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 93. Flow Duration, San Pedro River at Outlet (HSPF)
30%
-50%
Month
Figure 94. Average of Median Percent Change in Flow; NARCCAP Scenarios W1-W6 at all
Stations, Salt and San Pedro Basins (HSPF)
X-58
-------
Verde-Salt-San Pedro Basins, SWAT Model
Results at Downstream Station
°n
_, 20
u
I 15
LL.
"TO
I «•
c
ra
01
^ 5
n
Verde R Tangle Cr
m ^H
s
A
BASE
G(
$
A
1
0
±
o
ft
* *
ICLUS
;M
BASE
NAR(
ICLUS
;CAP
BASE
BC
it
ICLUS
SD
Figure 95. Mean Annual Flow, Verde River below Tangle Creek (SWAT)
X-59
-------
Verde R Tangle Cr
6 000
"uT
EC nnn
u
§ 4 000
£
ra
^" ? nnn
WD '
.2
^
£
— 1 nnn
i_
>
o
0 n
1H U
+ +
A A
Z_A *— *
+ + @
* * A + ~*
Q 3
8 8
•fr ^
BASE ICLUS BASE ICLUS BASE ICLUS
GCM NARCCAP BCSD
Figure 96. 100-yr Flow Peak, Verde River below Tangle Creek (SWAT)
Average Annual 7-day Low Flow (cms)
OOOO _k _k _k _».
Dho^cnbo-^Ko^cnboh
Verde R Tangle Cr
W ^ _!__!_
* *
A
A
A A ^ ^ © ©
BASE
G(
ICLUS
:M
A.
w
BASE
NAR(
A.
M
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 97. Average Annual 7-day Low Flow, Verde River below Tangle Creek (SWAT)
X-60
-------
«* n c; -
Richards-Baker Flashiness lnde>
3 O O O C
-^ ho co !u i;
Q
*
Verde R Tangle Cr
* +
*
A~
BASE
G(
ft
A
ICLUS
:M
*
BASE
NAR(
+
f
m
ICLUS
;CAP
BASE
BC
8
ICLUS
SD
Figure 98. Richards-Baker Flashiness Index, Verde River below Tangle Creek (SWAT)
£250
t/t
re
m
re
HI
>; 200
i
|150-
J=
HI
O
^ -inn -
Days to Flo
g i
Verde R Tangle Cr
* &
Q
A
BASE
G(
S A
A
ICLUS
:M
i
BASE
NAR(
J_
T
ICLUS
;CAP
J
21
BASE
BC
j
5l
ICLUS
SD
Figure 99. Days to Flow Centroid, Verde River below Tangle Creek (SWAT)
X-61
-------
1 600 000
1 400 000
ro Q(-\n nnn
LO
{^ cnn nnn
Verde R Tangle
Cr
4- +
.
W
o
A
Q
A
•£ w , _j_ *->
A
A
8
*
t?
BASE
G(
ICLUS
;M
o
*
BASE
NAR(
O
*
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 100. TSS Load, Verde River below Tangle Creek (SWAT)
Verde R Tangle
1 000
>
2
- 600
ra
5
Q.
l~ 4nn
9
A
8
A
+
^
O
Cr
+
"&
o
A
JL,
^4
O
o
A
A
w
BASE
ICLUS
GCM
BASE
ICLUS
NARCCAP
BASE
ICLUS
BCSD
Figure 101. TP Load, Verde River below Tangle Creek (SWAT)
X-62
-------
700
600
|
ra
5
Z 300
1-
200
100 -
Verde R Tangle
Cr
+ +
*
^
*
A
i
i
o
§
O
§
ft ft
0
O
BASE
G(
ICLUS
;M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 102. TN Load, Verde River below Tangle Creek (SWAT)
Results at Multiple Stations
Table 4. Summary of Range of Change Relative to Existing Conditions for NARCCAP
Dynamically Downscaled Scenarios, Verde-Salt-San Pedro Basins, SWAT Model
Results without LU Change
Min
Median Mean Max
Results with LU Change
Min Median Mean Max
West Clear Creek (gage 09505800)
Flow
TSS
TN
TP
-87.93%
543.91%
21.54%
-15.30%
-68.10%
3311.03%
123.92%
195.72%
-69.82%
3021.91
%
115.04%
169.16%
-54.47%
5220.10
%
190.03%
296.84%
-87.91%
552.58%
21.12%
-14.76%
-68.08%
3325.16
%
123.31%
196.78%
-69.79%
3034.34
%
114.49%
170.16%
-54.43%
5235.42
%
189.35%
298.27%
East Verde (gage 09507980)
Flow
TSS
-66.54%
-78.88%
-16.20%
-11.01%
-21.46%
-18.43%
4.52%
15.93%
-66.54%
-78.81%
-16.21%
-11.01%
-21.47%
-18.42%
4.51%
15.89%
X-63
-------
TN
TP
Results without LU Change
Min
-29.56%
-73.79%
Median
32.19%
-20.35%
Mean
24.06%
-25.56%
Max
47.16%
0.91%
Results with LU Change
Min
-29.53%
-73.77%
Median
32.12%
-20.22%
Mean
24.01%
-25.48%
Max
47.03%
0.97%
Oak Creek at Sedona (gage 09504430)
Flow
TSS
TN
TP
-26.58%
-12.39%
-14.21%
-10.82%
25.10%
54.46%
-5.00%
50.38%
23.80%
54.84%
-5.55%
51.52%
88.54%
144.80%
1.10%
128.63%
-26.55%
-12.97%
-14.16%
-11.36%
24.98%
52.63%
-4.96%
49.10%
23.69%
52.78%
-5.46%
50.05%
88.29%
140.72%
1.32%
125.88%
Verde R nr Camp Verde (gage 09506000)
Flow
TSS
TN
TP
-60.26%
-46.00%
10.21%
108.69%
5.11%
148.54%
72.87%
357.05%
-10.05%
108.65%
64.09%
327.69%
18.48%
233.25%
99.87%
495.44%
-60.38%
-47.03%
11.36%
108.58%
4.95%
145.93%
74.50%
354.07%
-10.21%
106.06%
65.57%
323.53%
18.22%
227.65%
100.72%
485.56%
Verde R nr Clarkdale (gage 09504000)
Flow
TSS
TN
TP
-41.14%
40.26%
25.20%
285.71%
40.02%
323.83%
111.17%
780.15%
19.88%
346.55%
109.75%
831.65%
51.52%
855.28%
191.14%
1480.95
%
-41.65%
38.05%
27.62%
275.32%
39.32%
318.14%
114.61%
749.04%
19.19%
336.94%
111.88%
794.13%
50.94%
824.50%
191.47%
1403.92
%
Verde R below Tangle Crk (gage 09508500)
Flow
TSS
TN
TP
-60.72%
-54.22%
10.57%
85.59%
1.99%
119.71%
56.75%
290.98%
-10.64%
91.02%
49.38%
267.10%
18.41%
231.21%
74.63%
403.13%
-60.81%
-55.24%
11.33%
85.91%
1.89%
117.36%
58.16%
289.64%
-10.75%
88.48%
50.29%
264.76%
18.22%
225.00%
75.01%
396.67%
Salt River nr Roosevelt (gage 09498500)
Flow
TSS
TN
TP
-57.18%
-72.06%
-16.91%
-10.18%
-10.92%
-14.60%
9.14%
43.82%
-3.41%
-4.17%
19.70%
59.75%
69.62%
100.90%
77.68%
161.34%
-57.18%
-72.07%
-16.91%
-10.35%
-10.95%
-14.65%
9.21%
43.58%
-3.43%
-4.23%
19.77%
59.45%
69.56%
100.71%
77.78%
160.79%
Aravaipa Cr nr Mammoth (gage 09473000)
X-64
-------
Flow
TSS
TN
TP
Results without LU Change
Min
-84.35%
-93.09%
-0.71%
-89.99%
Median
-12.33%
9.80%
37.01%
18.57%
Mean
1.99%
44.27%
44.84%
56.87%
Max
90.11%
220.47%
110.18%
252.59%
Results with LU Change
Min
-84.35%
-93.09%
-0.73%
-89.94%
Median
-12.33%
9.77%
36.82%
18.54%
Mean
1.99%
44.24%
44.68%
56.66%
Max
90.11%
220.40%
109.88%
251.78%
San Pedro R nr Redington (gage 0947200)
Flow
TSS
TN
TP
-70.72%
-65.76%
-17.50%
-84.13%
5.05%
11.84%
68.76%
57.68%
18.43%
47.17%
131.04%
73.97%
118.68%
212.38%
355.31%
322.07%
-70.83%
-65.93%
-17.86%
-84.06%
5.05%
11.39%
68.34%
57.10%
18.40%
47.37%
129.83%
73.46%
118.56%
213.42%
352.89%
320.27%
Clear Creek
12
LOWO
• LOW1
LOW2
• LOWS
•LOW4
LOWS
LOW6
Figure 103. Monthly Average Flows, Clear Creek (SWAT)
X-65
-------
Clear Creek
100
CO
I
0)
O)
03
0)
ro
Q
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.00001
Percent of Time that Flow is Equaled or Exceeded
Figure 104. Flow Duration, Clear Creek (SWAT)
East Verde
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 105. Monthly Average Flows, East Verde River (SWAT)
X-66
-------
East Verde
100
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.001
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 106. Flow Duration, East Verde River (SWAT)
Oak Creek nr Sedona
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 107. Monthly Average Flows, Oak Creek near Sedona (SWAT)
X-67
-------
Oak Creek nr Sedona
100
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 108. Flow Duration, Oak Creek near Sedona (SWAT)
Verde R nr Camp Verde
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 109. Monthly Average Flows, Verde River near Camp Verde (SWAT)
X-68
-------
Verde R nr Camp Verde
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.001
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 110. Flow Duration, Verde River near Camp Verde (SWAT)
Verde R nr Clarkdale
567
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 111. Monthly Average Flows, Verde River near Clarkdale (SWAT)
X-69
-------
Verde R nr Clarkdale
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 112. Flow Duration, Verde River near Clarkdale (SWAT)
Verde R Tangle Crk
34567
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 113. Monthly Average Flows, Verde River below Tangle Creek (SWAT)
X-70
-------
Verde R Tangle Crk
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 114. Flow Duration, Verde River below Tangle Creek (SWAT)
Salt River nr Roosevelt
567
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 115. Monthly Average Flows, Salt River near Roosevelt (SWAT)
X-71
-------
Salt River nr Roosevelt
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 116. Flow Duration, Salt River near Roosevelt (SWAT)
Aravaipa Cr nr Mammoth
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 117. Monthly Average Flows, Aravaipa Creek near Mammoth (SWAT)
X-72
-------
Aravaipa Cr nr Mammoth
100
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.001
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 118. Flow Duration, Aravaipa Creek near Mammoth (SWAT)
San Pedro R nr Redington
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 119. Monthly Average Flows, San Pedro River near Redington (SWAT)
X-73
-------
100 -i
1 10
o
1 -
CD
O)
CD
m 0.1
^
^,
'CD 0.01
Onm
San Pedro R nr Redington
1
t
P
•
»L
V^S.
t > ^ ^»
^^^_
^V ^*v
\ ^
^^
^v
'
LOWO
LOW1
LOW2
LOWS
LOW4
— LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 120. Flow Duration, San Pedro River near Redington (SWAT)
120%
100%
-40%
-60%
Month
Figure 121. Average of Median Percent Change in Flow; NARCCAP Scenarios W1-W6 at all
Stations, Verde-Salt-San Pedro Basins (SWAT)
X-74
-------
Minnesota River (Upper Mississippi Basin), HSPF Model
Results at Downstream Station
300 -i
250
_ 200
t/)
u
| 150
LL.
(D
3
i 100
<
c
ro
01
^ 50
0
MN River at Mouth
"& T^f
•
•
O
0
W
BASE
ICLUS
GCM
BASE
ICLUS
NARCCAP
BASE
ICLUS
BCSD
Figure 122. Mean Annual Flow, Minnesota River at Mouth (HSPF)
R nnn
5,000
V)
E
°_ 4,000
O
n 3,00°
01
1
d. 2,000
ro
01
Q.
| 1,000
LL.
1 0
O u
>H
O
MN
O
River at Mouth
S
1
O
A
o
A
A E ~ ~ ft it
• •
BASE
G(
ICLUS
;M
BASE
NARI
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 123. 100-yr Flow Peak, Minnesota River at Mouth (HSPF)
X-75
-------
oc
Average Annual 7-day Low Flow (cms)
-^ -^ N) N) CO C
DUiocnouioc
*
A
L-±
MN River at Mouth
*
A.
•&
A
*
A
*
A
A
O
0
A
£
$
?
+ T
BASE
G(
ICLUS
:M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 124. Average Annual 7-day Low Flow, Minnesota River at Mouth (HSPF)
0.08
X
01
•n 0.07
V. nn«
Richards-Baker Flashing
o o o o o c
D 2 S 8 g 8 S
MN River at Mouth
^ * *
1
S
A
i
A
O
I
I
T
¥
BASE
G(
ICLUS
:M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 125. Richards-Baker Flashiness Index, Minnesota River at Mouth (HSPF)
X-76
-------
250 -i
to
8 200
m
HI
ra 150
:lowCentroid
I
0
g, 50
0
MN River at Mouth
it
i
-------
MN River at Mouth
AT\ nnn
TT
1
TO
5
n
0 0
± ± + +
O O
BASE ICLUS BASE ICLUS BASE ICLUS
GCM NARCCAP BCSD
Figure 128. TN Load, Minnesota River at Mouth (HSPF)
3,500 -i
3 000 -
2,500
"£
h^1 2 000
•n
0 1,500
Q.
1 nnn
500 -
n
MN
River at Mouth
•A-
,&-
m ft o
A A
O
y
i
0
i
0
$
$
+ """
BASE
ICLUS
GCM
BASE
ICLUS
NARCCAP
BASE
ICLUS
BCSD
Figure 129. TP Load, Minnesota River at Mouth (HSPF)
X-78
-------
Results at Multiple Stations
Table 5. Summary of Range of Change Relative to Existing Conditions for NARCCAP
Dynamically Downscaled Scenarios, Minnesota River Basin (HSPF)
Results without LU Change
Min
Median
Mean
Max
Results with LU Change
Min
Median
Mean
Max
Yellow Medicine River (gage 05313500)
Flow
TSS
TN
TP
-10.06%
-3.34%
-6.39%
-1.76%
9.40%
46.70%
11.21%
31.49%
7.36%
35.03%
9.90%
24.62%
20.56%
59.07%
25.60%
45.60%
-10.06%
-3.34%
-6.39%
-1.76%
9.40%
46.70%
11.21%
31.49%
7.36%
35.03%
9.90%
24.62%
20.56%
59.07%
25.60%
45.60%
Redwood River (gage 05316500)
Flow
TSS
TN
TP
-4.49%
16.60%
-5.15%
4.81%
6.06%
57.24%
13.05%
27.67%
7.85%
46.60%
11.20%
25.86%
19.63%
60.37%
27.70%
45.04%
-4.49%
16.60%
-5.15%
4.81%
6.06%
57.23%
13.05%
27.67%
7.85%
46.60%
11.20%
25.86%
19.63%
60.37%
27.70%
45.04%
Cottonwood River (gage 05317000)
Flow
TSS
TN
TP
-6.44%
-10.25%
-5.65%
-9.23%
5.79%
52.99%
11.50%
42.09%
7.01%
45.40%
13.10%
36.16%
20.75%
90.03%
31.28%
67.39%
-6.44%
-10.25%
-5.65%
-9.23%
5.79%
52.99%
11.50%
42.09%
7.01%
45.40%
13.10%
36.16%
20.75%
90.03%
31.28%
67.39%
Watonwan River (gage 05319500)
Flow
TSS
TN
TP
-14.52%
-51.07%
-12.09%
-30.36%
13.51%
75.47%
22.67%
42.97%
7.94%
52.86%
18.00%
28.82%
24.23%
119.88
%
39.28%
59.21%
-14.52%
-51.07%
-12.09%
-30.36%
13.51%
75.46%
22.67%
42.97%
7.94%
52.86%
18.00%
28.82%
24.23%
119.87
%
39.28%
59.21%
Blue Earth River (gage 05320000)
Flow
TSS
TN
-17.47%
-42.34%
-14.64%
11.89%
70.22%
14.34%
6.06%
47.96%
10.30%
19.42%
93.08%
26.52%
-17.46%
-42.34%
-14.64%
11.89%
70.22%
14.34%
6.06%
47.96%
10.30%
19.42%
93.08%
26.52%
X-79
-------
TP
Results without LU Change
Min
-26.44%
Median
35.32%
Mean
22.11%
Max
42.29%
Results with LU Change
Min
-26.44%
Median
35.32%
Mean
22.11%
Max
42.29%
LeSueur River (gage 05320500)
Flow
TSS
TN
TP
-16.91%
-36.35%
-15.26%
-21.68%
14.00%
45.96%
10.17%
32.09%
7.53%
29.11%
5.53%
20.28%
22.37%
62.79%
19.82%
39.93%
-16.91%
-36.34%
-15.26%
-21.68%
14.00%
45.95%
10.17%
32.09%
7.53%
29.10%
5.53%
20.28%
22.36%
62.78%
19.81%
39.93%
Minnesota River at Mankato (gage 05325000)
Flow
TSS
TN
TP
-10.06%
-21.39%
-10.65%
-16.40%
7.15%
43.35%
10.70%
28.88%
6.94%
42.43%
9.44%
25.91%
21.84%
86.59%
23.23%
50.99%
-10.06%
-21.39%
-10.65%
-16.40%
7.15%
43.35%
10.70%
28.87%
6.94%
42.43%
9.44%
25.91%
21.84%
86.58%
23.23%
50.99%
Minnesota River nr Jordan (gage 05330000)
Flow
TSS
TN
TP
-10.66%
-21.05%
-10.80%
-17.31%
7.67%
46.52%
11.65%
29.59%
7.15%
46.52%
10.45%
26.31%
22.77%
97.42%
25.23%
53.21%
-10.65%
-21.04%
-10.79%
-17.29%
7.67%
46.37%
11.64%
29.55%
7.15%
46.37%
10.43%
26.28%
22.76%
97.12%
25.19%
53.16%
Minnesota River at Mouth
Flow
TSS
TN
TP
-10.83%
-22.00%
-10.57%
-17.10%
7.79%
43.74%
11.48%
28.68%
7.16%
44.32%
10.39%
25.64%
23.12%
94.35%
25.07%
52.72%
-10.79%
-22.04%
-10.54%
-16.95%
7.80%
42.85%
11.40%
28.44%
7.15%
43.43%
10.31%
25.43%
23.05%
92.79%
24.88%
52.40%
X-80
-------
Yellow Medicine
Figure 130. Monthly Average Flows, Yellow Medicine River (HSPF)
Yellow Medicine
1000
CO
I
0)
O)
03
CD
03
Q 0.001
0.0001
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 131. Flow Duration, Yellow Medicine River (HSPF)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
X-81
-------
Redwood
1 2 3 4 5 6 7 8 9 10 11 12
Figure 132. Monthly Average Flows, Redwood River (HSPF)
Redwood
1000
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 133. Flow Duration, Redwood River (HSPF)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
X-82
-------
Cottonwood
1 2 3 4 5 6 7 8 9 10 11 12
Figure 134. Monthly Average Flows, Cottonwood River (HSPF)
Cottonwood
1000
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 135. Flow Duration, Cottonwood River (HSPF)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
X-83
-------
Watonwan
6 7
Month
10 11 12
Figure 136. Monthly Average Flows, Watonwan River (HSPF)
Watonwan
1000
I
CD
O)
CD
CD
CD
Q 0.001
0.0001
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 137. Flow Duration, Watonwan River (HSPF)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
X-84
-------
Blue Earth
140
10 11 12
Figure 138. Monthly Average Flows, Blue Earth River (HSPF)
Blue Earth
1000
I
CD
O)
CD
CD
CD
Q 0.001
0.0001
LOWO
•LOW1
LOW2
•LOWS
• LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 139. Flow Duration, Blue Earth River (HSPF)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
X-85
-------
LeSueur
Figure 140. Monthly Average Flows, Le Sueur River (HSPF)
LeSueur
1000
I
CD
O)
CD
CD
0.1
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Q 0.01
0.001
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 141. Flow Duration, Le Sueur River (HSPF)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
X-86
-------
MN River Mankato
600
1 2 3 4 5 6 7 8 9 10 11 12
Figure 142. Monthly Average Flows, Minnesota River at Mankato (HSPF)
MN River Mankato
10000
tn
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 143. Flow Duration, Minnesota River at Mankato (HSPF)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
X-87
-------
MN River Jordan
600
1 2 3 4 5 6 7 8 9 10 11 12
Figure 144. Monthly Average Flows, Minnesota River near Jordan (HSPF)
MN River Jordan
10000
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 145. Flow Duration, Minnesota River near Jordan (HSPF)
X-88
-------
MN River at Mouth
600
1 2 3 4 5 6 7 8 9 10 11 12
Figure 146. Monthly Average Flows, Minnesota River at Mouth (HSPF)
MN River at Mouth
10000
tn
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 147. Flow Duration, Minnesota River at Mouth (HSPF)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
X-89
-------
140%
120%
-20%
Month
Figure 148. Average of Median Percent Change in Flow; NARCCAP Scenarios W1-W6 at all
Stations, Minnesota River Basin (HSPF)
X-90
-------
Minnesota River (Upper Mississippi Basin), SWAT Model
Results at Downstream Station
300
1" 250
u
> onn -
Mean Annual Flo
-J. _i r-
Ul O Ul C
D O O O C
Lower Minnesota
* *
&
A
O
A
O
m
m
A
0
A
0
w w
^^
+
o
o
BASE
G(
ICLUS
;M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 149. Mean Annual Flow, Lower Minnesota River (SWAT)
Lower Minnesota
R nnn
5,000
"wT
°^ 4,000
O
% 3,000
£
0
d. 2,000
ro
£
| 1,000
LL.
1 0
o u
tH
0 0
A A
W
+
BASE
G(
A
A
A
A
w £ i ¥ S
+
ICLUS
;M
BASE
NARI
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 150.100-yr Flow Peak, Lower Minnesota River (SWAT)
X-91
-------
fin
f 50
u
O
H 40
5
-n" 30
rage Annua
^ N)
D O
1
n
Lower Minnesota
* ft
A
+
BASE
G(
A
o
+.
ICLUS
:M
V
A
+
A
BASE
NAR(
A
+
A
o
kj
ICLUS
;CAP
A
0
A
Q
^^^i 3i
BASE
BC
ICLUS
SD
Figure 151. Average Annual 7-day Low Flow, Lower Minnesota River (SWAT)
Richards-Baker Flashiness Index
ooooooooo
bbbbbbbbb?
D-^hoco^oio^ioocD-
$
Lower Minnesota
$
£
^
BASE
G(
ICLUS
:M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 152. Richards-Baker Flashiness Index, Lower Minnesota River (SWAT)
X-92
-------
300 -i
f 250
re
m
^ 200
HI
Flow Centroid (Wa
I 8
0
1 50
0
Lower Minnesota
-A,
-4-
^
^
$~ W f $ W W
BASE
ICLUS
GCM
BASE
ICLUS
NARCCAP
BASE
ICLUS
BCSD
Figure 153. Days to Flow Centroid, Lower Minnesota River (SWAT)
Lower Minnesota
A ^nn nnn
3 500 000
.—
l~ o ^nn nnn
1 000 000
cnn nnn
n
* ^
m
A
•
A A
+
O O A
O
+
A
O
®
•fc
+
©
•&
+
+ +
BASE ICLUS BASE
GCM NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 154.TSS Load, Lower Minnesota River (SWAT)
X-93
-------
^
s
^
TO
5
1— 20 000
n
Lower Minnesota
* * 4 •
A
O
A
O
0
+ +
BASE
G(
ICLUS
:M
BASE
NAR(
n
ICLUS
;CAP
6
+
^^^^^™
BASE
BC
6
vv
ICLUS
SD
Figure 155.TN Load, Lower Minnesota River (SWAT)
R nnn
4 000
i
"O
TO
5
Q.
h- 9 nnn
n
Lower Minnesota
*
A
x^^^
v^^
+
BASE
G(
a
A
A
+
ill
+
ICLUS
:M
BASE
NAR(
A
+
ICLUS
;CAP
Q
{
BASE
BC
Q
J
ICLUS
SD
Figure 156.TP Load, Lower Minnesota River (SWAT)
X-94
-------
Results at Multiple Stations
Table 6. Summary of Range of Change Relative to Existing Conditions for NARCCAP
Dynamically Downscaled Scenarios, Minnesota River Basin SWAT Model
Results without LU Change
Min
Median
Mean
Max
Results with LU Change
Min
Median
Mean
Max
Upper Minnesota HUC 07020001
Flow
TSS
TN
TP
-3.95%
-17.42%
19.73%
-12.85%
39.98%
44.98%
70.17%
20.85%
36.17%
47.84%
67.01%
25.01%
85.38%
127.01%
122.77%
75.93%
-3.95%
-17.42%
19.73%
-12.85%
39.98%
44.98%
70.19%
20.87%
36.17%
47.84%
67.03%
25.03%
85.38%
127.01%
122.80%
75.97%
Pomme de Terre HUC 07020003
Flow
TSS
TN
TP
-6.62%
-21.07%
13.41%
-19.42%
33.48%
30.07%
53.98%
16.76%
32.66%
39.65%
56.56%
24.47%
77.84%
115.54%
107.51%
83.00%
-6.62%
-21.07%
13.41%
-19.42%
33.48%
30.07%
53.99%
16.79%
32.66%
39.65%
56.57%
24.50%
77.84%
115.54%
107.53%
83.04%
Lac qui Parle HUC 07020002
Flow
TSS
TN
TP
-7.31%
-17.12%
9.56%
-13.85%
36.35%
35.86%
54.59%
16.56%
32.40%
33.84%
48.18%
18.02%
71.11%
100.96%
83.64%
57.03%
-7.31%
-17.12%
9.56%
-13.86%
36.35%
35.86%
54.60%
16.57%
32.40%
33.84%
48.18%
18.03%
71.11%
100.96%
83.64%
57.05%
Yellow Medicine River HUC 07020004 (part)
Flow
TSS
TN
TP
-7.92%
-16.42%
0.03%
-20.52%
34.59%
41.72%
50.01%
18.54%
31.53%
36.16%
43.94%
18.61%
67.46%
94.48%
78.07%
54.03%
-7.92%
-16.42%
0.12%
-20.52%
34.59%
41.72%
50.02%
18.55%
31.53%
36.16%
43.97%
18.62%
67.46%
94.48%
78.07%
54.05%
Chippewa River HUC 07020005
Flow
TSS
TN
TP
-9.95%
-22.48%
-0.76%
-24.54%
25.69%
32.97%
36.51%
10.93%
23.79%
40.29%
36.79%
17.45%
65.55%
126.59%
78.51%
70.94%
-9.95%
-22.48%
-0.76%
-24.56%
25.68%
32.97%
36.53%
10.91%
23.79%
40.29%
36.81%
17.46%
65.55%
126.59%
78.50%
70.89%
X-95
-------
Results without LU Change
Min
Median
Mean
Max
Results with LU Change
Min
Median
Mean
Max
Redwood River HUC 07020006
Flow
TSS
TN
TP
-3.78%
-6.11%
-0.31%
-5.61%
24.23%
32.18%
36.42%
13.19%
25.92%
36.06%
34.70%
16.08%
53.95%
86.18%
64.01%
44.46%
-3.78%
-6.11%
-0.31%
-5.61%
24.23%
32.18%
36.43%
13.20%
25.92%
36.06%
34.70%
16.09%
53.94%
86.18%
64.02%
44.46%
Middle Minnesota HUC 07020007
Flow
TSS
TN
TP
-12.67%
-17.81%
4.32%
-5.46%
29.75%
52.83%
43.11%
25.85%
27.94%
47.38%
41.13%
24.83%
62.35%
101.14%
70.64%
59.67%
-12.67%
-17.81%
4.36%
-5.39%
29.74%
52.82%
43.18%
25.92%
27.94%
47.38%
41.21%
24.88%
62.34%
101.13%
70.72%
59.72%
Cottonwood HUC 07020008
Flow
TSS
TN
TP
-10.24%
-5.91%
4.22%
-5.03%
33.77%
43.69%
45.84%
20.93%
31.00%
48.11%
44.97%
25.34%
61.72%
104.61%
75.87%
63.13%
-10.24%
-5.91%
4.22%
-5.03%
33.77%
43.69%
45.84%
20.93%
31.00%
48.11%
44.97%
25.34%
61.72%
104.61%
75.87%
63.13%
Blue Earth HUC 07020009
Flow
TSS
TN
TP
-20.61%
-24.04%
-11.15%
-19.24%
33.56%
57.91%
45.50%
39.82%
22.74%
45.79%
35.07%
29.02%
49.17%
98.51%
63.82%
67.12%
-20.61%
-24.04%
-11.14%
-19.24%
33.55%
57.91%
45.55%
39.90%
22.74%
45.78%
35.09%
29.07%
49.17%
98.50%
63.86%
67.19%
Watonwan HUC 07020010
Flow
TSS
TN
TP
-23.39%
-30.63%
-23.34%
-31.51%
40.81%
62.21%
46.15%
38.20%
28.64%
52.72%
38.06%
31.53%
65.41%
124.99%
89.14%
92.71%
-23.39%
-30.63%
-23.34%
-31.52%
40.81%
62.21%
46.15%
38.19%
28.64%
52.72%
38.05%
31.51%
65.41%
124.98%
89.14%
92.70%
LeSueur HUC 07020011
Flow
TSS
TN
-13.79%
-5.12%
-0.63%
31.64%
52.63%
41.60%
21.83%
41.27%
33.83%
43.28%
75.30%
56.00%
-13.80%
-5.12%
-0.23%
31.63%
52.63%
42.25%
21.81%
41.27%
34.33%
43.26%
75.29%
56.43%
X-96
-------
TP
Results without LU Change
Min
-5.35%
Median
35.45%
Mean
28.50%
Max
57.54%
Results with LU Change
Min
-4.78%
Median
36.34%
Mean
29.17%
Max
58.15%
Lower Minnesota HUC 07020012
Flow
TSS
TN
TP
-14.28%
-22.81%
4.88%
-3.26%
29.82%
53.38%
43.83%
26.08%
27.39%
52.06%
42.08%
26.14%
62.41%
124.50%
71.13%
60.28%
-14.34%
-22.87%
4.88%
-2.52%
29.59%
52.90%
44.37%
27.28%
27.16%
51.41%
42.44%
27.15%
62.09%
122.93%
71.09%
60.74%
Minnesota River at Jordan (gage 05330000)
Flow
TSS
TN
TP
-13.54%
-19.49%
5.17%
-6.65%
30.38%
53.66%
42.75%
23.91%
28.14%
51.35%
41.55%
24.52%
63.25%
115.45%
70.98%
59.80%
-13.58%
-19.48%
5.17%
-6.28%
30.31%
53.58%
42.84%
24.19%
28.07%
51.27%
41.58%
24.68%
63.16%
115.24%
70.91%
59.53%
Minnesota River at Mankato (gage 05325000)
Flow
TSS
TN
TP
-13.01%
-17.21%
4.83%
-5.52%
30.10%
50.29%
42.80%
25.21%
28.11%
46.00%
41.32%
24.99%
62.93%
101.54%
70.91%
59.64%
-13.01%
-17.21%
4.85%
-5.49%
30.09%
50.28%
42.85%
25.27%
28.10%
46.00%
41.38%
25.02%
62.93%
101.52%
70.97%
59.65%
X-97
-------
Upper Minnesota
120
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 56. Monthly Average Flows, Upper Minnesota River (SWAT)
Upper Minnesota
1000
o
CD
O)
£
CD
Q 0.01
0.1
0.001
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 158. Flow Duration, Upper Minnesota River (SWAT)
X-98
-------
Pomme de Terre
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 159. Monthly Average Flows, Pomme de Terre (SWAT)
Pomme de Terre
1000
0.01
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 160. Flow Duration, Pomme de Terre (SWAT)
X-99
-------
Lac qui Parle
567
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 161. Monthly Average Flows, Minnesota River at Lac qui Parle (SWAT)
Lac qui Parle
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 162. Flow Duration, Minnesota River at Lac qui Parle (SWAT)
X-100
-------
Yellow Medicine River
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 163. Monthly Average Flows, Yellow Medicine River (SWAT)
Yellow Medicine River
1000
CO
^
o
0)
O)
03
CD
ro
Q
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.00001
Percent of Time that Flow is Equaled or Exceeded
Figure 164. Flow Duration, Yellow Medicine River (SWAT)
X-101
-------
Chippewa
6 7
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 165. Monthly Average Flows, Chippewa River (SWAT)
Chippewa
1000
CO
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 166. Flow Duration, Chippewa River (SWAT)
X-102
-------
Redwood
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 167. Monthly Average Flows, Redwood River (SWAT)
Redwood
1000
0.01
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 168. Flow Duration, Redwood River (SWAT)
X-103
-------
Middle Minnesota
700
600
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 169. Monthly Average Flows, Middle Minnesota River (SWAT)
Middle Minnesota
10000
0.001
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 170. Flow Duration, Middle Minnesota River (SWAT)
X-104
-------
Cottonwood
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 171. Monthly Average Flows, Cottonwood River (SWAT)
Cottonwood
1000
CO
^
o
0)
O)
£
0)
Q 0.01
0.001
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 172. Flow Duration, Cottonwood River (SWAT)
X-105
-------
Blue Earth
6 7
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 173. Monthly Average Flows, Blue Earth River (SWAT)
Blue Earth
1000
o
CD
O)
£
CD
Q 0.01
0.001
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 174. Flow Duration, Blue Earth River (SWAT)
X-106
-------
Watonwan
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 175. Monthly Average Flows, Watowan River (SWAT)
Watonwan
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 176. Flow Duration, Watowan River (SWAT)
X-107
-------
160%
11 12
Figure 177. Average of Median Percent Change in Flow; NARCCAP Scenarios W1-W6 at all
Stations, Minnesota River Basin (SWAT)
X-108
-------
Susquehanna River Basin, HSPF Model
Results at Downstream Station
Susq R Outlet
1,200
1,000
u ann
o
Li-
re
3
C
C
^ 400
ra
01
1
200
n
&~
+
©
^^
i
* *
_
H
BASE
G(
ICLUS
;M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 178. Mean Annual Flow, Susquehanna River Outlet (HSPF)
X-109
-------
30,000
1 25,000
u
g 20,000
ra
£
BO
5
"ra 10,000
3
£ 5,000
O
° 0
Susq R Outlet
+
g
8
A
t
V
A
±
V
^^
0
A
+
A
BASE
ICLUS
GCM
BASE
ICLUS
NARCCAP
BASE
ICLUS
BCSD
Figure 179. 100-yr Flow Peak, Susquehanna River Outlet (HSPF)
"i/T
I 200
_o
LL.
I 150
ra
-a
"ra -inn -
Average Annu
8 I
*
A
BASE
G(
Susq R Outlet
*
A
ICLUS
:M
X
•
BASE
NAR(
+
JL
"ir£ "
A +
•
Jk ifa
o
ICLUS
;CAP
BASE
BC
+
ICLUS
SD
Figure 180. Average Annual 7-day Low Flow, Susquehanna River Outlet (HSPF)
x-no
-------
0.25 -i
| 0.2-
c
3
01
-s °-15
ra
LL.
1
& 0.1
1/1
"S
ra
^
* 0.05
0
Susq R Outlet
©
4.
© |
A
A SB?
^D /"A
I
db
BASE
ICLUS
GCM
BASE
ICLUS
NARCCAP
BASE
ICLUS
BCSD
Figure 181. Richards-Baker Flashiness Index, Susquehanna River Outlet (HSPF)
180
"t/T
S 160
CO
0)
+•' -ion
s to FlowCentroid (Wa
^. O) 00 O h
3 0 0 0 C
20
+ J-
§ 9
Susq R Outlet
£
(i
i
0
s
0
BASE ICLUS
GCM
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 182. Days to Flow Centroid, Susquehanna River Outlet (HSPF)
X-lll
-------
t ^ nnn nnn
i
— *• A nnn nnn
ra
5
9 nnn nnn
Susq R Outlet
+ +
•^ ^
.A
Q
j\
®
»S t~i
y
•
A
w
4
y
V
A
M
2
2
BASE
G(
ICLUS
;ivi
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 183. TSS Load, Susquehanna River Outlet (HSPF)
Susq R Outlet
2 000
L.
§
•D
TO
O_
i ^ ^-x
* ' m
BASE
G(
W
ICLUS
;M
BASE
NAR(
® S
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 184. TP Load, Susquehanna River Outlet (HSPF)
X-112
-------
t en nnn
i
— *• Ar\ nnn
ra
5
1-
20 000
+
0-
Susq R Outlet
+
ff
t? W
A
BASE
G(
ICLUS
;M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 185. TN Load, Susquehanna River Outlet (HSPF)
X-113
-------
Results at Multiple Stations
Table 7. Summary of Range of Change Relative to Existing Conditions for NARCCAP
Dynamically Downscaled Scenarios, Susquehanna River Basin HSPF Model
Results without LU Change
Min
Median
Mean
Max
Results with LU Change
Min
Median
Mean
Max
Raystown Branch Juniata River at Saxton (gage 01562000)
Flow
TSS
TN
TP
-17.99%
-17.18%
-17.45%
-13.26%
-3.61%
42.95%
-2.53%
-6.66%
-1.61%
45.51%
-0.70%
-6.21%
14.36%
115.70
%
16.09%
2.03%
-17.99%
-17.18%
-17.45%
-13.26%
-3.61%
42.95%
-2.53%
-6.66%
-1.61%
45.51%
-0.70%
-6.21%
14.36%
115.70
%
16.09%
2.03%
WB Susquehanna River at Lewisberg
Flow
TSS
TN
TP
-20.17%
-4.00%
-15.17%
-7.37%
-2.47%
28.43%
-0.43%
-3.41%
-5.30%
26.18%
-3.23%
-3.88%
0.18%
44.79%
1.51%
-2.38%
-20.14%
-4.02%
-15.12%
-7.35%
-2.45%
28.31%
-0.40%
-3.39%
-5.29%
26.08%
-3.21%
-3.86%
0.20%
44.63%
1.51%
-2.36%
Susquehanna River at Danville (gage 01540500)
Flow
TSS
TN
TP
-14.40%
5.03%
-13.56%
-7.90%
-1.67%
30.55%
-1.86%
-4.19%
-3.58%
30.01%
-3.74%
-4.77%
-0.50%
57.81%
-1.10%
-3.53%
-14.40%
5.03%
-13.56%
-7.90%
-1.67%
30.55%
-1.85%
-4.19%
-3.58%
30.01%
-3.74%
-4.77%
-0.50%
57.81%
-1.10%
-3.53%
Susquehanna River at Marietta (gage 01576000)
Flow
TSS
TN
TP
-15.39%
0.59%
-13.82%
-8.00%
-1.56%
33.94%
-0.79%
-3.87%
-3.51%
30.88%
-2.74%
-4.28%
1.06%
53.97%
1.48%
-2.19%
-16.29%
-0.32%
-14.16%
-8.14%
-1.54%
33.81%
-0.77%
-3.83%
-3.65%
30.60%
-2.79%
-4.27%
1.06%
53.76%
1.47%
-2.16%
Susquehanna River Outlet
Flow
TSS
TN
TP
-15.14%
0.15%
-13.43%
-8.39%
-1.35%
33.95%
-0.36%
-3.98%
-3.33%
31.38%
-2.39%
-4.40%
1.40%
55.42%
1.99%
-2.02%
-15.95%
-0.62%
-13.71%
-8.41%
-1.32%
33.68%
-0.33%
-3.87%
-3.44%
30.94%
-2.44%
-4.33%
1.41%
54.89%
1.96%
-1.96%
X-114
-------
Juniata R Saxton
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
567
Month
8 9 10 11 12
Figure 186. Monthly Average Flow, Raystown Branch Juniata River at Saxton (HSPF)
Juniata R Saxton
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 187. Flow Duration, Raystown Branch Juniata River at Saxton (HSPF)
X-115
-------
WB Susq R Lewisberg
600
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 188. Monthly Average Flow, West Branch Susquehanna River at Lewisberg (HSPF)
WB Susq R Lewisberg
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 189. Flow Duration, West Branch Susquehanna River at Lewisberg (HSPF)
X-116
-------
Susq R Danville
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 190. Monthly Average Flow, Susquehanna River at Danville (HSPF)
Susq R Danville
10000
1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 191. Flow Duration, Susquehanna River at Danville (HSPF)
X-117
-------
Susq R Marietta
2500
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 192. Monthly Average Flow, Susquehanna River at Marietta (HSPF)
Susq R Marietta
100000
tn
1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 193. Flow Duration, Susquehanna River at Marietta (HSPF)
X-118
-------
Susq R Outlet
2500
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 194. Monthly Average Flow, Susquehanna River at Outlet (HSPF)
Susq R Outlet
100000
tn
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 195. Flow Duration, Susquehanna River at Outlet (HSPF)
X-119
-------
01
M
TO
U
0)
Q.
•a
OJ
0)
M
OJ
-30%
Month
Figure 196. Average of Median Percent Change in Flow; NARCCAP Scenarios W1-W6 at all
Stations, Susquehanna River Basin (HSPF)
X-120
-------
Susquehanna River Basin, SWAT Model
Results at Downstream Station
1,200 -i
1,000
_, 800
o
1 600
Li-
re
3
i 400
c
re
01
^ 200
0
Susquehanna mouth (2050306)
4
O
&
O
i
*
i
*
X
~1
BASE
ICLUS
GCM
BASE
ICLUS
NARCCAP
BASE
ICLUS
BCSD
Figure 197. Mean Annual Flow, Susquehanna River Mouth (SWAT)
X-121
-------
E9^ nnn
u
ro
£
BO
5.
— ^ nnn
o
° 0
Susquehanna mouth (2050306)
8
A
BASE
G(
8
A
ICLUS
:M
A
ft
BASE
NAR(
+ +
A
ft
f_
ICLUS
;CAP
*
0
ZA
BASE
BC
*
O
A
ICLUS
SD
Figure 198.100-yr Flow Peak, Susquehanna River Mouth (SWAT)
*i/r
I 200
_o
LL.
0 150
>
ra
TJ
ix
"ra -inn
Average Annu
8 I
*
A
£A
o
BASE
G(
Susquehanna mouth (2050306)
$ +
A A
ft
0 *
ICLUS
:M
BASE
NAR(
+
A *
•
Q
ICLUS
;CAP
BASE
BC
*
o
ICLUS
SD
Figure 199. Average Annual 7-day Low Flow, Susquehanna River Mouth (SWAT)
X-122
-------
x 0 25
01
•o
c
Richards-Baker Flashiness 1
o o
b ° ^ °
D O1 -* O1 M
Susquehanna mouth (2050306)
.
•4- 36 A* Jk.
ffl
BASE
G(
£^
»^
ICLUS
:M
1
BASE
NAR(
a
ICLUS
;CAP
A
®
BASE
BC
ICLUS
SD
Figure 200. Richards-Baker Flashiness Index, Susquehanna River Mouth (SWAT)
180
"t/T
'8 160
CO
S 140
0)
+•' -ion
Days to Flow Centroid (Wa
O -fc. O) 00 O h
D 0 0 0 0 C
Susquehanna mouth (2050306)
A
n
•
W
A
n
=A^
8
A
+
A
+
A
+
A
BASE
G(
ICLUS
;M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 201. Days to Flow Centroid, Susquehanna River Mouth (SWAT)
X-123
-------
8 000 000
7 000 000
0 4,000,000
LO
{^ -a nnn nnn
A
*
Susquehanna mouth (2050306)
A
-k
o
£ i
_
*
_
*
n
n
BASE
G(
ICLUS
;M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 202. TSS Load, Susquehanna River Mouth (SWAT)
6,000 -i
5,000
_ 4,000
i_
1
~ 3,000
ra
5
Q.
"~ 2,000
1,000
0
Susquehanna mouth (2050306)
n
l_
o
*
«
$
®
f
A
A
sV W
BASE
ICLUS
GCM
BASE
ICLUS
NARCCAP
BASE
ICLUS
BCSD
Figure 203. TP Load, Susquehanna River Mouth (SWAT)
X-124
-------
on nnn
i
— *• en nnn
ra
5
z
1— Art nnn
20 000 -
A
Susquehanna mouth (2050306)
A
• ^ ^P
r/
A
£T)
1
BASE
G(
ICLUS
;M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 204. TN Load, Susquehanna River Mouth (SWAT)
X-125
-------
Results at Multiple Stations
Table 8. Summary of Range of Change Relative to Existing Conditions for NARCCAP
Dynamically Downscaled Scenarios, Susquehanna River Basin (SWAT)
Results without LU Change
Min
Median
Mean
Max
Results with LU Change
Min
Median
Mean
Max
Upper Susquehanna HUC 02050101
Flow
TSS
TN
TP
-2.93%
-5.62%
45.88%
3.04%
12.56%
18.17%
75.36%
27.68%
10.45%
14.02%
76.83%
27.36%
15.50%
21.68%
99.64%
59.62%
-2.93%
-5.62%
45.88%
3.05%
12.56%
18.17%
75.36%
27.69%
10.45%
14.02%
76.83%
27.37%
15.50%
21.67%
99.64%
59.64%
Chenango HUC 02050102
Flow
TSS
TN
TP
-1.33%
-6.39%
32.00%
6.83%
13.06%
15.86%
48.75%
14.34%
10.39%
13.94%
51.04%
12.96%
16.13%
24.18%
68.74%
16.77%
-1.32%
-6.42%
31.98%
6.78%
13.06%
15.82%
48.73%
14.30%
10.39%
13.90%
51.02%
12.92%
16.13%
24.14%
68.71%
16.72%
Owego-Wappasening HUC 02050103
Flow
TSS
TN
TP
-2.16%
-8.14%
39.16%
1.55%
12.58%
17.35%
61.49%
14.40%
10.35%
13.91%
63.16%
13.10%
16.13%
25.19%
82.59%
24.10%
-2.16%
-8.13%
39.15%
1.53%
12.58%
17.36%
61.46%
14.36%
10.35%
13.91%
63.13%
13.07%
16.13%
25.20%
82.56%
24.07%
Tioga HUC 02050104
Flow
TSS
TN
TP
-5.57%
-9.02%
21.38%
-6.17%
12.57%
17.47%
42.38%
10.06%
10.25%
14.23%
39.90%
7.86%
16.43%
25.28%
52.61%
13.87%
-5.57%
-9.02%
21.37%
-6.17%
12.57%
17.48%
42.37%
10.05%
10.25%
14.23%
39.90%
7.85%
16.43%
25.28%
52.60%
13.87%
Chemung HUC 02050105
Flow
TSS
TN
TP
-4.38%
12.77%
28.01%
-1.02%
12.63%
19.73%
49.52%
13.10%
10.18%
18.69%
46.02%
12.22%
19.43%
44.35%
57.58%
20.53%
-4.38%
12.77%
28.01%
-1.02%
12.63%
19.73%
49.52%
13.10%
10.18%
18.69%
46.02%
12.22%
19.43%
44.35%
57.58%
20.53%
X-126
-------
Results without LU Change
Min
Median
Mean
Max
Results with LU Change
Min
Median
Mean
Max
Upper Susquehanna - Lackawanna HUC 02050107
Flow
TSS
TN
TP
-5.85%
10.13%
25.64%
3.34%
8.79%
17.56%
44.63%
14.91%
7.24%
13.91%
43.97%
13.94%
14.08%
24.94%
58.47%
20.04%
-5.85%
10.13%
25.64%
3.33%
8.79%
17.56%
44.62%
14.90%
7.24%
13.91%
43.96%
13.93%
14.08%
24.95%
58.46%
20.03%
Upper West Branch Susquehanna HUC 02050201
Flow
TSS
TN
TP
23.80%
4.05%
67.29%
33.51%
-2.72%
63.07%
94.34%
45.30%
-4.76%
57.82%
90.48%
44.90%
2.53%
82.45%
99.00%
57.08%
23.81%
4.02%
67.27%
33.74%
-2.73%
63.02%
94.17%
45.08%
-4.77%
57.77%
90.42%
44.99%
2.52%
82.39%
99.04%
57.26%
Sinnemahoning HUC 02050202
Flow
TSS
TN
TP
14.29%
14.32%
36.24%
-4.98%
5.33%
17.04%
62.62%
23.38%
3.08%
14.25%
62.69%
22.14%
9.54%
26.19%
84.45%
44.63%
14.29%
14.32%
36.15%
-5.09%
5.33%
17.04%
62.48%
23.17%
3.08%
14.25%
62.56%
21.94%
9.54%
26.19%
84.28%
44.37%
Pine HUC 02050205
Flow
TSS
TN
TP
-8.36%
10.51%
-2.43%
24.19%
9.77%
16.48%
23.34%
9.79%
6.66%
12.22%
22.34%
5.12%
11.67%
20.93%
37.97%
22.56%
-8.36%
10.51%
-2.43%
24.13%
9.77%
16.48%
23.31%
9.77%
6.66%
12.22%
22.32%
5.11%
11.67%
20.93%
37.95%
22.53%
Lower West Branch Susquehanna HUC 02050206
Flow
TSS
14.99%
20.99%
3.29%
3.46%
0.92%
0.98%
7.08%
9.63%
14.94%
20.89%
3.32%
3.57%
0.95%
1.09%
7.11%
9.73%
X-127
-------
TN
TP
Results without LU Change
Min
20.95%
5.23%
Median
38.10%
19.50%
Mean
37.14%
25.19%
Max
48.13%
46.70%
Results with LU Change
Min
20.71%
4.95%
Median
37.79%
19.21%
Mean
36.90%
25.15%
Max
47.89%
47.18%
Lower Susquehanna - Penns HUC 02050301
Flow
TSS
TN
TP
-9.65%
16.11%
25.20%
7.40%
6.11%
11.76%
42.75%
14.40%
4.62%
8.58%
42.26%
18.75%
11.36%
19.88%
55.42%
32.05%
-9.64%
16.10%
25.13%
7.24%
6.12%
11.77%
42.68%
14.23%
4.63%
8.58%
42.20%
18.66%
11.37%
19.89%
55.35%
32.22%
Raystown HUC 02050303
Flow
TSS
TN
TP
14.94%
16.49%
57.89%
-1.88%
9.00%
32.88%
80.61%
21.52%
7.75%
34.95%
84.77%
25.52%
25.79%
82.05%
110.76%
47.14%
14.94%
16.49%
57.90%
-1.86%
9.00%
32.88%
80.62%
21.55%
7.75%
34.95%
84.78%
25.55%
25.79%
82.05%
110.78%
47.19%
Lower Juniata HUC 02050304
Flow
TSS
TN
TP
10.44%
11.79%
41.30%
-2.88%
10.01%
17.24%
55.01%
5.01%
7.44%
13.86%
57.70%
5.03%
14.53%
24.60%
76.08%
14.45%
10.43%
11.78%
41.29%
-2.90%
10.02%
17.29%
55.01%
5.02%
7.44%
13.89%
57.69%
5.03%
14.53%
24.64%
76.09%
14.45%
Susquehanna mouth HUC 02050306
Flow
TSS
TN
TP
10.08%
15.67%
32.30%
6.27%
7.19%
11.82%
48.76%
12.74%
4.92%
8.49%
49.18%
15.92%
10.98%
17.75%
62.17%
27.94%
-9.97%
15.55%
32.07%
6.28%
7.27%
11.90%
48.51%
12.63%
5.00%
8.59%
48.96%
15.92%
11.04%
17.84%
61.94%
28.09%
Susquehanna River at Marietta (gage 01576000)
Flow
-9.97%
6.97%
4.96%
11.18% I! -9.94%
6.99%
4.98%
11.19%
X-128
-------
TSS
TN
TP
Results without LU Change
Min
15.13%
29.75%
6.04%
Median
12.16%
46.51%
12.82%
Mean
8.77%
46.61%
16.08%
Max
18.47%
59.51%
28.35%
Results with LU Change
Min
15.12%
29.62%
5.91%
Median
12.12%
46.37%
12.68%
Mean
8.74%
46.49%
16.02%
Max
18.43%
59.40%
28.45%
350
300
Upper Susq (2050101)
10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 205. Monthly Average Flows, Upper Susquehanna River (SWAT)
X-129
-------
Upper Susq (2050101)
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 206. Flow Duration, Upper Susquehanna River (SWAT)
Chenango (2050102)
250
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 207. Monthly Average Flows, Chenango (SWAT)
X-130
-------
Chenango (2050102)
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 208. Flow Duration, Chenango (SWAT)
Owego-Wappasening (2050103)
700
34567
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 209. Monthly Average Flows, Owego-Wappasening (SWAT)
X-131
-------
Owego-Wappasening (2050103)
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 210. Flow Duration, Owego-Wappasening (SWAT)
Tioga (2050104)
6 7
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 211. Monthly Average Flows, Tioga (SWAT)
X-132
-------
Tioga (2050104)
10000
CO
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 212. Flow Duration, Tioga (SWAT)
Chemung (2050105)
250
co
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 213. Monthly Average Flows, Chemung (SWAT)
X-133
-------
Chemung (2050105)
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 214. Flow Duration, Chemung (SWAT)
Upper Susq - Lackawanna (2050107)
1400
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 215. Monthly Average Flows, Upper Susquehanna River - Lackawanna (SWAT)
X-134
-------
Upper Susq - Lackawanna (2050107)
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 216. Flow Duration, Upper Susquehanna River - Lackawanna (SWAT)
Upper WBSusa (2050201)
5678
Month
9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 217. Monthly Average Flows, Upper West Branch Susquehanna River (SWAT)
X-135
-------
Upper WB Susa (2050201)
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 218. Flow Duration, Upper West Branch Susquehanna River (SWAT)
Sinnemahoning (2050202)
120
100
to
^
o
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 219. Monthly Average Flows, Sinnemahoning (SWAT)
X-136
-------
Sinnemahoning (2050202)
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 220. Flow Duration, Sinnemahoning (SWAT)
Pine (2050205)
140
120
10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 221. Monthly Average Flows, Pine (SWAT)
X-137
-------
Pine (2050205)
10000
CO
^
o
0)
O)
03
0)
ro
Q
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 222. Flow Duration, Pine (SWAT)
Lower WB Susq (2050206)
600
500
LOWO
LOW1
LOW2
LOWS
LOW4
LOWS
LOW6
6 7
Month
8 9 10 11 12
Figure 223. Monthly Average Flows, Lower West Branch Susquehanna River (SWAT)
X-138
-------
Lower WB Susq (2050206)
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 224. Flow Duration, Lower West Branch Susquehanna River (SWAT)
120%
Month
Figure 225. Average of Median Percent Change in Flow; NARCCAP Scenarios W1-W6 at all
Stations, Susquehanna River Basin (SWAT)
X-139
-------
Willamette River Basin, HSPF Model
Results at Downstream Station
Willamette R Outlet
1,000
_, 800
u
1 600
Li-
re
3
i 400
c
re
01
^ 200
n
A
*
A
A
6
A
8
A
Q
4
A
Q
5
BASE
G(
ICLUS
;M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 226. Mean Annual Flow, Willamette River Outlet (HSPF)
X-140
-------
1 6 000
"uT
E
12,000
c
TO
o. ft nnn
BO
5
— • c nnn
ra
OJ
4 000 -
_o
>
o
0 n
1H U
Willamette R Outlet
0
1
O
1
*
4
h
i^
O
•
4
-A-
O
0
A
*
0
A
*
BASE
G(
ICLUS
:M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 227.100-yr Flow Peak, Willamette River Outlet (HSPF)
"£• 300
u
1 250
LL.
3
° 200
ra
TJ
^ 150
ra
c
c
^ 100
01
BO
OJ
£ 50
n
Willamette R Outlet
A
0
A
0
4
A
4
A
*
4
BASE
G(
ICLUS
:M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 228. Average Annual 7-day Low Flow, Willamette River Outlet (HSPF)
X-141
-------
ashiness Index
o o
- P k>
O1 N) O1
LL.
1
& 0.1
I/I
ra
.c
* 0.05
0
Willamette R Outlet
A A
1 £ 1 4 o
O O T
BASE ICLUS BASE ICLUS BASE
A
A
Q
*
ICLUS
GCM NARCCAP BCSD
Figure 229. Richards-Baker Flashiness Index, Willamette River Outlet (HSPF)
Willamette R Outlet
Water Year Basis)
3 § I 1 I
w Centroid (
§ g s
u_
0
« 40
TO
Q
on
* *
s
$
f
^fc.
"^
m W ° °
BASE ICLUS
GCM
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 230. Days to Flow Centroid, Willamette River Outlet (HSPF)
X-142
-------
Willamette R Outlet
1 Knn nnn
1 600 000
1 4nn nnn
-^ 1 200 000
i
— *• -i nnn nnn
TO
""" 800 000
A A
A A A A " "
, JL ^ • O O
+ T^
0 0
BASE ICLUS BASE ICLUS BASE ICLUS
GCM NARCCAP BCSD
Figure231. TSS Load, Willamette River Outlet (HSPF)
A ^nn
4 000
"^"
^" Q nnn
i
— *• o cnn
•a ^,ouu
TO
"" 2 000
o_ '
1-
1 000
Willamette R Outlet
A
g-
A
*,
^J
1
0
|
0
A A
A
i
L\
~i
BASE
G(
ICLUS
:M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 232. TP Load, Willamette River Outlet (HSPF)
X-143
-------
35 000
t oc nnn
h^1
— *• on nnn
ra
5
1-
1 0 000
A
Willamette R Outlet
A
A
*
A
A
* * S 8 $ $
w
BASE
G(
ICLUS
;M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 233. TN Load, Willamette River Outlet (HSPF)
X-144
-------
Results at Multiple Stations
Table 9. Summary of Range of Change Relative to Existing Conditions for NARCCAP
Dynamically Downscale Scenarios, Willamette River Basin HSPF Model
Results without LU Change
Min
Median
Mean
Max
Results with LU Change
Min
Median
Mean
Max
Tualatin River at West Linn (gage 14207500)
Flow
TSS
TN
TP
-13.42%
-25.17%
-10.72%
-14.21%
-4.50%
10.27%
-3.96%
0.17%
-1.79%
15.01%
-2.59%
1.56%
12.35%
48.58%
6.34%
15.23%
13.02%
23.87%
-9.95%
12.79%
-4.18%
8.88%
-3.88%
-0.07%
-1.62%
13.31%
-2.55%
1.24%
12.02%
44.17%
5.71%
13.34%
Pudding River at Aurora (gage 1402000)
Flow
TSS
TN
TP
-16.00%
-30.22%
-15.76%
-19.31%
-2.54%
28.41%
-2.60%
8.63%
-2.68%
16.43%
-2.92%
3.67%
11.19%
46.12%
10.07%
19.98%
15.86%
29.58%
15.31%
18.64%
-2.48%
27.24%
-2.64%
8.16%
-2.63%
15.64%
-2.90%
3.42%
11.10%
44.45%
9.73%
19.06%
South Yamhill River at McMinnville (gage 14194150)
Flow
TSS
TN
TP
-13.06%
-25.38%
-13.03%
-17.48%
-3.46%
15.18%
-2.56%
4.54%
-1.22%
13.24%
-1.42%
4.02%
11.94%
44.55%
9.66%
21.59%
13.05%
25.29%
12.98%
17.38%
-3.45%
15.03%
-2.57%
4.49%
-1.22%
13.12%
-1.42%
3.99%
11.93%
44.25%
9.62%
21.44%
Mohawk River nr Springfield (gage 14165000)
Flow
TSS
-16.75%
-31.73%
-3.37%
16.15%
-2.88%
10.93%
10.17%
36.72%
16.74%
-
-3.37%
16.10%
-2.88%
10.90%
10.16%
36.66%
X-145
-------
TN
TP
Results without LU Change
Min
-18.86%
-26.14%
Median
-2.17%
8.44%
Mean
-3.61%
3.67%
Max
8.92%
20.10%
Results with LU Change
Min
31.69%
18.81%
26.05%
Median
-2.18%
8.39%
Mean
-3.60%
3.64%
Max
8.90%
20.02%
Willamette River at Salem (gage 14191000)
Flow
TSS
TN
TP
-21.09%
-17.43%
-13.71%
-12.82%
-8.49%
30.36%
-2.48%
6.46%
-8.99%
25.97%
-2.43%
4.40%
1.03%
50.61%
7.65%
15.17%
21.07%
17.40%
13.66%
12.77%
-8.48%
30.24%
-2.49%
6.41%
-8.98%
25.88%
-2.43%
4.37%
1.04%
50.47%
7.61%
15.09%
Willamette River Outlet
Flow
TSS
TN
TP
-20.16%
-18.90%
-12.36%
-9.71%
-8.10%
27.60%
-2.72%
3.12%
-8.42%
23.54%
-2.46%
1.96%
1.96%
49.49%
6.96%
10.17%
20.10%
18.75%
12.19%
-9.58%
-8.06%
27.05%
-2.73%
3.03%
-8.38%
23.10%
-2.45%
1.90%
1.98%
48.68%
6.82%
9.95%
X-146
-------
Tualatin R West Linn
120
I
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
• LOW1
•LOW2
•LOWS
•LOW4
•LOWS
•LOW6
•LOW7
LOWS
•LOW9
•LOW10
LOW11
•LOW12
•LOW13
LOW14
•L1WO
L1W1
L1W2
L1W3
L1W4
L1W5
L1W6
L1W7
Figure 234. Monthly Average Flows, Tualatin River at West Linn (HSPF)
Tualatin R West Linn
1000
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
• LOWO
•LOW1
•LOW2
•LOWS
•LOW4
•LOWS
•LOW6
•LOW7
LOWS
•LOW9
•LOW10
LOW11
•LOW12
•LOW13
LOW14
Figure 235. Flow Duration, Tualatin River at West Linn (HSPF)
X-147
-------
Pudding R Aurora
8 9 10 11 12
LOWO
•LOW1
•LOW2
•LOWS
•LOW4
•LOWS
•LOW6
•LOW7
LOWS
•LOW9
•LOW10
LOW11
•LOW12
•LOW13
LOW14
•L1WO
L1W1
L1W2
L1W3
L1W4
L1W5
L1W6
L1W7
Figure 236. Monthly Average Flows, Pudding River at Aurora (HSPF)
Pudding R Aurora
1000
0.1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
• LOWO
•LOW1
•LOW2
•LOWS
•LOW4
•LOWS
•LOW6
•LOW7
LOWS
•LOW9
•LOW10
LOW11
•LOW12
•LOW13
LOW14
Figure 237. Flow Duration, Pudding River at Aurora (HSPF)
X-148
-------
S. Yamhill R
120
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
•LOW2
•LOWS
•LOW4
-LOWS
•LOW6
•LOW7
LOWS
•LOW9
•LOW10
LOW11
•LOW12
•LOW13
LOW14
• L1WO
L1W1
L1W2
L1W3
L1W4
L1W5
L1W6
L1W7
Figure 238. Monthly Average Flows, South Yamhill River (HSPF)
S. Yamhill R
1000
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
• LOWO
•LOW1
•LOW2
•LOWS
•LOW4
•LOWS
•LOW6
•LOW7
LOWS
•LOW9
•LOW10
LOW11
•LOW12
•LOW13
LOW14
Figure 239. Flow Duration, South Yamhill River (HSPF)
X-149
-------
Mohawk R
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
•LOW2
•LOWS
•LOW4
•LOWS
•LOW6
•LOW7
LOWS
•LOW9
•LOW10
LOW11
•LOW12
•LOW13
LOW14
•L1WO
L1W1
L1W2
L1W3
L1W4
L1W5
L1W6
L1W7
Figure 240. Monthly Average Flows, Mohawk River (HSPF)
Mohawk R
1000
20%
40%
60%
80%
100%
Percent of Time that Flow is Equaled or Exceeded
• LOWO
•LOW1
•LOW2
•LOWS
•LOW4
•LOWS
•LOW6
•LOW7
LOWS
•LOW9
•LOW10
LOW11
•LOW12
•LOW13
LOW14
Figure 241. Flow Duration, Mohawk River (HSPF)
X-150
-------
Willamette R Salem
1400
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
•LOW2
•LOWS
•LOW4
•LOWS
•LOW6
•LOW7
LOWS
•LOW9
•LOW10
LOW11
•LOW12
• LOW13
LOW14
•L1WO
L1W1
L1W2
L1W3
L1W4
L1W5
L1W6
L1W7
Figure 242. Monthly Average Flows, Willamette River at Salem (HSPF)
Willamette R Salem
10000
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
• LOWO
•LOW1
•LOW2
•LOWS
•LOW4
•LOWS
•LOW6
•LOW7
LOWS
•LOW9
•LOW10
LOW11
•LOW12
•LOW13
LOW14
Figure 243. Flow Duration, Willamette River at Salem (HSPF)
X-151
-------
Willamette R Outlet
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
•LOW2
•LOWS
•LOW4
• LOWS
•LOW6
•LOW7
LOWS
•LOW9
•LOW10
LOW11
•LOW12
•LOW13
LOW14
•L1WO
L1W1
L1W2
L1W3
L1W4
L1W5
L1W6
L1W7
Figure 244. Monthly Average Flows, Willamette River Outlet (HSPF)
Willamette R Outlet
1 UUUUU
p 10000,
0, \
%
0 1000
CD
O)
CD
0) 100
H 10
Q
•1
•
• ^
•— ~— - ••-
-** I*VA.
• > r~ - .~) — ^~^_
i^^^^^
\
\
}
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
• i own
— •— LOW1
— *— LOW2
— 1»— LOWS
— •— LOW4
—•—LOWS
— i — LOW6
— LOW7
— LOWS
— •— LOW9
— •— LOW10
— *— LOW1 1
— *— LOW12
— •— LOW13
— •— LOW14
Figure 245. Flow Duration, Willamette River Outlet (HSPF)
X-152
-------
01
M
TO
U
0)
Q.
•a
OJ
0)
M
OJ
-30%
Month
Figure 246. Average of Median Percent Change in Flow; NARCCAP Scenarios W1-W6 at all
Stations, Willamette River Basin (HSPF)
X-153
-------
Willamette River Basin, SWAT Model
Results at Downstream Station
1,200 -i
1,000
_, 800
u
0 600
LL.
"TO
3
< 40°
C
re
01
200
0
Lower Willamette
A
A
ft
A
A
^^
A
9 0 K X ^ U
^^
^>» "
IK
BASE
ICLUS
GCM
BASE
ICLUS
NARCCAP
BASE
ICLUS
BCSD
Figure 247. Mean Annual Flow, Lower Willamette River (SWAT)
X-154
-------
tn
c 10 000
u
tn
TO
£
tl/j 6 000
O
(R
Lower Willamette
+
+
o
O
f t * ^ * *
T4 TH
* *
O
O
BASE ICLUS
GCM
BASE ICLUS
NARCCAP
BASE ICLUS
BCSD
Figure 248.100-yr Flow Peak, Lower Willamette River (SWAT)
Lower Willamette
rage Annual 7-day Low Flow (cms)
-i-ihJhJCJCJ4^4
DU1OU1OU1OC
1
5
A A ^ ^
J| 9. A A
db dh O O
88 O 0
* *
BASE ICLUS BASE ICLUS BASE ICLUS
GCM NARCCAP BCSD
Figure 249. Average Annual 7-day Low Flow, Lower Willamette River (SWAT)
X-155
-------
0.25 -i
-§ 0.2
c
U)
U)
Ol
_c
•5 0.15
re
LL.
re
"S
re
u
ce.
0.05
0
Lower Willamette
i
BASE
O
ICLUS
GCM
BASE
$ *
£?
ICLUS
NARCCAP
BASE
*
ICLUS
BCSD
Figure 250. Richards-Baker Flashiness Index, Lower Willamette River (SWAT)
160 -,
140
"ST
Days to Flow Centroid (Water Year Bas
N> -^ O CO O N>
O O O O O O O
Lower Willamette
&
*
*
&
f
t
BASE
ICLUS
GCM
BASE
ICLUS
NARCCAP
BASE
ICLUS
BCSD
Figure 251. Days to Flow Centroid, Lower Willamette River (SWAT)
X-156
-------
A ^nn nnn
4 000 000
_ 3,000,000
t 2 500 000
•0
JS o nnn nnn
1
I— -] 500 000
500 000
n
Lower Willamette
A
4.
s5^
A
A
A
*
8
A
*
g
A
O
*
A
0
~i
BASE
G(
ICLUS
:M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 252. TSS Load, Lower Willamette River (SWAT)
R nnn
5,000
4 000
h^1
**"** ^ nnn
ra
5
Q_
2 000
1 000
n
*
BASE
G(
Lower Willamette
*
ICLUS
:M
1
BASE
NAR(
ICLUS
;CAP
$
BASE
BC
*
ICLUS
SD
X-157
-------
Figure 253. TP Load, Lower Willamette River (SWAT)
en nnn
45 000
35 000
i
TO
h-
5,000
n
A
U
Lower Willamette
A
*
A
*
4
*
* *
,i.
A
i*
O ^ <&> ^
BASE
G(
ICLUS
:M
BASE
NAR(
ICLUS
;CAP
BASE
BC
ICLUS
SD
Figure 254. TN Load, Lower Willamette River (SWAT)
X-158
-------
Results at Multiple Stations
Table 10. Summary of Range of Change Relative to Existing Conditions for NARCCAP
Dynamically Downscaled Scenarios, Willamette River Basin (SWAT)
Results without LU Change
Min
Median
Mean
Max
Results with LU Change
Min
Median
Mean
Max
Middle Fork Willamette HUC 17090001
Flow
TSS
TN
TP
-17.51%
91.02%
49.94%
86.17%
0.03%
132.96%
84.32%
142.56%
-1.70%
135.57%
82.64%
140.77%
10.39%
177.73%
110.23%
183.53%
-17.51%
91.33%
48.72%
83.96%
0.03%
133.31%
82.46%
139.23%
-1.69%
135.95%
80.90%
137.48%
10.40%
178.19%
108.15%
179.48%
Coast Fork Willamette HUC 17090002
Flow
TSS
TN
TP
-4.95%
-4.93%
-14.77%
-19.61%
6.37%
29.49%
2.84%
5.31%
7.28%
31.31%
4.93%
6.77%
19.55%
63.92%
24.75%
31.82%
-4.98%
-4.65%
-15.48%
-19.46%
6.34%
29.73%
1.50%
5.08%
7.26%
31.54%
3.68%
6.60%
19.53%
64.12%
23.30%
31.67%
Upper Willamette HUC 17090003
Flow
TSS
TN
TP
-10.42%
-1.01%
-7.23%
-10.15%
3.89%
13.58%
2.07%
-0.27%
2.92%
15.18%
3.41%
-0.29%
12.86%
31.82%
13.70%
7.20%
-10.43%
-0.99%
-7.45%
-10.21%
3.89%
13.61%
1.78%
-0.32%
2.91%
15.20%
3.13%
-0.32%
12.85%
31.85%
13.40%
7.20%
McKenzie HUC 17090004
Flow
TSS
TN
TP
-12.39%
-0.95%
-1.52%
-10.56%
5.32%
114.25%
27.74%
35.02%
2.24%
105.67%
22.02%
24.38%
10.76%
218.42%
36.83%
48.46%
-12.39%
-0.94%
-1.70%
-10.63%
5.32%
114.53%
27.11%
34.55%
2.24%
105.97%
21.48%
24.02%
10.76%
219.11%
35.94%
47.80%
North Santiam HUC 17090005
Flow
TSS
TN
TP
-9.28%
-7.60%
-14.08%
-19.42%
5.21%
6.79%
-3.80%
-7.42%
5.32%
9.26%
-3.19%
-7.20%
16.94%
29.41%
5.72%
1.44%
-9.29%
-7.14%
-14.56%
-19.68%
5.21%
7.29%
-4.43%
-7.63%
5.31%
9.76%
-3.77%
-7.41%
16.94%
29.96%
5.07%
1.15%
South Santiam HUC 17090006
X-159
-------
Flow
TSS
TN
TP
Results without LU Change
Min
-7.77%
-8.09%
-15.63%
-19.81%
Median
4.90%
9.51%
-5.85%
-8.02%
Mean
6.28%
11.57%
-4.99%
-7.75%
Max
18.21%
32.22%
5.14%
3.02%
Results with LU Change
Min
-7.77%
-8.09%
-15.62%
-19.76%
Median
4.90%
9.51%
-5.90%
-8.05%
Mean
6.28%
11.57%
-5.02%
-7.74%
Max
18.21%
32.23%
5.08%
3.01%
Middle Willamette HUC 17090007
Flow
TSS
TN
TP
-8.45%
-9.71%
-11.45%
-7.59%
4.70%
8.31%
-4.78%
-3.62%
4.53%
8.06%
-3.64%
-3.50%
15.28%
24.61%
3.94%
-0.47%
-8.49%
-9.65%
-11.71%
-7.42%
4.69%
8.42%
-5.36%
-3.43%
4.51%
8.17%
-4.16%
-3.33%
15.27%
24.70%
3.37%
-0.31%
Yamhill HUC 17090008
Flow
TSS
TN
TP
-3.30%
-8.27%
-14.22%
-18.00%
7.65%
9.81%
-6.23%
-8.13%
9.71%
11.30%
-5.40%
-7.82%
23.21%
30.05%
3.24%
0.31%
-3.33%
-8.26%
-14.41%
-18.03%
7.64%
9.86%
-6.57%
-8.09%
9.69%
11.35%
-5.69%
-7.75%
23.20%
30.13%
2.98%
0.39%
Pudding River at Aurora (gage 14202000)
Flow
TSS
TN
TP
1.31%
0.39%
-23.37%
-20.73%
5.72%
5.98%
-19.63%
-13.83%
6.17%
6.80%
-19.35%
-14.92%
12.04%
15.12%
-15.64%
-11.96%
1.19%
0.24%
-23.55%
-19.81%
5.65%
5.94%
-20.07%
-12.77%
6.09%
6.62%
-19.75%
-13.98%
11.96%
14.68%
-16.04%
-10.77%
Tualatin HUC 17090010
Flow
TSS
TN
TP
-5.04%
-7.91%
-12.32%
-11.04%
6.49%
8.21%
-7.81%
-2.89%
8.24%
10.43%
-6.10%
-2.18%
22.16%
29.32%
0.58%
5.07%
-5.34%
-8.05%
-11.86%
-10.48%
6.31%
8.02%
-8.34%
-2.70%
8.07%
10.51%
-6.41%
-1.79%
22.08%
29.99%
0.15%
5.34%
Clackamas HUC 17090011
Flow
TSS
TN
-8.10%
-8.89%
-7.16%
9.04%
10.94%
1.37%
7.61%
9.11%
1.83%
20.18%
27.42%
9.21%
-8.13%
-8.86%
-7.05%
9.01%
10.91%
1.34%
7.58%
9.10%
1.88%
20.15%
27.41%
9.16%
X-160
-------
TP
Results without LU Change
Min
-7.68%
Median
0.66%
Mean
-0.08%
Max
4.89%
Results with LU Change
Min
-7.28%
Median
1.31%
Mean
0.55%
Max
5.60%
Lower Willamette HUC 17090012
Flow
TSS
TN
TP
-8.35%
-10.38%
-10.62%
-6.34%
5.20%
9.78%
-4.47%
-3.02%
4.91%
8.57%
-3.33%
-2.86%
15.89%
24.33%
3.87%
-0.32%
-8.39%
-10.31%
-10.86%
-6.16%
5.18%
9.90%
-4.99%
-2.84%
4.89%
8.68%
-3.80%
-2.68%
15.87%
24.49%
3.36%
-0.14%
Williamette River at Salem (gage 14191000)
Flow
TSS
TN
TP
-9.99%
-9.58%
-9.63%
-13.20%
4.30%
11.57%
-0.42%
-3.34%
3.70%
9.93%
0.79%
-3.23%
14.23%
24.93%
10.54%
4.13%
-10.00%
-9.51%
-9.95%
-13.30%
4.30%
11.67%
-0.83%
-3.34%
3.70%
10.03%
0.41%
-3.24%
14.23%
25.04%
10.15%
4.19%
Mohawk River nr Springfield (gage 14165000)
Flow
TSS
TN
TP
-8.80%
6.65%
8.19%
-1.08%
3.09%
2391.16
%
80.12%
132.86%
2.34%
2875.55
%
63.54%
101.95%
12.39%
8244.24
%
102.34%
165.98%
-8.80%
17.29%
6.78%
-2.24%
3.09%
2396.01
%
76.20%
128.33%
2.34%
2878.21
%
60.19%
98.19%
12.39%
8232.58
%
97.67%
160.61%
X-161
-------
Middle Fork Willamette
34567
Month
8 9 10 11 12
Figure 255. Monthly Average Flows, Middle Fork Willamette River (SWAT)
Middle Fork Willamette
1000
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 256. Flow Duration, Middle Fork Willamette River (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
X-162
-------
Coast Fork Willamette
120
100
I
1 2 3 4 5 6 7 8 9 10 11 12
Figure 257. Monthly Average Flows, Coast Fork Willamette River (SWAT)
Coast Fork Willamette
1000
0.01
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 258. Flow Duration, Coast Fork Willamette River (SWAT)
X-163
-------
Upper Willamette
34567
Month
8 9 10 11 12
Figure 259. Monthly Average Flows, Upper Willamette River (SWAT)
Upper Willamette
10000
tn
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 260. Flow Duration, Upper Willamette River (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
X-164
-------
McKenzie
6 7
Month
10 11 12
Figure 261. Monthly Average Flows, McKenzie River (SWAT)
McKenzie
10000
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.001
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 262. Flow Duration, McKenzie River (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
X-165
-------
North Santiam
34567
Month
8 9 10 11 12
Figure 263. Monthly Average Flows, North Santiam River (SWAT)
North Santiam
10000
tn
LOWO
•LOW1
LOW2
•LOWS
• LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 264. Flow Duration, North Santiam River (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
X-166
-------
South Santiam
6 7
Month
10 11 12
Figure 265. Monthly Average Flows, South Santiam River (SWAT)
South Santiam
1000
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 266. Flow Duration, South Santiam River (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
X-167
-------
Middle Willamette
6 7
Month
10 11 12
Figure 267. Monthly Average Flows, Middle Willamette River (SWAT)
Middle Willamette
10000
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 268. Flow Duration, Middle Willamette River (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
X-168
-------
Yamhill gage
6 7
Month
Figure 269. Monthly Average Flows, Yamhill River (SWAT)
10 11 12
Yamhill gage
1000
o
CD
O)
CD
CD
Q 0.01
0.001
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 270. Flow Duration, Yamhill River (SWAT)
X-169
-------
Pudding gage
6 7
Month
Figure 271. Monthly Average Flows, Pudding River (SWAT)
10 11 12
Pudding gage
1000
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 272. Flow Duration, Pudding River (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
X-170
-------
Tualatin
6 7
Month
Figure 273. Monthly Average Flows, Tualatin River (SWAT)
10 11 12
Tualatin
1000
LOWO
•LOW1
LOW2
•LOWS
• LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 274. Flow Duration, Tualatin River (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
X-171
-------
-10%
Month
Figure 275. Average of Median Percent Change in Flow; NARCCAP Scenarios W1-W6 at all
Stations, Willamette River Basin (SWAT)
X-172
-------
Appendix Y
Scenario Results for the Non-Pilot
Watersheds
Y-l
-------
Southern California Coastal Watersheds Y-3
Cook Inlet Basin Y-21
Georgia-Florida Coastal Basins Y-34
Illinois River Basin Y-51
Lake Erie Drainages Y-68
Nebraska: Loup and Elkhorn River Basins Y-85
Tar andNeuse River Basins Y-102
New England Coastal Basins Y-118
Lake Pontchartrain Drainage Y-135
Rio Grande Valley Y-149
Sacramento River Watershed Y-166
South Platte River Basin Y-181
Powder and Tongue Rivers Basins Y-198
Trinity River Watershed Y-215
Upper Colorado River Basin Y-232
LEGEND:
GCM
A CGCM3
+ HadCMS
0 GFDL
CCSM
RCM
A RCM3
A CRCM
+ HRM3
0 GFDL
• RCM3
WRFP
Y-2
-------
Southern California Coastal Watersheds
Santa Margarita
1 A
1 2
"F 10
E
u
J 0.8
LL.
"nj
1 0.6
|
I °-4
0.2
n n
+ _L
i T
t f
o o
* #
BASE ICLUS
NARCCAP
Figure 1. Mean annual flow (cms), Santa Margarita River (SWAT)
GCM
A CGCM3
+ HadCMS
0 GFDL
CCSM
RCM
A RCM3
A CRCM
+ HRM3
0 GFDL
• RCM3
ft WRFP
Y-3
-------
Santa Margarita
300
1 250
o
0 200
S2
ra
£
BO 150
5
3 100
Q.
O
t 50
6
o
rt o
+
+
II
* ~
8
BASE
ICLUS
NARCCAP
Figure 2.
100-yr Flow Peak (Log-Pearson III, cms), Santa Margarita River (SWAT)
n n^ -,
0.045
"uT
u 0 04
O n r\ii=.
s
o n rn
ra
"D Q Q25 -
S 002
c
c
rf
n m^
BO
2
ni n m
0 005
Santa Margarita
*
A
•
A
A
f^~ ^^x
o
BASE ICLUS
NARCCAP
Figure 3.
Average Annual 7-day Low Flow (cms), Santa Margarita River (SWAT)
Y-4
-------
0.6 -i
X 0.5
01
•o
(/)
«
01 04
(C
IE
(/)
ra
fc 0.3
ji
ra
CO
^
o
ce.
0.1
0
Santa Margarita
ft
1 i
BASE ICLUS
NARCCAP
Figure 4. Richards-Baker Flashiness Index, Average Annual 7-day Low Flow (cms), Santa
Margarita River (SWAT)
-i on
•— -icn
TO
CO
ra i4n
-ion -
i° mn
o
t
« on
1
[Z Rn
0
>» 4n
TO ^u
Q
20
n
Santa Margarita
i i
JS^^ ^^TBr
BASE ICLUS
NARCCAP
Figure 5.
Days to Flow Centroid (Water Year Basis), Santa Margarita River (SWAT)
Y-5
-------
Santa Margarita
60,000 -i
_ 40,000
1
^ 30,000
1
8
1- 20,000
* *
I *
£ *
BASE ICLUS
NARCCAP
Figure 6. TSS Load (MT/yr), Santa Margarita River (SWAT)
'\ACl-,
"Z"
^ an
•a
TO
° 60
"Z.
40
20
Santa Margarita
+ +
1 T
A
A
.
O O
JC ^
M X
BASE ICLUS
NARCCAP
Figure 7. TN Load (MT/yr), Santa Margarita River (SWAT)
Y-6
-------
an _,
80
7n
60
L.
>
£ 50
•o
O 40
Q.
"~ 30
20
m
Santa Margarita
, +
+
I *
Q ^
& w
BASE ICLUS
NARCCAP
Figure 8.
TP Load (MT/yr), Santa Margarita River (SWAT)
Y-7
-------
Table 1. Summary of range of change relative to existing conditions for NARCCAP dynamically
downscaled scenarios, Southern California Coastal basins SWAT model
Results without LU change
Min
Median
Mean
Max
Results with LU change
Min
Median
Mean
Max
Ventura River HUC 1807010
Flow
TSS
TN
TP
-6.97%
-18.83%
1.40%
-10.69%
-4.45%
-9.61%
16.53%
6.90%
1.93%
4.66%
19.95%
1 1 .25%
24.41%
61.06%
41.86%
36.48%
-8.98%
-11.51%
-4.32%
-9.83%
-6.25%
-3.99%
15.73%
12.42%
-0.14%
10.99%
14.87%
10.00%
21.98%
70.69%
34.38%
32.22%
Santa Clara HUC 18070102
Flow
TSS
TN
TP
-9.40%
-31.29%
-10.46%
-70.02%
0.69%
-19.22%
-6.48%
-66.08%
2.81%
-17.82%
-6.37%
-65.68%
21.67%
5.53%
-0.23%
-57.92%
-13.62%
-36.07%
-13.77%
-70.63%
-4.41%
-25.28%
-10.86%
-67.29%
-2.00%
-23.92%
-9.61%
-66.64%
16.57%
-2.31%
0.15%
-58.83%
Calleguas HUC 18070103
Flow
TSS
TN
TP
-11.30%
-4.20%
-0.79%
-10.49%
0.42%
2.99%
3.25%
2.85%
2.33%
12.58%
8.41%
9.60%
24.17%
70.34%
38.23%
59.30%
-17.87%
-9.45%
-1.50%
-17.10%
-7.20%
-2.67%
2.41%
-4.53%
-5.14%
5.66%
5.36%
1.04%
14.72%
56.17%
23.31%
40.75%
Los Angeles HUC 18070105
Flow
TSS
TN
TP
-15.87%
-35.29%
-9.75%
-46.85%
2.80%
-19.33%
-0.09%
-38.86%
4.30%
-18.02%
9.06%
-35.25%
37.89%
10.98%
39.81%
-12.03%
-18.50%
-36.71%
-17.07%
-47.04%
-0.63%
-21.50%
-2.52%
-35.50%
1.11%
-19.74%
5.67%
-31.18%
34.64%
10.97%
36.96%
-3.79%
San Gabriel HUC 18070106
Flow
TSS
TN
TP
-10.97%
-46.48%
-4.55%
-31.36%
2.25%
-33.68%
-3.02%
-18.19%
3.71%
-30.44%
-2.06%
-15.99%
25.91%
-0.95%
3.79%
9.90%
-12.23%
-46.72%
-5.60%
-33.83%
0.68%
-34.15%
-4.09%
-21.16%
2.20%
-30.68%
-3.12%
-18.30%
24.32%
-0.68%
2.80%
9.20%
San Jacinto HUC 18070202
Flow
TSS
TN
TP
-26.91%
-29.47%
52.93%
-16.17%
13.96%
30.18%
114.81%
40.99%
13.70%
32.03%
163.48%
48.26%
62.19%
115.28%
473.72%
148.01%
Santa Ana HUC 18070203
Flow
TSS
TN
TP
-17.57%
-17.88%
14.95%
-1 1 .22%
1 1 .28%
14.58%
49.96%
25.87%
13.34%
16.54%
56.12%
28.55%
56.22%
64.12%
144.14%
88.59%
-33.66%
-38.69%
43.09%
-18.05%
-22.45%
-22.71%
32.98%
-9.68%
4.16%
14.54%
128.20%
22.10%
4.51%
6.92%
77.79%
30.87%
3.95%
16.01%
164.27%
31.58%
6.64%
9.40%
88.47%
31.92%
48.71%
89.84%
466.14%
124.47%
46.75%
52.57%
208.75%
97.83%
Newport Bay HUC 18070204
Flow
-16.15%
0.13%
3.74%
37.46% 0 -18.48%
-3.44%
-0.03%
31.52%
Y-8
-------
TSS
TN
TP
Results without LU change
Min
-21.33%
7.48%
-10.91%
Median
3.64%
23.29%
12.80%
Mean
8.69%
45.09%
25.25%
Max
55.79%
174.31%
111.24%
Results with LU change
Min
-21.50%
9.53%
-6.18%
Median
-1.30%
24.60%
11.34%
Mean
3.39%
50.29%
26.90%
Max
44.39%
190.28%
122.82%
Santa Margarita HUC 18070302
Flow
TSS
TN
TP
Santa Ana at MWD (
Flow
TSS
TN
TP
Santa Clara at Piru
Flow
TSS
TN
TP
-21.03%
-20.57%
-10.04%
-19.56%
12.76%
58.05%
24.47%
53.00%
9.79%
49.64%
60.42%
42.96%
44.30%
129.12%
174.21%
121.01%
-24.27%
-24.97%
-17.61%
-21.32%
8.77%
47.01%
8.10%
45.06%
5.84%
39.83%
35.06%
37.06%
39.81%
115.71%
144.07%
113.56%
Gage 11 066460)
-10.41%
-22.83%
0.13%
-10.55%
2.18%
-2.88%
3.40%
0.59%
3.62%
-1.15%
3.29%
2.69%
21.08%
21.77%
5.51%
20.77%
-12.23%
-23.01%
1.65%
-7.75%
0.43%
-2.98%
6.08%
2.48%
1.73%
-1 .69%
6.13%
4.30%
18.88%
20.54%
8.87%
21.85%
Gage 11 109000)
-13.99%
-18.30%
-3.07%
-21.57%
3.78%
3.45%
2.19%
1.53%
4.73%
8.02%
6.65%
10.96%
34.04%
53.98%
33.14%
69.37%
-18.83%
-21.49%
-12.87%
-25.97%
-1.35%
-0.85%
-7.62%
-4.58%
-0.26%
3.55%
-0.49%
4.45%
29.08%
47.19%
40.07%
61.09%
Santa Margarita nr Temecula (Gage 1104400)
Flow
TSS
TN
TP
-21.99%
-20.54%
-14.68%
-22.14%
31.46%
49.13%
70.49%
61.06%
26.94%
41.71%
96.33%
53.75%
81.69%
110.32%
289.14%
141.38%
-28.25%
-24.68%
-24.59%
-22.61%
21.51%
36.25%
46.71%
56.67%
17.40%
30.04%
68.34%
50.73%
69.34%
91.47%
231.33%
142.63%
Y-9
-------
Ventura River
10 11 12
LOWO
• LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 9.
Monthly average flows, Ventura River (SWAT)
Ventura River
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 10. Flow duration, Ventura River (SWAT)
Y-10
-------
Santa Clara
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 11. Monthly average flows, Santa Clara River (SWAT)
Santa Clara
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 12. Flow duration, Santa Clara River (SWAT)
Y-ll
-------
Calleguas
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 13. Monthly average flows, Calleguas River (SWAT)
Calleguas
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 14. Flow duration, Calleguas River (SWAT)
Y-12
-------
Los Angeles
12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 15. Monthly average flows, Los Angeles River (SWAT)
Los Angeles
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 16. Flow duration, Los Angeles River (SWAT)
Y-13
-------
San Gabriel
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 17. Monthly average flows, San Gabriel River (SWAT)
San Gabriel
1000
to
^
3; 100
o
CD
O)
CD
CD
'CD
Q
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 18. Flow duration, San Gabriel River (SWAT)
Y-14
-------
San Jacinto
1 2 3 4 5 6 7 8 9 10 11 12
0.5
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 19. Monthly average flows, San Jacinto River (SWAT)
San Jacinto
to
^
o
CD
O)
CD
CD
100
10
0.1 =
'CD 0.01
Q
0.001
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 20. Flow duration, San Jacinto River (SWAT)
Y-15
-------
Santa Ana
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 21. Monthly average flows, Santa Ana River (SWAT)
Santa Ana
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 22. Flow duration, Santa Ana River (SWAT)
Y-16
-------
Newport Bay
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 23. Monthly average flows, Newport Bay HUC8 (SWAT)
Newport Bay
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 24. Flow duration, Newport Bay HUC8 (SWAT)
Y-17
-------
Santa Margarita
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 25. Monthly average flows, Santa Margarita River (SWAT)
Santa Margarita
100
to
^
o
CD
O)
CD
CD
'CD
Q
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.00001
Percent of Time that Flow is Equaled or Exceeded
Figure 26. Flow duration, Santa Margarita River (SWAT)
Y-18
-------
Santa Ana at MWD (11066460)
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 27. Monthly average flows, Santa Ana River at MWD (SWAT)
Santa Ana at MWD (11066460)
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 28. Flow duration, Santa Ana River at MWD (SWAT)
Y-19
-------
50%
40%
30%
20%
ro
.c
u
OJ
•a
OJ
M- 0%
o
01
8?
§ -10%
-20%
-30%
Month
Figure 29. Average of median percent change in flow; NARCCAP Scenarios W1-W6 at all stations,
Southern California Coastal basins (SWAT)
Y-20
-------
Cook Inlet Basin
Note: Coverage for the Cook Inlet basin is provided by only three of the six NARCCAP downscaled
scenario outputs. In addition, the ICLUS land use change analysis does not cover Alaska.
250 -i
200
"i/T
E 150
_o
LL.
1 100
c
c
c
ra
01
1 50
0
Kenai R at Soldotna (Kenai)
+
o
=
BASE
NARCCAP
Figure 30. Mean annual flow (cms), Kenai River at Soldotna (SWAT)
Y-21
-------
Kenai R at Soldotna (Kenai)
-5 nnn
"uT
LJ
2 000
o
2
a •
M
5.
v 1 nnn
ra
£
- 500
k.
6
o
rt o
£
(J
NARCCAP
Figure 31. 100-yr Flow Peak (Log-Pearson III, cms), Kenai River at Soldotna (SWAT)
Kenai R at Soldotna (Kenai)
AC.
40
-ac
30
I/)
~ 25
3 25
o
"~ 20
3 20
5
> 1"^
CO l3
?
IX
— -in -
ra lu
3
C
5 5
01
BO
2 0
£
¥
gj 0
£ BASE
NARCCAP
Figure 30. Average Annual 7-day Low Flow (cms), Kenai River at Soldotna (SWAT)
Y-22
-------
0.12 -i
x 0.1
01
•o
(/)
in
g 0.08
IE
to
ra
£ 0.06
ji
ra
CO
"ra °-04
^
o
£
0.02
0
Kenai R at Soldotna (Kenai)
0
BASE
NARCCAP
Figure 31. Richards-Baker Flashiness Index, Average Annual 7-day Low Flow (cms), Kenai
River at Soldotna (SWAT)
300 -i
•£•250
to
ra
m
to
> 200
b
Centroid (Wa
8
1 100
u.
0
S2,
Q 50
0
Kenai R at Soldotna (Kenai)
*
BASE
NARCCAP
Figure 32. Days to Flow Centroid (Water Year Basis), Kenai River at Soldotna (SWAT)
Y-23
-------
•dnn nnn -,
350 000
^
1
Kenai R at Soldotna (Kenai)
±
+
0
BASE
NARCCAP
Figure 33. TSS Load (MT/yr), Kenai River at Soldotna (SWAT)
Kenai R at Soldotna (Kenai)
3,000 -i
2,500
2 000 -
1
TO
5
2
1 nnn
tt
+
o
—
BASE
NARCCAP
Figure 34. TN Load (MT/yr), Kenai River at Soldotna (SWAT)
Y-24
-------
S.
•o
35
30
25
20
15
10
Kenai R at Soldotna (Kenai)
BASE
NARCCAP
Figure 35. TP Load (MT/yr), Kenai River at Soldotna (SWAT)
Y-25
-------
Table 2. Summary of range of change relative to existing conditions for NARCCAP dynamically
downscaled scenarios, Cook Inlet basin SWAT model
Results without LU change
Min
Median
Mean
Max
Kenai R at Soldotna (Kenai)
Flow
TSS
TN
TP
31.74%
95.66%
75.39%
-11.23%
54.35%
133.64%
100.25%
-10.13%
50.89%
124.56%
99.69%
-2.72%
66.56%
144.37%
123.44%
13.18%
Talkeetna R. near Talkeetna
Flow
TSS
TN
TP
19.47%
79.65%
11.74%
-13.52%
27.66%
97.68%
17.74%
-8.28%
26.18%
94.21%
17.20%
-9.72%
31.42%
105.28%
22.13%
-7.37%
Upper Susitna HUC 19020501
Flow
TSS
TN
TP
9.12%
60.09%
1.83%
-21.37%
26.59%
97.28%
7.02%
-11.77%
23.33%
91.60%
10.31%
-14.06%
34.27%
117.41%
22.07%
-9.04%
Matanuska HUC 19020402
Flow
TSS
TN
TP
10.16%
70.16%
39.83%
-20.66%
19.90%
90.58%
49.58%
-9.76%
16.81%
84.52%
46.85%
-12.96%
20.35%
92.81%
51.12%
-8.47%
Lower Susitna HUC 19020505
Flow
TSS
TN
TP
12.06%
75.34%
15.63%
-22.65%
19.42%
94.99%
17.88%
-14.35%
18.59%
91.77%
19.96%
-16.84%
24.29%
104.99%
26.36%
-13.52%
Chulitna HUC 19020502
Flow
TSS
TN
TP
8.57%
45.01%
23.36%
-19.45%
14.12%
57.93%
25.87%
-12.57%
13.78%
54.52%
27.33%
-14.57%
18.65%
60.63%
32.77%
-1 1 .69%
Talkeetna River Mouth HUC 19020503
Flow
TSS
TN
TP
16.46%
76.83%
12.13%
-14.71%
23.87%
92.96%
17.86%
-9.99%
22.75%
90.59%
17.60%
-11.39%
27.92%
101.99%
22.81%
-9.46%
Y-26
-------
Kenai R at Soldotna (Kenai)
600
500
,400
o
300
200
100
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW2
LOW4
•LOW6
Figure 36. Monthly average flows, Kenai River at Soldotna (SWAT)
Kenai R at Soldotna (Kenai)
10000
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 37. Flow duration, Kenai River at Soldotna (SWAT)
Y-27
-------
Talkeetna R. nearTalkeetna
LOWO
•LOW2
LOW4
•LOW6
6 7
Month
10 11 12
Figure 38. Monthly average flows, Talkeetna River near Talkeetna (SWAT)
Talkeetna R. near Talkeetna
10000
to
^
I
CD
O)
CD
CD
'CD
Q
1000
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 39. Flow duration, Talkeetna River near Talkeetna (SWAT)
Y-28
-------
Upper Susitna
700
1 2 3 4 5 6 7 8 9 10 11 12
100
LOWO
'LOW2
LOW4
•LOW6
Figure 40. Monthly average flows, Upper Susitna River (SWAT)
Upper Susitna
10000
1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 41. Flow duration, Upper Susitna River (SWAT)
Y-29
-------
Matanuska
600
500
to
^
o
,400
300
200
100
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW2
LOW4
•LOW6
Figure 42. Monthly average flows, Matanuska River (SWAT)
Matanuska
10000
tn
0.1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 43. Flow duration, Matanuska River (SWAT)
Y-30
-------
Lower Susitna
2500
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW2
LOW4
•LOW6
Figure 44. Monthly average flows, Lower Susitna River (SWAT)
Lower Susitna
10000
1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 45. Flow duration, Lower Susitna River (SWAT)
Y-31
-------
Chulitna
6 7
Month
10 11 12
LOWO
•LOW2
LOW4
•LOW6
Figure 46. Monthly average flows, Chulitna River (SWAT)
Chulitna
10000
1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 47. Flow duration, Chulitna River (SWAT)
Y-32
-------
Talkeetna River
600
500
,400
o
300
200
100
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
'LOW2
LOW4
•LOW6
Figure 48. Monthly average flows, Talkeetna River Mouth (SWAT)
Talkeetna River
10000
1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 49. Flow duration, Talkeetna River Mouth (SWAT)
Y-33
-------
Georgia-Florida Coastal Basins
Hillsborough
30
25
"uT
u 20
_O
LJ_
ra 15-
3
C
5
c 10
ra
01
1
C
n
+ +
m •
•
o o
jt
J\ ^^
w
BASE ICLUS
NARCCAP
Figure 50. Mean annual flow (cms), Hillsborough River (SWAT)
Hillsborough
finn
"uT
LJ
400
1
TO ^nn
QJ OUU
Q.
bo
5.
TO
£
O
— mn
L.
>
8
rt o
+
+
5
A A
&
"w
BASE ICLUS
NARCCAP
Figure 51. 100-yr Flow Peak (Log-Pearson III, cms), Hillsborough River (SWAT)
Y-34
-------
Hillsborough
c
Average Annual 7-day Low Flow (cms)
3 -*• N) CO ^ O1 C
+
! I
^m
£\
2 °
# #
BASE ICLUS
NARCCAP
Figure 52. Average Annual 7-day Low Flow (cms), Hillsborough River (SWAT)
Baker Flashiness Index
o o o
§b P ^
00 -* M
l/>
ra
u
£
0.02
0
Hillsborough
1 +
BASE ICLUS
NARCCAP
Figure 53. Richards-Baker Flashiness Index, Average Annual 7-day Low Flow (cms),
Hillsborough River (SWAT)
Y-35
-------
Hillsborough
onn
Days to FlowCentroid (Water Year Basis)
-» -» 10 10 I
8 8 8 8 8 I
i ±
* i
BASE ICLUS
NARCCAP
Figure 54. Days to Flow Centroid (Water Year Basis), Hillsborough River (SWAT)
Hillsborough
•o
TO
O
1 n nnn
5 000
n
+
*
+
A
O
o
BASE
NAR(
M
ICLUS
;CAP
Figure 55. TSS Load (MT/yr), Hillsborough River (SWAT)
Y-36
-------
A nnn -,
3 500
-a nnn
f="
""""" o nnn
TO
5
Hillsborough
+
+ 8
£
A .
ft *
Js^
£
BASE ICLUS
NARCCAP
Figure 56. TN Load (MT/yr), Hillsborough River (SWAT)
Hillsborough
1 000
>
2
TO
5
Q.
400
n
BASE
NARI
+
!
M
ICLUS
;CAP
Figure 57. TP Load (MT/yr), Hillsborough River (SWAT)
Y-37
-------
Table 3. Summary of range of change relative to existing conditions for NARCCAP dynamically
downscaled scenarios, Georgia-Florida basins SWAT model
Results without LU change
Min
Median
Mean
Max
Results with LU change
Min
Median
Mean
Max
AucillaHUC 03110103
Flow
TSS
TN
TP
-23.75%
-20.42%
-1.99%
-18.40%
12.24%
18.12%
12.71%
22.65%
13.58%
22.64%
13.99%
33.35%
45.52%
66.75%
24.82%
89.73%
23.75%
20.40%
-2.06%
18.71%
12.23%
18.10%
12.69%
22.48%
13.58%
22.62%
13.99%
33.32%
45.52%
66.73%
24.97%
90.36%
Upper Suwanee HUC 03110201
Flow
TSS
TN
TP
-24.84%
-24.26%
-23.50%
-24.98%
28.79%
39.56%
23.01%
25.16%
26.01%
37.78%
22.03%
24.20%
68.81%
98.24%
61.67%
69.24%
24.81%
24.23%
22.71%
23.70%
28.73%
39.49%
23.87%
25.90%
25.97%
37.72%
22.88%
25.02%
68.71%
98.07%
62.43%
71.25%
Alapaha HUC 031 10202
Flow
TSS
TN
TP
-15.81%
-18.00%
-11.17%
-23.30%
25.16%
31.27%
23.77%
13.11%
21.59%
28.14%
20.63%
10.13%
57.95%
76.14%
52.80%
43.22%
15.78%
17.97%
-9.70%
21.52%
25.09%
31.12%
24.31%
13.97%
21.53%
28.01%
21.62%
1 1 .48%
57.81%
75.79%
53.11%
43.43%
Withlacoochee (nr Pinetta, FL)
Flow
TSS
TN
TP
-21.49%
-25.76%
-10.73%
-21.61%
24.44%
28.45%
32.23%
21.51%
21.42%
26.79%
29.50%
20.29%
69.85%
85.95%
85.42%
70.56%
21.48%
25.72%
11.02%
21.56%
24.34%
28.36%
31.22%
21.14%
21.32%
26.72%
28.96%
20.42%
69.60%
85.78%
84.79%
71.62%
Little HUC 031 10204
Flow
TSS
TN
TP
-25.92%
-27.77%
-11.90%
-23.93%
24.76%
24.16%
32.94%
23.99%
20.44%
20.64%
27.78%
19.96%
67.92%
69.75%
71.48%
59.29%
25.89%
27.74%
12.55%
24.05%
24.69%
24.08%
32.45%
24.13%
20.37%
20.61%
28.19%
21.23%
67.74%
69.67%
73.81%
63.09%
Lower Suwanee HUC 03110205
Flow
TSS
TN
-24.77%
-26.02%
-15.26%
21.27%
29.60%
30.76%
19.99%
30.11%
30.75%
56.44%
81.11%
66.08%
24.81%
26.06%
15.75%
21.19%
29.50%
30.60%
19.89%
29.97%
31.00%
56.27%
80.86%
66.62%
Y-38
-------
TP
Results without LU change
Min
-24.37%
Median
25.22%
Mean
26.51%
Max
73.14%
Results with LU change
Min
25.00%
Median
25.58%
Mean
27.00%
Max
74.17%
Santa Fe HUC 031 10206
Flow
TSS
TN
TP
-26.50%
-34.97%
0.76%
-30.13%
15.44%
24.34%
34.95%
19.26%
12.31%
21.88%
34.17%
16.36%
37.17%
63.75%
52.87%
52.01%
26.64%
34.48%
-2.79%
32.41%
15.29%
23.78%
33.01%
21.55%
12.14%
21.33%
31.08%
15.81%
36.95%
62.17%
50.12%
49.06%
Apalachee Bay - St. Marks HUC 03120001
Flow
TSS
TN
TP
-25.80%
-20.78%
-3.33%
-27.82%
10.87%
13.04%
18.02%
11.78%
12.84%
17.60%
18.73%
15.72%
45.59%
55.25%
38.17%
59.93%
25.92%
20.01%
-7.19%
30.89%
10.36%
10.39%
15.56%
8.01%
12.20%
13.52%
17.39%
13.45%
44.44%
45.25%
39.98%
57.31%
Upper Ochlockonee HUC 03120002
Flow
TSS
TN
TP
-25.04%
-38.65%
-22.80%
-31.13%
26.32%
34.32%
29.48%
13.95%
19.62%
27.93%
24.54%
9.86%
61.25%
82.94%
77.46%
51.03%
25.03%
38.65%
22.12%
30.52%
26.30%
34.31%
30.39%
14.76%
19.61%
27.92%
25.48%
10.75%
61.22%
82.92%
79.43%
53.16%
Lower Ochlockonee HUC 03120003
Flow
TSS
TN
TP
-30.24%
-31.40%
-16.07%
-34.99%
14.32%
20.86%
22.06%
16.34%
13.81%
21.26%
22.91%
18.35%
50.41%
69.71%
59.18%
74.66%
30.23%
31.52%
17.77%
36.39%
14.11%
20.58%
21.60%
15.54%
13.56%
20.90%
22.00%
16.71%
49.94%
69.17%
57.25%
69.16%
Little Manatee HUC 03100203
Flow
TSS
TN
TP
-39.73%
-45.87%
-35.64%
-42.97%
-1.06%
-2.45%
17.99%
7.04%
0.44%
3.26%
27.11%
17.11%
47.18%
67.43%
109.29%
98.78%
39.16%
41.56%
33.32%
37.23%
-1 .25%
-2.06%
0.81%
-3.02%
-0.05%
0.73%
11.16%
7.99%
44.98%
52.77%
74.13%
75.35%
Alafia HUC 03100204
Flow
TSS
TN
TP
-35.01%
-35.63%
-20.18%
-26.24%
4.37%
6.13%
13.29%
5.15%
1.61%
3.28%
30.43%
1 .69%
46.53%
51.25%
123.07%
31.06%
34.18%
33.45%
-7.94%
24.25%
4.30%
5.77%
14.11%
5.79%
1.29%
2.78%
30.90%
3.49%
43.89%
46.56%
95.91%
32.00%
Y-39
-------
Results without LU change
Min
Median
Mean
Max
Results with LU change
Min
Median
Mean
Max
Hillsborough HUC 03100205
Flow
TSS
TN
TP
-37.66%
-34.82%
-29.58%
-26.00%
6.42%
6.04%
13.41%
12.06%
4.82%
4.86%
17.23%
15.92%
59.53%
55.70%
84.71%
81.45%
36.21%
30.98%
24.74%
25.62%
6.31%
5.05%
3.81%
1.54%
4.77%
3.58%
12.50%
10.93%
56.79%
45.17%
80.96%
80.06%
Aucilla
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
567
Month
10 11 12
Figure 58. Monthly average flows, Aucilla River (SWAT)
Y-40
-------
Aucilla
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 59. Flow duration, Aucilla River (SWAT)
350
300
Upper Suwanee
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 60. Monthly average flows, Upper Suwanee River (SWAT)
Y-41
-------
Upper Suwanee
10000
03
Q 0.1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 61. Flow duration, Upper Suwanee River (SWAT)
Alapaha
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 62. Monthly average flows, Alapaha River (SWAT)
Y-42
-------
Alapaha
1000
CO
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 63. Flow duration, Alapaha River (SWAT)
Withlacoochee (nr Pinetta, FL)
LOWO
•LOW1
LOW2
• LOWS
LOW4
LOWS
LOW6
6 7
Month
10 11 12
Figure 64. Monthly average flows, Withlacoochee River near Pinetta, FL (SWAT)
Y-43
-------
Withlacoochee (nr Pinetta, FL)
10000
S 0.1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 65. Flow duration, Withlacoochee River near Pinetta, FL (SWAT)
Little
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 66. Monthly average flows, Little River (SWAT)
Y-44
-------
Little
1000
CO
^
o
03
CD
03
Q 0.001
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.0001
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 67. Flow duration, Little River (SWAT)
700
600
Lower Suwanee
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 68. Monthly average flows, Lower Suwanee River (SWAT)
Y-45
-------
Lower Suwanee
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 69. Flow duration, Lower Suwanee River (SWAT)
Santa Fe
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
'LOWS
LOW4
LOWS
LOW6
Figure 70. Monthly average flows, Santa Fe River (SWAT)
Y-46
-------
Santa Fe
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 71. Flow duration, Santa Fe River (SWAT)
Apalachee Bay - St. Marks
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
'LOWS
LOW4
LOWS
LOW6
Figure 72. Monthly average flows, Apalachee Bay at St. Marks HUC8 (SWAT)
Y-47
-------
Apalachee Bay - St. Marks
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 73. Flow duration, Apalachee Bay at St. Marks HUC8 (SWAT)
Upper Ochlockonee
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
'LOWS
LOW4
LOWS
LOW6
Figure 74. Monthly average flows, Upper Ochlockonee River (SWAT)
Y-48
-------
Upper Ochlockonee
1000
CO
^
o
0)
O)
03
CD
Q 0.01
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.001
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 75. Flow duration, Upper Ochlockonee River (SWAT)
300
250
200
co
^
O
150
100
Lower Ochlockonee
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
LOW1
LOW2
LOWS
LOW4
LOWS
LOW6
Figure 76. Monthly average flows, Lower Ochlockonee River (SWAT)
Y-49
-------
Lower Ochlockonee
10000
CO
o 1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 77. Flow duration, Lower Ochlockonee River (SWAT)
30%
0%
10 11 12
Figure 78. Average of median percent change in flow; NARCCAP Scenarios W1-W6 at all
stations, Georgia-Florida basins (SWAT)
Y-50
-------
Illinois River Basin
Illinois gage
-acn
300
"l" 250
u
_0 200
LL.
"TO
1 150
5
c
8 luu
1
50
n
+ +
fc #
BASE ICLUS
NARCCAP
Figure 79. Mean annual flow (cms), Illinois River at Marseilles Gage (SWAT)
~i/T 3,uuu
u
O
i/)
L.
nj o nnn
gj ^,UUU
Q.
bo
^£
ra
O
LL.
o
o
rt o
Illinois gage
n o
1 A
^§^ ^^^
* f
BASE ICLUS
NARCCAP
Figure 80. 100-yr Flow Peak (Log-Pearson III, cms), Illinois River at Marseilles Gage (SWAT)
Y-51
-------
120 -i
| 100
u
1
"- 80
5
ra
"ra
c
01
BO
£
01
^ 20
0
Illinois gage
*
i v
i 5
BASE
ICLUS
NARCCAP
Figure 81. Average Annual 7-day Low Flow (cms), Illinois River at Marseilles Gage (SWAT)
Illinois gage
n IJ°L
0.16
9
^ n 14
i/)
i/)
QJ n 19
.E
IE
i/)
E °-1
"ra 0.08
(/)
•n
2- n OR
ra u-ub
^
_o
2 0.04
0.02
i *
BASE ICLUS
NARCCAP
Figure 82. Richards-Baker Flashiness Index, Average Annual 7-day Low Flow (cms), Illinois
River at Marseilles Gage (SWAT)
Y-52
-------
onn -i
180
"t/T
•— -icn -
?
m -140 -
>-
Hi
43 -ion
!° mn
o luu
t
JS 80
O
u- 60
o
>» 4n
& 40
20
Illinois gage
+ +
I *
BASE ICLUS
NARCCAP
Figure 83. Days to Flow Centroid (Water Year Basis), Illinois River at Marseilles Gage (SWAT)
c nnn nnn -,
4 000 000
-a cnn nnn
TT
•^ 3 000 000
1
""*" o 500 000
TO
12
1 nnn nnn
Illinois gage
+ +
1 0
•
* *
BASE ICLUS
NARCCAP
Figure 84. TSS Load (MT/yr), Illinois River at Marseilles Gage (SWAT)
Y-53
-------
Illinois gage
fin nnn
i
TO
5
z
on nnn
+
i
BASE
NAR(
+
i
&
ICLUS
;CAP
Figure 85. TN Load (MT/yr), Illinois River at Marseilles Gage (SWAT)
Illinois gage
c; nnn
4,500
4 000
-?" ^ nnn
TO
3
i-
1 500
1 nnn
1 £ —
^u^_^_ ^^^?
>?~ ~Tr
BASE ICLUS
NARCCAP
Figure 86. TP Load (MT/yr), Illinois River at Marseilles Gage (SWAT)
Y-54
-------
Table 4. Summary of range of change relative to existing conditions for NARCCAP dynamically
downscaled scenarios, Illinois River basin SWAT model
Results without LU change
Min
Median
Mean
Max
Results with LU change
Min
Median
Mean
Max
KankakeeHUC 07120001
Flow
TSS
TN
TP
-21.45%
-4.70%
-11.22%
1.71%
2.47%
23.89%
4.31%
20.63%
3.59%
25.87%
6.39%
22.77%
29.22%
55.44%
24.79%
41.20%
-21.59%
-4.99%
-10.86%
1 .25%
2.08%
23.13%
4.52%
19.38%
3.19%
25.18%
6.74%
21.99%
28.55%
54.44%
25.60%
41.12%
IroquoisHUC 07120002
Flow
TSS
TN
TP
-19.02%
-1.36%
-9.61%
6.98%
1.08%
19.70%
3.63%
25.45%
2.38%
22.64%
5.93%
27.23%
24.71%
46.79%
22.43%
43.90%
-19.02%
-1.36%
-9.63%
6.94%
1.08%
19.69%
3.62%
25.43%
2.38%
22.63%
5.93%
27.21%
24.71%
46.77%
22.44%
43.90%
Des Plaines HUC 07120004
Flow
TSS
TN
TP
-15.37%
-7.59%
1.54%
-0.95%
1.06%
3.38%
3.17%
0.97%
2.90%
3.87%
4.01%
1.77%
30.24%
19.58%
8.33%
6.25%
-16.38%
-7.27%
2.72%
-0.10%
-0.98%
3.11%
3.89%
1 .29%
0.87%
3.40%
4.89%
2.27%
26.80%
17.52%
9.12%
6.67%
Upper Illinois HUC 07120005
Flow
TSS
TN
TP
-13.76%
-3.34%
-4.84%
-0.78%
1.92%
23.58%
3.04%
6.23%
1.90%
23.64%
3.59%
6.67%
18.95%
48.18%
12.71%
13.81%
-14.07%
-3.11%
-4.34%
-0.91%
1.39%
22.89%
3.28%
5.75%
1.35%
22.91%
3.90%
6.22%
18.13%
46.20%
13.11%
13.47%
Upper Fox HUC 0120006
Flow
TSS
TN
TP
-15.30%
-0.89%
-3.41%
-3.84%
6.91%
27.24%
5.50%
4.68%
7.15%
28.25%
5.56%
4.58%
34.00%
59.62%
15.65%
13.37%
-16.54%
-6.85%
-2.54%
-4.29%
3.58%
16.85%
5.69%
3.56%
3.85%
17.17%
5.87%
3.52%
28.15%
43.56%
15.56%
11.83%
Lower Fox HUC 0120007
Flow
TSS
TN
TP
-19.24%
-5.98%
-3.68%
-4.31%
3.40%
28.20%
8.72%
8.80%
3.91%
27.28%
8.80%
8.09%
29.74%
58.02%
21.69%
19.62%
-20.41%
-6.48%
-2.70%
-5.22%
0.54%
26.05%
9.00%
6.76%
1.06%
25.05%
9.09%
6.12%
24.97%
53.73%
21.61%
17.14%
Lower Illinois-Senachwine HUC 07130001
Flow
TSS
TN
TP
-15.36%
-5.82%
-4.57%
-1.58%
3.02%
25.70%
5.67%
6.95%
2.39%
24.77%
5.54%
7.40%
21.65%
52.66%
16.21%
15.77%
-15.85%
-5.81%
-4.02%
-2.01%
2.05%
24.84%
5.89%
6.15%
1.45%
23.91%
5.83%
6.56%
20.22%
50.68%
16.48%
14.90%
Vermillion HUC 07130002
Flow
TSS
-17.02%
-0.45%
5.33%
20.30%
4.19%
22.43%
25.88%
46.88%
-17.02%
-0.53%
5.33%
20.21%
4.18%
22.34%
25.88%
46.82%
Y-55
-------
TN
TP
Results without LU change
Min
-7.19%
9.87%
Median
9.15%
24.75%
Mean
9.13%
26.19%
Max
24.45%
38.87%
Results with LU change
Min
-7.19%
9.85%
Median
9.09%
24.68%
Mean
9.10%
26.13%
Max
24.48%
38.88%
Lower Illinois-Chatauqua HUC 07130003
Flow
TSS
TN
TP
-22.22%
-9.74%
-7.11%
-0.59%
1 .49%
18.25%
6.95%
7.55%
0.68%
18.44%
6.38%
7.84%
24.55%
41.52%
18.30%
13.24%
-22.56%
-9.81%
-6.71%
-0.73%
0.73%
18.21%
7.09%
7.25%
-0.06%
18.27%
6.59%
7.56%
23.37%
41.03%
18.40%
12.97%
Mackinaw HUC 07130004
Flow
TSS
TN
TP
-14.96%
-0.53%
-0.05%
8.38%
5.97%
20.39%
12.48%
23.27%
5.18%
21.41%
12.33%
24.11%
21.81%
34.77%
20.57%
38.81%
-14.97%
-0.41%
0.30%
8.52%
5.92%
20.45%
12.61%
23.15%
5.13%
21.44%
12.60%
24.06%
21.73%
34.76%
20.92%
38.53%
Kankakee gage 05520500
Flow
TSS
TN
TP
-23.05%
-8.32%
-12.85%
-2.31%
3.44%
20.60%
4.24%
18.25%
5.13%
24.03%
6.82%
21.51%
33.23%
51.31%
26.37%
42.18%
-23.18%
-8.56%
-12.08%
-2.24%
3.03%
19.71%
4.78%
17.52%
4.68%
23.18%
7.56%
21.12%
32.45%
50.16%
27.42%
42.26%
Illinois River at Marseilles gage 05543500
Flow
TSS
TN
TP
-14.29%
-5.10%
-4.61%
-0.61%
2.23%
23.52%
4.07%
6.93%
2.05%
23.10%
4.64%
7.41%
19.70%
48.70%
14.55%
15.05%
-14.56%
-4.89%
-4.14%
-0.80%
1.73%
23.03%
4.31%
6.51%
1.54%
22.59%
4.96%
6.97%
18.92%
47.25%
14.94%
14.71%
Y-56
-------
Kankakee 7120001
300
250
,200
o
150
1 2 3 4 5 6 7 8 9 10 11 12
100
LOWO
• LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 87. Monthly average flows, Kankakee River (SWAT)
Kankakee 7120001
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 88. Flow duration, Kankakee River (SWAT)
Y-57
-------
Iroquois 7120002
2345
6 7
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 89. Monthly average flows, Iroquois River (SWAT)
Iroquois 7120002
1000
0.01
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 90. Flow duration, Iroquois River (SWAT)
Y-58
-------
Des Plaines River
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 91. Monthly average flows, Des Plaines River (SWAT)
Des Plaines River
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 92. Flow duration, Des Plaines River (SWAT)
Y-59
-------
Upper Illinois 7120005
4567
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 93. Monthly average flows, Upper Illinois River (SWAT)
Upper Illinois 7120005
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 94. Flow duration, Upper Illinois River (SWAT)
Y-60
-------
Upper Fox 7120006
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 95. Monthly average flows, Upper Fox River (SWAT)
Upper Fox 7120006
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 96. Flow duration, Upper Fox River (SWAT)
Y-61
-------
Lower Fox 7120007
140
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 97. Monthly average flows, Lower Fox River (SWAT)
Lower Fox 7120007
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 98. Flow duration, Lower Fox River (SWAT)
Y-62
-------
Lower Illinois-Senachwine
700
2345
6 7
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 99. Monthly average flows, Lower Illinois-Senachwine River (SWAT)
Lower Illinois-Senachwine
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
20%
40%
60%
80%
100%
Percent of Time that Flow is Equaled or Exceeded
Figure 100. Flow duration, Lower Illinois-Senachwine River (SWAT)
Y-63
-------
Vermillion 7130002
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 101. Monthly average flows, Vermillion River (SWAT)
Vermillion 7130002
1000
o
CD
O)
£
CD
CD
Q
0.001
0.0001
0.00001
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Percent of Time that Flow is Equaled or Exceeded
Figure 102. Flow duration, Vermillion River (SWAT)
Y-64
-------
Lower Illinois-Chatauqua
2345
6 7
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 103. Monthly average flows, Lower Illinois River - Lake Chatauqua (SWAT)
Lower Illinois-Chatauqua
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
20%
40%
60%
80%
100%
Percent of Time that Flow is Equaled or Exceeded
Figure 104. Flow duration, Lower Illinois River - Lake Chatauqua (SWAT)
Y-65
-------
Mackinaw
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 105. Monthly average flows, Mackinaw River (SWAT)
Mackinaw
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 106. Flow duration, Mackinaw River (SWAT)
Y-66
-------
40%
01
M
ro
c
OJ
30%
20%
HI
I 10%
•a
OJ
OJ
OJ
0%
-10%
-20%
Month
Figure 107. Average of median percent change in flow; NARCCAP Scenarios W1-W6 at all
stations, Illinois River basin (SWAT)
Y-67
-------
Lake Erie Drainages
an
on
7n
E 60
b bu
1 50
LL.
™ an
c
< on
c JU
TO
OJ
^ 90
10
n
Upper Maumee
+ +
0 0
1 *
* *
BASE ICLUS
NARCCAP
Figure 108. Mean annual flow (cms), Upper Maumee River (SWAT)
1,200 -i
1,000
"at
", 800
O
3 600
Q.
BO
3
js: 400
ra
£
JZ 200
O
O
0
Upper Maumee
BASE ICLUS
NARCCAP
Figure 109. 100-yr Flow Peak (Log-Pearson III, cms), Upper Maumee River (SWAT)
Y-68
-------
Upper Maumee
R -
Average Annual 7-day Low Flow (cms)
^-^roco^oio^ic
+ +
*
^ ^
* *
BASE ICLUS
NARCCAP
Figure 110. Average Annual 7-day Low Flow (cms), Upper Maumee River (SWAT)
0.25 -i
| 0.2
c
u)
u)
01
_c
1 0.15
ra
LL.
1
! 0-
ra
u
£
0.05
0
Upper Maumee
fS *
f? ~fl
BASE
ICLUS
NARCCAP
Figure 111. Richards-Baker Flashiness Index, Average Annual 7-day Low Flow (cms), Upper
Maumee River (SWAT)
Y-69
-------
Upper Maumee
?nn
-i«n -
d (Water Year Basis)
3 § 1 8 I
0 '""
J=
O 80
o
u_ 60
0
% 40
Q
20
8^^^^^*
s
BASE ICLUS
NARCCAP
Figure 112. Days to Flow Centroid (Water Year Basis), Upper Maumee River (SWAT)
Upper Maumee
4c;n nnn
t 250 000
•o
+ +
0 0
tt
X
BASE ICLUS
NARCCAP
Figure 113. TSS Load (MT/yr), Upper Maumee River (SWAT)
Y-70
-------
25,000 -i
20,000
k.
^! 15,000
1
•o
ra
^ 10,000
h-
5,000
0
Upper Maumee
0 0
9 ¥
A 4
* &
BASE ICLUS
NARCCAP
Figure 114. TN Load (MT/yr), Upper Maumee River (SWAT)
Upper Maumee
1 ?nn
800
i
TO
5
Q.
+ +
•^ ^^
BASE ICLUS
NARCCAP
Figure 115. TP Load (MT/yr), Upper Maumee River (SWAT)
Y-71
-------
Table 5. Summary of range of change relative to existing conditions for NARCCAP dynamically
downscaled scenarios, Lake Erie Drainages SWAT model
Results without LU change
Min
Median
Mean
Max
Results with LU change
Min
Median
Mean
Max
St. Joseph HUC 04100003
Flow
TSS
TN
TP
-22.89%
-25.64%
25.95%
-8.66%
12.80%
16.90%
77.70%
23.49%
10.51%
14.96%
77.53%
21.83%
33.57%
43.07%
105.26%
42.97%
-22.79%
-25.50%
26.28%
-7.76%
12.71%
16.73%
77.52%
24.07%
10.43%
14.94%
77.11%
22.14%
33.34%
43.09%
104.36%
43.21%
St. Marys HUC 04100004
Flow
TSS
TN
TP
-4.24%
-3.73%
-16.79%
-6.10%
28.80%
41.27%
18.83%
43.32%
30.53%
44.47%
34.44%
37.39%
72.13%
91.08%
104.97%
61.91%
-4.34%
-2.98%
-15.82%
-3.21%
28.42%
41.22%
20.67%
48.43%
30.13%
44.40%
35.94%
42.73%
71.35%
90.43%
105.37%
68.63%
Upper Maumee HUC 04100005
Flow
TSS
TN
TP
-15.25%
-18.67%
7.69%
-4.90%
16.41%
26.44%
62.93%
36.28%
16.93%
29.55%
61.35%
33.75%
44.67%
68.27%
108.23%
57.36%
-15.22%
-18.62%
8.41%
-3.07%
16.21%
26.10%
64.07%
40.80%
16.71%
29.16%
62.32%
37.14%
44.22%
67.68%
107.87%
62.09%
Tiffin HUC 04100006
Flow
TSS
TN
TP
-13.80%
-17.44%
-9.81%
-17.53%
21.85%
27.37%
62.88%
32.57%
20.13%
25.32%
58.46%
27.38%
44.42%
50.60%
111.70%
52.46%
-13.86%
-17.48%
-9.58%
-17.64%
21.85%
27.35%
63.13%
32.17%
20.11%
25.30%
58.57%
27.09%
44.37%
50.56%
111.71%
51.51%
Lower Maumee HUC 04100009
Flow
TSS
TN
TP
-11.71%
-13.77%
-6.13%
-12.30%
21.11%
27.92%
42.66%
25.08%
22.12%
31.24%
42.80%
25.60%
50.16%
69.34%
90.76%
50.32%
-11.68%
-13.79%
-6.02%
-11.70%
20.98%
27.75%
43.03%
26.38%
21.98%
31.05%
42.95%
26.85%
49.87%
68.96%
90.45%
52.88%
Sandusky HUC 04100011
Flow
TSS
TN
TP
-0.95%
-8.31%
-31.81%
-12.38%
25.32%
29.94%
1.59%
17.28%
25.96%
33.12%
7.79%
16.17%
52.06%
67.69%
53.03%
33.74%
-0.95%
-8.31%
-31.81%
-12.38%
25.32%
29.94%
1.59%
17.28%
25.96%
33.12%
7.79%
16.17%
52.06%
67.69%
53.03%
33.74%
Huron-Vermillion HUC 04100012
Flow
TSS
TN
TP
-1.54%
-4.89%
-11.76%
-6.58%
23.92%
27.39%
21.83%
19.66%
25.09%
28.15%
25.71%
17.61%
52.60%
62.82%
60.68%
32.56%
-1.54%
-4.88%
-11.76%
-6.58%
23.92%
27.40%
21.88%
19.66%
25.09%
28.15%
25.73%
17.61%
52.60%
62.82%
60.70%
32.57%
Black-Rocky HUC 04110001
Flow
TSS
-3.80%
-4.59%
15.11%
24.78%
13.38%
22.34%
28.15%
46.53%
-3.45%
-4.30%
15.31%
25.05%
13.58%
22.45%
28.20%
46.22%
Y-72
-------
TN
TP
Results without LU change
Min
5.43%
8.64%
Median
23.11%
29.40%
Mean
24.65%
30.01%
Max
47.06%
60.10%
Results with LU change
Min
10.90%
15.60%
Median
28.63%
36.56%
Mean
30.45%
37.41%
Max
54.90%
69.39%
CuyahogaHUC 041 10002
Flow
TSS
TN
TP
-3.53%
-4.69%
25.29%
19.28%
8.57%
12.57%
39.35%
33.47%
8.88%
13.09%
42.79%
35.85%
21.12%
30.95%
61.17%
58.22%
-3.54%
-4.72%
25.92%
20.90%
8.52%
12.49%
39.69%
35.07%
8.84%
13.03%
43.26%
37.31%
21.06%
30.83%
63.33%
61.39%
Grand HUC 041 10004
Flow
TSS
TN
TP
-0.30%
-2.32%
9.67%
5.95%
8.34%
9.20%
19.91%
17.69%
10.17%
10.80%
21.34%
19.39%
26.75%
31.03%
39.53%
42.00%
-0.28%
-2.31%
9.98%
6.56%
8.35%
9.14%
19.53%
17.82%
10.18%
10.78%
21.27%
19.46%
26.76%
31.07%
40.15%
42.58%
Auglaize HUC 04100007
Flow
TSS
TN
TP
-7.88%
-9.63%
-8.36%
-12.16%
22.58%
30.52%
33.77%
23.93%
25.44%
36.15%
35.11%
25.75%
55.81%
76.22%
84.13%
53.96%
-7.86%
-9.60%
-8.30%
-12.13%
22.57%
30.54%
33.71%
23.76%
25.44%
36.11%
35.04%
25.70%
55.79%
75.94%
84.06%
54.09%
Blanchard HUC 04100008
Flow
TSS
TN
TP
-10.08%
-7.81%
0.65%
-7.93%
20.91%
31.52%
41.36%
31.74%
22.64%
35.53%
41.39%
30.58%
51.46%
74.47%
83.21%
55.72%
-10.03%
-7.76%
0.83%
-7.84%
20.90%
31.50%
41.31%
31.43%
22.63%
35.52%
41.34%
30.52%
51.42%
74.43%
83.25%
55.81%
Y-73
-------
St. Joseph
10 11 12
LOWO
• LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 116. Monthly average flows, St. Joseph River (SWAT)
St. Joseph
1000
Q 0.01
0.001
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 117. Flow duration, St. Joseph River (SWAT)
Y-74
-------
St. Marys
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 118. Monthly average flows, St. Marys River (SWAT)
St. Marys
1000
to
I
CD
O)
CD
CD
Q 0.01
0.001
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 119. Flow duration, St. Marys River (SWAT)
Y-75
-------
Upper Maumee
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 120. Monthly average flows, Upper Maumee River (SWAT)
Upper Maumee
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 121. Flow duration, Upper Maumee River (SWAT)
Y-76
-------
Tiffin
567
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 122. Monthly average flows, Tiffin River (SWAT)
Tiffin
1000
to
^
o
CD
O)
CD
CD
Q
0.1
0.01
0.001
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 123. Flow duration, Tiffin River (SWAT)
Y-77
-------
Lower Maumee
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 124. Monthly average flows, Lower Maumee River (SWAT)
Lower Maumee
10000
CO
^
I
CD
O)
CD
CD
CD
Q
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 125. Flow duration, Lower Maumee River (SWAT)
Y-78
-------
Sandusky
120
100
o
1 2 3 4 5 6 7 8 9 10 11 12
40
20
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 126. Monthly average flows, Sandusky River (SWAT)
Sandusky
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 127. Flow duration, Sandusky River (SWAT)
Y-79
-------
Huron
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 128. Monthly average flows, Huron River (SWAT)
Huron
1000
0.01
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 129. Flow duration, Huron River (SWAT)
Y-80
-------
Black
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 130. Monthly average flows, Black River (SWAT)
Black
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 131. Flow duration, Black River (SWAT)
Y-81
-------
Cuyahoga
1 2 3 4 5 6 7 8 9 10 11 12
10
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 132. Monthly average flows, Cuyahoga River (SWAT)
Cuyahoga
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 133. Flow duration, Cuyahoga River (SWAT)
Y-82
-------
Grand
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 134. Monthly average flows, Grand River (SWAT)
Grand
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 135. Flow duration, Grand River (SWAT)
Y-83
-------
60%
-20%
-30%
Month
Figure 136. Average of median percent change in flow; NARCCAP Scenarios W1-W6 at all
stations, Lake Erie drainages (SWAT)
Y-84
-------
Nebraska: Loup and Elkhorn River Basins
on
70
en
1
o
_O
ra
3
C
C 3Q
< JU
c
ra
QJ on
1 20
10
n
Lower Elkhorn (06800500)
i i
A A
O O
BASE ICLUS
NARCCAP
Figure 137. Mean annual flow (cms), Lower Elkhorn River (SWAT)
Lower Elkhorn (06800500)
1 000 -
"at
", 800
O
i_
ra Rnn
flj Duu
Q.
M
3
^ 400
ra
£
1Z 200
O
o
rt o
A
+
$
O
BASE
NARI
A
+
J
O
ICLUS
;CAP
Figure 138. 100-yr Flow Peak (Log-Pearson III, cms), Lower Elkhorn River (SWAT)
Y-85
-------
Lower Elkhorn (06800500)
on
Average Annual 7-day Low Flow (cms)
-^ -^ N) N) C
3 O1 O O1 O O1 C
* *
H ~
m m
O O
BASE ICLUS
NARCCAP
Figure 139. Average Annual 7-day Low Flow (cms), Lower Elkhorn River (SWAT)
Lower Elkhorn (06800500)
n 1
0.09
.g 0.08
_c
« 0.07
01
_c
•^ n nfi
_ra
& 0.05
^
A °-04
•s
^ 0.03
u
5
0.02
0.01
^^^^^— -I-
j^
BASE ICLUS
NARCCAP
Figure 140. Richards-Baker Flashiness Index, Average Annual 7-day Low Flow (cms), Lower
Elkhorn River (SWAT)
Y-86
-------
250 -i
"t/T
« 200
&
(Water Year
8
bwCentroic
1
u_
0
i? 50
Q
0
Lower Elkhorn (06800500)
, ,
i i
BASE
ICLUS
NARCCAP
Figure 141. Days to Flow Centroid (Water Year Basis), Lower Elkhorn River (SWAT)
Lower Elkhorn (06800500)
6,000,000 -i
5,000,000
_ 4,000,000
1
""*" 3 000 000
TO
5
1- 2,000,000
-------
Lower Elkhorn (06800500)
m nnn
9 000
8 000
7 000
i
^3 '
TO
""" 4 000
1-
3 000
9 nnn
1,000
• A
A A
^^^^
$
^L
$
BASE
NARI
ICLUS
;CAP
Figure 143. TN Load (MT/yr), Lower Elkhorn River (SWAT)
Lower Elkhorn (06800500)
1 400
|
^
TO
5
I—
400
200
t t
A A
0 O
BASE ICLUS
NARCCAP
Figure 144. TP Load (MT/yr), Lower Elkhorn River (SWAT)
Y-88
-------
Table 6. Summary of range of change relative to existing conditions for NARCCAP dynamically
downscaled scenarios, Loup and Elkhorn basins SWAT model
Results without LU change
Min
Median
Mean
Max
Results with LU change
Min
Median
Mean
Max
Upper Elkhorn gage 06799500
Flow
TSS
TN
TP
-37.25%
-44.79%
-43.55%
-46.94%
21.57%
24.89%
20.04%
26.75%
15.15%
17.52%
12.19%
18.98%
41.00%
49.08%
34.53%
52.58%
-37.25%
-44.78%
-43.55%
-46.94%
21.57%
24.89%
20.04%
26.75%
15.16%
17.52%
12.20%
18.98%
41.00%
49.08%
34.53%
52.58%
Lower Elkhorn gage 06800500
Flow
TSS
TN
TP
-32.28%
-40.44%
-12.24%
-34.81%
30.92%
39.06%
0.60%
31.02%
21.17%
30.60%
5.73%
22.92%
42.81%
61.82%
45.26%
47.40%
-32.43%
-40.27%
-12.14%
-34.77%
30.87%
39.05%
0.67%
30.99%
21.07%
30.61%
5.78%
22.89%
42.68%
61.79%
45.29%
47.34%
N Fork Elkhorn HUC 10220002
Flow
TSS
TN
TP
-31.27%
-24.86%
-18.98%
-17.12%
29.04%
33.74%
-10.03%
16.74%
20.62%
26.77%
-2.74%
12.21%
45.97%
58.16%
38.44%
27.10%
-31.27%
-24.86%
-18.99%
-17.14%
29.04%
33.73%
-10.04%
16.74%
20.61%
26.76%
-2.75%
12.21%
45.97%
58.15%
38.44%
27.11%
Logan HUC 10220004
Flow
TSS
TN
TP
-19.06%
-13.69%
-2.33%
-16.58%
42.44%
51.04%
9.36%
43.36%
37.19%
46.72%
16.70%
39.08%
65.62%
82.44%
63.74%
68.01%
-19.08%
-13.62%
-2.33%
-16.64%
42.43%
51.03%
9.36%
43.33%
37.17%
46.73%
16.71%
39.04%
65.60%
82.41%
63.74%
67.96%
Upper Middle Loup HUC 10210001
Flow
TSS
TN
TP
-37.61%
-52.72%
-37.66%
-44.06%
-2.34%
-3.93%
2.22%
-3.10%
-2.01%
-2.13%
3.01%
-2.28%
19.17%
32.01%
28.82%
25.00%
-37.61%
-52.72%
-37.66%
-44.06%
-2.34%
-3.93%
2.22%
-3.10%
-2.01%
-2.13%
3.01%
-2.28%
19.17%
32.01%
28.82%
25.00%
Dismal HUC 10210002
Flow
TSS
TN
TP
-37.63%
-50.64%
-33.01%
-43.47%
-4.76%
-8.78%
16.90%
-7.38%
-2.46%
-2.99%
9.58%
-3.30%
22.50%
34.85%
29.34%
27.23%
-37.63%
-50.64%
-33.01%
-43.47%
-4.76%
-8.78%
16.90%
-7.38%
-2.46%
-2.99%
9.58%
-3.30%
22.50%
34.85%
29.34%
27.23%
Lower Middle Loup HUC 10210003
Flow
TSS
TN
TP
-46.33%
-58.53%
-48.22%
-50.20%
-0.33%
-1.37%
3.34%
-2.70%
0.47%
3.49%
1 .68%
0.27%
30.89%
52.73%
36.42%
35.49%
-46.33%
-58.53%
-48.23%
-50.20%
-0.32%
-1.35%
3.38%
-2.68%
0.48%
3.50%
1 .66%
0.28%
30.89%
52.72%
36.36%
35.52%
South Loup HUC 10210004
Flow
-54.05%
11.97%
7.96%
49.17%
-54.05%
11.99%
7.97%
49.17%
Y-89
-------
TSS
TN
TP
Results without LU change
Min
-66.37%
-51.14%
-53.12%
Median
18.77%
3.28%
9.04%
Mean
19.46%
-3.42%
6.19%
Max
109.14%
31.16%
49.11%
Results with LU change
Min
-66.39%
-51.32%
-53.09%
Median
18.68%
3.02%
9.14%
Mean
19.52%
-3.73%
6.18%
Max
108.95%
30.26%
49.15%
Mud HUC 10210005
Flow
TSS
TN
TP
-73.10%
-80.86%
-55.02%
-73.53%
17.49%
17.09%
-1.14%
15.40%
10.94%
22.51%
-5.38%
10.07%
62.36%
139.63%
23.61%
67.06%
-73.10%
-81.00%
-55.33%
-73.34%
17.50%
16.21%
-2.01%
15.90%
10.95%
21.59%
-6.19%
10.57%
62.36%
137.83%
22.60%
67.80%
Upper North Loup HUC 10210006
Flow
TSS
TN
TP
-30.62%
-40.02%
-14.84%
-3.50%
0.69%
11.50%
34.51%
3.97%
1.14%
9.20%
30.67%
3.47%
21.30%
38.49%
74.21%
7.56%
-30.62%
-40.02%
-14.84%
-3.50%
0.69%
11.50%
34.51%
3.97%
1.14%
9.20%
30.67%
3.47%
21.30%
38.49%
74.21%
7.56%
Lower North Loup HUC 10210007
Flow
TSS
TN
TP
-52.42%
-57.69%
-46.98%
-52.85%
3.86%
5.14%
19.48%
2.62%
1.34%
5.07%
11.79%
3.18%
33.63%
49.86%
48.95%
41.57%
-52.42%
-57.69%
-46.94%
-52.84%
3.86%
5.14%
19.51%
2.63%
1.34%
5.07%
11.84%
3.19%
33.64%
49.85%
49.11%
41.59%
Calamus HUC 10210008
Flow
TSS
TN
TP
-38.44%
-49.76%
-29.90%
-40.65%
2.00%
2.13%
25.52%
1.14%
0.37%
2.00%
14.71%
0.63%
23.41%
36.20%
32.65%
26.62%
-38.44%
-49.76%
-29.90%
-40.65%
2.00%
2.13%
25.52%
1.14%
0.37%
2.00%
14.71%
0.63%
23.41%
36.20%
32.65%
26.62%
Loup HUC 10210009
Flow
TSS
TN
TP
-79.17%
-88.67%
-65.58%
-69.16%
12.46%
24.15%
17.51%
8.65%
11.64%
25.48%
11.07%
10.32%
72.64%
116.93%
58.25%
67.53%
-79.17%
-88.67%
-65.59%
-69.16%
12.47%
24.17%
17.34%
8.66%
11.64%
25.49%
11.03%
10.33%
72.64%
116.94%
58.30%
67.55%
Cedar HUC 10210010
Flow
TSS
TN
TP
-47.10%
-58.97%
-47.53%
-47.44%
15.83%
26.40%
4.25%
15.72%
7.21%
13.94%
-2.56%
7.19%
28.46%
42.11%
16.16%
28.71%
-47.10%
-58.96%
-47.53%
-47.43%
15.83%
26.41%
4.19%
15.73%
7.21%
13.94%
-2.37%
7.19%
28.46%
42.11%
16.06%
28.71%
Y-90
-------
Upper Elkhorn (06799500)
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
-LOW4
LOWS
LOW6
Figure 145. Monthly average flows, Upper Elkhorn River (SWAT)
Upper Elkhorn (06799500)
1000
0.01 4
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 146. Flow duration, Upper Elkhorn River (SWAT)
Y-91
-------
Lower Elkhorn (06800500)
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
• LOWS
•LOW4
LOWS
LOW6
Figure 147. Monthly average flows, Lower Elkhorn River (SWAT)
Lower Elkhorn (06800500)
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 148. Flow duration, Lower Elkhorn River (SWAT)
Y-92
-------
N Fork Elkhorn
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
-LOWS
-LOW4
LOWS
LOW6
Figure 149. Monthly average flows, North Fork Elkhorn River (SWAT)
N Fork Elkhorn
1000
0.01
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 150. Flow duration, North Fork Elkhorn River (SWAT)
Y-93
-------
Logan
567
Month
10 11 12
LOWO
•LOW1
LOW2
• LOWS
-LOW4
LOWS
LOW6
Figure 151. Monthly average flows, Logan River (SWAT)
Logan
100
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 152. Flow duration, Logan River (SWAT)
Y-94
-------
Upper Middle Loup
6789
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
• LOW4
LOWS
LOW6
Figure 153. Monthly average flows, Upper Middle Loup River (SWAT)
Upper Middle Loup
100
o
CD
CD
Q
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 154. Flow duration, Upper Middle Loup River (SWAT)
Y-95
-------
Dismal
567
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 155. Monthly average flows, Dismal River (SWAT)
Dismal
100
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 156. Flow duration, Dismal River (SWAT)
Y-96
-------
Lower Middle Loup
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
• LOWS
-LOW4
LOWS
LOW6
Figure 157. Monthly average flows, Lower Middle Loup River (SWAT)
Lower Middle Loup
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 158. Flow duration, Lower Middle Loup River (SWAT)
Y-97
-------
South Loup
6789
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
• LOW4
LOWS
LOW6
Figure 159. Monthly average flows, South Loup River (SWAT)
South Loup
1000
0.01
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 160. Flow duration, South Loup River (SWAT)
Y-98
-------
Mud
8 9 10 11 12
LOWO
•LOW1
LOW2
• LOWS
-LOW4
LOWS
LOW6
Figure 161. Monthly average flows, Mud River (SWAT)
Mud
100
0.01
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 162. Flow duration, Mud River (SWAT)
Y-99
-------
Upper North Loup
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
•LOW4
LOWS
LOW6
Figure 163. Monthly average flows, Upper North Loup River (SWAT)
Upper North Loup
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 164. Flow duration, Upper North Loup River (SWAT)
Y-100
-------
35%
0%
10 11 12
Figure 165. Average of median percent change in flow; NARCCAP Scenarios W1-W6 at all
stations, Loup-Elkhorn basins (SWAT)
Y-101
-------
Tar and Neuse River Basins
Neuse_Mouth
Qnn
Mean Annual Flow (cms)
-i. -i. hJ hJ t
Ul O Ul O Ul C
D O O O O O C
+ +
• •
A A
9 2
BASE ICLUS
NARCCAP
Figure 166. Mean annual flow (cms), mouth of Neuse River (SWAT)
Neuse_Mouth
m nnn
Q nnn
"in*
~ c nnn
o
tfl
ro c nnn
Q_
tin
O 4 nnn
g 3,000
a.
> o nnn
LL.
>• 1 ,000
o
o
""" o
+ _L.
T
A A
A A
• A
0 0
A A
?? T?
BASE ICLUS
NARCCAP
Figure 167. 100-yr Flow Peak (Log-Pearson III, cms), mouth of Neuse River (SWAT)
Y-102
-------
Neuse_Mouth
en
Average Annual 7-day Low Flow (cms)
_i._i.hON>tOCO-^.-k.C
nouiouiouiouic
+ +
A •
§
P) ^
\J
& x
BASE ICLUS
NARCCAP
Figure 168. Average Annual 7-day Low Flow (cms), mouth of Neuse River (SWAT)
0.12 -i
c
3
g 0.08
IE
-------
(Water Year Basis)
-* M M
8 8 8
bwCentroic
8
u_
0
i? 50
Q
0
Neuse_Mouth
£ •
f ~~^
BASE ICLUS
NARCCAP
Figure 170. Days to Flow Centroid (Water Year Basis), mouth of Neuse River (SWAT)
1 200 000
>. 1 000 000
1
TO
o
Jj
n
Neuse_Mouth
T
A A
L\L *-*
n O
2 x
* *
BASE ICLUS
NARCCAP
Figure 171. TSS Load (MT/yr), mouth of Neuse River (SWAT)
Y-104
-------
on nnn -,
1 4 000
i
TO
o
""" 8 000
1-
6 000
4 nnn
2,000
Neuse_Mouth
A
Q 2
A
>^^
BASE ICLUS
NARCCAP
Figure 172. TN Load (MT/yr), mouth of Neuse River (SWAT)
9 nnn
L.
§.
•n
ra
~* 1 nnn
I-
n
Neuse_Mouth
+
+
m "
Q
A ^
,'V-
T^~
BASE ICLUS
NARCCAP
Figure 173. TP Load (MT/yr), mouth of Neuse River (SWAT)
Y-105
-------
Table 7. Summary of range of change relative to existing conditions for NARCCAP dynamically
downscaled scenarios, Tar-Neuse watershed SWAT model
Results without LU change
Min
Median
Mean
Max
Results with LU change
Min
Median
Mean
Max
Contentnea Creek (gage 02091500)
Flow
TSS
TN
TP
-11.95%
-22.15%
-3.11%
-10.80%
20.12%
49.06%
44.81%
47.10%
21.63%
58.45%
53.26%
57.60%
61.83%
190.18%
141.32%
166.11%
-11.94%
-22.13%
-2.79%
-10.45%
19.88%
48.83%
46.27%
47.50%
21.40%
58.05%
54.87%
57.59%
61.34%
188.94%
143.00%
164.93%
Neuse River at Kinston (gage 02089500 )
Flow
TSS
TN
TP
-7.77%
-10.52%
5.65%
1.69%
20.31%
30.82%
40.64%
43.42%
22.01%
36.31%
43.90%
50.77%
59.34%
107.02%
97.93%
120.14%
-8.12%
-10.56%
14.33%
7.85%
19.10%
30.00%
52.17%
53.00%
20.76%
35.40%
57.70%
61.64%
56.94%
105.08%
127.50%
146.07%
Neuse at Mouth HUC 03020204
Flow
TSS
TN
TP
-13.65%
-18.16%
-1.17%
-6.45%
17.69%
29.00%
31.19%
42.94%
Tar River at Tarboro (Upper Tar, Fishing
Flow
TSS
TN
TP
-4.58%
-7.85%
8.65%
2.23%
20.27%
31.12%
27.11%
29.91%
19.90%
34.52%
35.80%
48.33%
58.08%
99.03%
88.54%
129.77%
-13.71%
-18.37%
1.51%
-3.23%
17.15%
28.07%
35.54%
48.33%
19.35%
33.50%
41.61%
53.95%
56.99%
96.92%
100.72%
142.56%
; HUCs 03020101 and 03020102
23.15%
42.58%
31.52%
31.09%
61.50%
126.63%
61.86%
71.26%
-4.61%
-7.79%
8.00%
1.98%
20.18%
31.07%
28.37%
31.64%
23.05%
42.51%
32.58%
32.55%
61.30%
126.44%
64.49%
74.33%
Pamlico (Tar Mouth) HUC 03020104
Flow
TSS
TN
TP
-9.47%
-13.29%
-1.21%
-8.97%
19.57%
31.58%
28.35%
31.05%
20.91%
34.71%
30.96%
34.74%
57.03%
96.24%
58.91%
77.35%
-9.49%
-13.32%
-0.27%
-7.62%
19.45%
31.41%
30.32%
33.21%
20.79%
34.52%
34.42%
38.10%
56.79%
95.77%
64.57%
83.03%
Lower Tar HUC 03020103
Flow
TSS
TN
TP
-7.29%
-14.95%
3.23%
-6.89%
20.69%
40.55%
30.57%
31.72%
22.12%
46.67%
34.36%
35.75%
60.12%
144.39%
72.61%
92.03%
-7.32%
-14.96%
5.11%
-3.73%
20.51%
40.38%
36.84%
39.27%
21.94%
46.43%
40.96%
42.98%
59.75%
143.72%
82.45%
103.36%
Upper Neuse HUC 03020201
Flow
TSS
TN
TP
-6.73%
-8.38%
7.82%
5.87%
20.35%
31.16%
36.82%
39.80%
21.98%
40.77%
43.72%
49.61%
58.69%
118.63%
102.65%
122.46%
-7.16%
-8.99%
22.37%
18.41%
19.03%
29.65%
55.25%
54.61%
20.56%
38.81%
64.51%
67.38%
55.96%
114.27%
143.10%
158.17%
Middle Neuse HUC 03020202
Flow
-1 1 .22%
19.38%
21.18%
59.87%
-11.38%
18.63%
20.42%
58.38%
Y-106
-------
TSS
TN
TP
Results without LU change
Min
-18.29%
0.24%
-5.32%
Median
38.51%
35.61%
39.93%
Mean
45.11%
38.59%
42.78%
Max
140.27%
89.70%
108.29%
Results with LU change
Min
-18.42%
5.27%
-0.99%
Median
37.61%
41.60%
46.18%
Mean
44.13%
47.52%
50.37%
Max
137.90%
107.46%
124.56%
Contentnea HUC 03020203
Flow
TSS
TN
TP
-12.43%
-20.96%
-4.11%
-9.95%
20.47%
42.94%
37.92%
38.83%
21.42%
49.15%
43.31%
43.97%
61.60%
160.75%
111.71%
124.92%
-12.43%
-20.88%
-3.82%
-9.31%
20.17%
42.63%
38.47%
37.67%
21.13%
48.77%
45.60%
44.44%
60.97%
159.54%
114.03%
123.57%
Y-107
-------
Contentnea Creek (gage)
10 11 12
LOWO
• LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 174. Monthly average flows, Contentnea Creek gage (SWAT)
Contentnea Creek (gage)
10000
CO
^
o
0)
O)
03
(D
ro
0.01
0.001
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 175. Flow duration, Contentnea Creek gage (SWAT)
Y-108
-------
Kinston (gage)
250
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 176. Monthly average flows, Neuse River at Kinston (SWAT)
Kinston (gage)
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 177. Flow duration, Neuse River at Kinston (SWAT)
Y-109
-------
Neuse Mouth
700
1 2 3 4 5 6 7 8 9 10 11 12
100
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 178. Monthly average flows, mouth of Neuse River (SWAT)
Neuse Mouth
10000
1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 179. Flow duration, mouth of Neuse River (SWAT)
Y-110
-------
Tarboro (Upper Tar, Fishing)
250
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 180. Monthly average flows, Upper Tar River at Tarboro (SWAT)
Tarboro (Upper Tar, Fishing)
10000
1000
I
CD
O)
CD
CD
-^ 0.01
CD
Q
0.001
0.0001
0.1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 181. Flow duration, Upper Tar River at Tarboro (SWAT)
Y-lll
-------
Pamlico (Tar Mouth)
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 182. Monthly average flows, mouth of Tar River at Pamlico (SWAT)
Pamlico (Tar Mouth)
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 183. Flow duration, mouth of Tar River at Pamlico (SWAT)
Y-112
-------
Lower Tar
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 184. Monthly average flows, Lower Tar River (SWAT)
Lower Tar
10000
S 0.1
0.01
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 185. Flow duration, Lower Tar River (SWAT)
Y-113
-------
Upper Neuse
250
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 186. Monthly average flows, Upper Neuse River (SWAT)
Upper Neuse
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 187. Flow duration, Upper Neuse River (SWAT)
Y-114
-------
Middle Neuse
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 188. Monthly average flows, Middle Neuse River (SWAT)
Middle Neuse
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 189. Flow duration, Middle Neuse River (SWAT)
Y-115
-------
Contentnea
140
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 190. Monthly average flows, Contentnea Creek (SWAT)
Contentnea
10000
S 0.1
0.01
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 191. Flow duration, Contentnea Creek (SWAT)
Y-116
-------
40%
35%
0%
10 11 12
Figure 192. Average of median percent change in flow; NARCCAP Scenarios W1-W6 at all
stations, Neuse-Tar watershed (SWAT)
Y-117
-------
New England Coastal Basins
300 -i
250
_ 200
in
u
_0 150
LL.
"ra
3
| 10°
c
ra
01
1 50
0
Merrimack River at Mouth
i i
* &
BASE ICLUS
NARCCAP
Figure 193. Mean annual flow (cms), mouth of the Merrimack River (SWAT)
Merrimack River at
3,000 -i
2,500
I
C
o
S2
ra 1 ^nn -
01 ' >ouu
0.
BO
5.
ra
01
0.
o
LL. ^00
I
rt n
O
T
4
A
vv
*^*
Mouth
5
4
A
Jk
^S
BASE
ICLUS
NARCCAP
Figure 194. 100-yr Flow Peak (Log-Pearson III, cms), mouth of the Merrimack River (SWAT)
Y-118
-------
Merrimack River at Mouth
<~n
verage Annual 7-day Low Flow (cms)
; s o1 g g i
<
+ +
t t
| A
BASE ICLUS
NARCCAP
Figure 195. Average Annual 7-day Low Flow (cms), mouth of the Merrimack River (SWAT)
Merrimack River at Mouth
n nc -
Richards-Baker Flashiness Index
o o o o o c
2 S 8 g S \
BASE ICLUS
NARCCAP
Figure 196. Richards-Baker Flashiness Index, Average Annual 7-day Low Flow (cms), mouth of
the Merrimack River (SWAT)
Y-119
-------
Merrimack River at Mouth
?nn
180 -
"t/T
re
m
8 140 -
>-
15 120
g
lioo-
.fc
c
O 80 -
5
0
0
Q
20 -
A A
4u tt
BASE ICLUS
NARCCAP
Figure 197. Days to Flow Centroid (Water Year Basis), mouth of the Merrimack River (SWAT)
k.
>
•D
TO
o -ic nnn -
in
£
c, nnn -
n .
Merrimack River at Mouth
i i
* *
A *
BASE ICLUS
NARCCAP
Figure 198. TSS Load (MT/yr), mouth of the Merrimack River (SWAT)
Y-120
-------
Merrimack River at
1? nnn
8 000
i
""""" R nnn
TO
5
z
4 nnn
*
Mouth
•^£^^^^^^^_ ^"^
BASE
NAR(
ICLUS
;CAP
Figure 199. TN Load (MT/yr), mouth of the Merrimack River (SWAT)
Merrimack River at Mouth
700
>
•^
^ 500
JS ^nn
_i
Q.
300
n
*
8 s
^r w
BASE ICLUS
NARCCAP
Figure 200. TP Load (MT/yr), mouth of the Merrimack River (SWAT)
Y-121
-------
Table 8. Summary of range of change relative to existing conditions for NARCCAP dynamically
downscaled scenarios, New England Coastal basins SWAT model
Results without LU change
Min
Median
Mean
Max
Results with LU change
Min
Median
Mean
Max
Presumpscot HUC 01060001
Flow
TSS
TN
TP
-7.72%
-12.76%
-19.40%
-12.75%
14.77%
11.99%
-6.11%
15.74%
10.95%
7.38%
-8.73%
16.55%
18.05%
15.10%
-0.88%
39.60%
-7.63%
-12.60%
-19.60%
-12.11%
14.73%
1 1 .68%
-6.61%
13.87%
10.92%
7.27%
-9.41%
14.07%
17.95%
15.01%
-3.23%
31.27%
Saco River at Mouth HUC 01060002
Flow
TSS
TN
TP
-4.46%
33.26%
6.42%
-19.05%
9.65%
55.66%
19.19%
-6.74%
7.57%
55.69%
17.66%
-7.37%
12.18%
68.36%
23.73%
-0.59%
-4.45%
32.52%
6.10%
-18.81%
9.64%
54.78%
18.81%
-6.54%
7.57%
54.82%
17.28%
-7.16%
12.18%
67.47%
23.29%
-0.43%
Piscataqua HUC 01060003
Flow
TSS
TN
TP
-6.78%
-13.21%
21.60%
3.69%
9.38%
12.41%
40.51%
35.38%
7.23%
8.98%
41.00%
34.78%
14.73%
21.98%
60.85%
64.83%
-6.65%
-12.85%
19.49%
1.36%
9.48%
12.83%
40.86%
35.37%
7.32%
9.36%
39.95%
32.89%
14.83%
22.47%
57.88%
56.81%
Merrimack River at Mouth HUC 01070006
Flow
TSS
TN
TP
-6.14%
-15.05%
1.14%
-6.32%
9.19%
17.65%
18.37%
10.57%
7.37%
13.49%
16.74%
9.09%
14.98%
27.68%
27.55%
18.08%
-6.04%
-14.87%
0.97%
-6.15%
9.25%
17.60%
19.24%
12.02%
7.44%
13.44%
17.37%
10.27%
15.08%
27.62%
28.90%
20.51%
Pemigewasset HUC 01070001
Flow
TSS
TN
TP
-1.75%
-27.46%
-11.26%
-21.25%
8.56%
-12.23%
-8.86%
-18.27%
7.53%
-14.35%
-6.92%
-17.02%
12.70%
-9.88%
5.57%
-5.49%
-1.74%
-27.37%
-11.14%
-20.95%
8.57%
-12.12%
-8.79%
-17.94%
7.54%
-14.23%
-6.89%
-16.77%
12.71%
-9.74%
5.29%
-5.63%
Concord River at Lowell HUC 01070005
Flow
TSS
TN
TP
-12.55%
-15.37%
5.83%
-1.17%
10.86%
15.26%
14.53%
4.47%
7.20%
10.51%
15.28%
4.09%
16.86%
25.17%
24.74%
9.02%
-12.39%
-15.49%
5.14%
-1 .64%
10.88%
14.90%
14.19%
4.56%
7.27%
10.41%
14.77%
4.15%
17.02%
25.57%
24.48%
9.92%
Charles River at Mouth HUC 01090001
Flow
TSS
TN
TP
-4.45%
-5.04%
-0.09%
-0.05%
5.63%
6.85%
0.64%
0.31%
4.62%
5.78%
0.60%
0.26%
9.16%
11.94%
1.31%
0.41%
-4.41%
-5.01%
-0.10%
-0.05%
5.63%
6.90%
0.63%
0.32%
4.63%
5.87%
0.59%
0.27%
9.21%
12.20%
1.29%
0.43%
Winnepesaukee River HUC 01070002
Flow
TSS
0.03%
-16.55%
9.88%
-7.60%
9.07%
-7.98%
15.35%
-0.32%
0.05%
-16.27%
9.89%
-7.27%
9.08%
-7.68%
15.37%
0.02%
Y-122
-------
TN
TP
Results without LU change
Min
5.98%
-14.58%
Median
22.17%
3.20%
Mean
24.11%
9.24%
Max
54.33%
55.01%
Results with LU change
Min
5.00%
-15.18%
Median
21.40%
2.89%
Mean
23.31%
8.43%
Max
53.52%
53.36%
Contoocook HUC 01070003
Flow
TSS
TN
TP
-8.43%
-28.90%
9.99%
-12.86%
8.37%
-3.23%
30.28%
7.33%
6.64%
-5.16%
28.91%
4.32%
15.65%
10.29%
42.92%
14.57%
-8.32%
-28.72%
7.95%
-13.04%
8.45%
-3.07%
28.21%
8.78%
6.72%
-4.97%
27.29%
4.94%
15.75%
10.65%
42.17%
13.78%
Nashua HUC 01070004
Flow
TSS
TN
TP
-11.94%
-29.14%
-19.12%
-31.88%
11.04%
4.66%
1.95%
-7.56%
8.60%
0.81%
4.28%
-2.98%
19.79%
14.13%
31.83%
24.15%
-11.80%
-28.65%
-19.59%
-31.27%
11.06%
4.81%
2.34%
-5.86%
8.65%
1.11%
5.62%
-0.54%
19.93%
14.54%
35.57%
30.43%
Nashua River at East Pepperell gage 01096500
Flow
TSS
TN
TP
-11.75%
-27.21%
-20.27%
-34.88%
10.34%
10.65%
-4.91%
-16.51%
8.32%
6.67%
-1.04%
-13.16%
20.33%
23.02%
26.21%
12.60%
-11.65%
-26.93%
-21.80%
-35.84%
10.35%
10.66%
-6.15%
-17.26%
8.36%
6.79%
-2.21%
-13.77%
20.42%
23.44%
25.33%
12.04%
Saco River at Cornish gage 01066000
Flow
TSS
TN
TP
-4.01%
-30.52%
-0.09%
-30.86%
8.86%
-22.56%
11.06%
-15.88%
6.87%
-23.95%
9.71%
-16.73%
10.80%
20.95%
15.89%
-9.98%
-4.01%
-30.45%
-0.18%
-30.22%
8.86%
-22.48%
10.94%
-15.43%
6.87%
-23.88%
9.58%
-16.29%
10.80%
20.88%
15.73%
-9.72%
Y-123
-------
Presumpscot
10 11 12
LOWO
• LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 201. Monthly average flows, Presumpscot River (SWAT)
Presumpscot
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 202. Flow duration, Presumpscot River (SWAT)
Y-124
-------
Saco River at Mouth
300
250
200
to
^
o
150
1 2 3 4 5 6 7 8 9 10 11 12
100
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 203. Monthly average flows, mouth of the Saco River (SWAT)
Saco River at Mouth
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 204. Flow duration, mouth of the Saco River (SWAT)
Y-125
-------
Piscataqua
120
100
o
1 2 3 4 5 6 7 8 9 10 11 12
40
20
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 205. Monthly average flows, Piscataqua River (SWAT)
Piscataqua
1000
0.1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 206. Flow duration, Piscataqua River (SWAT)
Y-126
-------
Merrimack River at Mouth
1 2 3 4 5 6 7 8 9 10 11 12
100
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 207. Monthly average flows, mouth of the Merrimack River (SWAT)
Merrimack River at Mouth
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 208. Flow duration, mouth of the Merrimack River (SWAT)
Y-127
-------
Pemigewasset
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 209. Monthly average flows, Pemigewasset River (SWAT)
Pemigewasset
10000
CO
^
o
CD
O)
CD
CD
CD
Q
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 210. Flow duration, Pemigewasset River (SWAT)
Y-128
-------
Concord River at Lowell
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 211. Monthly average flows, Concord River at Lowell (SWAT)
Concord River at Lowell
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 212. Flow duration, Concord River at Lowell (SWAT)
Y-129
-------
Charles River at Mouth
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 213. Monthly average flows, mouth of the Charles River (SWAT)
Charles River at Mouth
1000
CO
0)
O)
03
CD
03
Q
100
10
1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 214. Flow duration, mouth of the Charles River (SWAT)
Y-130
-------
Winnepesaukee
4567
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 215. Monthly average flows, Winnepesaukee River (SWAT)
Winnepesaukee
1000
0.1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 216. Flow duration, Winnepesaukee River (SWAT)
Y-131
-------
Contoocook
120
100
o
1 2 3 4 5 6 7 8 9 10 11 12
40
20
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 217. Monthly average flows, Contoocook River (SWAT)
Contoocook
1000
0.1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 218. Flow duration, Contoocook River (SWAT)
Y-132
-------
Nashua
567
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 219. Monthly average flows, Nashua River (SWAT)
Nashua
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 220. Flow duration, Nashua River (SWAT)
Y-133
-------
100%
-40%
Month
Figure 221. Average of median percent change in flow; NARCCAP Scenarios W1-W6 at all
stations, New England Coastal basins (SWAT)
Y-134
-------
Lake Pontchartrain Drainage
250 -i
200
"i/T
_o
LL.
1 100
1
c
ra
01
1 50
0
Lake Maurepas (Amite Mouth)
*
A ~~1
O
BASE
NARI
W
ICLUS
;CAP
Figure 222. Mean annual flow (cms), mouth of the Amite River at Lake Maurepas (SWAT)
Y-135
-------
Lake Maurepas (Amite Mouth)
7,000 -i
6,000
*i/r
E 5 000
o
O 4 000 -
S2
ra
£
oj3 3 000
5
QJ 2 000
O
LL" 1 000
6
o
rt n
+
+
t
* *
BASE
ICLUS
NARCCAP
Figure 223. 100-yr Flow Peak (Log-Pearson III, cms), mouth of the Amite River at Lake
Maurepas (SWAT)
Lake Maurepas (Amite Mouth)
ra
•? 20
ix
"ra
1 15
5
01
M m
ra lu
1
n
+ +
*
ffi fi
* *
BASE ICLUS
NARCCAP
Figure 224. Average Annual 7-day Low Flow (cms), mouth of the Amite River at Lake Maurepas
(SWAT)
Y-136
-------
ashiness Index
o o
- P k>
O1 M O1
LL.
1
ra
i_
ra
u
£
0.05
0
Lake Maurepas (Amite Mouth)
i
BASE
i
ICLUS
NARCCAP
Figure 225. Richards-Baker Flashiness Index, Average Annual 7-day Low Flow (cms), mouth of
the Amite River at Lake Maurepas (SWAT)
(Water Year Basis)
-^ N) N)
8 8 8
•o
o
.fc
& 100
1
0
to
& 50
Q
0
Lake Maurepas (Amite Mouth)
+ +
^~
BASE
|
ICLUS
NARCCAP
Figure 226. Days to Flow Centroid (Water Year Basis), mouth of the Amite River at Lake
Maurepas (SWAT)
Y-137
-------
Lake Maurepas (Amite Mouth)
cnn nnn
350 000
L.
i
TO
£
o
_IL
TT
•
^A
O
-fr
BASE
NARI
ICLUS
;CAP
Figure 227. TSS Load (MT/yr), mouth of the Amite River at Lake Maurepas (SWAT)
Lake Maurepas (Amite Mouth)
m nnn
Q nnn
i
""""" ^ nnn
TO
""" A nnn
3 000
2 000
1 ,000
n
^
*
A *
0
a- *
BASE ICLUS
NARCCAP
Figure 228. TN Load (MT/yr), mouth of the Amite River at Lake Maurepas (SWAT)
Y-138
-------
Lake Maurepas (Amite Mouth)
1 °nn
1 600
1 200
^ 1 ,000
•o
S 800
Q.
400
A
? A
o
O -ru
A
T7
BASE ICLUS
NARCCAP
Figure 229. TP Load (MT/yr), mouth of the Amite River at Lake Maurepas (SWAT)
Y-139
-------
Table 9. Summary of range of change relative to existing conditions for NARCCAP dynamically
downscaled scenarios, Neuse-Tar watershed SWAT model
Results without LU change
Min
Median
Mean
Max
Results with LU change
Min
Median
Mean
Max
Amite HUC 08070202
Flow
TSS
TN
TP
-23.29%
-29.86%
-11.70%
-12.27%
2.10%
7.74%
21.07%
15.19%
-1.82%
1.98%
21.52%
16.20%
16.57%
30.26%
49.95%
41.72%
-23.15%
-29.85%
-10.61%
-10.96%
2.10%
7.63%
23.62%
17.45%
-1.81%
1.83%
25.14%
19.30%
16.43%
29.43%
54.81%
46.13%
Tickfaw HUC 08070203
Flow
TSS
TN
TP
-24.75%
-26.07%
-20.05%
-26.81%
3.41%
5.13%
32.79%
17.23%
-1.67%
0.10%
28.86%
12.52%
15.82%
20.01%
60.83%
39.32%
-24.71%
-25.91%
-22.31%
-27.78%
3.40%
4.91%
32.67%
17.23%
-1.66%
-0.01%
30.15%
13.31%
15.79%
19.80%
65.51%
41.83%
Lake Maurepas (Amite Mouth) HUC 08070204
Flow
TSS
TN
TP
-23.02%
-28.77%
-9.04%
-17.45%
0.74%
5.89%
21.42%
14.34%
-1.98%
1.57%
20.56%
11.87%
15.38%
27.60%
42.96%
35.37%
-22.87%
-28.67%
-8.62%
-16.30%
0.71%
5.68%
23.61%
16.20%
-1.96%
1.33%
23.48%
14.15%
15.27%
26.78%
47.41%
37.83%
Tangipahoa R at Robert (gage 07375500)
Flow
TSS
TN
TP
-20.43%
-26.21%
-6.20%
-14.39%
3.61%
9.45%
36.55%
29.94%
-0.44%
3.89%
26.90%
20.14%
14.97%
26.57%
52.22%
46.75%
-20.43%
-26.11%
-6.27%
-14.44%
3.59%
9.61%
39.25%
32.14%
-0.45%
4.17%
28.94%
21.73%
14.96%
27.03%
55.30%
48.91%
Tchefuncte HUC 08090201
Flow
TSS
TN
TP
-24.68%
-25.36%
-14.60%
-19.34%
-0.53%
0.12%
12.84%
5.16%
-4.23%
-2.99%
11.39%
3.52%
15.81%
18.15%
32.27%
24.08%
-24.58%
-24.63%
-16.77%
-19.97%
-0.70%
-0.40%
13.50%
6.36%
-4.30%
-3.27%
12.23%
4.76%
15.58%
16.74%
36.97%
28.24%
Tickfaw R at Holden (gage 07376000)
Flow
TSS
TN
TP
-22.01%
-29.60%
-9.61%
-18.53%
7.63%
15.24%
17.93%
11.44%
2.06%
9.61%
12.33%
5.48%
21.82%
49.72%
29.33%
24.36%
-22.01%
-29.60%
-9.56%
-18.52%
7.63%
15.24%
17.98%
11.48%
2.06%
9.61%
12.38%
5.51%
21.82%
49.69%
29.41%
24.42%
Amite R nr Denham Sprs (gage 07378500)
Flow
TSS
TN
TP
-22.67%
-30.85%
-10.47%
-8.75%
3.18%
4.21%
16.89%
13.46%
-1.05%
-1.11%
15.87%
14.21%
16.97%
24.24%
43.35%
39.65%
-22.60%
-30.64%
-8.72%
-6.61%
3.16%
4.33%
18.39%
15.27%
-1.05%
-0.88%
18.66%
17.41%
16.90%
24.28%
47.57%
44.42%
Y-140
-------
Amite (8070202)
300
250
,200
o
150
1 2 3 4 5 6 7 8 9 10 11 12
100
LOWO
• LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 230. Monthly average flows, Amite River (SWAT)
Amite (8070202)
10000
1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 231. Flow duration, Amite River (SWAT)
Y-141
-------
Tickfaw (8070203)
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 232. Monthly average flows, Tickfaw River (SWAT)
Tickfaw (8070203)
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 233. Flow duration, Tickfaw River (SWAT)
Y-142
-------
Lake Maurepas (Amite Mouth)
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 234. Monthly average flows, Amite River Mouth (SWAT)
Lake Maurepas (Amite Mouth)
10000
1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 235. Flow duration, Amite River Mouth (SWAT)
Y-143
-------
Tangipahoa R at Robert (07375500)
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
6 7
Month
10 11 12
Figure 236. Monthly average flows, Tangipahoa River at Robert (SWAT)
Tangipahoa R at Robert (07375500)
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 237. Flow duration, Tangipahoa River at Robert (SWAT)
Y-144
-------
Teh efu note
34567
Month
8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 238. Monthly average flows, Tchefuncte River (SWAT)
Tchefuncte
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 239. Flow duration, Tchefuncte River (SWAT)
Y-145
-------
Tickfaw R at Holden (07376000)
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 240. Monthly average flows, Tickfaw River at Holden (SWAT)
Tickfaw R at Holden (07376000)
1000
0.1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 241. Flow duration, Tickfaw River at Holden (SWAT)
Y-146
-------
Amite R nr Denham Sprs (07378500)
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 242. Monthly average flows, Amite River near Denham Springs (SWAT)
Amite R nr Denham Sprs (07378500)
10000
CO
^
o
CD
O)
CD
CD
CD
Q
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 243. Flow duration, Amite River near Denham Springs (SWAT)
Y-147
-------
01
OB
ro
.c
u
4-t
C
OJ
u
OJ
o.
20%
10%
0%
-20%
•a
-------
Rio Grande Valley
60 -i
50
_ 40
u
1 30
LL.
"ra
3
< 20
c
ra
01
10
0
RG downstream 13020203
• A
£
A
BASE
*
A
ICLUS
NARCCAP
Figure 245. Mean annual flow (cms), Rio Grande downstream (SWAT)
120
i/>
E 100
u
O 80
u)
L.
ra
QJ
Q.
J. en
3
Ji
S 40
0 4U
O
u- pn
k. ^u
O
O
rt o
RG downstream 13020203
• •
t -&•
A X
A A
BASE ICLUS
NARCCAP
Figure 246. 100-yr Flow Peak (Log-Pearson III, cms), Rio Grande downstream (SWAT)
Y-149
-------
RG downstream 13020203
on
Average Annual 7-day Low Flow (cms)
-^ -^ N) N) C
3 O1 O O1 O O1 C
• •
tiz
I
A A
BASE ICLUS
NARCCAP
Figure 247. Average Annual 7-day Low Flow (cms), Rio Grande downstream (SWAT)
RG downstream 13020203
n nno
0.008
9
•n 0.007
(/)
£ 0.006
IE
u)
ro n nn^
"ra 0.004
CQ
(/)
"2 0.003
.c
_o
2 0.002
0.001
$ $
A A
n A
• •
BASE ICLUS
NARCCAP
Figure 248. Richards-Baker Flashiness Index, Average Annual 7-day Low Flow (cms), Rio
Grande downstream (SWAT)
Y-150
-------
RG downstream 13020203
1 on
160
"t/T
2 -|4n
re 14U
re
at
>- 120
5
< -inn
jwCentroid (\
I) 00 C
DOC
u_
0
g, 40
2
20
i A
| |
BASE ICLUS
NARCCAP
Figure 249. Days to Flow Centroid (Water Year Basis), Rio Grande downstream (SWAT)
an nnn -,
i_
>
•o
t^
(/i
RG downstream 13020203
• •
if &
S o
BASE ICLUS
NARCCAP
Figure 250. TSS Load (MT/yr), Rio Grande downstream (SWAT)
Y-151
-------
RG downstream 13020203
1 ?nn
800
i
-o °uu
ra
5
z
4nn
•
*
*
I
BASE
NAR
*
±
ICLUS
:CAP
Figure 251. TN Load (MT/yr), Rio Grande downstream (SWAT)
ifin -,
-ion
t -inn
h^1
**** on
ra
5
o. en
40
on
RG downstream 13020203
• •
* *
A
A
-|-
BASE ICLUS
NARCCAP
Figure 252. TP Load (MT/yr), Rio Grande downstream (SWAT)
Y-152
-------
Table 10. Summary of range of change relative to existing conditions for NARCCAP dynamically
downscaled scenarios, Rio Grande Valley SWAT model
Results without LU change
Min
Median
Mean
Max
Results with LU change
Min
Median
Mean
Max
Rio Grande Headwaters HUC 13010001
Flow
TSS
TN
TP
-41.70%
-63.67%
-41.87%
-52.61%
-35.70%
-50.43%
-29.09%
-32.42%
-29.64%
-42.25%
-24.31%
-29.60%
9.07%
15.46%
5.88%
5.96%
-41.70%
-63.66%
-41.87%
-52.58%
-35.70%
-50.43%
-29.10%
-32.41%
-29.64%
-42.25%
-24.31%
-29.59%
9.07%
15.46%
5.88%
5.96%
Alamosa-Trinchera HUC 13010002
Flow
TSS
TN
TP
-40.07%
-57.38%
-71.34%
-61.33%
-36.14%
-45.99%
-58.93%
-50.08%
-27.91%
-38.83%
-41.48%
-34.79%
9.87%
12.28%
51.02%
46.04%
-40.07%
-57.33%
-73.16%
-65.50%
-36.15%
-45.97%
-59.77%
-52.36%
-27.91%
-38.80%
-41.57%
-36.08%
9.87%
12.26%
55.40%
51.65%
Saguache HUC 13010004
Flow
TSS
TN
TP
-42.18%
-62.77%
-34.54%
-23.33%
-36.97%
-38.71%
-21.18%
-11.96%
-29.84%
-36.89%
-19.94%
-12.07%
8.55%
12.34%
4.40%
4.13%
-42.18%
-62.69%
-34.52%
-23.38%
-36.97%
-38.67%
-21.18%
-12.10%
-29.84%
-36.86%
-19.93%
-12.14%
8.55%
12.32%
4.40%
4.11%
Conejos HUC13010005
Flow
TSS
TN
TP
-45.38%
-42.45%
-78.09%
-70.37%
-34.01%
-35.94%
-60.90%
-46.54%
-27.42%
-32.35%
-45.43%
-37.75%
9.88%
-10.53%
1.30%
8.32%
-45.38%
-39.84%
-78.04%
-70.40%
-34.01%
-34.05%
-61.06%
-46.86%
-27.42%
-30.27%
-45.56%
-37.95%
9.87%
-9.05%
0.42%
8.49%
Rio Grande at Otowi Br (Upper RG - HUC 13020101)
Flow
TSS
TN
TP
-38.59%
-54.20%
-63.74%
-55.72%
-33.16%
-43.49%
-51.99%
-44.23%
-24.68%
-35.54%
-40.58%
-34.55%
12.64%
14.20%
22.83%
22.51%
-38.60%
-54.11%
-64.49%
-59.91%
-33.17%
-43.43%
-51.84%
-45.38%
-24.69%
-35.48%
-40.21%
-35.22%
12.64%
14.17%
25.09%
26.21%
Rio Chama HUC13020102
Flow
TSS
TN
TP
-37.82%
-25.35%
-48.02%
-41.88%
-29.36%
-18.19%
-35.10%
-28.21%
-22.58%
-12.24%
-27.69%
-22.28%
19.86%
22.72%
16.32%
22.51%
-37.83%
-25.34%
-47.94%
-41.81%
-29.37%
-18.19%
-35.11%
-28.20%
-22.58%
-12.25%
-27.72%
-22.29%
19.86%
22.65%
16.26%
22.39%
Rio Grande-Santa Fe HUC 13020201
Flow
TSS
TN
TP
-37.74%
-53.44%
-63.48%
-58.17%
-32.46%
-43.76%
-52.32%
-47.21%
-23.89%
-34.99%
-40.32%
-36.06%
12.53%
14.55%
25.45%
26.77%
-37.76%
-53.24%
-63.86%
-61.85%
-32.48%
-43.61%
-51.88%
-47.93%
-23.90%
-34.84%
-39.80%
-36.38%
12.53%
14.48%
27.72%
30.85%
Jemez HUC 13020202
Flow
TSS
-31.19%
-60.50%
-15.25%
-32.22%
-13.97%
-28.38%
1 1 .66%
3.78%
-31.23%
-59.73%
-15.28%
-31.87%
-13.98%
-28.05%
11.68%
3.53%
Y-153
-------
TN
TP
Results without LU change
Min
-10.06%
-19.00%
Median
1.78%
-0.67%
Mean
6.61%
-4.09%
Max
45.15%
5.32%
Results with LU change
Min
-12.22%
-10.21%
Median
1.09%
0.50%
Mean
0.21%
-0.76%
Max
6.36%
5.17%
Rio Grande Albuquerque (downstream) HUC 13020203
Flow
TSS
TN
TP
-34.27%
-51.16%
-62.68%
-58.63%
-29.35%
-40.58%
-51.75%
-47.32%
-21.38%
-32.46%
-39.76%
-36.02%
11.94%
13.77%
25.50%
27.09%
-34.26%
-50.78%
-62.85%
-61.33%
-29.35%
-40.32%
-50.88%
-47.61%
-21.38%
-32.21%
-38.95%
-35.97%
11.92%
13.63%
27.46%
30.97%
Saguache Creek HUC 13010004
Flow
TSS
TN
TP
-40.98%
-70.73%
-34.66%
-35.28%
-35.20%
-47.18%
-26.24%
-21.06%
-29.12%
-43.79%
-21.98%
-20.06%
7.83%
2.23%
3.51%
3.34%
-40.98%
-70.73%
-34.66%
-35.28%
-35.20%
-47.18%
-26.24%
-21.06%
-29.12%
-43.79%
-21.98%
-20.06%
7.83%
2.23%
3.51%
3.34%
RG near Lobatos (gage 08251500)
Flow
TSS
TN
TP
-40.04%
-57.43%
-70.93%
-60.88%
-36.01%
-45.68%
-58.84%
-49.85%
-27.86%
-38.80%
-41.24%
-34.42%
9.81%
12.26%
51.49%
45.97%
-40.04%
-57.38%
-72.92%
-64.89%
-36.01%
-45.66%
-59.72%
-52.14%
-27.86%
-38.77%
-41.37%
-35.75%
9.81%
12.25%
55.93%
51.63%
RG near Taos (gage 08276500)
Flow
TSS
TN
TP
-39.40%
-56.65%
-67.53%
-57.67%
-34.47%
-45.24%
-54.17%
-45.76%
-25.67%
-37.43%
-41.83%
-35.16%
11.72%
13.35%
26.31%
25.51%
-39.40%
-56.60%
-68.66%
-62.55%
-34.47%
-45.21%
-54.18%
-47.22%
-25.67%
-37.40%
-41.63%
-36.14%
11.72%
13.33%
28.55%
29.20%
RG at Albuquerque (gage 08330000)
Flow
TSS
TN
TP
-36.80%
-52.87%
-61.72%
-57.89%
-31.36%
-41.39%
-51.00%
-46.82%
-22.95%
-33.72%
-39.20%
-35.70%
12.68%
13.67%
25.36%
26.81%
-36.79%
-52.56%
-62.24%
-61.16%
-31.37%
-41.18%
-50.29%
-47.29%
-22.95%
-33.52%
-38.57%
-35.83%
12.66%
13.57%
27.44%
30.90%
Y-154
-------
Rio Grande Headwaters 13010001
1 2 3 4 5 6 7 8 9 10 11 12
Figure 253. Monthly average flows, Rio Grande Headwaters (SWAT)
Rio Grande Headwaters 13010001
100
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 254. Flow duration, Rio Grande Headwaters (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Y-155
-------
Alamosa-Trinchra 13010002
6 7
Month
Figure 255. Monthly average flows, Alamosa River (SWAT)
10 11 12
Alamosa-Trinchra 13010002
100
LOWO
•LOW1
LOW2
• LOWS
LOW4
LOWS
LOW6
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 256. Flow duration, Alamosa River (SWAT)
Y-156
-------
Saguache 13010004
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 257. Monthly average flows, Saguache River (SWAT)
Saguache 13010004
co
I
CD
O)
CO
CD
'co
Q
100
10
1
0.1
0.01
0.001
0.0001
(
0.00001
60
80
Percent of Time that Flow is Equaled or Exceeded
Figure 258. Flow duration, Saguache River (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Y-157
-------
Conejos 13010005
6 7
Month
10 11 12
Figure 259. Monthly average flows, Conejos River (SWAT)
Conejos 13010005
100
to
I
CD
O)
CD
CD
Q 0.001
0.0001
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 260. Flow duration, Conejos River (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Y-158
-------
RG at Otowi Br (Upper RG - 13020101)
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 261. Monthly average flows, Upper Rio Grande at Otowi Bridge (SWAT)
RG at Otowi Br (Upper RG -13020101)
100
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 262. Flow duration, Upper Rio Grande at Otowi Bridge (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Y-159
-------
RioChama 13020102
1 2 3 4 5 6 7 8 9 10 11 12
Figure 263. Monthly average flows, Rio Chama (SWAT)
Rio Chama 13020102
100
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 264. Flow duration, Rio Chama (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Y-160
-------
Rio Grande-Santa Fe 13020201
1 2 3 4 5 6 7 8 9 10 11 12
Figure 265. Monthly average flows, Rio Grande at Santa Fe (SWAT)
Rio Grande-Santa Fe 13020201
100
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 266. Flow duration, Rio Grande at Santa Fe (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Y-161
-------
Jemez 13020202
23456789
Month
Figure 267. Monthly average flows, Jemez River (SWAT)
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Jemez 13020202
100
0.001
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 268. Flow duration, Jemez River (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Y-162
-------
RG downstream 13020203
Figure 269. Monthly average flows, Rio Grande downstream (SWAT)
RG downstream 13020203
1000
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 270. Flow duration, Rio Grande downstream (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Y-163
-------
Saguache Creek
1 2 3 4 5 6 7 8 9 10 11 12
Figure 271. Monthly average flows, Saguache Creek (SWAT)
Saguache Creek
100
1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 272. Flow duration, Saguache Creek (SWAT)
LOWO
•LOW1
LOW2
•LOWS
-LOW4
LOWS
LOW6
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Y-164
-------
-40%
-45%
Month
Figure 273. Average of median percent change in flow; NARCCAP Scenarios W1-W6 at all
stations, Rio Grande Valley (SWAT)
Y-165
-------
Sacramento River Watershed
Sacramento-Stone Coral
-acn
300 -
"* 9^n -
u
_0 200 -
LJ_
"ro
5 i^n -
5
c
s iuu
i
50 -
n
i i
•^ ^*
+ +
BASE ICLUS
NARCCAP
Figure 274. Mean annual flow (cms), Sacramento River at Stone Corral (SWAT)
Sacramento-Stone
1 n nnn
LJ
Q nnn -
o
S2
gj D.UUU
Q.
M
3
^ 4 000
TO
£
O
6
O
rt n -
*
,
r^
BASE
NARI
Coral
A-
i
-------
Sacramento-Stone Coral
180 -
*wT
J_ 160 -
g
° idn
_ 140
g
° 120
ro
"? 100
"ro
c
c
$A
ft
T
BASE ICLUS
NARCCAP
Figure 276. Average Annual 7-day Low Flow (cms), Sacramento River at Stone Corral (SWAT)
Sacramento-Stone Coral
0.18 -
X
•o u- ID
_c
8 0.14 -
Ol
_c
ards-Baker Flas
D ° o ?
I) 00 1^ h
u
5
0.04
i i
2 5
BASE ICLUS
NARCCAP
Figure 277. Richards-Baker Flashiness Index, Average Annual 7-day Low Flow (cms), Sacramento
River at Stone Corral (SWAT)
Y-167
-------
Sacramento-Stone Coral
1 on
"t/T
2 -|4n
m 14U
re
d)
>- 120
oj
5
< -inn
jwCentroid (\
I) 00 C
DOC
u_
0
g, 40
2
20
A
ff
BASE ICLUS
NARCCAP
Figure 278. Days to Flow Centroid (Water Year Basis), Sacramento River at Stone Corral
(SWAT)
Sacramento-Stone Coral
3,000,000 -i
2 500 000 -
_ 2,000,000
1
^ 1,500,000
1
1— 1 000 000
500,000
0 -
A A
§ i
*£? *£$
J^^^^^^^^ ^^^^^^A
+ +
BASE ICLUS
NARCCAP
Figure 279. TSS Load (MT/yr), Sacramento River at Stone Corral (SWAT)
Y-168
-------
Sacramento-Stone Coral
1.4 nnn
19 nnn
1 0 000
L.
>
I— 8 000
•0
TO
O g QQQ _
Z
4 000
2 000
e I
M~ ~r
+ +
BASE ICLUS
NARCCAP
Figure 280. TN Load (MT/yr), Sacramento River at Stone Corral (SWAT)
Sacramento-Stone
3,000 -i
2,500
_ 2,000
h^1
ra
5
Q.
1 nnn
Coral
o
i
^ zx
+
BASE
+
ICLUS
NARCCAP
Figure 281. TP Load (MT/yr), Sacramento River at Stone Corral (SWAT)
Y-169
-------
Table 11. Summary of range of change relative to existing conditions for NARCCAP dynamically
downscaled scenarios, Sacramento River watershed SWAT model
Results without LU change
Min
Median
Mean
Max
Results with LU change
Min
Median
Mean
Max
Sacramento-Stone Corral HUC 18020104
Flow
TSS
TN
TP
-11.03%
-5.89%
-11.49%
-14.30%
-1 .23%
13.09%
-0.49%
1.78%
-1.80%
13.25%
0.42%
1.18%
4.45%
38.51%
9.80%
14.99%
-11.05%
-5.76%
-10.50%
-14.08%
-1.25%
13.33%
-0.46%
1.83%
-1.82%
13.38%
0.31%
1.06%
4.41%
38.72%
8.08%
14.14%
Lower Cow HUC 18020101 (Bend Bridge)
Flow
TSS
TN
TP
-5.09%
-10.94%
3.12%
1.00%
-0.95%
-1.09%
6.47%
7.26%
-1 .28%
1.53%
5.95%
6.24%
1.76%
20.73%
9.12%
10.78%
-5.09%
-10.89%
2.46%
0.44%
-0.94%
-0.82%
5.85%
7.13%
-1.27%
1.72%
5.57%
5.97%
1.76%
20.78%
8.62%
10.27%
Lower Cottonwood HUC 18020102
Flow
TSS
TN
TP
-16.68%
-15.40%
3.76%
-2.35%
-2.01%
4.05%
10.14%
7.52%
-4.01%
3.22%
12.05%
8.98%
5.32%
18.85%
27.24%
24.84%
-16.68%
-15.36%
3.62%
-2.20%
-2.01%
4.07%
10.07%
7.52%
-4.02%
3.22%
11.96%
8.96%
5.31%
18.80%
26.97%
24.61%
Sacramento-Lower Thomes HUC 18020103
Flow
TSS
TN
TP
-10.37%
-1.83%
-11.31%
-14.39%
-1 .24%
16.40%
-1.74%
0.25%
-1.77%
16.85%
-0.43%
-0.21%
4.18%
43.90%
9.19%
14.44%
-10.38%
-1.54%
-10.33%
-14.27%
-1.24%
16.67%
-1.68%
0.16%
-1.77%
17.10%
-0.43%
-0.27%
4.16%
44.16%
7.88%
14.02%
Lower Butte HUC 18020105
Flow
TSS
TN
TP
-20.79%
-30.11%
-16.48%
-13.12%
-1.86%
6.33%
3.62%
7.94%
-3.05%
3.17%
2.23%
6.42%
8.08%
24.54%
15.25%
19.27%
-20.94%
-30.00%
-15.09%
-12.03%
-2.09%
7.06%
1.49%
5.69%
-3.28%
3.71%
1.84%
5.81%
7.80%
24.88%
18.61%
22.50%
Upper Stony HUC 180201 15
Flow
TSS
TN
TP
-14.78%
-4.98%
5.30%
16.29%
3.16%
47.36%
31.33%
58.23%
1.86%
43.92%
29.96%
55.38%
10.29%
68.51%
48.92%
91.07%
-14.76%
-4.98%
4.87%
15.67%
3.18%
47.34%
30.64%
57.29%
1.88%
43.91%
29.28%
54.50%
10.32%
68.49%
47.91%
89.89%
Upper Cow HUC 18020118
Flow
TSS
TN
TP
-14.84%
-35.74%
-11.70%
-18.32%
-6.38%
-17.12%
-4.16%
-9.79%
-5.66%
-11.08%
-3.37%
-9.29%
3.04%
19.48%
5.96%
-1.08%
-14.81%
-35.73%
-14.94%
-19.03%
-6.35%
-17.15%
-9.06%
-11.34%
-5.63%
-11.11%
-8.87%
-10.47%
3.06%
19.44%
-2.00%
-2.76%
Sacramento River, Keswick gage
Flow
TSS
-0.31%
-14.51%
-0.09%
3.16%
-0.11%
2.19%
0.04%
14.94%
-0.31%
-14.13%
-0.09%
3.52%
-0.12%
2.56%
0.04%
15.33%
Y-170
-------
TN
TP
Results without LU change
Min
-0.40%
-0.80%
Median
0.23%
0.26%
Mean
0.19%
0.18%
Max
0.49%
0.73%
Results with LU change
Min
-0.63%
-1.21%
Median
0.20%
0.14%
Mean
0.07%
-0.03%
Max
0.31%
0.39%
Y-171
-------
Sacramento-Stone Corral
1 2 3 4 5 6 7 8 9 10 11 12
100
LOWO
• LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 282. Monthly average flows, Sacramento River at Stone Corral (SWAT)
Sacramento-Stone Corral
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 283. Flow duration, Sacramento River at Stone Corral (SWAT)
Y-172
-------
Lower Cow (Bend Bridge)
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 284. Monthly average flows, Sacramento River at Bend Bridge (SWAT)
10000 i
'u>
0 1000
1
Li-
ra 10°
03
CD
<
^ 10
CD
Q
1
Lower Cow (Bend Bridge)
V
— —
"• -^
•* — ,
— -^
^^LOWO
LOW1
— LOW2
LOWS
LOW4
LOWS
— LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 285. Flow duration, Sacramento River at Bend Bridge (SWAT)
Y-173
-------
Lower Cottonwood
5678
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 286. Monthly average flows, Lower Cottonwood (SWAT)
Lower Cottonwood
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 287. Flow duration, Lower Cottonwood (SWAT)
Y-174
-------
Sacramento-Lower Thomes
700
1 2 3 4 5 6 7 8 9 10 11 12
100
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 288. Monthly average flows, Sacramento - Lower Thomes (SWAT)
Sacramento-Lower Thomes
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 289. Flow duration, Sacramento -Lower Thomes (SWAT)
Y-175
-------
Lower Butte
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 290. Monthly average flows, Lower Butte (SWAT)
Lower Butte
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 291. Flow duration, Lower Butte (SWAT)
Y-176
-------
Upper Stony
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 292. Monthly average flows, Upper Stony (SWAT)
Upper Stony
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 293. Flow duration, Upper Stony (SWAT)
Y-177
-------
Upper Cow
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 294. Monthly average flows, Upper Cow (SWAT)
Upper Cow
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 295. Flow duration, Upper Cow (SWAT)
Y-178
-------
Keswick gage
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 296. Monthly average flows, Sacramento River at Keswick Gage (SWAT)
10000 -1
'u>
0 1000
1
LL
8, 10°
03
CD
^
>> 10
CO
Q
H
Keswick gage
k.
^v
=5
•• —
===:
^^^^^^
^^^^r
=====
**iiii
~"^==::\
3
^^LOWO
LOW1
LOWS
LOW4
LOWS
— LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 297. Flow duration, Sacramento River at Keswick Gage (SWAT)
Y-179
-------
01
OB
ro
.c
u
4-t
c
OJ
OJ
Q.
C
.2
•5
OJ
0)
0)
15%
10%
5%
0%
-5%
-10%
-15%
Month
Figure 298. Average of median percent change in flow; NARCCAP Scenarios W1-W6 at all
stations, Sacramento River watershed (SWAT)
Y-180
-------
South Platte River Basin
on
u
1 15
LL.
"ra
3
C
c
ra
01
n
Middle S Platte-Cherry Crk 10190003
ft *
A A
ft
o u
BASE ICLUS
NARCCAP
Figure 299. Mean annual flow (cms), Middle South Platte River at Cherry Creek (SWAT)
Middle S Platte-Cherry Crk 10190003
I/)
E
", 800 -
c
O
S2
bo
5.
ra
3
o
8
rt n
A
^L. ^>
^np IP I
•TT
A *
_ ^
BASE ICLUS
NARCCAP
Figure 300. 100-yr Flow Peak (Log-Pearson III, cms), Middle South Platte River at Cherry Creek
(SWAT)
Y-181
-------
Average Annual 7-day Low Flow (cms)
O-^MCO^O1O)~-l
Middle S Platte-Cherry Crk 10190003
-fr w
f f
O u
BASE ICLUS
NARCCAP
Figure 301. Average Annual 7-day Low Flow (cms), Middle South Platte River at Cherry Creek
(SWAT)
0.06 -i
X 0.05
•o
c
8
£ 0.04
IE
U)
ra
fe 0.03
^
ra
CO
"ra °-02
^
u
£
0.01
0
Middle S Platte-Cherry Crk 10190003
A
^ •
* *
BASE ICLUS
NARCCAP
Figure 302. Richards-Baker Flashiness Index, Average Annual 7-day Low Flow (cms), Middle
South Platte River at Cherry Creek (SWAT)
Y-182
-------
300 -i
^-250 -
'to
re
m
re
>• 200
! 60,000 -
^
TO
""" AC\ nnn -
30 000 -
10,000
n -
A ft
I |
0 °
BASE ICLUS
NARCCAP
Figure 304. TSS Load (MT/yr), Middle South Platte River at Cherry Creek (SWAT)
Y-183
-------
Middle S Platte-Cherry Crk 10190003
6,000 -i
5,000
_ 4,000
k.
1
^ 3,000
ra
5
"~ 2,000
A A
I 1
•
. +
j*"*i. C ^
u u
BASE ICLUS
NARCCAP
Figure 305. TN Load (MT/yr), Middle South Platte River at Cherry Creek (SWAT)
4nn
-ann
L.
•n
™ onn
Q.
"~ 150
-inn
50
n
Middle S Platte-Cherry Crk 10190003
* *
/\
^ •
A A
S 5
BASE ICLUS
NARCCAP
Figure 306. TP Load (MT/yr), Middle South Platte River at Cherry Creek (SWAT)
Y-184
-------
Table 12. Summary of range of change relative to existing conditions for NARCCAP dynamically
downscaled scenarios, South Platte River watershed SWAT model
Results without LU change
Min
Median
Mean
Max
Results with LU change
Min
Median
Mean
Max
S Platte Headwaters HUC 10190001
Flow
TSS
TN
TP
-36.89%
-65.57%
-65.33%
-44.99%
-20.62%
-54.01%
-47.10%
-31.42%
-20.43%
-53.01%
-48.90%
-28.35%
0.84%
-36.33%
-31.36%
-3.42%
-36.93%
-65.71%
-65.35%
-44.93%
-20.69%
-54.51%
-47.33%
-31.00%
-20.51%
-53.56%
-48.97%
-28.14%
0.73%
-37.43%
-31.43%
-3.24%
Uppers Platte HUC 10190002
Flow
TSS
TN
TP
-35.38%
-41.89%
-37.36%
-30.52%
-13.10%
-15.38%
-19.15%
-15.11%
-15.35%
-19.00%
-19.82%
-15.22%
-0.68%
-1.21%
-4.91%
-1.63%
-35.01%
-39.28%
-37.33%
-30.26%
-13.20%
-14.59%
-19.33%
-14.82%
-15.34%
-17.98%
-19.67%
-14.66%
-1.02%
-1.77%
-3.59%
0.19%
Middle S Platte-Cherry Crk HUC 10190003
Flow
TSS
TN
TP
-35.48%
-19.67%
-37.48%
-27.54%
-10.22%
-7.10%
-1 1 .45%
-5.46%
-9.34%
-7.12%
-10.98%
-7.04%
18.52%
4.16%
16.28%
10.56%
-35.22%
-19.08%
-37.35%
-27.59%
-10.26%
-6.82%
-11.20%
-4.92%
-9.43%
-6.90%
-10.98%
-6.92%
18.01%
4.06%
15.73%
10.07%
Clear HUC 10190004
Flow
TSS
TN
TP
-19.59%
-68.91%
-19.10%
-28.01%
-8.28%
-57.70%
-10.15%
-14.79%
-7.78%
-58.62%
-8.96%
-12.76%
6.48%
-49.02%
5.27%
9.07%
-19.56%
-68.76%
-19.10%
-27.94%
-8.37%
-57.54%
-10.20%
-14.75%
-7.80%
-58.48%
-8.92%
-12.58%
6.53%
-48.91%
5.48%
9.58%
StVrain HUC 10190005
Flow
TSS
TN
TP
-18.59%
-25.25%
-21.67%
-5.28%
-4.98%
-12.08%
-8.97%
-1.28%
-6.21%
-13.12%
-10.09%
-1.89%
5.80%
-2.27%
0.52%
1.15%
-18.69%
-25.27%
-21.23%
-5.39%
-5.10%
-11.94%
-8.43%
-1.19%
-6.34%
-13.16%
-9.69%
-1.86%
5.62%
-2.75%
0.92%
1.19%
Big Thompson HUC 10190006
Flow
TSS
TN
TP
-10.44%
-25.70%
-1 1 .23%
-6.65%
-2.98%
-19.72%
-4.28%
-3.26%
-3.31%
-19.67%
-3.87%
-2.64%
3.12%
-14.31%
2.26%
1.88%
-10.49%
-25.37%
-1 1 .62%
-6.57%
-2.98%
-19.32%
-4.66%
-3.12%
-3.33%
-19.32%
-4.31%
-2.56%
3.12%
-13.94%
1.92%
1.42%
Cache La Poudre HUC 10190007
Flow
TSS
TN
TP
-26.87%
-39.12%
-23.68%
-4.55%
-5.02%
-18.89%
-1.27%
2.79%
-6.79%
-20.53%
-4.69%
3.36%
9.01%
-4.98%
15.28%
13.28%
-26.28%
-37.64%
-21.68%
-4.30%
-4.89%
-18.03%
-0.65%
2.41%
-6.59%
-19.68%
-3.58%
3.97%
9.00%
-4.65%
18.09%
15.42%
Lone Tree-Owl HUC 10190008
Flow
TSS
-29.85%
-30.31%
-2.46%
-0.91%
1.93%
5.36%
32.81%
39.21%
-29.79%
-29.94%
-2.45%
-1.06%
1.92%
5.04%
32.73%
38.13%
Y-185
-------
TN
TP
Results without LU change
Min
-31.92%
-29.46%
Median
-7.24%
-7.93%
Mean
-3.57%
-3.07%
Max
33.36%
30.23%
Results with LU change
Min
-31.77%
-29.21%
Median
-7.12%
-7.78%
Mean
-3.44%
-3.06%
Max
33.46%
29.70%
Crow HUC 10190009
Flow
TSS
TN
TP
-53.04%
-47.25%
-35.94%
-7.34%
-27.95%
-18.53%
-17.32%
-2.93%
-16.36%
-9.23%
-8.94%
-1.72%
25.26%
27.06%
18.73%
4.22%
-53.04%
-47.16%
-35.93%
-7.34%
-27.95%
-18.49%
-17.31%
-2.93%
-16.36%
-9.21%
-8.90%
-1.72%
25.26%
27.01%
18.93%
4.23%
Kiowa HUC 10190010
Flow
TSS
TN
TP
-20.20%
-17.09%
1 1 .20%
-3.07%
7.02%
6.16%
24.69%
2.35%
13.17%
14.66%
26.70%
3.96%
59.23%
60.82%
42.80%
14.21%
-20.20%
-17.09%
1 1 .20%
-3.07%
7.02%
6.16%
24.69%
2.35%
13.17%
14.65%
26.70%
3.96%
59.23%
60.81%
42.80%
14.21%
Bijou HUC 10190011
Flow
TSS
TN
TP
-44.46%
-57.62%
-59.50%
-43.73%
-5.63%
-14.14%
-40.45%
-8.30%
-6.93%
-14.33%
-38.73%
-9.95%
31.10%
27.79%
-22.22%
23.06%
-44.41%
-56.46%
-59.50%
-43.25%
-5.65%
-13.94%
-40.42%
-8.14%
-6.95%
-14.20%
-38.74%
-9.78%
31.01%
27.13%
-22.28%
22.76%
S Platte at Henderson (gage 06720500)
Flow
TSS
TN
TP
-26.13%
-45.93%
-21.50%
-16.10%
-7.87%
-27.09%
-9.07%
-6.24%
-10.17%
-29.11%
-9.18%
-6.35%
1.35%
-19.72%
2.21%
2.56%
-26.12%
-43.64%
-21.73%
-16.42%
-8.32%
-26.28%
-9.19%
-6.34%
-10.46%
-27.83%
-9.02%
-6.22%
0.66%
-18.24%
3.48%
3.71%
Box Elder Creek
Flow
TSS
TN
TP
S Platte at Denver (gac
Flow
TSS
TN
TP
-60.45%
-59.18%
-45.23%
-46.68%
-11.87%
-12.68%
-34.61%
-7.23%
-16.04%
-15.39%
-32.63%
-10.58%
17.00%
20.95%
-19.49%
17.85%
-59.20%
-51.55%
-45.66%
-43.35%
-11.53%
-10.83%
-34.34%
-5.83%
-15.60%
-13.45%
-32.33%
-9.03%
17.00%
17.55%
-19.12%
18.07%
je 0671 4000)
-35.68%
-40.75%
-36.98%
-30.63%
-10.19%
-9.77%
-17.73%
-14.00%
-13.61%
-16.06%
-18.82%
-14.57%
3.07%
1.64%
-4.22%
-1.40%
-34.68%
-37.20%
-36.48%
-30.04%
-10.75%
-10.76%
-17.70%
-13.35%
-13.81%
-14.94%
-18.46%
-13.68%
1.71%
3.33%
-3.44%
0.43%
Y-186
-------
S Platte Headware 10190001
10 11 12
LOWO
• LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 307. Monthly average flows, South Platte Headwaters (SWAT)
S Platte Headware 10190001
1000
o
CD
O)
CD
CD
CD
Q
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.00001
Percent of Time that Flow is Equaled or Exceeded
Figure 308. Flow duration, South Platte Headwaters (SWAT)
Y-187
-------
Upper SPIatte 10190002
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 309. Monthly average flows, Upper South Platte River (SWAT)
Upper SPIatte 10190002
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 310. Flow duration, Upper South Platte River (SWAT)
Y-188
-------
Middle S Platte-Cherry Crk 10190003
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 311. Monthly average flows, Middle South Platte River at Cherry Creek (SWAT)
Middle S Platte-Cherry Crk 10190003
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 312. Flow duration, Middle South Platte River at Cherry Creek (SWAT)
Y-189
-------
Clear 10190004
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 313. Monthly average flows, Clear Creek (SWAT)
Clear 10190004
100
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 314. Flow duration, Clear Creek (SWAT)
Y-190
-------
StVrain 10190005
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 315. Monthly average flows, St Vrain River (SWAT)
StVrain 10190005
100
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 316. Flow duration, St Vrain River (SWAT)
Y-191
-------
Big Thompson 10190006
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 317. Monthly average flows, Big Thompson River (SWAT)
Big Thompson 10190006
1000
0.1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 318. Flow duration, Big Thompson River (SWAT)
Y-192
-------
Cache La Poudre 10190007
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 319. Monthly average flows, Cache La Poudre River (SWAT)
Cache La Poudre 10190007
1000
0.1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 320. Flow duration, Cache La Poudre River (SWAT)
Y-193
-------
Lone Tree-Owl 10190008
1 2 3 4 5 6 7 8 9 10 11 12
0.5
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 321. Monthly average flows, Lone Tree - Owl (SWAT)
Lone Tree-Owl 10190008
100
o
CD
O)
CD
CD
Q 0.001
0.0001
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 322. Flow duration, Lone Tree - Owl (SWAT)
Y-194
-------
Crow 10190009
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 323. Monthly average flows, Crow River (SWAT)
Crow 10190009
100
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 324. Flow duration, Crow River (SWAT)
Y-195
-------
Kiowa 10190010
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 325. Monthly average flows, Kiowa River (SWAT)
Kiowa 101 90010
m -i
to
o
1
t
O)
£
£
<
^
03
Q
1 _
[
I
LOWO
LOW1
— LOW2
LOWS
LOW4
— LOWS
— LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 326. Flow duration, Kiowa River (SWAT)
Y-196
-------
40%
-30%
Month
Figure 327. Average of median percent change in flow; NARCCAP Scenarios W1-W6 at all
stations, South Platte River watershed (SWAT)
Y-197
-------
Powder and Tongue Rivers Basins
Lower Tongue
25
on
i
o
3
0 15
LL.
TO
3
£
C
TO
OJ
c
n
JW
w
o
BASE
_«_
M
O
ICLUS
NARCCAP
Figure 328. Mean annual flow (cms), Lower Tongue River (SWAT)
cnn -,
400
"uT
5 oou
~ -ann
§
a! ^^
Q.
bo
^£
S 1bu
Q.
LL.
k.
> 50
3
rt o
Lower Tongue
& &
• •
$ $
O O
BASE ICLUS
NARCCAP
Figure 329. 100-yr Flow Peak (Log-Pearson III, cms), Lower Tongue River (SWAT)
Y-198
-------
A -i
(/)
U
1
LL.
3 25
0 2.5
>
ra
•n j
r«.
"ra
i 1.5
<
01
BO .
ra 1
1
n ^
Lower Tongue
* *
m m
A A
A A
* *
O O
BASE ICLUS
NARCCAP
Figure 330. Average Annual 7-day Low Flow (cms), Lower Tongue River (SWAT)
Lower Tongue
Richards-Baker Flashiness Index
O O O O O C
2 S 8 g S 8
§ i
BASE ICLUS
NARCCAP
Figure 331. Richards-Baker Flashiness Index, Average Annual 7-day Low Flow (cms), Lower
Tongue River (SWAT)
Y-199
-------
Lower Tongue
onn
Days to FlowCentroid (Water Year Basis)
-^ -^ N) N) C
8 8 8 8 8 I
i *
8 8
BASE ICLUS
NARCCAP
Figure 332. Days to Flow Centroid (Water Year Basis), Lower Tongue River (SWAT)
1 9nn nnn
i
TO
5
•dnn nnn
n
Lower Tongue
ft ft
t t
J S
BASE ICLUS
NARCCAP
Figure 333. TSS Load (MT/yr), Lower Tongue River (SWAT)
Y-200
-------
1 400
>. 1 000
i
""""" Kon
TO
5
Lower Tongue
A
• •
A A
T T
O O
BASE ICLUS
NARCCAP
Figure 334. TN Load (MT/yr), Lower Tongue River (SWAT)
Qcn
300
250
•o
TO
O 150
Q_
100
en
n
Lower Tongue
*
• f
A A
O O
BASE ICLUS
NARCCAP
Figure 335. TP Load (MT/yr), Lower Tongue River (SWAT)
Y-201
-------
Table 13. Summary of range of change relative to existing conditions for NARCCAP dynamically
downscaled scenarios, Tongue and Powder Rivers watershed SWAT model
Results without LU change
Min
Median
Mean
Max
Results with LU change
Min
Median
Mean
Max
Lower Powder, Mizpah HUCs 10090209 and 10090210
Flow
TSS
TN
TP
-40.04%
-59.86%
-56.12%
-56.62%
39.41%
49.18%
50.82%
50.83%
54.43%
85.32%
84.84%
84.75%
206.01%
334.52%
330.85%
330.05%
-40.04%
-59.86%
-56.12%
-56.62%
39.41%
49.18%
50.82%
50.83%
54.43%
85.32%
84.84%
84.75%
206.01%
334.52%
330.85%
330.05%
Clear
Flow
TSS
TN
TP
-19.63%
-5.88%
11.60%
15.41%
4.99%
12.61%
25.40%
35.94%
4.82%
62.14%
27.42%
34.72%
29.73%
301.52%
48.89%
49.70%
-19.63%
-5.88%
11.60%
15.41%
4.99%
12.61%
25.40%
35.94%
4.82%
62.14%
27.42%
34.72%
29.73%
301.52%
48.89%
49.70%
Little Powder HUC 10090208
Flow
TSS
TN
TP
-42.49%
78.30%
10.22%
24.32%
35.37%
118.86
%
60.13%
53.31%
36.98%
151.59%
92.36%
76.22%
126.23%
256.12%
309.15%
217.07%
-42.49%
78.30%
10.22%
24.32%
35.37%
118.86
%
60.13%
53.31%
36.98%
151.59%
92.36%
76.22%
126.23%
256.12%
309.15%
217.07%
Middle Powder HUC 10090207
Flow
TSS
TN
TP
-30.51%
124.73%
41.30%
47.19%
12.35%
401.48
%
54.59%
56.45%
16.60%
476.52%
72.29%
67.49%
78.80%
1109.71
%
168.12%
122.45%
Crazy Woman HUC 10090205
-28.57%
-34.19%
-19.99%
-35.99%
12.33%
4.31%
-11.98%
-6.97%
15.34%
5.63%
1.35%
1.35%
72.52%
49.86%
63.51%
59.84%
-28.57%
-34.19%
-19.99%
-35.99%
-30.51%
124.73
%
41.30%
47.19%
12.33%
4.31%
-11.98%
-6.97%
12.35%
401.48
%
54.59%
56.45%
15.34%
5.63%
1.35%
1.35%
16.60%
476.52%
72.29%
67.49%
72.52%
49.86%
63.51%
59.84%
78.80%
1109.71
%
168.12%
122.45%
-28.57%
-34.19%
-19.99%
-35.99%
Upper Powder HUC 10090202
Flow
TSS
TN
TP
-32.36%
27.05%
31.22%
28.92%
7.45%
832.76
%
59.06%
65.83%
10.07%
1020.30
%
74.50%
68.23%
61.02%
2449.84
%
135.51%
133.36%
-32.36%
27.05%
31.22%
28.92%
7.45%
832.76
%
59.06%
65.83%
10.07%
1020.30
%
74.50%
68.23%
61.02%
2449.84
%
135.51%
133.36%
Middle Fork Powder HUC 10090201
Flow
TSS
TN
TP
-21.44%
-48.68%
-17.76%
-20.86%
4.03%
-32.66%
-4.76%
-5.68%
2.38%
-31.99%
3.38%
1.39%
23.91%
-9.05%
34.70%
38.72%
-21.44%
-48.68%
-17.76%
-20.86%
4.03%
-32.66%
-4.76%
-5.68%
2.38%
-31.99%
3.38%
1.39%
23.91%
-9.05%
34.70%
38.72%
Y-202
-------
S Fork Powder, Salt HUCs 10090203 and 10090204
Flow
TSS
TN
TP
-30.63%
-20.89%
3.98%
0.86%
6.62%
30.09%
39.37%
49.86%
7.66%
363.91%
32.25%
37.93%
50.62%
2018.84
%
62.00%
67.49%
-30.63%
-20.89%
3.98%
0.86%
6.62%
30.09%
39.37%
49.86%
7.66%
363.91%
32.25%
37.93%
50.62%
2018.84
%
62.00%
67.49%
Lower Tongue HUC10090102
Flow
TSS
TN
TP
-30.29%
-34.27%
-29.42%
-33.26%
15.13%
30.86%
28.19%
27.48%
27.64%
55.34%
50.61%
49.13%
140.14%
251.20%
219.77%
224.40%
-30.29%
-34.27%
-29.42%
-33.26%
15.13%
30.86%
28.19%
27.48%
27.64%
55.34%
50.61%
49.13%
140.14%
251.20%
219.77%
224.40%
Upper Tongue HUC 10090101
Flow
TSS
TN
TP
-21.39%
6.84%
-19.96%
-15.22%
0.45%
32.12%
-4.64%
-1.98%
1.96%
43.35%
3.63%
8.30%
38.77%
110.24%
51.81%
63.19%
Powder at Moorhead (gage 06324500)
Flow
TSS
TN
TP
Powder at Locate
Flow
TSS
TN
TP
-26.90%
121.64%
31.50%
33.81%
6.09%
375.40
%
46.23%
56.51%
7.84%
510.94%
51.80%
53.93%
47.48%
1317.64
%
90.53%
73.64%
-21.39%
6.84%
-19.96%
-15.22%
-26.90%
121.64
%
31.50%
33.81%
0.45%
32.12%
-4.64%
-1.98%
6.09%
375.40
%
46.23%
56.51%
1.96%
43.35%
3.63%
8.30%
7.84%
510.94%
51.80%
53.93%
38.77%
110.24%
51.81%
63.19%
47.48%
1317.64
%
90.53%
73.64%
gage 06326500)
-40.81%
-58.53%
-54.12%
-54.91%
36.36%
46.17%
47.52%
47.42%
50.16%
79.14%
78.69%
78.49%
192.44%
309.85%
304.85%
303.61%
-40.81%
-58.53%
-54.12%
-54.91%
36.36%
46.17%
47.52%
47.42%
50.16%
79.14%
78.69%
78.49%
192.44%
309.85%
304.85%
303.61%
Tongue at State Line (gage 06306300)
Flow
TSS
TN
TP
-19.50%
-78.56%
-51.92%
-47.25%
-4.37%
-73.08%
-33.13%
-28.61%
-3.34%
-69.76%
-26.82%
-20.26%
21.00%
-57.96%
18.82%
33.03%
-19.50%
-78.56%
-51.92%
-47.25%
-4.37%
-73.08%
-33.13%
-28.61%
-3.34%
-69.76%
-26.82%
-20.26%
21.00%
-57.96%
18.82%
33.03%
Y-203
-------
Lower Powder, Mizpah
120
o
1 2 3 4 5 6 7 8 9 10 11 12
40
20
Figure 336. Monthly average flows, Lower Powder River (SWAT)
Lower Powder, Mizpah
1000
LOWO
•LOW1
LOW2
• LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 337. Flow duration, Lower Powder River (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Y-204
-------
Clear
1 2 3 4 5 6 7 8 9 10 11 12
Figure 338. Monthly average flows, Clear Creek (SWAT)
Clear
1000
LOWO
•LOW1
LOW2
• LOWS
LOW4
LOWS
LOW6
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 339. Flow duration, Clear Creek (SWAT)
Y-205
-------
Little Powder
1 2 3 4 5 6 7 8 9 10 11 12
Figure 340. Monthly average flows, Little Powder River (SWAT)
Little Powder
1000
tn
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 341. Flow duration, Little Powder River (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Y-206
-------
Middle Powder
1 2 3 4 5 6 7 8 9 10 11 12
Figure 342. Monthly average flows, Middle Powder River (SWAT)
Middle Powder
1000
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 343. Flow duration, Middle Powder River (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Y-207
-------
Crazy Woman
6 7
Month
10 11 12
Figure 344. Monthly average flows, Crazy Woman Creek (SWAT)
Crazy Woman
100
CO
^
o
0)
O)
£
0)
ro 0.01
Q
0.001
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 345. Flow duration, Crazy Woman Creek (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Y-208
-------
Upper Powder
1 2 3 4 5 6 7 8 9 10 11 12
Figure 346. Monthly average flows, Upper Powder River (SWAT)
Upper Powder
1000
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.0001
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 347. Flow duration, Upper Powder River (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Y-209
-------
Middle Fork Powder
1 2 3 4 5 6 7 8 9 10 11 12
Figure 348. Monthly average flows, Middle Fork Powder River (SWAT)
Middle Fork Powder
1000
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 349. Flow duration, Middle Fork Powder River (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Y-210
-------
SF Powder, Salt
6 7
Month
10 11 12
Figure 350. Monthly average flows, South Fork Powder River (SWAT)
SF Powder, Salt
1000
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 351. Flow duration, South Fork Powder River (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Y-211
-------
Lower Tongue
6 7
Month
10 11 12
Figure 352. Monthly average flows, Lower Tongue River (SWAT)
Lower Tongue
1000
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 353. Flow duration, Lower Tongue River (SWAT)
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Y-212
-------
Upper Tongue
1 2 3 4 5 6 7 8 9 10 11 12
Figure 354. Monthly average flows, Upper Tongue River (SWAT)
Upper Tongue
1000
1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 355. Flow duration, Upper Tongue River (SWAT)
LOWO
•LOW1
LOW2
•LOWS
-LOW4
LOWS
LOW6
• LOWO
•LOW1
LOW2
• LOWS
LOW4
LOWS
LOW6
Y-213
-------
60%
50%
-20%
Month
Figure 356. Average of median percent change in flow; NARCCAP Scenarios W1-W6 at all
stations, Tongue and Powder Rivers watershed (SWAT)
Y-214
-------
Trinity River Watershed
AC,r\
350 -
E 300
u
LJ_
™ pnn
c
c
TO
OJ
5 100
n
Lower Trinity
JL
ITP
A
t t
I A
0 ^
BASE ICLUS
NARCCAP
Figure 357. Mean annual flow (cms), Lower Trinity River (SWAT)
Lower Trinity
"uT
LJ
1
nj Q nnn -
o_
WD
5.
TO
£
o
L.
>
8
rt o
t t
A A
O O
BASE ICLUS
NARCCAP
Figure 358. 100-yr Flow Peak (Log-Pearson III, cms), Lower Trinity River (SWAT)
Y-215
-------
Lower Trinity
14
"77" 19
E
u
3
O m
LL.
3
o
8
ra
c
c
Oi 4 -
M
2
_l_
+
„
ffi
A
BASE ICLUS
NARCCAP
Figure 359. Average Annual 7-day Low Flow (cms), Lower Trinity River (SWAT)
n 14 -
9
•D
8
01
- 01 -
£ U.I
V)
ra
LL.
i_ n nft
^
ra
CO
-o u-ub
i_
ra
^
LJ
•— n n4 -
n -
Lower Trinity
• n
w w
BASE ICLUS
NARCCAP
Figure 360. Richards-Baker Flashiness Index, Average Annual 7-day Low Flow (cms), Lower
Trinity River (SWAT)
Y-216
-------
250 -i
"t/T
« 200
&
(Water Year
8
bwCentroic
1
u_
0
i? 50
Q
0
Lower Trinity
5
r\
BASE
ICLUS
NARCCAP
Figure 361. Days to Flow Centroid (Water Year Basis), Lower Trinity River (SWAT)
Lower Trinity
1,200,000 -i
1 000 000 -
_ 800,000
TO
5
200,000
n
+ +
, *
A A
gl 0
0 0
BASE ICLUS
NARCCAP
Figure 362. TSS Load (MT/yr), Lower Trinity River (SWAT)
Y-217
-------
Lower Trinity
25,000 -i
^! 15,000
•o
ra
^ 10,000
1-
'
M
0
BASE ICLUS
NARCCAP
Figure 363. TN Load (MT/yr), Lower Trinity River (SWAT)
Lower Trinity
3,000 -i
_ 2,000
>
|
ra
0
1 000
n
^
* t
i
O
BASE ICLUS
NARCCAP
Figure 364. TP Load (MT/yr), Lower Trinity River (SWAT)
Y-218
-------
Table 14. Summary of range of change relative to existing conditions for NARCCAP dynamically
downscaled scenarios, Neuse-Tar watershed SWAT model
Results without LU change
Min
Median
Mean
Max
Results with LU change
Min
Median
Mean
Max
Upper WF Trinity HUC12030101
Flow
TSS
TN
TP
-60.57%
-70.32%
-45.92%
-46.01%
27.48%
-5.70%
77.86%
50.23%
38.78%
8.70%
128.00%
87.80%
125.65%
93.85%
350.37%
255.66%
-61.73%
-61.87%
-48.06%
-43.74%
9.22%
16.81%
54.80%
46.22%
17.28%
34.02%
79.90%
67.76%
84.08%
136.66%
235.62%
203.40%
Lower W. Fk Trinity River HUC 12030102
Flow
TSS
TN
TP
-48.48%
-69.49%
3.18%
0.13%
7.94%
-9.50%
35.04%
23.81%
12.11%
-0.92%
30.56%
20.82%
59.28%
61.32%
45.98%
33.52%
-45.86%
-58.95%
6.44%
1.34%
7.22%
15.37%
41.92%
27.54%
10.72%
25.41%
37.42%
24.49%
53.80%
100.51%
57.61%
38.39%
Elm Fork Trinity HUC 12030103
Flow
TSS
TN
TP
-52.26%
-55.37%
-26.19%
-29.16%
9.17%
10.34%
45.35%
47.68%
8.56%
9.74%
51.60%
55.43%
51.00%
57.00%
136.69%
151.86%
-48.83%
-52.17%
-18.99%
-24.85%
6.67%
7.82%
50.11%
49.68%
5.96%
7.25%
56.27%
56.59%
43.74%
50.47%
139.44%
149.99%
Denton HUC 12030104
Flow
TSS
TN
TP
-50.19%
-52.29%
-26.26%
-28.75%
15.40%
14.44%
68.20%
57.76%
16.70%
16.07%
57.38%
50.63%
64.10%
65.77%
102.85%
98.07%
-48.72%
-50.30%
-31.30%
-33.30%
8.72%
9.58%
70.11%
57.86%
9.88%
11.07%
56.69%
49.04%
50.85%
55.79%
118.52%
111.53%
Upper Trinity HUC 12030105
Flow
TSS
TN
TP
-43.89%
-65.26%
-13.18%
-13.87%
6.25%
-13.20%
12.31%
8.86%
8.45%
-9.53%
13.12%
9.82%
43.88%
34.16%
35.69%
29.72%
-41.40%
-53.88%
-7.31%
-10.00%
5.10%
10.22%
20.08%
14.56%
7.19%
15.09%
20.89%
15.78%
39.68%
69.42%
44.30%
36.11%
East Fork Trinity HUC 12030106
Flow
TSS
TN
TP
-41.81%
-60.79%
-6.13%
-6.82%
3.94%
-2.11%
10.22%
6.94%
6.03%
0.38%
11.36%
7.87%
40.43%
47.87%
27.67%
21.76%
-38.87%
-54.02%
-1.96%
-4.36%
4.89%
8.76%
16.18%
10.46%
6.63%
11.36%
17.14%
11.54%
38.77%
62.36%
34.34%
26.38%
Cedar HUC 12030107
Flow
TSS
TN
TP
-38.90%
-60.77%
-2.97%
-3.75%
13.92%
-17.57%
18.60%
13.31%
16.65%
-14.89%
17.67%
12.57%
51.89%
16.22%
32.43%
24.60%
-40.35%
-48.05%
-0.38%
-2.66%
7.41%
7.30%
23.77%
16.78%
9.92%
10.90%
22.29%
15.64%
41.82%
50.55%
38.59%
28.27%
Richland HUC 12030108
Flow
TSS
-42.84%
-59.43%
21.39%
-15.89%
26.82%
-9.71%
86.52%
37.07%
-46.58%
-46.56%
11.32%
10.09%
16.22%
18.00%
70.02%
78.52%
Y-219
-------
TN
TP
Results without LU change
Min
-32.71%
-30.63%
Median
30.99%
33.38%
Mean
31.84%
34.04%
Max
80.39%
81.94%
Results with LU change
Min
-26.36%
-24.58%
Median
46.01%
47.05%
Mean
45.48%
46.47%
Max
94.84%
94.70%
Chambers HUC 12030109
Flow
TSS
TN
TP
-41.69%
-59.45%
-32.92%
-30.47%
21.72%
-16.76%
30.34%
31.72%
27.12%
-10.10%
30.34%
32.13%
84.62%
37.50%
68.18%
70.61%
-44.71%
-44.92%
-20.65%
-19.05%
1 1 .62%
11.87%
56.62%
55.36%
16.42%
20.66%
55.72%
54.92%
67.47%
83.60%
100.53%
99.28%
Lower Trinity-Tehuacana HUC 12030201
Flow
TSS
TN
TP
-44.39%
-69.01%
-15.65%
-15.88%
7.60%
-20.17%
14.82%
10.31%
10.82%
-15.73%
17.33%
12.50%
51.69%
24.85%
42.46%
33.51%
-43.33%
-54.89%
-9.75%
-11.88%
5.00%
12.46%
22.50%
15.68%
8.03%
18.79%
24.90%
18.07%
45.99%
75.04%
50.98%
39.58%
Lower Trinity-Kickapoo HUC 12030202
Flow
TSS
TN
TP
-40.67%
-69.05%
-24.79%
-26.83%
10.74%
-20.58%
25.51%
24.45%
10.86%
-16.24%
27.06%
24.40%
47.61%
25.93%
64.32%
60.68%
-40.54%
-54.16%
-19.76%
-24.03%
8.09%
14.74%
30.19%
25.86%
8.11%
21.02%
32.84%
26.87%
42.79%
81.13%
71.60%
64.21%
Lower Trinity (at mouth) HUC 12030203
Flow
TSS
TN
TP
-37.90%
-73.07%
-19.77%
-17.32%
11.73%
-27.12%
30.25%
32.30%
10.53%
-21.23%
31.77%
32.42%
45.51%
24.42%
65.38%
63.07%
-37.98%
-54.90%
-12.28%
-10.76%
9.23%
19.88%
39.02%
39.98%
8.01%
29.60%
41.39%
40.71%
41.24%
103.95%
75.79%
70.90%
Clear Creek nr Sanger (gage 05317000)
Flow
TSS
TN
TP
-38.27%
-59.35%
-1.43%
-8.77%
14.86%
-17.40%
30.94%
24.70%
17.51%
-15.83%
48.37%
40.07%
61.53%
22.09%
118.20%
103.05%
-39.47%
-46.40%
-13.68%
-15.96%
5.66%
8.19%
21.07%
20.54%
7.64%
10.18%
33.83%
32.24%
44.03%
58.96%
119.75%
112.33%
Y-220
-------
Upper WF Trinity
10 11 12
LOWO
• LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 365. Monthly average flows, Upper WF Trinity River (SWAT)
Upper WF Trinity
o
CD
O)
CD
CD
1000
100
10
0.1
Q 0.01
0.001
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 366. Flow duration, Upper WF Trinity River (SWAT)
Y-221
-------
W. Fk Trinity River at Grand Prairie
120
34567
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 367. Monthly average flows, Lower West Fork Trinity River at Grand Prairie (SWAT)
W. Fk Trinity River at Grand Prairie
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 368. Flow duration, Lower West Fork Trinity River at Grand Prairie (SWAT)
Y-222
-------
Elm Fork Trinity
120
100
to
^
o
1 2 3 4 5 6 7 8 9 10 11 12
40
20
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 369. Monthly average flows, Elm Fork Trinity River (SWAT)
Elm Fork Trinity
10000
1000
o
CD
O)
CD
CD
-^ 0.01
CD
Q
0.001
0.0001
0.1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 370. Flow duration, Elm Fork Trinity River (SWAT)
Y-223
-------
Denton
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 371. Monthly average flows, Denton River (SWAT)
Denton
1000
o
CD
O)
CD
CD
Q 0.01
0.001
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 372. Flow duration, Denton River (SWAT)
Y-224
-------
Upper Trinity
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 373. Monthly average flows, Upper Trinity River (SWAT)
Upper Trinity
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 374. Flow duration, Upper Trinity River (SWAT)
Y-225
-------
East Fork Trinity
300
250
,200
o
150
1 2 3 4 5 6 7 8 9 10 11 12
100
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 375. Monthly average flows, East Fork Trinity River (SWAT)
East Fork Trinity
10000
1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 376. Flow duration, East Fork Trinity River (SWAT)
Y-226
-------
Cedar
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 377. Monthly average flows, Cedarr (SWAT)
Cedar
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 378. Flow duration, Cedar River (SWAT)
Y-227
-------
Richland
11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 379. Monthly average flows, Richland (SWAT)
Richland
10000
to
I
CD
O)
CD
CD
'CD
Q
0.001
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 380. Flow duration, Richland (SWAT)
Y-228
-------
Chambers
1 2 3 4 5 6 7 8 9 10 11 12
10
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 381. Monthly average flows, Chambers (SWAT)
Chambers
1000
CO
^
o
0)
O)
£
0)
Q 0.01
0.001
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 382. Flow duration, Chambers (SWAT)
Y-229
-------
Lower Trinity-Tehuacana
600
500
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 383. Monthly average flows, Lower Trinity River - Tehuacana (SWAT)
Lower Trinity-Tehuacana
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 384. Flow duration, Lower Trinity River - Tehuacana (SWAT)
Y-230
-------
50%
40%
30%
20%
ro
.c
u
OJ
•a
OJ
M- 0%
o
01
8?
§ -10%
-20%
-30%
Month
Figure 385. Average of median percent change in flow; NARCCAP Scenarios W1-W6 at all
stations, Trinity River watershed (SWAT)
Y-231
-------
Upper Colorado River Basin
Colorado R (nr State Line 09163500) 14010005
9nn
"i/T
^
_o
U.
•3 100
c
5
c
TO
OJ
5 50
n
^
0
g
BASE
{
ICLUS
NARCCAP
Figure 386. Mean annual flow (cms), Colorado River near State Line (SWAT)
1 finn -,
u
o
S2
3 °uu
Q.
M
•3 600 -
Ji
ra
01
_o
LL.
8
rt o
Colorado R (nr State Line 09163500) 14010005
§A
5
& &
A A
BASE ICLUS
NARCCAP
Figure 387. 100-yr Flow Peak (Log-Pearson III, cms), Colorado River near State Line (SWAT)
Y-232
-------
Upper Colorado River Basin
Colorado R (nr State Line 09163500) 14010005
onn -
"uT
1 15°
_o
LL.
1 100 -
1
c
ra
01
1 50
0 -
• •
$ 5
BASE ICLUS
NARCCAP
Figure 386. Mean annual flow (cms), Colorado River near State Line (SWAT)
1 finn -i
T* 1 9nn -
u
o
S2
3 80°
Q.
M
j Rnn -
ji
ra
01
_o
LL.
. onn -
8
rt o
Colorado R (nr State Line 09163500) 14010005
§tt
5
* *
A A
BASE ICLUS
NARCCAP
Figure 387. 100-yr Flow Peak (Log-Pearson III, cms), Colorado River near State Line (SWAT)
Y-232
-------
300 -i
-J-250
to
re
m
re
>• 200
b
FlowCentroid (Wa
1 8
0
s,
Q 50
0
Colorado R (nr State Line 09163500) 14010005
f f
BASE ICLUS
NARCCAP
Figure 390. Days to Flow Centroid (Water Year Basis), Colorado River near State Line (SWAT)
Colorado R (nr State Line 09163500) 14010005
3,000,000 -i
2,500,000
1
^ 1,500,000
1
1— 1 000 000
500,000
n
• ^
*
*
BASE
ICLUS
NARCCAP
Figure 391. TSS Load (MT/yr), Colorado River near State Line (SWAT)
Y-234
-------
Colorado R (nr State Line 09163500) 14010005
1.4 nnn
19 nnn
1 0 000
I— s 000
•o
TO
O g QQQ _
Z
1-
4 000
2 000
•
•
$
$
BASE
NAR(
ICLUS
;CAP
Figure 392. TN Load (MT/yr), Colorado River near State Line (SWAT)
finn
>
""*" AC\C\
TO
5
Q- 300
n
Colorado R (nr State Line 09163500) 14010005
• •
+ ±
| M
BASE ICLUS
NARCCAP
Figure 393. TP Load (MT/yr), Colorado River near State Line (SWAT)
Y-235
-------
Table 15. Summary of range of change relative to existing conditions for NARCCAP dynamically
downscaled scenarios, Upper Colorado River watershed SWAT model
Results without LU change
Min
Median
Mean
Max
Results with LU change
Min
Median
Mean
Max
Colorado Headwaters HUC 14010001
Flow
TSS
TN
TP
-11.21%
-17.71%
-33.36%
-17.27%
-5.12%
-11.15%
-31.12%
-11.84%
-3.76%
-8.49%
-30.12%
-10.16%
9.05%
8.20%
-24.81%
1.22%
-11.33%
-17.84%
-33.56%
-17.73%
-5.23%
-1 1 .27%
-31.36%
-12.40%
-3.87%
-8.60%
-30.34%
-10.64%
8.95%
8.08%
-24.89%
0.91%
Blue HUC 14010002
Flow
TSS
TN
TP
-13.20%
-20.76%
-38.27%
-58.02%
-9.03%
-14.55%
-29.63%
-46.02%
-6.90%
-1 1 .22%
-29.88%
-44.24%
5.58%
8.17%
-22.73%
-23.75%
-13.41%
-21.03%
-39.68%
-56.95%
-9.25%
-14.84%
-30.50%
-45.27%
-7.10%
-1 1 .49%
-30.91%
-43.54%
5.42%
7.94%
-23.50%
-23.53%
Eagle HUC 14010003
Flow
TSS
TN
TP
-13.35%
-19.38%
-40.80%
-1.23%
Roaring Fork (at Glenwood Sps, c
Flow
TSS
TN
TP
-15.92%
-21.86%
-48.72%
-26.80%
-5.88%
-12.82%
-39.56%
4.70%
-4.09%
-9.29%
-38.81%
4.27%
10.95%
10.22%
-35.18%
10.10%
-13.64%
-19.69%
-40.80%
-2.39%
-6.16%
-13.02%
-39.52%
3.69%
-4.36%
-9.53%
-38.72%
3.31%
10.66%
9.91%
-34.87%
9.62%
age 09085000) HUC 14010004
-8.48%
-13.97%
-43.87%
-18.36%
-6.20%
-10.93%
-43.00%
-17.13%
9.98%
11.56%
-35.69%
-0.78%
-15.99%
-21.95%
-48.71%
-26.65%
-8.53%
-14.01%
-43.92%
-18.13%
-6.28%
-11.01%
-43.05%
-16.99%
9.88%
11.43%
-35.79%
-0.77%
Colorado R (nr State Line gage 09163500) HUC 14010005
Flow
TSS
TN
TP
-14.31%
-20.13%
-27.32%
-21.03%
-8.68%
-13.07%
-20.18%
-16.41%
-5.19%
-8.36%
-16.46%
-11.00%
16.29%
23.96%
10.47%
19.13%
-14.36%
-20.21%
-27.39%
-21.07%
-8.74%
-13.13%
-20.25%
-16.45%
-5.25%
-8.42%
-16.54%
-11.06%
16.23%
23.90%
10.37%
18.95%
East-Taylor HUC 14020001
Flow
TSS
TN
TP
-12.05%
-96.05%
-61.79%
-21.02%
-6.00%
-95.62%
-57.01%
-14.28%
-4.71%
-95.27%
-57.31%
-13.31%
11.13%
-93.33%
-53.71%
1.00%
-12.05%
-96.05%
-61.79%
-21.02%
-6.00%
-95.62%
-57.01%
-14.28%
-4.71%
-95.27%
-57.31%
-13.31%
11.13%
-93.33%
-53.71%
1.00%
Upper Gunnison HUC 14020002
Flow
TSS
TN
TP
-19.75%
-27.09%
-37.82%
-28.38%
-11.28%
-16.35%
-29.31%
-18.48%
-7.83%
-11.04%
-25.21%
-14.75%
20.26%
31.07%
6.29%
17.46%
-19.75%
-27.09%
-37.82%
-28.38%
-1 1 .28%
-16.35%
-29.31%
-18.48%
-7.83%
-11.04%
-25.21%
-14.75%
20.26%
31.07%
6.29%
17.46%
Tomichi Cr (at Gunnison gage 09119000) HUC 14020003
Flow
-20.21%
-10.45%
-7.52%
11.50%|| -20.21%
-10.45%
-7.52%
11.50%
Y-236
-------
TSS
TN
TP
Results without LU change
Min
-41.56%
-86.43%
-37.34%
Median
-26.71%
-76.90%
-20.21%
Mean
-24.85%
-78.26%
-18.35%
Max
-5.57%
-71.70%
6.59%
Results with LU change
Min
-41.56%
-86.43%
-37.33%
Median
-26.70%
-76.90%
-20.21%
Mean
-24.84%
-78.26%
-18.35%
Max
-5.57%
-71.70%
6.59%
N Fork Gunnison HUC 14020004
Flow
TSS
TN
TP
-1 1 .63%
-19.14%
-28.19%
-33.29%
-10.02%
-17.67%
-25.74%
-30.86%
-3.15%
-8.98%
-18.51%
-23.76%
22.93%
24.86%
10.54%
5.35%
-11.63%
-19.14%
-28.19%
-33.28%
-10.02%
-17.67%
-25.74%
-30.86%
-3.15%
-8.98%
-18.51%
-23.75%
22.93%
24.85%
10.54%
5.35%
Lower Gunnison HUC 1402000
Flow
TSS
TN
TP
-18.22%
-25.82%
-31.05%
-25.82%
-12.13%
-18.37%
-23.11%
-18.81%
-7.15%
-11.08%
-18.04%
-14.10%
22.10%
34.02%
16.64%
17.91%
-18.21%
-25.82%
-31.05%
-25.82%
-12.13%
-18.37%
-23.11%
-18.81%
-7.15%
-11.08%
-18.04%
-14.10%
22.10%
34.02%
16.64%
17.91%
Uncompahgre HUC 14020006
Flow
TSS
TN
TP
-14.92%
-21.21%
-22.65%
-12.34%
-10.82%
-16.35%
-18.37%
-7.34%
-5.51%
-9.45%
-11.12%
-2.57%
20.62%
27.31%
25.59%
27.48%
-14.92%
-21.21%
-22.65%
-12.35%
-10.82%
-16.35%
-18.37%
-7.36%
-5.51%
-9.45%
-11.11%
-2.58%
20.62%
27.32%
25.59%
27.45%
Gunnison R nr Gunnison (gage 09114500)
Flow
TSS
TN
TP
-13.60%
-84.04%
-63.22%
-20.21%
-6.85%
-82.63%
-58.80%
-12.47%
-5.47%
-81.80%
-59.06%
-11.32%
1 1 .45%
-76.18%
-55.93%
4.71%
-13.60%
-84.03%
-63.22%
-20.21%
-6.85%
-82.63%
-58.80%
-12.47%
-5.47%
-81.80%
-59.06%
-11.32%
11.45%
-76.18%
-55.93%
4.71%
Colorado R nr Cameo (gage 09095500)
Flow
TSS
TN
TP
-11.74%
-15.94%
-30.28%
-12.74%
-5.96%
-9.86%
-25.67%
-9.90%
-3.89%
-6.37%
-23.62%
-5.57%
11.57%
14.89%
-10.87%
18.60%
-11.84%
-16.05%
-30.44%
-13.00%
-6.05%
-9.96%
-25.87%
-10.14%
-3.98%
-6.47%
-23.81%
-5.89%
11.47%
14.77%
-11.06%
18.10%
Y-237
-------
Colorado Headwaters 14010001
10 11 12
LOWO
• LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 394. Monthly average flows, Colorado River Headwaters (SWAT)
Colorado Headwaters 14010001
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 395. Flow duration, Colorado River Headwaters (SWAT)
Y-238
-------
Blue 14010002
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 396. Monthly average flows, Blue River (SWAT)
Blue 14010002
100
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 397. Flow duration, Blue River (SWAT)
Y-239
-------
Eagle 14010003
1 2 3 4 5 6 7 8 9 10 11 12
10
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 398. Monthly average flows, Eagle River (SWAT)
Eagle 14010003
1000
0.1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 399. Flow duration, Eagle River (SWAT)
Y-240
-------
Roaring Fork (at Glenwood Sps 09085000) 14010004
o
1 2 3 4 5 6 7 8 9 10 11 12
40
20
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 400. Monthly average flows, Roaring Fork at Glenwood Springs (SWAT)
Roaring Fork (at Glenwood Sps 09085000) 14010004
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 401. Flow duration, Roaring Fork at Glenwood Springs (SWAT)
Y-241
-------
Colorado R (nr State Line 09163500) 14010005
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 402. Monthly average flows, Colorado River near State Line (SWAT)
Colorado R (nr State Line 09163500) 14010005
10000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
1
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 403. Flow duration, Colorado River near State Line (SWAT)
Y-242
-------
East-Taylor 14020001
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 404. Monthly average flows, East - Taylor (SWAT)
East-Taylor 14020001
1000
0.1
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 405. Flow duration, East - Taylor (SWAT)
Y-243
-------
Upper Gunnison 14020002
140
1 2 3 4 5 6 7 8 9 10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 406. Monthly average flows, Upper Gunnison River (SWAT)
Upper Gunnison 14020002
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 407. Flow duration, Upper Gunnison River (SWAT)
Y-244
-------
Tomichi Cr (at Gunnison 09119000) 14020003
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 408. Monthly average flows, Tomichi Creek at Gunnison (SWAT)
Tomichi Cr (at Gunnison 09119000) 14020003
100
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0.01
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 409. Flow duration, Tomichi Creek at Gunnison (SWAT)
Y-245
-------
N Fork Gunnison
6 7
Month
10 11 12
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 410. Monthly average flows, North Fork Gunnison River (SWAT)
N Fork Gunnison
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 411. Flow duration, North Fork Gunnison River (SWAT)
Y-246
-------
Lower Gunnison
LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
Figure 412. Monthly average flows, Lower Gunnison River (SWAT)
Lower Gunnison
1000
• LOWO
•LOW1
LOW2
•LOWS
LOW4
LOWS
LOW6
0% 20% 40% 60% 80% 100%
Percent of Time that Flow is Equaled or Exceeded
Figure 413. Flow duration, Lower Gunnison River (SWAT)
Y-247
-------
60%
50%
-40%
Month
Figure 414. Average of median percent change in flow; NARCCAP Scenarios W1-W6 at all
stations, Upper Colorado River watershed (SWAT)
Y-248
-------
APPENDIX Z.
OVERVIEW OF CLIMATE SCENARIO
MONTHLY TEMPERATURE,
PRECIPITATION, AND POTENTIAL
EVAPOTRANSPIRATION
Table Z-1. Climate change scenarios evaluated
NARCCAP dynamically downscaled scenarios
1 CGCM3 / CRCM
2 HadCM3 / HRM3
3 GFDL / RCM3
4 GFDL / GFDL high res
5 CGCM3 / RCM3
6 CCSM / WRFP
Driving GCMs of the NARCCAP scenarios (without downscaling)
7 CGCM3
8 HadCMS
9 GFDL
10 CCSM
BCSD statistically downscaled scenarios
11 CGCM3
12 HadCM3
13 GFDL
14 CCSM
Z-1
-------
ACF - Temperature (degF)
95
=:
85
SC
75
7C
55
60
55
5C
45
40
35
3C
25
2C
15
10
I Max mum
Min mum
• 75%
25%
• Med an
ACF
456789
Month
10 11 12
Base
•W1
•W2
•W3
•W4
•W5
•W6
•W7
•W8
•W9
W10
•W11
W12
W13
W14
Figure Z-1. ACF: Comparison of climate scenario temperature for the ACF basin
Note: See Table Z-1 for key to climate scenarios.
Z-2
-------
Central Arizona - Temperature (degF)
-
T
•
T
I
I
•
ZILI
Maximum
75%
25%
Median
Central AZ
D)
^
2
LU
12
Figure Z-2. Ariz: Comparison of climate scenario temperature for the Salt, Verde, and San Pedro
basins
Note: See Table Z-1 for key to climate scenarios.
Z-3
-------
Central Nebraska - Temperature (F)
I
Maximun
Minimurr
75%
25%
Median
Central Nebraska
20 -
10 -
n
6 7
Month
Base
-W1
-W2
-W3
-W4
•W5
-W6
10
11
12
Figure Z-3. CenNeb: Comparison of climate scenario temperature for the Loup/Elkhorn River
basins
Note: See Table Z-1 for key to climate scenarios.
Z-4
-------
Cook Inlet - Temperature (F)
60
55
50
45
40
35
3C
25
20
15
10
E
0
-5
I Maximum
Minimum
675%
• Median
Cook Inlet
Base
-W2
•W4
Figure Z-4. Cook: Comparison of climate scenario temperature for the Cook Inlet basin
Note: See Table Z-1 for key to climate scenarios.
Z-5
-------
Lake Erie - Temperature (F)
T
•
M axi mum
75%
25%
Median
Lake Erie
Base
-W1
-W2
•W3
-W4
W5
•W6
9 10 11 12
Figure Z-5. Erie: Comparison of climate scenario temperature for the Lake Erie Drainages
Note: See Table Z-1 for key to climate scenarios.
Z-6
-------
GA FL Coast - Temperature (F)
I
Maximun
Minimurr
75%
25%
Median
GA FL Coast
100
30
20 -
10 -
0
1
Base
-W1
-W2
-W3
-W4
-W5
W6
5 6 7 8 9 10 11 12
Month
Figure Z-6. GaFIa: Comparison of climate scenario temperature for the Georgia-Florida Coastal
Plain
Note: See Table Z-1 for key to climate scenarios.
Z-7
-------
Illinois - Temperature (F)
T
•
M axi mum
75%
25%
Median
Illinois
Base
•W1
-W2
-W3
-W4
W5
-W6
9 10 11 12
Figure Z- 7. Illin: Comparison of climate scenario temperature for the Illinois River basin
Note: See Table Z-1 for key to climate scenarios.
ZQ
-o
-------
Minnesota - Temperature (degF)
I
Maximun
M i n i m u rr
Upper MS
12
Figure Z-8. Minn: Comparison of climate scenario temperature for the Minnesota River (Upper
Mississippi) basin
Note: See Table Z-1 for key to climate scenarios.
Z-9
-------
New England Coastal - Temperature (F)
I
Maximun
Minimurr
75%
25%
Median
New England Coastal
20 -
10 -
n
*r
6 7
Month
10 11
Base
•W1
•W2
-W3
•W4
W5
-W6
12
Figure Z- 9. NewEng: Comparison of climate scenario temperature for the New England Coastal
basins
Note: See Table Z-1 for key to climate scenarios.
Z-10
-------
Lake Pontchartrain - Temperature (F)
I
Maximun
Minimurr
75%
25%
Median
Lake Pontchartrain
30
20 -
10 -
0
1
Base
•W1
-W2
-W3
-W4
W5
•W6
5 6 7 8 9 10 11 12
Month
Figure Z-10. Pont: Comparison of climate scenario temperature for the Lake Pontchartrain
drainage
Note: See Table Z-1 for key to climate scenarios.
Z-ll
-------
Rio Grande - Temperature (F)
~]
-
.
T
•
M axi mum
75%
25%
Median
Rio Grande
Base
•W1
-W2
•W3
•W4
W5
-W6
Figure Z-11. RioGra: Comparison of climate scenario temperature for the Rio Grande Valley
Note: See Table Z-1 for key to climate scenarios.
Z-12
-------
Sacramento - Temperature (F)
T
•
M axi mum
75%
25%
Median
Sacramento
40 -
30 -
20 -
10
n
Base
•W1
•W2
-W3
«W4
W5
-W6
567
Month
10 11
12
Figure Z-12. Sac: Comparison of climate scenario temperature for the Sacramento River basin
Note: See Table Z-1 for key to climate scenarios.
Z-13
-------
Coastal Southern CA - Temperature (F)
96
= C
25
80
75
70
65
S3
EE
EC
49
JC
3E
30
25
20
IE
1C
I Maximum
Minimum
• 75%
25%
• Median
Coastal Southern CA
I50 -
m 40 -
h-
30
20
10
n
Base
Q 7
Month
10
11
12
-W2
-W3
-W4
-W5
W6
Figure Z-13. SoCal: Comparison of climate scenario temperature for the Coastal Southern
California basins
Note: See Table Z-1 for key to climate scenarios.
Z-14
-------
South Platte - Temperature (F)
T
•
•
M axi mum
75%
25%
Median
South Platte
Base
-W1
-W2
•W3
•W4
Wo
•W6
10 11 12
Figure Z-14. SoPlat: Comparison of climate scenario temperature for the South Platte River basin
Note: See Table Z-1 for key to climate scenarios.
Z-15
-------
Susquehanna - Temperature (degF)
I
Maximun
Minimurr
D)
2,
2
LU
Susquehanna
12
Figure Z-15. Susq: Comparison of climate scenario temperature for the Susquehanna River basin
Note: See Table Z-1 for key to climate scenarios.
Z-16
-------
Albemarle-Pamlico - Temperature (F)
T
•
•
M axi mum
75%
25%
Median
Albemarle-Pamlico
Base
•W1
•W2
•W3
•W4
•W5
W6
1 2 3 4 5 6 7 8 9 10 11 12
Month
Figure Z-16. TarNeu: Comparison of climate scenario temperature for the Tar and Neuse River
basins
Note: See Table Z-1 for key to climate scenarios.
40 -
30 -
20 -
10
n
»
Z-17
-------
Trinity - Temperature (F)
T
•
M axi mum
75%
25%
Median
30
20 -
10 -
0
1
Trinity
Base
•W1
•W2
•W3
«W4
W5
-W6
567
Month
10 11
12
Figure Z-17. Trin: Comparison of climate scenario temperature for the Trinity River basin
Note: See Table Z-1 for key to climate scenarios.
Z-18
-------
Upper Colorado - Temperature (F)
I
Maximun
Minimurr
75%
25%
Median
Upper Colorado
Base
•W1
-W2
-W3
•W4
W5
-W6
Figure Z-18. UppCol: Comparison of climate scenario temperature for the Upper Colorado River
basin
Note: See Table Z-1 for key to climate scenarios.
Z-19
-------
Willamette - Temperature (degF)
130
95
so
85
s:
75
7C
65
60
55
EC
45
4C
35
30
25
20
15
10
5
T
I
Maximun
M i n i m u rr
LU
Willamette
Figure Z-19. Willa: Comparison of climate scenario temperature for the Willamette River basin
Note: See Table Z-1 for key to climate scenarios.
Z-20
-------
Powder/Tongue - Temperature (F)
I
M axi m u n
Minimurr
75%
25%
Median
Powder/Tongue
Base
-W1
-W2
•W3
•W4
•W6
10 11 12
Figure Z-20. Yellow: Comparison of climate scenario temperature for the Powder/Tongue River
basins
Note: See Table Z-1 for key to climate scenarios.
Z-21
-------
ACF - Precipitation (in)
105
1SC
=E
=c
85
a;;
75
72
65
sc
55
EC
45
42
35
I
M axi mum
Minimum
75%
25%
Median
W1 W2 W3 W4 W5 W6 W7 VIS V\S W10 W11 W12 W13 W14
ACF
1 2 3 4 5 6 7 8 9 10 11 12
Figure Z-21. ACF: Comparison of climate scenario precipitation for the ACF basin
Note: See Table Z-1 for key to climate scenarios.
Z-22
-------
Central Arizona - Precipitation (in)
I
M axi mum
Minimum
75%
25%
Median
¥
•
W2 W3 W4 W5 W6 W7 WS W9 W10
W12 W13 W14
Central AZ
Figure Z-22. Ariz: Comparison of climate scenario precipitation for the Salt, Verde, and San Pedro
basins
Note: See Table Z-1 for key to climate scenarios.
Z-23
-------
Central Nebraska - Precipitation (in)
II
-.
T
4
T
iz
T
•
T
1
ii
__
I Maximum
Minimum
• 75%
25%
T
^
T
±
Central Nebraska
Base
-W1
-W2
-W3
-W4
>W5
-W6
Figure Z-23. CenNeb: Comparison of climate scenario precipitation for the Loup/Elkhorn River
basin
Note: See Table Z-1 for key to climate scenarios.
Z-24
-------
Cook Inlet - Precipitation (in)
1=5
ICC
as
sc
85
75
7C
65
EC
EE
EC
JE
35
3C
2E
15
10
5
T
T
Maximum
I Minimum
25%
• Median
Cook Inlet
07
r
Base
•W2
•W4
11 12
Figure Z-24. Cook: Comparison of climate scenario precipitation for the Cook Inlet basin
Note: See Table Z-1 for key to climate scenarios.
Z-25
-------
Erie St. Clair - Precipitation (in)
2:
T
1
JL
T
I
1
• i
-*-
li
T
1
T
• 1
,
II
Minimum
25%
• Median
Lake Erie
Base
-W1
-W2
•W3
-W4
W5
-W6
Figure Z-25. Erie: Comparison of climate scenario precipitation for the Lake Erie drainages
Note: See Table Z-1 for key to climate scenarios.
Z-26
-------
GA FL Coast - Precipitation (in)
1
—
1
1
f-
.
T
1
T
1
I Maximum
Minimum
• 75%
25%
T
1
!
GA FL Coast
Base
-W1
-W2
-W3
-W4
W5
-W6
Figure Z-26. GaFIa: Comparison of climate scenario precipitation for the Georgia-Florida Coastal
Plain
Note: See Table Z-1 for key to climate scenarios.
Z-27
-------
Illinois - Precipitation (in)
1
2:
ff
I
i
_L
~r
1
I
-r
f
T
t
±
-r
J
i
I
Minirnum
25%
• Median
Illinois
Base
•W1
-W2
•W3
-W4
W5
-W6
10 11 12
Figure Z-27. Illin: Comparison of climate scenario precipitation for the Illinois River basin
Note: See Table Z-1 for key to climate scenarios.
Z-28
-------
Minnesota - Precipitation (in)
I
M axi mum
Minimum
75%
25%
Median
i
i
I
I
I
i
I
T
i I
W2 W3 W4 W5 W6 W7
W10 W11 W12 W13 W14
Upper MS
67
Month
10 11 12
Figure Z-28. Minn: Comparison of climate scenario precipitation for the Minnesota River (Upper
Mississippi) basin
Note: See Table Z-1 for key to climate scenarios.
Z-29
-------
New England Coastal - Precipitation (in)
110
105
too
95
SO
85
sc
75
70
65
60
55
EC
45
43
35
33
T
i
T
_i_
,,
I
^L
i
•
T
j_
m
I
.
T
1
•
_j__
T
I
f
_L
I Maximum
Minimum
• 75%
25%
New England Coastal
2 3 4 5 6 7 8 9 10 11 12
Base
•W1
-W2
-W3
•W4
W5
-W6
Figure Z-29. NewEng: Comparison of climate scenario precipitation for the New England Coastal
basins
Note: See Table Z-1 for key to climate scenarios.
Z-30
-------
Lake Pontchartrain - Precipitation (in)
2:
• 1
T
I
•
~T
T
1
J
ii
i
T
1
T
Minirnum
25%
• Median
Lake Pontchartrain
o
fc!
Base
•W1
-W2
•W3
•W4
W5
-W6
9 10 11 12
Figure Z-30. Pont: Comparison of climate scenario precipitation for the Lake Pontchartrain
drainage
Note: See Table Z-1 for key to climate scenarios.
Z-31
-------
Rio Grande - Precipitation (in)
ICC
95
SO
= 5
2C
75
re
55
EC
55
E3
45
JO
35
30
25
20
15
10
1
I
T
-j-
1
•
T
T
I
I
_L
T
T
•
I
T
T
•
T
1 Minimum
• 75%
25%
1
1
I
I
Rio Grande
•n
i i
in
IU
9,
0 .
7
2_
^
jyt
S3
fe,.
*C-H
.-.
Base
•W1
•W2
-W3
•W4
W5
-W6
Figure Z-31. RioGra: Comparison of climate scenario precipitation for the Rio Grande Valley
Note: See Table Z-1 for key to climate scenarios.
Z-32
-------
Sacramento - Precipitation (in)
,
T
_l_
Maximum
Minimum
75%
25%
o
ft!
0.
Sacramento
Base
•W2
-W3
•W4
W5
-W6
Figure Z-32. Sac: Comparison of climate scenario precipitation for the Sacramento River basin
Note: See Table Z-1 for key to climate scenarios.
Z-33
-------
Coastal Southern CA - Precipitation (in)
105
ICC
95
SO
85
80
75
70
65
I
Maximum
Minimum
75%
25%
Median
i
Coastal Southern CA
Base
•W1
•W2
-W3
-W4
W5
-W6
10 11 12
Figure Z-33. SoCal: Comparison of climate scenario precipitation for the Coastal Southern
California basin
Note: See Table Z-1 for key to climate scenarios.
Z-34
-------
South Platte - Precipitation (in)
~r
I
i
~T
rr
•
T
IT
1
f
T
1
I
II
T
j
zT_
I Maximum
Minimum
• 75%
25%
T
I
I
I
South Platte
Base
•W2
-W3
-W4
-W5
W6
Figure Z-34. SoPlat: Comparison of climate scenario precipitation for the South Platte River basin
Note: See Table Z-1 for key to climate scenarios.
Z-35
-------
Susquehanna - Precipitation (in)
i
I
M axi mum
Minimum
75%
25%
Median
m
o
LU
o:
D.
W2 W3 W4 W5 W6 W7 WS W9 W10 W11 W12 W13 W14
Susquehanna
Figure Z-35. Susq: Comparison of climate scenario precipitation for the Susquehanna River basin
Note: See Table Z-1 for key to climate scenarios.
Z-36
-------
Albemarle-Pamlico - Precipitation (in)
H
T
T
1
_L
T
i
T
1
1
I
T
1
I
T
T
1
1
_L
-J-
II
— j-
I
I
Minimum
25%
• Median
Albemarle-Pamlico
o
Base
•W1
•W2
•W3
•W4
•W5
-W6
Figure Z-36. TarNeu: Comparison of climate scenario precipitation for the Tar and Neuse River
basin
Note: See Table Z-1 for key to climate scenarios.
Z-37
-------
Trinity - Precipitation (in)
110
105
too
95
SO
85
s:
75
70
65
60
55
EC
45
4:
35
33
I
Maximum
Minimum
75%
25%
Median
Trinity
itj'
tt!
Base
•W1
-W2
-W3
•W4
W5
-W6
10 11 12
Figure Z-37. Trin: Comparison of climate scenario precipitation for the Trinity River basin
Note: See Table Z-1 for key to climate scenarios.
Z-38
-------
Upper Colorado - Precipitation (in)
105
ICC
95
SO
85
80
75
70
65
I
Maximum
Minimum
75%
25%
Median
i
i
i
Upper Colorado
Base
•W2
-W3
W4
W5
-W6
Figure Z-38. UppCol: Comparison of climate scenario precipitation for the Loup/Elkhorn River
basin
Note: See Table Z-1 for key to climate scenarios.
Z-39
-------
Willamette - Precipitation (in)
110
105
i:c
SE
sc
85
a;;
75
I
Maximum
Minimum
75%
25%
Med an
o
LU
OH
a.
0
Willamette
Base
8
10
11
12
234567
Month
Figure Z-39. Willa: Comparison of climate scenario precipitation for the Willamette River basin
Note: See Table Z-1 for key to climate scenarios.
Z-40
-------
Powder/Tongue - Precipitation (in)
—
m*
• •
T
T
_T~
11
T
T
m
J_
I Maximum
Minimum
• 75%
25%
T
4
T
Powder/Tongue
Base
•W1
•W2
-W3
•W4
W5
-W6
Figure Z-40. Yellow: Comparison of climate scenario precipitation for the Powder/Tongue River
basin
Note: See Table Z-1 for key to climate scenarios.
Z-41
-------
ACF - PMET (in)
Maximum
Minimum
75%
25%
Median
Base W1 W2 W3 W4 W5 W6 W7 W8 W9 W10 W11 W12 W13 W14
ACF
LJJ
Q_
5678
Month
10
11
12
Figure Z-41. ACF: Comparison of climate scenario Penman-Monteith reference ET for the ACF
basin
Note: See Table Z-1 for key to climate scenarios.
Z-42
-------
CeAZ- PMET(in)
Maximum
Minimum
75%
25%
Median
Base W1 W2 W3 W4 W5 W6 W7 W8 W9 W10 W11 W12 W13 W14
Central AZ
LU
2
D.
12
Figure Z-42. Ariz: Comparison of climate scenario Penman-Monteith reference ET for the Salt,
Verde, and San Pedro basins
Note: See Table Z-1 for key to climate scenarios.
Z-43
-------
Central Nebraska - PMET (in)
Maximum
Minimum
75%
25%
Median
WE
we
Central Nebraska
Base
-W1
-W2
-W3
-W4
-W5
>W6
Month
Figure Z-43. CenNeb: Comparison of climate scenario Penman-Monteith reference ET for the
Loup/Elkhorn River basin
Note: See Table Z-1 for key to climate scenarios.
Z-44
-------
Cook Inlet - PMET (in)
Jfl
^
"
1
T
2
T
ac
t
_L
I Maximum
Minimum
• 75%
25%
• Median
i
5
-^
W4
Cook Inlet
Base
•W2
•W4
-W6
Figure Z-44. Cook: Comparison of climate scenario Penman-Monteith reference ET for the Cook
Inlet basin
Note: See Table Z-1 for key to climate scenarios.
Z-45
-------
Erie St. Clair - PMET (in)
•ice
JC
3C
70
SC
ID
40
3D
2C
10
D
I
Maximum
Minimum
75%
25%
Median
W3
Lake Erie
1MB
Base
-VV1
-W2
-W3
-W4
W5
*W6
Figure Z-45. Erie: Comparison of climate scenario Penman-Monteith reference ET for the Lake Erie
drainages
Note: See Table Z-1 for key to climate scenarios.
Z-46
-------
GA FL Coast - PMET (in)
Maximum
Minimum
75%
25%
Median
1MB
GA FL Coast
1 2 3 4 5 6 7 8 9 10 11 12
Base
-W1
-W2
-W3
-W4
-W5
W6
Figure Z-46. GaFIa: Comparison of climate scenario Penman-Monteith reference ET for the
Georgia-Florida Coastal Plain
Note: See Table Z-1 for key to climate scenarios.
Z-47
-------
Illinois -PMET (in)
Maximum
Minimum
75%
25%
Median
1MB
Illinois
LJJ
2 3 4 5 6 7 8 9 10 11 12
Base
-W1
-W2
-W3
•W4
W5
-W6
Figure Z-47. Illin: Comparison of climate scenario Penman-Monteith reference ET for the Illinois
River basin
Note: See Table Z-1 for key to climate scenarios.
Z-48
-------
Upper MS - PMET (in)
Maximum
Minimum
75%
25%
Median
Base W1 W2 W3 W4 W5 W6 W7 W8 W9 W10 W11 W12 W13 W14
Upper MS
LU
^
Q.
12
Figure Z-48. Minn: Comparison of climate scenario Penman-Monteith reference ET for the
Minnesota River (Upper Mississippi) basin
Note: See Table Z-1 for key to climate scenarios.
Z-49
-------
New England Coastal - PMET (in)
Maximum
Minimum
75%
25%
Median
we
New England Coastal
LLJ
23456789 10 12
Base
•W1
•W2
-W3
•W4
W5
-W6
Figure Z-49. NewEng: Comparison of climate scenario Penman-Monteith reference ET for the New
England Coastal basins
Note: See Table Z-1 for key to climate scenarios.
Z-50
-------
Lake Pontchartrain - PMET (in)
1SC
&:
3C
70
6C'
ID
40
3D
2C
10
D
I
Maximum
Minimum
75%
25%
Median
W3
Lake Pontchartrain
1MB
1 2 3 4 5 6 7 8 9 10 11 12
Base
•W1
-W2
•W3
-W4
W5
-W6
Figure Z-50. Pont: Comparison of climate scenario Penman-Monteith reference ET for the Lake
Pontchartrain drainage
Note: See Table Z-1 for key to climate scenarios.
Z-51
-------
Rio Grande - PMET (in)
Maximum
Minimum
75%
25%
Median
1MB
Rio Grande
Base
-W1
-W2
-W3
-W4
W5
-W6
Figure Z-51. RioGra: Comparison of climate scenario Penman-Monteith reference ET for the Rio
Grande Valley
Note: See Table Z-1 for key to climate scenarios.
Z-52
-------
Sacramento - PMET (in)
Maximum
Minimum
75%
25%
Median
1MB
Sacramento
LJJ
5 6 7 8 9 10 11 12
Base
•W1
-W2
•W3
-W4
W5
-W6
Figure Z-52. Sac: Comparison of climate scenario Penman-Monteith reference ET for the
Sacramento River basin
Note: See Table Z-1 for key to climate scenarios.
Z-53
-------
Coastal Southern CA - PMET (in)
Maximum
Minimum
75%
25%
Median
Coastal Southern CA
Base
•W1
•W2
-W3
-W4
W5
-W6
Figure Z-53. SoCal: Comparison of climate scenario Penman-Monteith reference ET for the
Coastal Southern California basins
Note: See Table Z-1 for key to climate scenarios.
Z-54
-------
South Platte - PMET (in)
Maximum
Minimum
75%
25%
Median
South Platte
Base
•W2
-W3
•W4
W5
-W6
Figure Z-54. SoPlat: Comparison of climate scenario Penman-Monteith reference ET for the South
Platte River basin
Note: See Table Z-1 for key to climate scenarios.
Z-55
-------
Susquehanna - PMET (in)
Maximum
Minimum
75%
25%
Median
Base W1 W2 W3 W4 W5 W6 W7 W8 W9 W10 W11 W12 W13 W14
Susquehanna
5678
Month
10 11
12
Figure Z-55. Susq: Comparison of climate scenario Penman-Monteith reference ET for the
Susquehanna River basin
Note: See Table Z-1 for key to climate scenarios.
Z-56
-------
Albemarle-Pamlico - PMET (in)
m
m
7:
5C>
50
40
3C
2D
10
D
I
Maximum
Minimum
75%
25%
Median
W3
W5
Albemarle-Pamlico
Base
•W1
•W2
•W3
•W4
•W6
Figure Z-56. TarNeu: Comparison of climate scenario Penman-Monteith reference ET for the Tar
and Neuse River basins
Note: See Table Z-1 for key to climate scenarios.
Z-57
-------
Trinity - PMET (in)
Maximum
Minimum
75%
25%
Median
1MB
Trinity
LJJ
Base
•W1
-W2
-W3
-W4
W5
-W6
10
11 12
Figure Z-57. Trin: Comparison of climate scenario Penman-Monteith reference ET for the Trinity
River basin
Note: See Table Z-1 for key to climate scenarios.
Z-58
-------
Upper Colorado - PMET (in)
Maximum
Minimum
75%
25%
Median
1MB
Upper Colorado
5 6 7 8 9 10 11 12
Base
•W1
-W2
-W3
-W4
W5
-W6
Figure Z-58. UppCol: Comparison of climate scenario Penman-Monteith reference ET for the
Upper Colorado basin
Note: See Table Z-1 for key to climate scenarios.
Z-59
-------
Willamette - PMET (in)
100
90
80
70
60
50
40
30
20
10
ISI
Maximum
Minimum
75%
25%
Median
Base W1 W2 W3 W4 W5 W6 W7 W8 W9 W10 W11 W12 W13 W14
Willamette
LU
Figure Z-59. Willa: Comparison of climate scenario Penman-Monteith reference ET for the
Willamette River basin
Note: See Table Z-1 for key to climate scenarios.
Z-60
-------
Powder/Tongue - PMET (in)
Maximum
Minimum
75%
25%
Median
Powder/Tongue
LJJ
5
Q_
Base
•W1
•W2
-W3
W5
-W6
11 12
Figure Z-60. Yellow: Comparison of climate scenario Penman-Monteith reference ET for the
Powder/Tongue River basins
Note: See Table Z-1 for key to climate scenarios.
Z-61
------- |