Modeling the Tongue River
Watershed with LSPC and CE-
QUAL-W2
June 25, 2007
FINAL DRAFT
Prepared by:
U.S. Environmental Protection Agency
Montana Operations Office and Tetra Tech, inc.
Project Manager: Ron Steg
$
<
33
\
(U
w.
**1
%
UU
o
J
-------
-------
Modeling the Tongue River
Watershed with LSPC and CE-
QUAL-W2
FINAL DRAFT
June 25, 2007
Prepared by:
U.S. Environmental Protection Agency
Montana Operations Office
and Tetra Tech, Inc.
Project Manager: Ron Steg
Cover photo by USGS
-------
-------
Contents
Acronyms
AMPP
Agronomic Monitoring and Protection Program
ARM
Administrative Rules of Montana
BLM
Bureau of Land Management
CBM
Coal bed methane
CFS
Cubic feet per second
DNRC
Montana Department of Natural Resources and Conservation
DO
Dissolved oxygen
EC
Electrical conductivity
GIS
Geographic information system
HSPF
Hydrological Simulation Program-Fortran
HUC
Hydrologic Unit Code
IEMB
Industrial Energy and Minerals Bureau, Montana DEQ
LSPC
Loading Simulation Program C++
MBOG
Montana Board of Oil and Gas
MDEQ
Montana Department of Environmental Quality
MFWP
Montana Fish, Wildlife, and Parks
mg/L
Milligrams per liter
MRLC
Multi-Resolution Land Characterization
MUID
Map Unit Identifier
NCDC
National Climatic Data Center
NCEPD
Northern Cheyenne Environmental Protection Department
NLCD
National Land Cover Data
NOAA
National Oceanic and Atmospheric Administration
NOx
Nitrate + Nitrite
NRCS
Natural Resources Conservation Service
NTU
Nephelometric turbidity units
NWIS
National Water Information System
PCS
Permit Compliance System
SAR
Sodium adsorption ratio
SC
Specific conductance
SNOTEL
SNOwpack TELemetry
STATSGO
State Soil Geographic Database
TDS
Total dissolved solids
TKN
Total Kjeldahl Nitrogen
TMDL
Total maximum daily load
TN
Total nitrogen
TP
Total phosphorus
TP A
TMDL planning area
TR
Total Recoverable
TRR
Tongue River Reservoir
TRWU
Tongue River Water Users
TSS
Total suspended solids
T&Y
Tongue and Yellowstone Irrigation District
|jS/cm
Micro Siemens per centimeter
USDI
United States Department of Interior
USEPA
United State Environmental Protection Agency
USFS
United States Forest Service
USGS
United States Geological Survey
WDEQ
Wyoming Department of Environmental Quality
WRCC
Western Regional Climate Center
WWDC
Wyoming Water Development Commission
WTP
Water Treatment Plant
WWTP
Wastewater Treatment Plan
i
-------
-------
Contents
Table of Contents
Executive Summary ix
1.0 Introduction 1
2.0 Model Selection 3
2.1 Technical Criteria 3
2.2 Regulatory Criteria 4
2.3 User Criteria 5
2.4 Loading Simulation Program C++ (LSPC) Overview 5
2.5 CE-QUAL-W2 Overview 6
2.6 Model Linkage 7
3.0 LSPC Model Setup 9
3.1 Modeled Parameters 9
3.2 Watershed Segmentation 12
3.3 Waterbody Representation 15
3.3.1 Streams 15
3.3.2 Tongue River Reservoir 16
3.4 Weather Data 19
3.4.1 Patching and Disaggregating Rainfall Data 22
3.4.2 Potential Evapotranspiration 23
3.5 Land Cover Representation 24
3.5.1 Impervious Lands 26
3.5.2 Irrigated Lands 26
3.5.3 Final Modeled Land Uses 32
3.6 Watershed Grouping 33
3.7 Point Sources 35
3.7.1 Wastewater Treatment Facilities 37
3.7.2 Coalbed Methane 39
3.7.2.1 CBM Outfalls 39
3.7.2.2 CBM Ponds 43
3.7.2.3 Modeling CBM Outfalls and Ponds 45
3.7.3 Coal Mines 50
3.8 Stock Ponds 51
3.9 High-Altitude Reservoirs and Diversions 54
3.10 Other Diversions 56
3.10.1 Mead-Coffeen, Piney-Cruse, and Prairie Dog Diversions 56
3.10.2 Tongue and Yellowstone Diversion 58
3.10.3 Sheridan Water Treatment Plant. 58
3.11 Irrigation 59
3.12 Surface and Subsurface Water Chemistry Concentrations 61
3.12.1 Surface Runoff Concentrations 61
3.12.2 Subsurface Concentrations 61
3.12.2.1 Literature Review 61
3.12.2.2 Modeled Subsurface Concentrations 64
4.0 LSPC Calibration 67
4.1 Hydrologic Calibration Methodology 68
4.1.1 High-Altitude Hydrologic Calibration Methodology 70
4.1.2 Prairie Hydrologic Calibration Methodology 73
4.2 Water Chemistry Calibration Methodology 74
4.2.1 Salinity and SAR 74
4.2.2 Nutrients 75
4.2.3 Water Temperature 77
5.0 Performance and Evaluation of the Calibrated Model 79
5.1 Flow 80
iii
-------
Contents
5.2 Specific Conductance 82
5.3 SAR 83
5.4 Total Nitrogen 84
5.5 Total Phosphorus 84
5.6 Salinity and SAR Calibration in the Tongue River Reservoir 85
5.7 Model Limitations 85
5.7.1 Weather Data 86
5.7.2 Flow Alterations 86
5.7.3 Point Sources Discharges 86
5.7.4 Physiographic Characteristics 86
5.7.5 Observation Data 86
5.7.6 Hydrology Calibration Data 87
5.7.7 Water Chemistry Calibration Data 87
6.0 CE-QUAL-W2 Model Setup and Calibration 89
6.1 CE-QUAL-W2 Model Configuration 89
6.1.1 CE-QUAL-W2 Segmentation/Computational Grid Setup 91
6.1.2 Initial Conditions 94
6.1.3 Boundary Conditions 94
6.1.4 Reservoir Outflo w 94
6.1.5 Point Sources 94
6.1.6 Meteorological Data 96
6.1.7 CE-QUAL-W2 Calibration Time Period 96
6.2 Upstream Conditions 97
6.2.1 Flow 97
6.2.2 Water Chemistry. 97
6.2.3 Temperature 98
6.3 CE-QUAL-W2 Model Calibration and Validation 99
6.4 CE-QUAL-W2 Modeling Coefficients 107
6.5 CE-QUAL-W2 Data Limitations 109
6.5.1 Point Sources Discharges 109
6.5.2 Observation Data 109
6.5.3 Water Chemistry Calibration Data 109
7.0 Uncertainty 111
7.1 Model Use 111
7.2 Conclusions 115
8.0 Steps to Improve Model Performance 117
9.0 References 119
APPENDIX A - Point Sources A-1
APPENDIX B - LSPC Model Evaluation Results B-1
iv
-------
Contents
Tables
Table 2-1.
Table 2-2.
Table 3-1.
Table 3-2.
Table 3-3.
Table 3-4.
Table 3-5.
Table 3-6.
Table 3-7.
Table 3-8.
Table 3-9.
Table 3-10
Table 3-11
Table 3-12
Table 3-13
Table 3-14
Table 3-15
Table 3-16
Table 3-17
Table 3-18
Table 4-1.
Table 4-2.
Table 4-3.
Table 4-4.
Table 5-1.
Table 5-2.
Table 5-3.
Table 5-4.
Table 6-1.
Table 6-2.
Table 7-1.
Table 7-2.
Montana's numeric criteria for salinity (measured as electrical conductivity) applicable to
the Tongue River watershed 4
Montana's numeric SAR criteria for the Tongue River watershed 4
Relationships between observed EC and the sum of cations (Ca, Mg, and Na in
milliequivalents) at selected USGS gages in the Tongue River watershed 11
Weatherstations used in the Tongue River watershed modeling 21
NLCD land cover in the Tongue River watershed 24
Percentage of impervious cover for developed land use classes 26
Total acres of irrigated land in the Tongue River watershed in Montana 28
LSPC land use groupings and their associated NLCD land use classes 32
Modeled land uses in the Tongue River watershed 32
Watershed groups for the Tongue River LSPC model 33
Summary of permitted wastewater discharge data in the Tongue River watershed
(average values for the available period of record are reported) 37
Summary of modeled CBM flow and concentrations 45
Summary of modeled CBM outfalls per LSPC subbasin 48
Summary of modeled coal mine outfalls and data 50
Total volume of stock ponds per modeling subbasin 52
Summary of modeled high-altitude reservoirs and their associated input parameters 55
Average monthly flow (cfs) for the three diversions from the Powder River watershed. ..57
Crop irrigation requirements for irrigated land in the Tongue River watershed 60
Summary of Discharge Concentrations (mg/L) from Base of C Horizon, AMPP Alluvial
Soils (calculated from data in Schafer et al., 2006) 63
Calculated discharge concentrations from alluvial soils 63
Recommended criteria for the Tongue River watershed hydrology calibration 67
Recommended criteria for the Tongue River watershed chemistry calibration 68
USGS monitoring sites used for the Tongue River LSPC model hydrology calibration. ..68
USGS monitoring sites used for the Tongue River LSPC model water chemistry
calibration 74
Model performance criteria (Adapted from Donigian, 2000) 79
Model performance - average flow at various timescales 81
Model performance - average SC at various time scales 82
Model performance - average SAR at various timescales 83
Annual average water chemistry concentrations (mg/L) at the USGS gage along the
Tongue River at Decker, Montana (gage ID 06306300) 98
Kinetic Coefficients used in the calibration of Tongue River Reservoir 108
Qualitative comparison of predicted and observed minimum and maximum SC values
(Symbols indicate under or over prediction) 113
Qualitative comparison of predicted and observed minimum and maximum SAR values
(Symbols indicate under or over prediction) 114
v
-------
Contents
Figures
Figure 1-1. Location of the Tongue River watershed 2
Figure 3-1. Observed cations in the Tongue River at the State Line (USGS Gage 06306300). All
data through September 30, 2006 are shown 10
Figure 3-2. Linear regression of observed cations to observed EC at USGS gage 06306300 (Tongue
River at the State Line) 10
Figure 3-3. Modeling subwatersheds for the Upper Tongue River watershed and Tongue River
Reservoir 13
Figure 3-4. Modeling subwatersheds for the lower Tongue River watershed 14
Figure 3-5. Stream channel representation in the LSPC model 16
Figure 3-6. Stage-volume-area relationships for the Tongue River Reservoir 17
Figure 3-7. Sensitivity analysis of five model runs for water released from the Tongue River
Reservoir 18
Figure 3-8. Weatherstation mapping for the Tongue River watershed 20
Figure 3-9. Twenty-three years of annual rainfall for weather station 242266 (Decker, Montana) 22
Figure 3-10. NLCD land cover in the Tongue River watershed 25
Figure 3-11. Example area where irrigated land is classified as "hay/pasture" or "cropland" by the
NLCD data. Tongue River near Ashland is shown 28
Figure 3-12. Example area where irrigated land is classified as "emergent herbaceous wetlands."
Otter Creek near Ashland is shown 29
Figure 3-13. Location of irrigated land in the Tongue River watershed 30
Figure 3-14. Watershed groups for the Tongue River LSPC model 34
Figure 3-15. Location of the permitted surface water discharges in the Tongue River watershed (as of
September 30, 2006) 36
Figure 3-16. Location of known CBM outfalls in the Tongue River watershed as of September 30,
2006 41
Figure 3-17. Summary of total monthly CBM discharge from Montana and Wyoming facilities (all
available data from 1997-2006) 42
Figure 3-18. Location of CBM ponds and pond groups 47
Figure 3-19. CBM pond processes modeled in LSPC 49
Figure 3-20. Relationship between watershed area and stock pond volume 51
Figure 3-21. Location of stock ponds in the Tongue River watershed 53
Figure 3-22. Simplified schematic of high-altitude reservoirs and diversions in the Goose Creek
watershed (not to scale) 54
Figure 3-23. Transbasin diversions from the Powder River watershed to the Tongue River watershed.
57
Figure 3-24. Histogram of Discharge Concentrations from Base of C Horizon, AMPP Alluvial Soils
(calculated from data in Schafer et al., 2006) 62
Figure 4-1. LSPC calibration sites for the Tongue River watershed 69
Figure 4-2. Snow budget for the Dome Lake SNOTEL gage 71
Figure 4-3. Snow budget for the Sucker Creek SNOTEL gage 72
Figure 4-4. Water versus air temperature regression at State Line (06306300) 77
Figure 5-1. Time series of daily mean predicted and observed flow for the Tongue River at State Line
near Decker, Montana (USGS gage 06306300) 80
Figure 5-2. Time series of TN data for the Tongue River at State Line near Decker, Montana (USGS
gage 06306300) 84
Figure 5-3. Time series of TP data for the Tongue River at State Line near Decker, Montana (USGS
gage 06306300) 85
Figure 6-1. Tongue River Reservoir bathymetry 90
Figure 6-2. Tongue River Reservoir segmentation 92
vi
-------
Contents
Figure 6-3. Longitudinal profile of the Tongue River Reservoir 93
Figure 6-4. Point source locations in Tongue River Reservoir model 95
Figure 6-5. Location of weather station used for Tongue River Reservoir modeling 96
Figure 6-6. Polynomial fit used to derive temperature time series from observed historical data 98
Figure 6-7. Water Surface Elevation Calibration 99
Figure 6-8. Monitoring station locations 100
Figure 6-9. Temperature (deg C) calibration (2001) 102
Figure 6-10. Temperature (deg C) validation (2003) 103
Figure 6-11. Dissolved oxygen calibration (2001) 104
Figure 6-12. Dissolved Oxygen validation (2003) 105
Figure 6-13. Chlorophyll-a calibration/validation at Station 1 106
Figure 6-14. P04 calibration/validation at Station 1 106
vii
-------
-------
Executive Summary
EXECUTIVE SUMMARY
This document is a companion to the document titled, "Water Quality Assessment for the Tongue River
Planning Area, Montana" (EPA, 2007 - hereafter referred to as the "Assessment Report"). Its purpose is
to explain the process for and report the results of selecting, setting up, and calibrating the computer
models that were created to support development of the Assessment Report. This report is not intended to
be a user's manual. A user's manual may be prepared at a future date if and when it is decided how the
modeling tools described in this document will be used by others.
Two models were developed to simulate the Tongue River Watershed:
1. A Loading Simulation Program C+ + (LSPC) model to simulate watershed processes in the
Tongue River watershed, and fate and transport of selected chemical constituents (i.e., salinity,
SAR, total nitrogen, and total phosphorus) in the Tongue River, Tongue River tributaries, and
the Tongue River Reservoir.
2. A CE-QUAL-W2 model to simulate hydrology and nutrients, dissolved oxygen, and temperature
in the Tongue River Reservoir.
USGS data were used to calibrate both models at various locations in the Tongue River and tributaries to
the Tongue River (e.g., Hanging Woman, Otter, and Pumpkin Creeks). The calibration encompassed the
time period between 1991 and 2006 and varied for the various chemical constituents and hydrology
based on the availability of data at a given USGS gage station. Because of limited weather data,
calibration was not performed prior to 1991. A formal model validation was not performed due to
limited data.
In general, for the Tongue River, model performance is good to very good for the prediction of average
flow, specific conductance, and SAR at time scales of one month or greater and model performance is
better during the growing season months than the non-growing season months. Model performance for
the prediction of average specific conductance and SAR was good to very good for the growing season
and longer time periods in all tributaries (where sufficient data were available to evaluate model
performance). With the exception of Hanging Woman Creek where flow prediction was fair to good for
the growing season and longer time periods, model performance for the prediction of flow in the
tributaries was poor.
The primary use of the model to date relates to the prediction of what water quality might have been like
in the absence of anthropogenic influence (i.e., predicting the "natural" condition). The model has also
been used to evaluate the potential magnitude of water quality effects associated with historical and
future discharges from individual sources or categories of sources (e.g., CBM discharge, irrigation, etc.).
Without other assessment methods and/or data to define the "natural" condition, and/or monitoring data
to spatially or temporally isolate individual sources, model simulation is the only option for prediction of
the water quality condition. So long as the model results are used with caution and uncertainty is
acknowledged, the model is well suited for this purpose.
-------
-------
Introduction
1.0 INTRODUCTION
In 2003, Montana DEQ published a report titled, Total Maximum Daily Load (TMDL) Status Report -
Tongue River TMDL Planning Area (referred to herein as the Status Report) (MDEQ, 2003). The
purpose of that document was to provide a summary of TMDL-related work that had been performed
through March 2003, and to facilitate further watershed study and TMDL development in the Tongue
River watershed. Two phases of the TMDL process were documented in the Status Report. First, a
watershed characterization was completed to obtain a better understanding of the Tongue River watershed
and its environmental/socioeconomic setting. Second, preliminary water quality impairment assessments
were completed using data available at the time of the report. However, in some streams, the water
quality impairment status could not be determined due to lack of data or pending changes to Montana's
water quality standards.
After completion of the Status Report, it was determined that a computer model of the Tongue River
watershed could help to refine and complete the water quality impairment assessments. The model could
be used to fill data gaps, and could be used to compare natural versus existing conditions within the
303(d) listed streams. Also, the model could then be used for other TMDL related activities, such as
calculating pollutant loads, allocating loads, calculating load reductions, and predicting future loads under
various management scenarios. Model selection and initialization began in May of 2003.
As part of the modeling process, the Tongue/Powder/Rosebud TMDL Modeling Committee was formed to
provide input and recommendations for the model selection, setup and calibration. The Committee met
seven times between July 2003 and August 2004, and was composed of individuals representing a variety
of interests including agriculture, Northern Cheyenne Tribe, Crow Tribe, Montana Department of
Environmental Quality (MDEQ), Wyoming Department of Environmental Quality (WDEQ), U.S.
Environmental Protection Agency (USEPA), Bureau of Land Management (BLM), Natural Resources
Conservation Service (NRCS), the Department of Natural Resources and Conservation (DNRC), and
industry. Members of the Committee were briefed on the status of the modeling effort, and members
provided insight and recommendations related to key water quality issues, characteristics of the
watershed, and sources of data. Meeting minutes are available online at:
http://deq.mt.gov/wqinfo/TMDL/TonguePowderRosebudTMDL.asp
This document is a companion to the document titled, "Water Quality Assessment for the Tongue River
Watershed, Montana" (hereafter referred to as the "Assessment Report") (USEPA, 2007) and its purpose
is to explain the process for selecting, setting up, and calibrating the computer models that were created to
support TMDL related activities for the Tongue River watershed (Figure 1-1). The model selection
process is described in Section 2.0. Model set-up, calibration, and performance for the Tongue River
watershed are described in Sections 3.0, 4.0, and 5.0, respectively. Setup and calibration of the Tongue
River Reservoir model are discussed in Section 6.0. Section 7.0 discusses model uncertainty.
1
-------
Introduction
Miles City
MOMA.NA
WYOMING
Brandenberg
Ashland
ytriern
feyenne
Birney
BIG HORN
COUNTY-
Dayton
CUSTER
COUNT
ROSEBUD*
COUNTY
Lame Deer
SHERIDAN
COUNTY
N
BIG HORN
COUNTY
POWDER RIVER
lCOUNTY
Moor head
u
19 Cstsu;. (trill towns.
Counties
Roads
Tongue River Mainstcm
Streams
20 Milf-, Tribal Land
] t niirjiir- H r'.'i'r W-3fc n.hi-d
Figure 1 -1. Location of the Tongue River watershed.
2
-------
Model Selection
2.0 MODEL SELECTION
The following criteria were considered and addressed in selecting the appropriate watershed and receiving
water models forthe Tongue River TMDL Planning Areas (expanding on classification of Mao, 1992):
• Technical Criteria
• Regulatory Criteria
• User Criteria
2.1 Technical Criteria
The pollutants of concern for the current modeling application are salinity (measured as specific
conductance, SC), sodium adsorption ratio (SAR), and nutrients. Salinity and SAR are composite
measures respectively representing the sum of cations and the ratio of major cations (sodium, calcium,
and magnesium), so one approach to simulating salinity and SAR is to simulate individual cations. Land
use in the Tongue River watershed includes extensive areas of grasslands, rangelands, shrublands, and
forest, with limited urban land uses. Most agricultural and urban land uses are concentrated along the
valley floors of perennial streams where much of the agriculture relies on irrigation due to the arid nature
of each watershed. Different potential sources of pollutants are associated with each of the various land
uses and each land use affects the hydrology of the watershed differently. Some sources contribute
relatively constant discharges of pollutants while others are heavily influenced by snowmelt and rain
events. The Tongue River Reservoir is also a significant factor in the Tongue River watershed because it
alters the timing and magnitude of flows from the upstream portion of the watershed.
Based on these considerations, the following technical factors were critical to selecting an appropriate
watershed model:
• The model should be able to address the pollutants of concern (e.g., salinity, SAR [or major
cations], and nutrients).
• The model should be able to address a watershed with primarily rural land uses.
• The model should be appropriate for simulating large watersheds.
• The model should provide adequate time-step estimation of flow and not over-simplify storm
events to provide accurate representation of rainfall events/snowmelt and resulting peak runoff.
• The model should be capable of simulating various pollutant transport mechanisms (e.g.,
groundwater contributions, sheet flow, etc.).
• The model should include an acceptable snowmelt routine.
• The model should be flexible enough to accommodate issues such as the arid nature of the
watershed and the extensive amount of irrigation activities.
• The model should be able to be linked to an appropriate reservoir model.
Other technical factors were important to consider in selecting an appropriate receiving water model for
the Tongue River Reservoir. The reservoir was built for irrigation, recreational, and flood control
purposes in 1940 and was re-habilitated from 1996 to 1999. It now has an active storage capacity of
approximately 79,000 acre-feet of water (Personal Communications, Kevin Smith, Montana DNRC, June
14, 2004). The reservoir is long and narrow with an average depth of 6.1 meters (20 feet). The reservoir
is listed as being impaired due to nutrients, organic enrichment, and total suspended solids, although
limited historic water chemistry data are available (MDEQ, 1996, 2006a). The reservoir is also a
significant factor in controlling flow and water chemistry in the Tongue River below the dam. Technical
criteria associated with modeling the Tongue River Reservoir model therefore included the following:
3
-------
Model Selection
• The model should be able to address the pollutants of concern (e.g., salinity, SAR, and nutrients).
• The model should be appropriate for a long and narrow reservoir with spatially varying depths.
• The model should provide output from upstream to downstream in the reservoir and at depth.
• The model should be able to be linked to the Tongue River watershed model.
2.2 Regulatory Criteria
Regulatory criteria were also a key consideration in selecting appropriate watershed and reservoir models.
A stream or reservoir's assimilative capacity is determined through adherence to numeric water chemistry
standards. Table 2-1 and Table 2-2 summarize the water quality standards applicable to the Tongue River
and Tongue River Reservoir (MDEQ, 2006b). These tables indicate that the salinity, measured as
electrical conductivity (EC), and SAR standards are applied as both monthly average values and
maximum "not-to-exceed" values. The selected model therefore needed to be able to provide output that
can be directly compared to these standards. For example, some models only provide annual or monthly
output and would therefore be inadequate for assessing compliance with the component of Montana's
standard that is expressed as an instantaneous maximum. Consistency with water quality standards was
the primary regulatory criterion that affected model selection.
Table 2-1. Montana's numeric criteria for salinity (measured as electrical conductivity)
applicable to the Tongue River watershed.
Waterbody
Season
Monthly Average EC
(|jS/cm)
Maximum EC (pS/cm)
Tongue River
Nov 1 - Mar 1
1,000
1,500
Mar 2-Oct 31
1,500
2,500
Tongue River Tributaries
Nov 1 - Mar 1
500
500
Mar2-Oct 31
500
500
Tongue River Reservoir
Nov 1 - Mar 1
1,000
1,500
Mar 2-Oct 31
1,000
1,500
MDEQ, 2006b
Table 2-2.
Montana's numeric SAR criteria for the Tongue River watershed.
Waterbody
Season
Monthly Average SAR
Maximum SAR
Tongue River
Nov 1 - Mar 1
3.0
4.5
Mar 2-Oct 31
5.0
7.5
Tongue River Tributaries
Nov 1 - Mar 1
5.0
7.5
Mar 2-Oct 31
3.0
4.5
Tongue River Reservoir
Nov 1 - Mar 1
3.0
4.5
Mar 2-Oct 31
3.0
4.5
MDEQ, 2006b
4
-------
Model Selection
2.3 User Criteria
User criteria are associated with the needs, expectations, and resources of the stakeholders involved in the
modeling project. Stakeholders expressed a strong desire (via the Modeling Committee) to base
management decisions on the best available data and science. Furthermore, the stakeholders indicated
that the modeling software should be non-proprietary, tested, and accepted. A further consideration was
that the two state agencies (Montana DEQ and Wyoming DEQ) indicated that they would be interested in
using the models for future applications (such as supporting NPDES permitting decisions). Each of these
criteria was considered during the model selection process.
2.4 Loading Simulation Program C++ (LSPC) Overview
Based on the considerations described above and previous modeling experience, the Loading Simulation
Program C++ (LSPC) was selected to address all of the modeling needs except nutrient response in the
Tongue River Reservoir (see Section 2.3 for a discussion of the model selected to address nutrients in the
Tongue River Reservoir). LSPC is a version of the Hydrologic Simulation Program FORTRAN (HSPF)
model that has been ported to the C++ programming language to improve efficiency and flexibility.
LSPC integrates a geographical information system (GIS), comprehensive data storage and management
capabilities, the original HSPF algorithms, and a data analysis/post-processing system into a convenient
PC-based windows interface. LSPC's algorithms are identical to a subset of those in the HSPF model.
LSPC is currently maintained by the EPA Office of Research and Development in Athens, Georgia. A
brief overview of the HSPF model is provided below and a detailed discussion of HSPF simulated
processes and model parameters are available in the HSPF User's Manual (Bicknell et al. 1996).
HSPF is a comprehensive, public domain, watershed and receiving water quality modeling framework
that was originally developed in the mid-1970's and is supported by USEPA and USGS. During the past
several years it has been used to develop hundreds of USEPA-approved TMDLs and it is generally
considered the most advanced hydrologic and watershed loading model 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 1960's. The HSPF framework is developed in a modular
fashion with many different components that can be assembled in different ways, depending on the
objectives of the individual 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 submodules that calculate the various hydrologic and water
chemistry processes in the watershed. Many options are available for both simplified and complex
process formulations. Spatially, the watershed is divided into a series of subbasins representing the
drainage areas that contribute to each of the stream reaches. These subbasins are then further subdivided
into segments representing different land uses. For the developed areas, the land use segments are further
divided into the pervious (PERLND) and impervious (IMPLND) fractions. The stream network
(RCHRES) links the surface runoff and groundwater flow contributions from each of the land segments
and subbasins and routes them through the waterbodies using storage routing techniques. The stream
model includes precipitation and evaporation from the water surfaces, as well as flow contributions from
the watershed, tributaries, and upstream stream reaches. Flow withdrawals can also be accommodated.
The stream network is constructed to represent all of the major tributary streams, as well as different
portions of stream reaches where significant changes in water chemistry occur.
5
-------
Model Selection
Advantages to choosing LSPC for this application include:
• Simulates all of the necessary constituents and applies to rural watersheds.
• Capable of simulating both stream and reservoir processes.
• A comprehensive modeling framework using the proposed LSPC approach facilitates
development of TMDLs not only for this project, but also for potential future projects to
address other impairments throughout the basin.
• The time-variable nature of the modeling enables a straightforward evaluation of the cause-
effect relationship between source contributions and waterbody response and direct
comparison to relevant water quality criteria.
• The proposed modeling tools are free and publicly available. This is advantageous for
distributing the model to interested stakeholders and amongst government agencies.
• The model simulates both surface and subsurface impacts to flow and water quality.
• LSPC provides storage of all geographic, modeling, and point source permit data in a
Microsoft Access database and text file formats to provide for efficient manipulation of data.
• LSPC presents no inherent limitations regarding the size and number of watersheds and
streams that can be modeled.
• LSPC provides post-processing and analytical tools designed specifically to support TMDL
development and reporting requirements.
• LSPC can be linked to the Tongue River Reservoir CE-QUAL-W2 model (see Section 2.5).
2.5 CE-QUAL-W2 Overview
LSPC simulates lakes and reservoirs, but only as 1-dimensional, completely mixed systems. This is
sufficient for the simulation of major cations, as geochemical analysis with MINTEQ suggests that losses
to mineralization and settling of cations in the reservoir are not significant. However, a more complex
reservoir model was needed to simulate nutrients, eutrophication processes, and reservoir stratification.
The U.S. Army Corps of Engineers CE-QUAL-W2 (W2) model was selected as the receiving water
model for simulating nutrients in the Tongue River Reservoir. W2 is a two-dimensional,
longitudinal/vertical (laterally averaged), hydrodynamic water quality model (Cole and Wells, 2003). The
model is applicable to lakes, rivers, and estuaries that do not exhibit significant lateral variability in water
quality conditions. It allows application to multiple branches for geometrically complex waterbodies with
variable grid spacing, time variable boundary conditions, and multiple inflows and outflows from
point/nonpoint sources and precipitation.
Advantages to choosing W2 for the Tongue River Reservoir modeling application include the following:
• W2 is able to address the pollutants of concern in the reservoir (e.g., phosphorus, nitrogen,
dissolved oxygen, and chlorophyll a).
• W2 is appropriate for a long and narrow reservoir with spatially varying depths.
• W2 is able to provide output from upstream to downstream in the reservoir and at depth.
• W2 has been successfully linked in previous applications to LSPC.
• Simpler receiving water models would be limited in their ability to address the characteristics of
the reservoir (long, narrow, and deep).
• Simpler receiving water models would also prove inadequate to support a more detailed analysis
should additional data become available.
• W2 is capable of simulating cause-and-effect relationship between loading and reservoir
response.
6
-------
Model Selection
The two major components of the W2 model include hydrodynamics and water quality kinetics. Both of
these components are coupled (i.e. the hydrodynamic output is used to drive the water quality at every
time step). This makes it very efficient to execute model runs. The hydrodynamic portion of the model
predicts water surface elevations, velocities, and temperature. The W2 model uses the ULTIMATE -
QUICKEST numerical scheme for advection - dispersion computation. The ULTIMATE - QUICKEST
numerical scheme is a third order finite difference scheme. This method reduces the numerical diffusion
in the vertical direction to a minimum. In areas of high gradients this scheme reduces undershoots and
overshoots which may produce small negative concentrations. The water quality portion of W2 can
simulate the constituents required for the Tongue River Reservoir, including dissolved oxygen (DO),
nutrients, and phytoplankton interactions.
2.6 Model Linkage
As described in Section 2.4, the LSPC model was adequate for simulating watershed and in-stream
processes for all of the pollutants of concern in the Tongue River watershed. LSPC's lake modeling
features (i.e., simple, 1-dimensional vertically mixed reservoir model) were also adequate for modeling
salinity and SAR. Therefore, LSPC was used to model salinity and SAR for the entire Tongue River
watershed, including the Tongue River Reservoir.
The LSPC lake modeling processes were not adequate for modeling the complex interactions of nutrients
in the Tongue River Reservoir. Therefore, the CE-QUAL-W2 model was employed to provide a more
robust analysis. LSPC was still used to model total nitrogen and total phosphorus loads coming into the
Tongue River Reservoir from the upstream watershed (referred to as the "Upper" Tongue River
watershed). Loads generated by LSPC were simply input to the W2 model as the upstream boundary
condition. Nutrients were not assessed downstream of the Tongue River Reservoir (herein referred to as
the "Lower" Tongue River watershed).
7
-------
-------
LSPC Model Setup
3.0 LSPC MODEL SETUP
As described previously, two models have been setup to simulate the Tongue River watershed (i.e., LSPC
and CE-QUAL-W2). This section describes the setup and calibration for the LSPC model. Setup and
calibration for the Tongue River Reservoir model is described in Section 6.0.
3.1 Modeled Parameters
As describe in Section 2.0, the pollutants of concern in the Tongue River watershed are salinity, sodium
adsorption ratio (SAR), and nutrients. SAR is the ratio of sodium (Na) to calcium (Ca) and magnesium
(Mg) in a waterbody, and therefore LSPC was setup to model calcium, magnesium, and sodium
concentrations.
Salinity (measured as specific conductance [SC] or electrical conductivity [EC]) is an indirect measure of
the total dissolved solids in a waterbody. EC is typically expressed in microsiemens per cm ((iS/cm) at
standard temperature and is approximately equivalent to the sum of either anions or cations in solution
(expressed as milliequivalents or meq) times 100 (APHA, 1992). Assuming that Na, Mg, and Ca are the
dominant cations in the waterbody (ignoring K), then EC ^ 100 is approximately equal to the sum of {Na
+ Ca + Mg} in milliequivalents. More generally,
EC (juS /cm) x (Na + Ca +Mg + Z)-100,
Where Z is the sum of other cations (predominantly K), and all cation concentrations are expressed on a
meq basis.
Because of the "Z" component, EC cannot be approximated from only the sum of individual Ca, Mg, and
Na cations (as milliequivalents). Also, the activity of individual cations varies with temperature.
Therefore, an alternative method was developed to calculate EC using regression equations based on the
observed relationship between EC and the sum of the Na, Ca, and Mg concentrations in milliequivalents.
To evaluate this approach, cation concentrations and electrical conductivity was evaluated at USGS
station 06306300 (Tongue River at the State Line).
Figure 3-1 presents the cation observations in the Tongue River at the State Line (November 1985 to
September 2006). Potassium (K) is always present at low concentrations while calcium (Ca) and
magnesium (Mg) are present at the highest concentrations. Also, Ca and Mg are typically near the same
value. The low and relatively constant K concentrations suggest this cation can be ignored in the
calculation of EC. The general annual trend is observed in this figure as salt concentrations reach a
minimum during the spring snowmelt period.
Figure 3-2 presents the linear regression of EC against the sum of the 3 major cations (Na, Ca, and Mg, in
milliequivalents per liter) at USGS Gage 06306300. The regression provides a strong linear fit (R2 =
98.3) and shows that an accurate prediction of observed EC can be obtained by using only the 3 major
cations. Thus, reconstruction of EC from Ca, Mg, and Na should be sufficient for the LSPC/W2
applications.
9
-------
LSPC Model Setup
at
E
ns
O
140
120
100
80
60
40
20
J
0
1985
~ Ca (mg/L)
o Mg (mg/L)
Na (mg/L)
~
~ ~
CP
~
0
e
O O
1
n
* V
e>
•*x
X>< X
1987
1989
1991
1993
~
~
o
o
o
[D
X
2L
~
~
ft
2^
X K(mg/L)
Cb ~
B R
B J
m & P
°
XX
Figure 3-1.
1995
1997
1999
2001
2003
2005
2007
Observed cations in the Tongue River at the State Line (USGS Gage 06306300). All
data through September 30, 2006 are shown.
E
o
w
o
LU
73
0)
>
5
(A
n
O
1,400
1,200
1,000
800
600
400
200
0
0.0
o Tongue River at Stateline
¦ Linear Trendline
278x + 23.86 o
2.0
4.0
6.0
.0
10.0
12.0
14.0
Sum of Observed Ca, Mg, and Na Cations (meq)
Figure 3-2. Linear regression of observed cations to observed EC at USGS gage 06306300
(Tongue River at the State Line).
Because of the previously discussed relationships for deriving EC from cation concentrations, there was
no need to directly model EC or TDS as separate state variables. Salinity (expressed as specific
conductance in (iS/cm) was simply computed using the observed relationship between EC and the sum of
the Na, Ca, and Mg cations. The relationships at selected USGS gages are reported in Table 3-1. As
evidenced by the R2 values, the regressions provide a good fit, indicating that this approximation is
satisfactory. The advantage of this approach is two-fold. First, this method reduces the total number of
modeled state variables, which reduces model run time and simplifies model structure. Second, this
method insures that salinity specifications are consistent with sodium, magnesium, calcium, and SAR
values.
10
-------
LSPC Model Setup
Nutrients in the LSPC were modeled as total nitrogen and total phosphorus. Too few calibration data
(i.e., stream, precipitation, interflow, groundwater, and point source concentrations) were available to
model nutrient species (e.g., nitrate, nitrite, ammonia, and orthophosphorus).
Table 3-1. Relationships between observed EC and the sum of cations (Ca, Mg, and Na in
milliequivalents) at selected USGS gac
ies in the Tongue River watershed.
USGS Station Name
USGS
Station ID
Modeling
Subbasin
Relationship
R2
Tongue River near Dayton,
Wyoming
06298000
3090
EC (pS/cm) = 84.704 (Ca + Mg + Na) + 24.105
0.79
Tongue River at the Montana-
Wyoming State Line, Montana
06306300
3006
EC (pS/cm) = 88.278 (Ca + Mg + Na) + 23.36
0.98
Prairie Dog Creek near Acme,
Wyoming
06306250
3007
EC (pS/cm) = 73.704 (Ca + Mg + Na) + 174.8
0.98
Goose Creek near Acme,
Wyoming
06305700
3022
EC (pS/cm) = 79.405 (Ca + Mg + Na) + 59.692
0.98
Tongue River at the Tongue
River Reservoir Dam near
Decker, Montana
06307500
3112
EC (pS/cm) = 83.923 (Ca + Mg + Na) + 49.005
0.97
Tongue River at the Birney
Day School Bridge, Montana
06307616
3088
EC (pS/cm) = 83.293 (Ca + Mg + Na) + 48.478
0.91
Tongue River near the
Brandenberg Bridge, Montana
06307830
1047
EC (pS/cm) = 83.895 (Ca + Mg + Na) + 60.936
0.97
Tongue River at Miles City,
Montana
06308500
1002
EC (pS/cm) = 83.096 (Ca + Mg + Na) + 69.933
0.90
Hanging Woman Creek near
Birney, Montana
06307600
1095
EC (pS/cm) = 78.801 (Ca + Mg + Na) + 172.9
0.95
Otter Creek at Ashland,
Montana
06307740
1059
EC (pS/cm) = 72.625 (Ca + Mg + Na) + 374.36
0.90
Pumpkin Creek near Miles
City, Montana
06308400
1007
EC (pS/cm) = 86.408 (Ca + Mg + Na) + 111.35
0.99
11
-------
LSPC Model Setup
3.2 Watershed Segmentation
LSPC was configured to simulate the Tongue River watershed as a series of hydrologically connected
subwatersheds. The spatial subdivision of the watersheds allowed for a more refined representation of
pollutant sources, and a more realistic description of hydrologic factors. Subwatershed delineation was
primarily based on Montana DNRC's Draft 6th Code HUCs, but also took into consideration spatial
variation in sources, hydrology, and jurisdictional boundaries (DNRC, 2006a). Output from LSPC is for
the most downstream point of each subwatershed (sometimes referred to as the "pour point").
Subwatersheds were therefore delineated to obtain modeling output at key flow or water quality stations
and at political boundaries (e.g., the Montana/Wyoming state line and at the upstream and downstream
boundaries of the Northern Cheyenne Reservation).
Because of the Tongue River Reservoir, the subwatersheds were classified into three distinct regions:
• Upper Tongue River - The portion of the watershed upstream of the Tongue River Reservoir (i.e.,
headwaters to the confluence with the reservoir). Numbered as LSPC subbasins 3001 to 3108.
• Lower Tongue River - The portion of the watershed located downstream of the Tongue River
Reservoir (i.e., Tongue River Reservoir Dam to the mouth). Numbered as LSPC subbasins 1001
to 1113.
• Tongue River Reservoir - The portion of the watershed draining directly to the Tongue River
Reservoir. Numbered as LSPC subbasin 3000.
Relatively large subwatersheds were specified in the model due to the homogenous land use/land cover
characteristics, and the fact that fewer subwatersheds reduced computational time. Smaller subwatersheds
were delineated in the Bighorn Mountains to reduce the elevation change per subwatershed (which
impacts the snow simulation process). Each of the subwatersheds was modeled as a single stream
segment that was assumed to be a completely mixed, one-dimensional segment with a trapezoidal cross-
section (see Section 3.3). The final subwatersheds and primary streams for the LSPC model are shown in
Figure 3-3 and Figure 3-4, as are the location of cities and political boundaries.
12
-------
Figure 3-3. Modeling subwatersheds for the Upper Tongue River watershed and Tongue River Reservoir.
-------
LSPC Model Setup
Cities
Counties
Tribal Land
Mo-deled Streams
) Lower Tongue Subbasins
Rosebud
County
Big Horn
County
Powder Rfvei
County
TONGUE HSVEK
RSfiSBVOm
WYOMING
j |
\ '' I ! \ v
N* j 1 \)1m\
\
tic& Vv-
Sheridan
County J
0
5
10
20 Miles
MONTANA
v—^
WVOMIHC
J I 1 I I I I I
Figure 3-4. Modeling subwatersheds for the lower Tongue River watershed.
14
-------
LSPC Model Setup
3.3 Waterbody Representation
LSPC was configured to model both streams and reservoirs in the Tongue River watershed. Section 3.3.1
discusses the information used to model the stream segments, and Section 3.3.2 discusses the setup of the
Tongue River Reservoir. High altitude reservoirs located in the Bighorn Mountains are further discussed
in Section 3.9.
3.3.1 Streams
Each subwatershed in LSPC was represented with a single stream assumed to be a completely mixed,
one-dimensional segment with a trapezoidal cross-section (Figure 3-5). Input parameters for the reaches
include initial depth, length, depth, width, slope, Manning's roughness coefficient, and coefficients to
describe the shape of the stream channel. The methodology for determining these parameters for the
Tongue River watershed is described below:
• IDEPTH (Reach Initial Water Depth) - Assumed to be half the bankfull depth.
• LENGTH (Reach Length) - Determined from the National Hydrography Dataset (NHD)
medium resolution stream reach network (available online at http://nhd.usgs.gov/).
• DEPTH (Reach Bankfull Depth) - Reach bankfull depth values were estimated based on
the equation 1 (below). The coefficients for "a" and "b" were determined based on cross
sections measured by USGS at nine sites in the Tongue River watershed (USGS, 2004).
eq. 1: BankfullDepth(ft) = a x (WatershedAreaf
• WIDTH (Reach Bankfull Width) - Reach bankfull width values were estimated based on
equation 2. The coefficients for "c" and "d" were determined based on cross sections
measured by USGS at nine sites in the Tongue River watershed (USGS, 2004).
eq.2 : BankfullWidth(ft) = c x (WatershedArea)d
• SLOPE (Reach Slope) - Calculated based on elevation data from the USGS 30-meter
National Elevation Dataset (USGS, 2002).
• MANN (Manning's Roughness Coefficient for the Stream Channel) - Estimated
coefficient of 0.02 was applied to each representative stream reach based on typical
literature values (Schwab et al., 1993)
• R1 (Reach ratio of Bottom Width to Bankfull Width) - Estimated from USGS, 2004.
• R2 (Reach Side Slope of Floodplain) - Estimated from USGS, 2004.
• W1 (Reach Floodplain Width Factor) - Estimated from USGS, 2004.
Reach dimensions for the Tongue River LSPC model are included in the LSPC model input file, which is
available upon request.
15
-------
LSPC Model Setup
Channel Cross-Section
Figure 3-5. Stream channel representation in the LSPC model.
3.3.2 Tongue River Reservoir
As discussed in Section 2.2, major cations (and thus salinity and SAR) were modeled in the Tongue River
Reservoir using the lake/reservoir features of LSPC. Nutrients were modeled in the reservoir using the
CE-QUAL-W2 program, which is discussed in Section 6.
The Tongue River Reservoir (TRR) was originally completed in 1940 by constructing an earthen dam on
the Tongue River north of Decker, Montana (DNRC, 2004). A 1996-1999 rehabilitation project
increased the reservoir's active storage capacity from approximately 68,000 acre-feet of water to 79,000
acre-feet of water. An additional spillway was also added during the rehabilitation project so that the
maximum potential discharge from the reservoir is now approximately 4,000 cfs. Hie average depth of
the reservoir is 6.1 meters (20 feet) with a length of approximately 12.5 kilometers (7.8 miles) (DNRC.
2005). The average volume of water in the reservoir between 1999 and 2006 was 40,432 acre-feet and
the median residence time during this period was approximately 88 days (with longer residence times
during the fall, winter, and spring and shorter
residence times during the summer) (DNRC,
2006).
The primary spillway for the Tongue River
Reservoir is a concrete labyrinth spillway
(weirwall spillway) with a crest of 3,428.4 feet
MSL, corresponding to a storage volume of
79,071 acre-feet of water in the reservoir. The
primary spillway was re-constructed in the late
1990's and the first full year of normal operation
was 2000 (Personal Communications, Kevin
Smith, Montana DNRC, June 14,2004). Very
little water has gone over the spillway since the
re-construction. The reservoir also has an
emergency spillway with a crest at 3,431.5 feet,
or when the reservoir volume is at 91,107 acre-feet
of water.
In addition to the primary and emergency spillways, the reservoir has two inlet structures. The first was
built in 1940 and the second in 1999 (Personal Communications, Kevin Smith, Montana DNRC, June 14,
2004). Each structure has inlets at two invert elevations (3,375 feet and 3,390 feet) with grills on all sides
Tongue River Reservoir Primary Spillway and Inlet
Release Structure
(Photo by Tetra Tech, inc.)
16
-------
LSPC Model Setup
and on top. Water flow through these grills is controlled through a central system located within the
earthen dam. There is no way to close one grate versus another and water intake through the individual
grills is therefore not regulated. At its fullest, the reservoir is drafting water through all grill inlets, the
emergency spillway, and the primary spillway. However, normal operation is to draft water over the
primary spillway and through the inlets. By the end of summer, water is typically only discharging
through the two inlet towers. The reservoir is almost never drawn down below an elevation of 3,404 feet
(Personal Communications, Kevin Smith, Montana DNRC, June 14, 2004).
The Tongue River Reservoir was set up in the LSPC model based on data provide by Montana DNRC.
The reservoir was not explicitly modeled as a reach segment with fixed dimensions, but was rather
simulated with the use of a stage-volume relationship (referred to as an F-Table in LSPC). DNRC
provided monthly observed stage and volume data for the time period between 1960 to present (DNRC,
2006). The data between 2000 and 2006 were used to create a stage-volume relationship for the post-
construction conditions in the Tongue River Reservoir. Pre-reconstruction conditions were not modeled,
which as described in Section 4, provided a limitation to the time period that could be used to calibrate the
mainstem Tongue River downstream of the Tongue River Reservoir. Reservoir surface area was
calculated using a GIS and CAD files of reservoir bathymetry (also provided by Montana DNRC). The
primary overflow spillway height was set at 54.4 feet (elevation of 3,428.4 feet), and the intake structure
for managed reservoir releases was set at 53.5 feet (elevation of 3375.0 feet). Figure 3-6 shows the stage-
volume-area relationship developed for the Tongue River Reservoir.
-Volume
-Surface Area
4,500
100,000
80,000
v
E
3
O
>
60,000
40,000
20,000
-- 4,000
-- 3,500
-- 3,000
-- 2,500
-- 2,000
-- 1,500
-- 1,000
-- 500
- 0
10 15 20 25 30 35 40 45 50 55 60 65
Stage (feet)
Stage-volume-area relationships for the Tongue River Reservoir.
Figure 3-6.
The measured stage and volume data provided the data needed to properly size the Tongue River
Reservoir in the model. However, the DNRC data did not provide information about reservoir discharge.
Releases from the Tongue River Reservoir are managed, and therefore a simple stage-discharge
relationship could not be developed. Water is stored throughout the winter and spring seasons and
released in the spring and summer to provide water to downstream irrigators. The USGS gage located in
the Tongue River downstream of the Tongue River Reservoir (USGS Gage 06307500) measures the
discharge from the reservoir, but it does not differentiate between "managed" water (i.e., water regulated
through the two inlet structures) and water that flows over the spillway.
17
-------
LSPC Model Setup
Daily discharge data at the USGS gage downstream of the Tongue River Reservoir Dam were assessed to
determine the reservoir releases due to management versus overflow over the spillways. Flow percentiles
were calculated for the daily discharge data from October 1, 2000 to September 30, 2006. Based on the
calculated flow percentiles, five different model sensitivity runs were conducted where flows less than the
60, 70, 80, 90, and 95th percentiles were defined as "management" (i.e., releases through the inlet
structure). The reservoir management time series was withdrawn from the reservoir and input into the
next downstream modeling subbasin - 1112. The remaining water volume was allowed to naturally
accumulate, evaporate, and discharge over the reservoir spillway. Figure 3-7 shows the results of the five
model runs, and the impact the various management scenarios have on reservoir stage.
• Measured Stage (ft) 95 90 80 70 60
Figure 3-7. Sensitivity analysis of five model runs for water released from the Tongue River
Reservoir.
Based on the sensitivity analysis, it was assumed that flows occurring below the 90th percentile were due
to managed releases from the reservoir. The time series of flows for USGS Gage 06307500 below the
90th percentile were input into the LSPC model as a managed discharge or ""w ithdrawal" from the
reservoir that is reintroduced into the downstream mainstem Tongue River in LSPC subbasin 1112. The
remainder of the water in the reservoir is allowed to overflow as needed. Discharge from the spillway
was also determined in the F-Table based on the stage of the reservoir and a spillway width of 100 feet.
This method for modeling discharge from the Tongue River Reservoir has one primary advantage - it
allows for potential scenarios where reservoir management can be altered or removed. However, as seen
in Figure 3-7, this approach does introduce some error into the model. Without observed data from the
reservoir outlets, it is impossible to determine what volume of water is manually released versus
discharged over the spillway. As seen in Section 4, this had an impact on the main stem Tongue River
hydrology and water chemistry calibration downstream of the Tongue River Reservoir Dam.
18
-------
LSPC Model Setup
3.4 Weather Data
The LSPC model is driven by precipitation and other climatologic data (e.g., temperature, cloud cover,
wind speed). As a result, meteorological data are a critical component of the watershed modeling effort.
Appropriate representation of precipitation, wind movement, solar radiation, potential evapotranspiration,
cloud cover, temperature, and dew point are required to develop a valid model.
A number of sources were consulted to determine the availability of weather data - the National Climatic
Data Center (NCDC), Natural Resources Conservation Service (NRCS), U.S. Forest Service, and the
Bureau of Land Management. Ideal weather stations had daily (or more frequent) data collected recently
(i.e., 1980 to present), were located within the watershed boundary, and recorded multiple parameters in
addition to precipitation. Appropriate data were determined to be available from 10 NCDC stations and 6
NRCS SNOwpack TELemetry (SNOTEL) stations. Only stations with daily rainfall totals were used in
the LSPC model, as NCDC personnel indicated that daily stations were more complete and had more
thorough quality control than hourly stations (Personal Communications, NOAA, December 13, 2006).
Data from four hourly rainfall stations were used for distribution only, to disaggregate the daily rainfall
totals to hourly. Precipitation data were obtained from the daily stations, and weather gages were
assigned to each subwatershed using the Thiessen polygon method (Theissen, 1911). At the time of the
model setup, weather data were available through September 30, 2006. Additional data may be added to
the model as needed.
The weather stations used in the modeling are summarized in Table 3-2 and their locations are shown in
Figure 3-8. Figure 3-8 illustrates that large areas must be simulated based on data from single weather
stations. In fact, the average area per weather station is 338 square miles. This relatively sparse coverage
of weather stations posed one of the most significant limitations of the modeling effort. Substantial
variability in meteorology is known to be present across the Tongue River watershed due to its size and
topography. Perhaps even more importantly, the timing of individual storms can also be significantly
different throughout the region, especially during convective summer thunderstorms. As explained in
Section 5.7.1, the extrapolation of precipitation from a limited number of points is believed to be one of
the largest sources of model error. Nevertheless, the relatively higher density and high quality of
SNOTEL gages for precipitation and temperature in the mountain regions significantly improved the
high-altitude snowfall/snowmelt representation, as described in Section 4.1.1. The mountain region is
responsible for about 63 percent of all the water in the Tongue River Watershed.
19
-------
LSPC Model Setup
Crow
DAYTON «
tW#r&£35 JUNCTION
story
DOOC
0 5 10 20 Milts
1 I I I M ' ' >
# NCDC Gage-s
I Sr-JOTEL Gag«&
Streams
f iibj'l Land
Counties
Weather Station Mapping
Decker
Dayton
| Rranttunbcrg
Mdiȣ. Q*y
SoflMtt*
Volberg
LvHsr
Shcdslan AP
Slwfldart Fi»td Sw
Story
Bl9 G04S4
Rnnn Springs
0ur9»i4 Junction
Dome LaNo
Suck«r C(9«»;
I Tib Crtiofc
MONTANA
WYOMING
DECKER 4 HNE
OtJWC SP-RlNftf> DiV
• LEI "TEW 9 N
IWLEi CITV Af Q
; <»
VO| R :"jRc«
*ONMC7TG3WNW
Figure 3-8. Weather station mapping for the Tongue River watershed.
20
-------
LSPC Model Setup
Table 3-2. Weather stations used in the Tongue River watershed modeling.
Station Name
Station ID
Agency
Data
Type
Temperature
Elevation (ft)
Available Data
Big Goose
07E32S
NRCS
SNOTEL
7990
Precip, Min/Max Temp
Bone Springs
07E18S
NRCS
SNOTEL
9350
Precip, Min/Max Temp
Burgess Junction
07E33S
NRCS
SNOTEL
7880
Precip, Min/Max Temp
Dome Lake
07E34S
NRCS
SNOTEL
8880
Precip, Min/Max Temp
Sucker Creek
07E12S
NRCS
SNOTEL
8880
Precip, Min/Max Temp
Tie Creek
07E39S
NRCS
SNOTEL
6870
Precip, Min/Max Temp
Brandenberg
241084
NCDC
Da
ly
2770
Precip, Min/Max Temp
Leiter 9 N
485506
NCDC
Da
ly
4160
Precip, Min/Max Temp
Miles City AP
245690
NCDC
Da
iy
2624
Precip, Min/Max Temp
Sheridan AP
488155
NCDC
Da
iy
3945
Precip, Min/Max Temp
Dayton
482399
NCDC
Da
iy
3945°
Precip
Sheridan Field Station
488160
NCDC
Da
iy
3750
Precip, Min/Max Temp, Evap
Decker 4 NNE
242266
NCDC
Da
iy
3750°
Precip
Sonnette 2 WNW
247740
NCDC
Da
iy
3900
Precip, Min/Max Temp
Story
488626
NCDC
Da
iy
5083
Precip, Min/Max Temp
Volborg
248607
NCDC
Da
iy
5083°
Precip
Sheridan AP
24029
NCDC
Hourly
3945
Dewpoint, Wind, Cloud/Solar
Ashland Ranger Station
MT0330
NCDC
Hourly
n/a
Precip
Sheridan AP
WY8155
NCDC
Hourly
n/a
Precip
Story
WY8626
NCDC
Hourly
n/a
Precip
3 NRCS is the National Resource Conservation Service; NCDC is the National Climatic Data Center
b SNOTEL are SNOwpackTELemetry stations (SnowWater Equivalent data are also available, but used
c Temperature data was either unavailable or largely missing. The nearest representative surrogate was
for calibration rather than
used (listed above entry).
input)
21
-------
LSPC Model Setup
3.4.1 Patching and Disaggregating Rainfall Data
The Tongue River LSPC models were run at an hourly time step to best capture the impact of individual
storm events. However, as shown in Figure 3-8 and Table 3-2, many of the precipitation stations only
recorded daily precipitation. In addition, many of the stations contained various intervals of accumulated,
missing, or deleted data. Missing or deleted intervals are periods over which either the gage malfunctions
or the data records were somehow lost. Accumulated intervals represent cumulative precipitation over
several hours or days, but the exact temporal distribution of the data is unknown. The normal-ratio
method (Dunn & Leopold, 1978) was used to compute accumulated, missing, and deleted data intervals.
The normal-ratio was also used to disaggregate the daily rainfall totals to hourly records based on the
hourly rainfall distributions at nearby gages.
The normal-ratio method estimates a missing rainfall value using a weighted average from surrounding
stations with similar rainfall patterns according to the relationship:
PA =
r n na ^
p,
J
where PA is the missing precipitation value at station A. n is the number of surrounding stations with valid
data at the same specific point in time, NA is the long term average precipitation at station^, TV, is the long
term average precipitation at nearby station and I', is the observed precipitation at nearby station For
each missing data record at station^, n consists of only the surrounding stations with valid data;
therefore, for each record, n varies from 1 to the maximum number of surrounding stations. When no
precipitation is available at the surrounding stations, zero precipitation is assumed at station^. The US
Weather Bureau has a long established practice of using the long-term average rainfall as the precipitation
normal. Since the normal ratio considers the long-term average rainfall as the weighting factor, this
method is adaptable to regions where there is large orographic variation in precipitation. Figure 3-9
below shows the 23-water-year annual rainfall totals at a representative station (Decker, Montana).
Percent Missing (Original)
- Patched Composite - - ~— Original - Decker, MT
c
to
£
15
3
C
C
<
100%
o
o
0)
£
LO CO h- 00 a>
0)0)0)0)0)
0)0)0)0)0)
O ^ CM CO t
o o o o o
o o o o o
CM CNJ CM CM CM
Figure 3-9. Twenty-three years of annual rainfall for weather station 242266 (Decker,
Montana).
22
-------
LSPC Model Setup
3.4.2 Potential Evapotranspiration
Evapotranspiration represents the sum of direct evaporation of surface and soil moisture and transpiration
of water by plants. In the Tongue River watershed, evapotranspiration is a key component of the
hydrologic balance. As a weather input, LSPC requires potential evapotranspiration (PEVT), which is the
maximum naturally achievable amount at any given moment.
Although there are some tests available for actually measuring evapotranspiration in the field, most
practitioners estimate evapotranspiration using empirical formulations that are a function of other related
(and more commonly observed) weather data. Three widely used methods are the Hamon method (1961),
the Jensen-Haise method (1963) and the Penman Pan-Evaporation method (1948). The Penman method
computes evaporation as a function of temperature, solar radiation, dew point or relative humidity, and
wind movement. The other two methods, Hamon and Jensen-Haise, are simplified empirical
representations that require fewer observed datasets to compute.
The various potential evapotranspiration methods were assessed in the Tongue River watershed. Based
on test of each method, the Penman method predicted evapotranspiration closest to measured data at
Sheridan Field Station in Wyoming. Therefore, the Penman equation was used to calculate potential
evapotranspiration for the entire Tongue River watershed.
Because PEVT is a key component of the hydrologic balance, it is also a potential source of model
uncertainty and error. Few measured PEVT data were available for the Tongue River watershed, and few
data were available for calculating PEVT using the previously described methods. Furthermore, the
various methods for calculating PEVT give differing results. As explained in Section 5.7.1, the
extrapolation of PEVT from a limited number of weather stations, combined with the varying methods for
calculating PEVT, is believed to be one of the largest sources of model error.
23
-------
LSPC Model Setup
3.5 Land Cover Representation
LSPC requires a basis for distributing hydrologic and pollutant loading parameters. This is necessary to
appropriately represent hydrologic variability throughout each watershed, which is influenced by land
surface and subsurface characteristics. It is also necessary to represent variability in pollutant loading,
which is highly correlated to soil characteristics and land practices.
Land cover data were obtained from the 2001 National Land Cover Database (NLCD). The NLCD GIS
coverage was derived from satellite imagery obtained around 2001 (i.e., late 1990s to 2003) and is the
most current and detailed land cover data known to be available at the time of this report (MRLC, 2007).
Each 30 m by 30 m (98-foot by 98)-foot pixel contained within the satellite image is classified according
to its reflective characteristics into 29 distinct land covers (see Homer et al., 2004). The NLCD land
cover data for the Tongue River are shown in Figure 3-10, and data are summarized in Table 3-3.
Table 3-3. NLCD land cover in the Tongue River watershed.
Area
Percent
Land Cover
Acres
Square Miles
of Watershed
Grassland
1,465,447
2,290
42.39%
Shrubland
1,132,203
1,769
32.75%
Evergreen Forest
645,990
1,009
18.69%
Woody Wetland
67,365
105
1.95%
Pasture and Hay
41,899
65
1.21%
Cropland
34,899
55
1.01%
Emergent Wetland
23,897
37
0.69%
Open Space, Developed
16,694
26
0.48%
Barren
10,240
16
0.30%
Deciduous Forest
8,407
13
0.24%
Low Intensity Development
5,118
8
0.15%
Open Water
2,763
4
0.08%
Medium Intensity Development
1,733
3
0.05%
High Intensity Development
375
1
0.01%
Total
3,457,031
5,402
100.00%
24
-------
LSPC Model Setup
SW»mf
_J Cogntiw
| 'ALaliti
~| Dwvrkipfld Cpon Sfuw:«
K] Diwtlopad. LO* IWMMV
| D»v*top*& llM«fi»AY
| Devalupsd High IrfBti&fty
~] Ednwn
¦ 0#e*irliMLR Fnrnw
| Eutfgrwn ror«*t
~ Stmjfe'SctUfc
O^utindiittortiiicMiKi
Pastw«JH«y
~ Cuttrvjlad Crop*
] Vwoedy Wottmcfc
Q Efn*rg«^ Hwba£«us W*l»«n®»
~ Tnt*fl Lm»4
Big Horn
County
WYOMIMO
Johnson
40 Miles
J I
Sheridan
County
MONTANA
Rosebud
County
Figure 3-10. NLCD land cover in the Tongue River watershed.
25
-------
LSPC Model Setup
It was important to evaluate the effects of impervious and irrigated land uses. These land uses are not
explicitly identified in the NLCD land cover, although certain NLCD classes (such as
"commercial/industrial/transportation") provide insight into the specific type of land use and cover (i.e.,
impervious cover). The following sections describe the modifications made to the NLCD land cover data
to represent the impervious and irrigated land uses.
3.5.1 Impervious Lands
LSPC requires that land cover categories be divided into separate pervious and impervious land units for
modeling. Separate model algorithms are then used to simulate major hydrologic and pollutant loading
processes for both land units - PERLND and IMPLND (respectively), which are further discussed in
Section 4.1. The percentage of impervious area for the four urban land use classes is shown in
Table 3-4. Values were obtained from the NLCD definitions of impervious cover (Homer et al., 2004).
The percent of impervious land per NLCD land use class was modified during the initial calibration of the
LSPC model, as the model is sensitive to the amount of impervious land in the watershed. Specifically,
the LSPC model routes water from impervious surfaces directly to the stream segment, and thus requires
as input the effective impervious area, rather than the total impervious area. This is typical for larger
cities with storm sewer systems and high intensity development, but does not necessarily describe the
land use classified by the NLCD data as "impervious surfaces" in the Tongue River watershed. It was
assumed that runoff from land areas with a small amount of impervious surface (i.e., small area of
developed open space and low intensity development) most likely flows on to adjacent pervious land
areas where infiltration can occur prior to discharge into a stream segment. These NLCD land uses were
set to 100 percent pervious surfaces. The medium and high intensity land use impervious areas were not
modified from the NLCD default values.
Table 3-4. Percentage of impervious cover for developed land use classes.
NLCD Land Use
NLCD Land Use Code
% Pervious
% Impervious
Open Space (developed)
21
100
0
Low Intensity Development
22
100
0
Medium Intensity Development
23
35
65
High Intensity Development
24
10
90
Total
100
100
3.5.2 Irrigated Lands
The NLCD land use coverage does not explicitly identify irrigated land (although some land use classes,
like cropland, can indicate the presence of irrigation). An analysis was conducted to specifically identity
the amount and location of irrigated land in the Tongue River watershed. Two different methods were
employed to identify irrigated land - one for irrigated land in Wyoming and another for irrigated land in
Montana.
The Wyoming Water Development Commission conducted a detailed inventory and analysis of water use
in the Tongue River watershed in Wyoming. As part of the inventory, aerial photos were obtained and
used to identify irrigated land. Irrigated land was delineated into a GIS coverage, and verified with field
surveys and interviews. Detailed information about the methodology used to identify irrigated land can
be found in the Basin Plan Technical Memoranda, "Appendix C - Irrigated Lands Mapping and Water
Rights Data," (WWDC, 2002e). A GIS coverage of irrigated land was obtained from the Powder/Tongue
River Basin Plan website, which identified the following irrigated land use classes:
26
-------
LSPC Model Setup
Full & Partial Supply Irrigation & Man-
Induced
Man-Induced Beneficial Use from Seepage
Full & Partial Supply Mix of Irrigation
Non-Irrigated Lands enclosed by Irrigated
Full Supply & Man-Induced Beneficial Use
Partial Supply & Man-Induced Beneficial Use
Full Supply Irrigation
Partial Supply Irrigation
Full Supply Irrigation with Development
Partial Supply Irrigation with Development
Idle Lands
Side Tributary - Kick-outs or Ditches
Idle Lands with Development
Spreader Dike - holds water
For modeling purposes, it was assumed that only land identified as having full or partial supply irrigation
actually receives supplemental water. The irrigated land was classified into two groups - full supply and
partial supply irrigation. The Powder/Tongue Basin Plan defines these as described below (WWDC,
2002e):
• Full supply irrigation - Typically receives a full water supply.
• Partial supply irrigation - Typically receives a reduced water supply due to limited water
availability or the inability to provide complete field coverage.
Both full and partial supply irrigated land uses were added as new land use classes in the LSPC model,
and the corresponding NLCD classified land was subtracted from the appropriate subwatershed. All
NLCD land uses within the identified irrigated land polygons were reclassified as either full or partial
supply irrigation. Based on the WWDC data, 54,402 acres of full supply irrigated land and 14,724 acres
of partial supply irrigated land are present in the Tongue River watershed in Wyoming.
No recent irrigated land analysis was available for the Tongue River watershed in Montana. The Montana
Agricultural Statistics Service reported that there were 55,000 acres of irrigated alfalfa hay in Big Horn,
Custer, Powder River, and Rosebud Counties in 2001 (MASS, 2002). However, only 22 percent of this
area is in the Tongue River watershed. The 1947 Water Resources Survey for Big Horn County, Montana
states that approximately 30,000 acres of land are irrigated from the Tongue River between the Tongue
River Reservoir Dam and the mouth (Montana State Engineer, 1947).
A GIS analysis was performed to obtain more detailed information about the amount and location of
irrigated land in the Tongue River watershed, Montana. The 2001 NLCD land use coverage was overlain
on the U.S. Farm Services Agency National Agricultural Imagery Program (NAIP) 1-meter resolution
images for Montana. Several irrigated fields in the Tongue River watershed were identified from the
aerial photos, and then overlain with the NLCD land use data. The NLCD data appeared to classify
irrigated land (as identified in the aerial photos) as one of three land use categories - pasture/hay (81),
cropland (82), and emergent herbaceous wetlands (95). This pattern was verified for 20 fields irrigated
from either the Tongue River or major tributaries to the Tongue River. Irrigated land along the Tongue
River was primarily classified as cropland or pasture/hay. Figure 3-11 illustrates this by showing the
aerial photo and NLCD land use for an irrigated field near Ashland, Montana. It was assumed that these
two land uses correspond to "full supply irrigation", as defined by the Powder/Tongue Basin Plan
(WWDC, 2002e). Irrigated land classified by the NLCD as "emergent herbaceous wetlands" primarily
occurred along the perennial tributaries (e.g., Hanging Woman Creek, Otter Creek, and Pumpkin Creek).
Figure 3-12 illustrates an irrigated field in the Otter Creek watershed that is classified as emergent
herbaceous wetlands. It was assumed that this land received a partial supply of irrigation water. Based
27
-------
LSPC Model Setup
on this analysis, 31,151 acres of irrigated land were identified in the Tongue River watershed, Montana
(Table 3-5). Figure 3-13 shows the location of the irrigated land in Montana and Wyoming.
Table 3-5, Total acres of irrigated land in the Tongue River watershed in Montana
Watershed
Full Supply
Partial Supply
Total
Pumpkin Creek
8,044
454
8,498
Otter Creek
1,868
24
1,892
Hanging Woman Creek
305
42
348
Tongue River (Other)
19,167
1,247
20,414
Total
29,384
1,767
31,151
Sources U S Farm Services Agency National Agrlciitural Imagery
Program (NAIP) 1 meter re&otirtinn images
-------
LSPC Model Setup
Sdutcts.. U S, Fdim Services Agency National Agjiguiiuial Imagery
FYogram {NAIF) 1-nwlet resolution images for Montana
I JLCD 2001 Land Use
Figure 3-12. Example area where irrigated land is classified as "emergent herbaceous
wetlands." Otter Creek near Ashland is shown.
29
-------
LSPC Model Setup
Custer
County
Powder River
County
Irrigated Land
| Full Supply liTigatk>n
| Kjiltsl Supply IritJJliOn
Stnrntt
_ Countfcs
Tiibnl Land
Tiinguf! Kivi^r 'JiiMd
County
Sheridan
County
MONTANA
WYOMIMO
Big Horn
County
Johnson
County
40 Milts
Figure 3-13. Location of irrigated land in the Tongue River watershed.
30
-------
LSPC Model Setup
The accuracy of this methodology was verified with the Wyoming Powder/Tongue Basin Plan irrigated
land coverage. The NLCD classified 55,385 acres of land as cropland, pasture/hay, or emergent
herbaceous wetlands in the Tongue River watershed in Wyoming. The Wyoming Powder/Tongue Basin
Plan indicated that there are 69,126 acres of irrigated land (WWDC, 2002e). This assessment suggests
that the current methodology underestimates irrigated land in the Tongue River watershed in Montana.
Additional data and field verification is recommended in the future to correct potential errors introduced
by this assessment.
31
-------
LSPC Model Setup
3.5.3 Final Modeled Land Uses
The NLCD land uses were grouped by similar characteristics to simplify the LSPC model and improve
model run time. The LSPC land use groups and associated NLCD classes are shown in Table 3-6. The
total acres per LSPC land use group are shown in Table 3-7.
Table 3-6. LSPC land use groupings and their associated NLCD land use classes-
LSPC Land Group
NLCD Land Use Name and Code
Barren
Bare Rock/Sand Clay (31)
Cropland (non-irrigated)
Cropland (82)
Forest
Deciduous Forest (41)
Evergreen Forest (42)
Full Supply Irrigation
NA- Identified Separately
Partial Supply Irrigation
NA - Identified Separately
Pasture/Grassland (non-irrigated)
Grasslands/Herbaceous (71)
Pasture/Hay (81)
Shrubland
Shrubland (52)
Urban Impervious
Open Space, Developed (21)
Low Intensity Development (22)
Medium Intensity Development (23)
High Intensity Development (24)
Urban Pervious
Open Space, Developed (21)
Low Intensity Development (22)
Medium Intensity Development (23)
High Intensity Development (24)
Water
Open Water (11)
Wetlands
Woody Wetlands (90)
Emergent Herbaceous Wetlands (95)
Table 3-7. Modeled land uses in the Tongue River watershed.
LSPC Group Name LSPC Group Number Total Acres Total Square Miles
Barren
1
10,166
16
Cropland
6
2,861
4
Forest
3
653,630
1,021
Full Supply Irrigation
9
83,787
131
Partial Supply Irrigation
10
16,491
26
Pasture/Grassland (non-
irrigated)
4
1,463,287
2,286
Shrubland
5
1,125,334
1,758
Urban Impervious
20
2,207
3
Urban Pervious
7
18,541
29
Water
11
2,744
4
Wetlands
8
77,988
122
Total
3,457,036
5,402
32
-------
LSPC Model Setup
3.6 Watershed Grouping
The LSPC model allows for variation of input parameters by land use (see Section 3.5). Input parameters
can also be varied by watershed groups. A watershed group is defined as a subset of modeling
subwatersheds that have similar soils and geology. The basis for watershed groups in LSPC was initially
defined by the NRCS hydrologic soils groups, as defined by the STATSGO data (see Section 2.1.9.2 of
the 2003 Status Report). However, initial calibration results showed that further group refinement was
needed, primarily because of the complex geochemistry in the Tongue River watershed. In LSPC,
groundwater concentrations are varied per watershed group (see Section 3.12.2), and hydrologic soil
groups did not always account for the large changes in groundwater quality concentrations from the
headwaters of the Tongue River to the mouth, or within the Tongue River tributaries. Groundwater in the
Tongue River watershed exhibits a pattern of increasing dissolved solids (i.e., TDS, calcium, magnesium,
and sodium) from the Bighorn Mountains to the mouth, with localized impacts from high salinity coal bed
aquifers (USDI, 2003). The initial watershed groups were refined to reflect changes in soil hydrology as
well as groundwater quality. Final watershed groups are described in Table 3-8, and are shown in Figure
3-14. These groups provided the primary basis for varying input parameters during model calibration
(further described in Sections 4.1 and 4.2 - Hydrology Calibration Methodology and Water Quality
Calibration Methodology).
Four additional watershed groups were set up to allow for calibration of the areas draining to CBM ponds
and stock ponds, and their associated downstream subirrigated areas (see Sections 3.7.2 and 3.8 for
additional details about the CBM and stock pond modeling). Groups 5 and 16 represent the areas
draining to CBM and stock ponds (i.e., the total watershed area of all of the ponds per subbasin) for the
Upper and Lower Tongue River watersheds, respectively. Groups 6 and 17 represent the areas that are
subirrigated downstream of CBM and stock ponds in the Upper and Lower Tongue River watersheds,
respectively.
Table 3-8. Watershed groups for the Tongue River LSPC model.
Region
Group Name
Group #
Corresponding Model Subbasins
Upper Tongue
River and
Tongue River
Reservoir
Bighorn Mountains -
Low Elevation
1
3025, 3029, 3044, 3047-3051, 3056, 3057, 3061-3063, 3071,
3072, 3074, 3076, 3082-3085, 3090-3108
Bighorn Mountains -
Middle Elevation
2
3030, 3031, 3034, 3035, 3036, 3053, 3054, 3058, 3059 ,
3064, 3065, 3073
Bighorn Mountains -
High Elevation
3
3032, 3033, 3052, 3055, 3060
Prairie
4
3000-3024, 3026-3028, 3037-3043, 3045, 3046, 3067-3070,
3075-3081, 3086-3089
Area Draining to CBM
and Stock Ponds
5
13002, 13006, 13007, 13008, 13010, 13012, 13013, 13014,
13017, 13019, 13021, 13022, 20504, 20604,
20804, 20902, 21904, 22004
Area Subirrigated by
CBM and Stock Ponds
6
30504, 30604, 30804, 30902, 31904, 32004
Lower Tongue
River
Tongue River Prairie -
Coal Region
11
1046-1057; 1079-1092; 1107-1113
Pumpkin Creek
12
1006-1025
Otter Creek
13
1058-1078
Hanging Woman Creek
14
1093-1106
Tongue River Prairie -
Non Coal Region
15
1001-1005; 1026-1045
Area Draining to CBM
and Stock Ponds
16
11104, 11105, 20111, 20113, 20115, 20211, 20212, 20213,
20312, 20315, 20413, 20414, 21812, 21815, 21911, 21914
Area Subirrigated by
CBM and Stock Ponds
17
30111, 30113, 30115, 30211, 30212, 30213, 30312, 30315,
30413, 30414, 31812, 31815, 31911, 31914
33
-------
LSPC Model Setup
D 5 10 20 Miles
Lj_l—i—L-i—L.i—l
| Tntxil Uinrt
SlfUillTE.
| Counties
LSPC Groups
S 3
4
11
12
13
= I H
IS
MONTANA
WVOMiwn
rJ
i
Crow
Figure 3-14. Watershed groups for the Tongue River LSPC model.
34
-------
LSPC Model Setup
3.7 Point Sources
There are a number of point source surface water discharges in the
Tongue River watershed, including wastewater treatment plants,
industrial facilities, coal mines, and coalbed methane. Data for these
point sources were compiled from a number of sources including
Wyoming DEQ, Montana DEQ, local industry, members of the
Modeling Committee, and EPA's Permit Compliance System (PCS).
Flows and pollutant concentrations from the point sources were input
to the LSPC model as time series and were combined with modeled
estimates of surface runoff and subsurface loads to simulate in-stream
water chemistry. Each point source category is summarized in the
following sections, and additional details are provided in Appendix A.
Figure 3-15 shows the location of all of the known point sources in the
Tongue River watershed as of September 30, 2006.
Point Sources
A detailed description
of all point sources in
the Tongue River
watershed is presented
in Appendix A.
35
-------
LSPC Model Setup
• CBM
¦ Coal Mtrw
» WMew«ter Tiealmenl
Wteter Treatment Plant
Streams
CouritlM
Tribal Lin-3
TonjLM fttver Watershed
N
MONTANA
WYOMING
D 5 10 20 Ml In
1 1 1 1 1 1 ' ' 1
Figure 3-15. Location of the permitted surface water discharges in the Tongue River watershed
(as of September 30, 2006).
36
-------
LSPC Model Setup
3.7.1 Wastewater Treatment Facilities
As of September 30, 2006, there were six permitted wastewater treatment plants that discharge in the
Tongue River watershed - City of Sheridan, WY; City of Ranchester, WY; City of Dayton, WY; Bighorn
Mountain KOA (located near Sheridan, WY), Burgess Junction Dump Station (at Burgess Junction, WY),
and the Powder Horn Ranch community (located southeast of Sheridan, WY). Three of the facilities
operate lagoons (Dayton, Ranchester, and Burgess Junction), and two of the facilities operate small
package plants (Powder Horn Ranch and Bighorn Mountain KOA) (WDEQ, 2003a; 2003b, 2004; 2005a;
2005b). The City of Sheridan operates a larger activated sludge (extended-aeration) facility with an
average discharge of 4.6 cubic feet per second (USEPA, 2006b). Plant discharge data for all six facilities
is summarized in Table 3-9 and in Appendix A.
Two additional wastewater treatment lagoons (City of Ashland, Montana and City of Birney, Montana)
are operated in the Tongue River watershed, but do not require permits from Montana DEQ because of
their size and because no direct discharge is anticipated (Personal Communications, USEPA, April 3,
2007). These two lagoons were not modeled.
Table 3-9. Summary of permitted wastewater discharge data in the Tongue River watershed
(average values for the available period of record are reported).
Facility
Outfall
Period of
Operation
Receiving
Waterbody
n2
Flow
(cfs)
sc
((jS/cm)
TDS
(mg/L)
SAR
Total
N
(mg/L)
Total
P
(mg/L)
Sheridan
WWTP
(WY0020010)
001
1988-
Present1
Goose
Creek
60
4.60
838
442
2.73
NA
3.7
Dayton
Lagoon
(WY0020435)
001
1988-
Present1
Tongue
River
57
0.12
NA
NA
NA
NA
NA
Ranchester
Lagoon
(WY0022161)
001
1990-
Present1
Tongue
River
61
0.16
NA
643
NA
NA
NA
Bighorn
Mountain
KOA
(WY0026441)
001
1990-
Present1
Goose
Creek
56
0.003
NA
NA
NA
NA
NA
Powder Horn
Ranch
(WY0036251)
001
2001 -
Present
Little Goose
Creek
56
0.03
NA
NA
NA
NA
NA
Burgess
Junction
Dump
Station
(WY0020931)
001
1981-
Present1
North Fork
Tongue
River
56
0.000
NA
NA
NA
NA
NA
002
56
0.063
NA
NA
NA
NA
NA
003
56
0.002
NA
NA
NA
NA
NA
City of
Ashland
Lagoon
None
1990-
Present1
Groundwater
NA
NA
NA
NA
NA
NA
NA
City of
Birney
Lagoon
None
1990-
Present1
Groundwater
NA
NA
NA
NA
NA
NA
NA
The actual construction date for the facility is unknown. The period of operation was determined by the first reported compliance monitoring in the
online PCS database, and may not reflect the actual construction date.
2The total number of samples varied per parameter. The largest sample count is reported.
NA - Not Available
37
-------
LSPC Model Setup
The six permitted treatment facilities were simulated in LSPC as monthly varying continuous loads based
on their reported discharge volumes and concentrations. Monthly average discharge and concentration
data were obtained from the EPA Permit Compliance System and from Wyoming DEQ (see Appendix
A). Few concentration data were available for the facilities; therefore monthly varying concentrations
were estimated for most parameters at most times. The assumptions for the estimated concentrations at
each facility are presented below:
City of Sheridan WWTP - Cation data (i.e., calcium, magnesium, and sodium) was available from
effluent toxicity testing on file with Wyoming DEQ and ranging from 1988 to 1997 (multiple hardcopy
reports, WDEQ, 1988-1997). Average monthly cation concentrations were calculated from the available
data (n=14) and input into the model as a monthly varying time series through September 30, 2006. Total
nitrogen concentrations were based on literature values and set at 14 mg/L (USEPA, 1997).
Total phosphorus concentrations were initially based on the one measured value reported by the facility
(3.7 mg/L on May 10, 1988). However, initial model results indicated that the implementation of this
value caused an unsatisfactory calibration of TP at the State Line USGS gage, particularly at base flow.
Rather than using the reported concentration or literature values, an effective phosphorus contribution was
calculated based on concentrations observed at two USGS gages located downstream of the treatment
plant. USGS Gage 06305500 (Goose Creek below Sheridan, Wyoming) and 06305700 (Goose Creek
near Acme, Wyoming) are 7.45 river miles apart, have paired phosphorus data, and both are downstream
of the treatment plant (gage 06305500 is located immediately downstream of the plant, gage 06305700
7.45 miles downstream). A simple mass balance for the two gages and wastewater treatment plant
(assuming an average flow of 4.6 cfs) indicated that the effective TP concentration from the treatment
plant outfall is 1.2 mg/L. This value was input into the LSPC time series. It should be noted that this
value may not represent the actual TP concentration in the treatment plant outfall. However, it represents
an effective concentration that takes into consideration the actual concentration discharged from the
outfall plus near-field uptake. Future measurements of TP in the treatment plant outfall plus an
assessment of algal uptake in Goose Creek could help to refine this assessment.
City of Ranchester Lagoons - Average monthly TDS was reported for the Ranchester Lagoon effluent,
but no other cations were reported. Calcium, sodium, and magnesium in the effluent were calculated
based on the relationship of the three cations to TDS as measured in the Tongue River at Dayton,
Wyoming (USGS gage 06298000), which is the closest gage to the Ranchester drinking water intake.
This method assumes that there is no change in the balance of cations as the water moves through the
public water system. No recent total phosphorus or total nitrogen data were available for the Ranchester
Lagoons. Total phosphorus concentrations were based on literature values and set at 5 (USEPA, 1997).
Total nitrogen concentrations were set at the observed ammonia concentrations and at the average
observed ammonia concentration where no ammonia data were available.
City of Dayton Lagoons - No recent cation, total phosphorus, or total nitrogen data were available for
the City of Dayton Lagoons. Effluent concentrations were therefore set to the same values as at the
Ranchester Lagoons (see above).
Bighorn Mountain KOA - No recent cation, total phosphorus, or total nitrogen data were available for
the City of Dayton Lagoons. Effluent concentrations were therefore set to the same values as at the
Ranchester Lagoons (see above).
Burgess Junction Dump Station - Data from the Burgess Junction Dump Station indicate that the
outfall from the lagoon system (outfall 001) does not discharge. Therefore, this outfall was not modeled
in LSPC. Outfalls 002 and 003 discharge water from two underdrain systems which do not necessarily
represent the effluent in the lagoon system. No relevant water chemistry data were available for this
38
-------
LSPC Model Setup
facility. Furthermore, the facility is only operational from May through September of each year.
Therefore, outfalls 002 and 003 were not modeled in LSPC.
Powder Horn Ranch - No recent cation, total phosphorus, or total nitrogen data were available for the
Powder Horn Ranch. Effluent concentrations were therefore set to the same values as at the Ranchester
Lagoons (see above).
3.7.2 Coalbed Methane
Coalbed methane (CBM) is methane gas found in coal seems and their associated aquifers (Keith et al.,
2003). The methane is generally held in the coal seams by pressure from the local aquifers. Pumping
water out of the aquifers releases the pressure from the coal seams thereby allowing the methane to rise to
the surface. Pumped water from the coal aquifer can be directly discharged to a stream, stored in a
containment unit, treated, used for irrigation or stock watering purposes, or re-injected. In the Tongue
River watershed, most water from CBM wells is pumped into a series of outfalls that discharge to ponds.
One outfall (or pond) can contain effluent from one to hundreds of individual CBM wells.
Montana DEQ and Wyoming DEQ require NPDES permits for CBM facilities discharging to surface
waterbodies. Permits are issued for a specified number of outfalls (not individual wells) at specific
locations. Facilities are required to monitor effluent from the outfalls and submit discharge monitoring
reports (DMRs) to the respective agencies, generally on a monthly basis. Effluent monitoring occurs at
the outfall discharge point before it enters a pond or stream, although some facilities are also required to
monitor downstream impacts at compliance points. Ponds and pond discharges are rarely considered
regulated NPDES compliance points.
The following sections summarize the CBM facilities, outfalls, and ponds located in the Tongue River
watershed. Measured discharges and water chemistry at the CBM outfalls are discussed in Section
3.7.2.1, while ponds, pond design, and pond processes are discussed in Section 3.7.2.2. The methods and
assumptions used to model the outfalls and ponds in LSPC are discussed in Section 3.7.2.3. Additional
information about the permitted CBM facilities is provided in Appendix A.
3.7.2.1 CBM Outfalls
CBM permits and DMR data were obtained from Montana DEQ and Wyoming DEQ. Data for Montana
CBM facilities were provided by Montana DEQ in Microsoft Excel spreadsheets. The spreadsheets
contained DMR data for three facilities having 18 CBM outfalls, all of which have direct discharges to the
mainstem Tongue River (MDEQ, 2006c). Figure 3-16 shows the location of the 18 direct discharges in
Montana. Data were available from January 5, 2000 to October 31, 2006. Additional details about the
Montana discharges are provided in Appendix A.
The Montana Board of Oil and Gas (MBOG) indicates that Fidelity Exploration and Production Company
(Fidelity) began CBM operations in the Tongue River watershed in March 1997 (MPDES Permit
MT0030457) (MBOG, 2006). At the time, no MPDES permit was required, and Montana DEQ did not
issue a permit until June 16, 2000 (MDEQ, 2000). Because of this, no DMR data was submitted between
March 1997 and January 2000. Several assumptions were therefore made about the timing and amount of
discharge occurring between March 1997 and January 2000 at the Fidelity facility:
• According to DMR records, outfalls 001 through 009 were discharging in January, 2000. It was
assumed that these 9 outfalls were also discharging from April 1, 1997 to January 1, 2000.
39
-------
LSPC Model Setup
• It was assumed that flows from April 1, 1997 to January 1, 2000 were equal to the average long-
term reported flow from each outfall, using all available data.
• It was assumed that water chemistry concentrations from April 1, 1997 to January 1, 2000 were
equal to the long-term average concentration from each outfall, using all available data.
It is acknowledged here that this is most likely an overestimate of the actual amount of discharged water
because new wells were drilled over time and not all at once. However, it is considered a conservative
estimate for this analysis.
Wyoming DEQ provided a Microsoft Access database (dated January 23, 2007) of data for CBM facilities
in the Tongue River watershed, Wyoming. The database contained over one million records of CBM
DMR data dating from March 1, 1999 to January 1, 2007. Wyoming DEQ personnel indicated that the
database contains information for all of the permitted CBM facilities in the Tongue River watershed
through January 1, 2007 (Personal Communications, Wyoming DEQ, March 19, 2007). The database
contained data for 420 CBM outfalls that discharge to ponds and 7 outfalls that discharge directly to
streams. An estimated 222 of the outfalls discharge to off-channel ponds, and 198 discharge to on-
channel ponds (WDEQ, 2007) (further discussed in Section 3.7.2.2). Of the 427 outfalls, only 410 had
reported flows. The other 10 outfalls either had no reported data or had reported discharges of zero.
Wyoming personnel indicated that this was common, as outfalls are often permitted and built but not used
(Personal Communications, Wyoming DEQ, March 19, 2007). Additional information about CBM
facilities located in Wyoming is provided in Appendix A, and Figure 3-16 shows the location of the
known CBM outfalls as of September 30, 2006
40
-------
LSPC Model Setup
20 Mite
_l
Slnjinrro
CoundM
Tnfanl Lqnd
_| T«nju« Rr.»r vVswrcrwd
MONTANA
WYOMING
CBMOuHalls
* Direct Discharg*
¦ 0*-Chnnrwl Rmwjrvoirs
• On-Charmal Retervuirs
Figure 3-16. Location of known CBM outfalls in the Tongue River watershed as of September
30, 2006.
41
-------
LSPC Model Setup
A summary of the reported flows for all Wyoming and Montana CBM outfalls is presented in Figure 3-
17. Flows from the Montana CBM facilities began in 1997 and remained relatively constant with an
average flow of 2.3 cubic feet per second. The average SC and SAR from the Montana facilities was
1,935 (iS/cm and 41.1, respectively. CBM facilities were first constructed in Wyoming in 1999, and
discharges have increased overtime to apeak of 13.0 cubic feet per second in November of 2004. The
average SC and SAR from the Wyoming facilities was 1,900 (iS/cm and 38.1, respectively. The total
yearly effluent volumes from Wyoming CBM outfalls within the Tongue River watershed (as reported by
the Wyoming Oil and Gas Commission) are also displayed in Figure 3-17 for comparison.
-Montana —a—Wyoming Oil and Gas Commission —x— Wyoming
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
Figure 3-17. Summary of total monthly CBM discharge from Montana and Wyoming facilities
(all available data from 1997-2006)
42
-------
LSPC Model Setup
3.7.2.2 CBM Ponds
Most CBM effluent is discharged into ponds. Ponds are designed to contain, infiltrate, and evaporate
discharge water. At times, the ponds may have a surface water discharge from overtopping or pond
failure. Wyoming DEQ allows both on-channel and off-channel CBM ponds in the Tongue River
watershed (Montana currently does not have any MPDES permitted CBM ponds in the Tongue River
watershed). On-channel ponds are defined by Wyoming DEQ as "an impoundment sited on or within a
distance of 500 feet of a designated water feature as defined on a United States Geologic Survey (USGS)
1:24,000 scale topographic map, or within a distance of 500 feet of the floodplain or mapped alluvium
(including alluvial mixtures) of a stream system as defined on a Wyoming State Geological Survey
1:100,000 scale surficial geology map," (WDEQ, 2006a). Off-channel ponds are defined as any pond not
meeting the requirements of an on-channel pond. Pond sizing and design varies per CBM facility.
Designs can vary from complete discharge of CBM effluent to complete containment of all CBM effluent
plus water from up to a 100-year storm event (Various permits, WDEQ, 2000-2006). Several examples
from the CBM NPDES permits are presented below to illustrate the variety of permitted pond designs:
• Permit WY0038628 - All eleven outfalls authorized under this permit will discharge into on-
channel reservoirs as described in Table 1 (Part I.B.I 2) of the permit. While the reservoirs
receive effluent from this CBM facility, the permit does not require containment of the effluent
within the reservoirs. This permit anticipates discharge of effluent to Goose Creek, with the
limits specified below. In addition, neither the reservoirs nor their spillways will constitute
regulated discharge points under this permit (WDEQ, 2005c).
• Permit WY0051811 - The permittee has committed to containment of all CBM effluent within a
series of on-channel reservoirs. The permittee is required to contain all effluent within the
reservoirs, and may not discharge except during periods of time in which stormwater runoff
enters the reservoir, causing it to overtop and spill (WDEQ, 2005d').
• Permit WY0052345 - The permittee has submitted information to demonstrate that all produced
effluent from this facility will be contained in 7 on-channel reservoirs. The water budget for this
facility confirms that these reservoirs will have sufficient capacity to contain all of the estimated
effluent from this facility as well as storm runofffrom up to a 100-year/24-hour event (WDEQ,
2005e).
• Permit WY0046540 - Under this permitting option, the produced water is immediately
discharged to a confined, off channel pit, stock pond or other man made containment unit (class
4 water) that will not flow into any other waters of the state. The permittee has demonstrated
through a water balance study that, considering CBMwell inflow, natural precipitation,
evaporation and infiltration, the off channel containment unit will be adequate to contain all
CBM discharge water and stormwater up to a 25 year 24 hour event. In addition, the permittee
has committed to the complete containment of all discharged water (WDEQ, 2005J).
• Permit WY0052043 - The permittee has committed to, and will be required to, contain all of the
CBM produced water and stormwater runoff up to and including the runofffrom a 100 year, 24
hour storm event in the playa. The permittee has also submitted a water balance which
demonstrates that, considering CBMwell production, natural precipitation, and evaporation
and infiltration, and other facilities that these wells are linked to in the immediate vicinity, the
playa will be adequate to contain all estimated CBM discharge water discharged to it. To allow
the permittee the maximum flexibility in the use of the playa lake for CBM produced water
containment, no flow limit will be established in this permit. However, the permittee has
committed to the complete containment of all discharged water within the topographically closed
basin (WDEQ, 2004c).
43
-------
LSPC Model Setup
A literature review was conducted to better understand the hydrological and chemical processes occurring
in the CBM ponds. The physical setup for CBM ponds is similar to stock ponds (and indeed some CBM
ponds are former stock ponds): they tend to be located on ephemeral channels and discharge directly to
downstream perennial waters only during storm events. The major difference is that CBM ponds receive
near continuous inputs of produced water, while stock ponds primarily rely on natural inflow events.
A major effect of CBM ponds is to raise the water table in the alluvial aquifer. Payne and Saffer (2005)
found that infiltration was a major part of the water budget; however, this infiltration occurred to a much
greater extent in the ephemeral stream channels than in the ponds themselves. Infiltration from the ponds
was limited by a variety of factors, including compaction, settling of fine-grained material to the pond
bottom, and presence of a mounded water table under the ponds. In 2003, analysis over a travel distance
of about 1500 meters partitioned the flow into 46 percent infiltration plus transpiration, 44 percent surface
flow, and 10 percent direct evaporation (Payne and Saffer, 2005). However, the balance will shift further
toward infiltration and away from surface flow with additional travel distance. Field calculated estimates
of infiltration ranged from 0.032 to 0.132 cfs/mi. The authors judged that a typical industry value for
infiltration losses of about 0.1 cfs/mi may be a slight overestimate. It is important to note that these
estimates of "infiltration" actually refer to total losses, and so are the sum of infiltration and
evapotranspirative losses, which may be high due to the presence of riparian vegetation.
Water discharged over the dams goes into what were ephemeral stream channels, where much of it
infiltrates into the alluvial aquifer or is lost to evapotranspiration (Payne and Saffer, 2005). Water also
exfiltrates through the bottom of the pond into the alluvial aquifer. The resulting increase in water
volume can establish perennial flow and wetlands in the receiving channel, but typical operation appears
to result in a condition in which continuous surface flow does not reach natural perennial streams (Payne
and Saffer, 2005). However, during storm events significant amounts of direct surface flow may proceed
through the system.
Flow associated with CBM ponds reaches perennial stream reaches by four potential pathways: direct
overflow (Op), direct infiltration (Ip), interflow (Ic), and groundwater (Gc). Each of these pathways will
have different water chemistry characteristics. Wheaton and Brown (2005) make a number of points
regarding these issues in relation to CBM ponds. Experiments reported by Wheaton and Brown first
indicate that the water chemistry of the groundwater associated with CBM ponds is more strongly related
to the chemistry of the receiving hydrogeology than to the characteristics of the source water. CBM-
produced water is typically high in bicarbonate, very low in sulfate, has low concentrations of calcium
and magnesium, and relatively high concentrations of sodium (and thus a high SAR). In general,
infiltration into the shallow groundwater results in increases in Ca and Mg due to dissolution of calcium
and magnesium carbonates in the formerly unsaturated overburden, potentially followed by exchange of
Ca and Mg for Na where sodic shales are contacted. Wheaton and Brown (2005) conclude that "the
increase in TDS is likely associated with the dissolution of minerals present in the overburden material
and is not directly associated with the [CBM] production water. In fact, similar TDS levels would be
expected if a surface water source, or rainwater, had been pumped into the pond." The increases in TDS
during infiltration are largely due to dissolution of Ca and Mg, while Na seems to be approximately
conservative.
A second point is that the effect of increased groundwater flows will change over time as available salts
are flushed from the system by exchange of one or more porewater volumes. In large aquifers, this effect
will take place over very long time frames; however, it may be an important consideration in small,
confined alluvial aquifers or along the flowpaths through overburden for infiltration from ponds.
Preliminary conclusions by Wheaton and Brown (2005) suggest that water chemistry immediately
beneath an infiltration pond can be approximated by a simple mixing model of pond water and ambient
44
-------
LSPC Model Setup
ground water. However, as flowpaths become longer, the resulting groundwater discharge will become
more and more similar to ambient groundwater. The primary impact of ponds on baseflow water
chemistry would thus seem to be an increase in groundwater flow, rather than a significant change in the
ionic concentrations. Impacts from individual ponds are likely to be highly variable, dependent on local
geology, and hard to generalize.
3.7.2.3 Modeling CBM Outfalls and Ponds
The DMR data obtained from Wyoming DEQ and Montana DEQ for CBM outfalls (as discussed in
Section 3.7.2.1) were entered into the LSPC model as monthly varying time series based on the reported
flow and water chemistry concentrations. Average monthly flows, salinity, and SAR were reported for
most Montana and Wyoming CBM outfalls. However, additional parameters required for the LSPC
modeling (i.e., calcium, magnesium, sodium, total nitrogen, and total phosphorus concentrations) were
not always reported. Where missing, these parameters were set to the average concentration for that
particular outfall or facility. If no data were reported for a facility, data from nearby CBM facilities were
used to supplement the missing information. No total nitrogen (TN) or total phosphorus (TP) data were
available for any of the Wyoming CBM wells. Wyoming DEQ personnel indicated that no nutrient
monitoring has ever been required of CBM facilities (Personal Communications, Wyoming DEQ, March
20, 2007). The average TN and TP concentrations from the 16 Fidelity outfalls (Montana MPDES
permit MT0030457) were therefore used for all Wyoming CBM outfalls (1.40 and 0.10 mg/L,
respectively). A summary of all modeled CBM outfall data are presented in Table 3-10.
Table 3-10. Summary of modeled CBM flow and concentrations.
Parameter
Average
Min
Max
Flow (cfs)
0.18
0.00
5.55
Calcium (mg/L)
7.8
0.2
131.9
Magnesium (mg/L)
4.0
0.1
114.3
Sodium (mg/L)
476.7
4.3
1,428.0
Total Nitrogen (mg/L)
1.49
0.10
3.30
Total Phosphorus (mg/L)
0.10
0.05
0.12
In all, 428 CBM outfalls had at least one month with a reported flow, and 428 time series (18 in Montana
and 410 in Wyoming) were input into the LSPC model. Discharges from the outfalls were either routed
into a stream or pond, depending on the permit. Nineteen of the outfalls were direct discharges, and 409
discharges were routed into ponds. It was assumed that one outfall discharged to one pond; therefore 409
ponds were input into the LSPC model.
As discussed in Section 3.7.2.2, there are a variety of CBM pond designs. Ponds can be located on or off-
channel, contain 0 to 100 percent of CBM effluent, and contain 0 to 100 percent of flow from major storm
events. Because of the varying pond designs, several assumptions were made to simplify pond modeling
processes. Each pond was assumed to be rectangular shaped with a triangular profile, and a full pool
depth of 10 feet. It was also assumed that infiltration from all ponds is 15 mm/day (based on literature
reviews: Payne and Saffer, 2005; Neff, 1980). These assumptions (along with the pond design, CBM
discharge, and precipitation data) provided the information for calculating pond sizes.
For modeling purposes, the ponds were classified into three groups based on the frequency of overflows.
The groups and number of ponds per group are described below and are shown in Figure 3-18:
45
-------
LSPC Model Setup
• Group 1 - "Complete Containment" - Group 1 contains both on channel and off channel ponds
that are designed for complete containment of CBM effluent plus water from storm events. This
assumes that infiltration and any resulting changes to the groundwater are the same from either on
or off channel ponds. This assumption is appropriate because of the scale of the LSPC model
(i.e., large, watershed scale model that is not appropriate for field scale assessments) (272 ponds).
• Group 2 - "No Containment" - Group 2 consists of CBM ponds that are designed to simply
"dissipate energy" from CBM discharges (25 ponds). Ponds in this group are not designed to
store CBM effluent or storm event runoff.
• Group 3 - "Partial Containment" - This group consists of on-channel ponds that are designed
for complete containment of CBM effluent, but allow overflow during storm events (112 ponds).
Each group was modeled with a different set of rules and assumptions. The primary difference among the
groups was the sizing of the ponds. Group 1 ponds were sized in the LSPC model large enough to
capture all effluent and storm events up to a 100-year flood, Group 2 were sized to contain flow from one
day of CBM effluent (based on best professional judgment), and Group 3 were designed to contain 100
percent of the CBM effluent, but overflow with storm events.
Individual CBM ponds were not modeled in LSPC; rather, ponds and CBM discharges to ponds were
lumped per modeling subbasin (as defined in Section 3.2). One surrogate pond was created for each Pond
Group, resulting in zero to three surrogate ponds per modeling subbasin (depending on the type of ponds
present in the subbasin). The total volume of effluent discharging to a specified group in a specified
subbasin was summed to create the surrogate pond. LSPC currently does not allow for varying pond sizes
over time, therefore the total volume of CBM ponds as of September 30, 2006 was used to compute pond
sizes. It is recognized that this introduces error into the model, particularly between 1999 and 2006 when
CBM experienced rapid growth in the watershed.
It should be noted that some modeling subbasins had no CBM outfalls, while others had one or more
particular group of ponds. The number of surrogate ponds per subbasin ranged from zero to three based
on the type of ponds present in the subbasin. The total number of ponds per modeling subbasin and Pond
Group is shown in Table 3-11. The CBM time series were routed into the surrogate pond for each
subbasin, and water was allowed to infiltrate, evaporate, and overflow.
46
-------
LSPC Model Setup
Northern
Cheyenne
Crow
MONTANA
WYOMING
MONTANA
CBM Ponds
* Unoup 1 - Complete Containment
* Group ? - Mo CooMin
-------
LSPC Model Setup
Table 3-11. Summary of modeled CBM outfalls per LSPC subbasin.
LSPC
Subbasin
ID
Subbasin Name
Direct
Discharge
Group 1
(Complete
Containment)
Group 2 (No
Containment)
Group 3
(Partial
Containment)
Total
1104
Hanging Woman Creek
0
10
0
0
10
1105
Hanging Woman Creek
0
8
0
0
8
1106
Hanging Woman Creek
0
2
0
0
2
1112
Tongue River
1
0
0
0
1
3002
Badger Creek
0
54
0
0
54
3005
Tongue River
1
0
0
0
1
3006
Tongue River
4
1
0
3
8
3007
Prairie Dog Creek
1
68
3
73
145
3008
Prairie Dog Creek
0
50
0
16
66
3009
Prairie Dog Creek
0
0
0
0
0
3010
Prairie Dog Creek
0
16
0
20
36
3012
Prairie Dog Creek
0
9
0
0
9
3013
Prairie Dog Creek
0
2
0
0
2
3014
Tongue River
3
2
0
0
5
3016
Tongue River
9
4
0
0
13
3017
Tongue River
0
13
0
0
13
3019
Tongue River
0
32
9
0
41
3021
Goose Creek
0
1
10
0
11
3022
Goose Creek
0
0
3
0
3
Total
19
272
25
112
428
On-channel ponds have an associated watershed area upstream of the pond. This area was estimated by
delineating watersheds for a subset of ponds in a GIS, which yielded an average watershed area of 120
acres per on-channel CBM pond. For modeling purposes, it was assumed that this area is composed of 50
percent grasslands and 50 percent shrubland (based on the GIS analysis). The CBM pond watershed area
was subtracted from the modeling subbasin watershed and routed directly into the surrogate ponds as
dictated by the location of the ponds and their associated permitted design.
Downstream of the ponds, it was assumed that infiltrated water is added to the active groundwater
storage, which then raises the water table and provides subirrigation (i.e., irrigation provided by the
presence of a higher than normal water table) to plants downstream of the ponds (Payne and Saffer, 2005;
Wheaton and Brown, 2005). The infiltrated water is also subject to loss to deep aquifers and evaporation,
and excess water is discharged back to the stream. Based on the research of Wheaton and Brown (2005),
it was assumed that water infiltrated from the CBM ponds equilibrates with the ambient groundwater.
Therefore, the water chemistry of the subirrigated area was set to the same interflow and groundwater
concentrations as the rest of the watershed group, which varied depending on the location within the
Tongue River watershed (see Section 3.12.2).
This subirrigated area (like the area subirrigated by stock ponds - see Section 3.8) was assumed to follow
the stream channel for a maximum distance of 1,000 meters, subirrigating a 30 meter wide area of land
downstream of the ponds. Based on a GIS analysis, it was assumed that all areas subirrigated by CBM
ponds were composed of NLCD land use classes 95 (emergent herbaceous wetlands) and 71 (grassland).
Using these assumptions, the subirrigated area was discretely represented in the LSPC model, and water
infiltrated from CBM ponds was routed into this area. It is recognized that this approach is an
oversimplification of pond infiltration and subirrigation processes, and that actual subirrigation amounts
48
-------
LSPC Model Setup
will vary based on the size of the pond, infiltration rate, soils, vegetation, groundwater, and in-stream
flow.
Overflow from stock ponds and type 3 CBM ponds are assumed to go directly to the stream segment.
This is because overflow from these ponds are mostly driven by storms. If a storm is large enough to
cause one of these types of ponds to overflow, then it is reasonable to assume that all pervious land
surfaces are saturated, with little opportunity for additional infiltration. However, Type 2 CBM ponds,
which are designed for continuous overflow, discharge to the sub-irrigated land surface. This water is
subject to additional infiltration because the timing of these controlled discharges does not necessarily
follow the timing of wet weather events that would otherwise saturate the ground.
Figure 3-19 presents a summary of the CBM pond processes described in the previous paragraphs and
modeled in LSPC.
Flow From Upstream Area
(On-Channel Ponds Only)
Evaporation Precipitation
\ ** \ \ \ \
Flow From CBM
Outfalls
* y y * * Y
ronu uveniuws
(Groups 2 and 3 Only)
/
Pond Overflows
Soil Substrate
Pond Storage Volume
Infiltrated Water to Downstream
Subirrigated Area
Figure 3-19. CBM pond processes modeled in LSPC.
49
-------
LSPC Model Setup
3.7.3 Coal Mines
Three strip coal mines are currently operating in the Tongue River watershed - Spring Creek, Decker
East, and Decker West. The Bighorn Mine Coal Mine was operational until 2000, and is now undergoing
reclamation. Additional details about all four strip mines are included in Appendix A.
Only two outfalls were modeled in LSPC - Decker West outfall 007 and Decker East outfall 002. Both
outfalls discharge to the Tongue River Reservoir. According to the discharge monitoring reports (DMRs)
and information from the Montana DEQ Industrial and Energy Minerals Bureau (IEMB), these are the
only two outfalls that continuously discharge (see Appendix A). The remaining outfalls have intermittent,
unpredictable discharges and therefore were not included in the LSPC model. None of the Spring Creek
outfalls discharge on a regular basis. Decker West outfall 007 and Decker East outfall 002 were modeled
as monthly varying continuous loads based on their reported discharge volumes and concentrations
(obtained from the online PCS database and from the IEMB). Summary statistics from the time series is
presented in Table 3-12. Total nitrogen data for both outfalls was calculated as the sum of the ammonia
plus nitrate plus nitrite data.
Table 3-12. Summary of modeled coal mine outfalls and data.
Outfall
Parameter
Average
Min
Max
Decker West #007
Flow (cfs)
2.25
0.03
26.90
Calcium (mg/L)
158.8
58.3
242.1
Magnesium (mg/L)
117.5
53.6
155.0
Sodium (mg/L)
297.2
90.8
510.2
Total Nitrogen (mg/L)
1.92
0.03
7.56
Total Phosphorus (mg/L)
0.05
0.01
0.23
Decker East #002
Flow (cfs)
1.39
0.69
2.37
Calcium (mg/L)
72.9
27.0
102.0
Magnesium (mg/L)
60.6
44.7
76.7
Sodium (mg/L)
538.9
296.2
791.4
Total Nitrogen (mg/L)
1.88
0.43
3.45
Total Phosphorus (mg/L)
0.04
0.01
0.18
50
-------
LSPC Model Setup
3.8 Stock Ponds
Stock ponds are small human-made water impoundments that serve as water supply for livestock and
crops. Stock ponds are found throughout the Tongue River watershed (Figure 3-21) and have the ability
to affect hydrologic processes in the following ways:
• Delay response to storms by capturing runoff and then releasing via overflow.
• Reduce overall stream flows due to loss of water from evaporation or use of water.
• Reduce downstream sediment loads through settling.
• Infiltrate water, thereby increasing downstream baseflow, and potentially increasing downstream
dissolved solid concentrations.
In the LSPC model, stock pond setup was similar to the CBM ponds. One surrogate pond was created for
every subbasin with a stock pond. Table 3-13 summarizes the LSPC modeling subbasins with stock
ponds and the total volume per subbasin. The surrogate pond was sized to be the sum of the volumes of
the stock ponds in the subbasin based on information provided by Montana DNRC and the Wyoming
State Engineers Office (DNRC, 2003; WWDC, 2002c). Sizing and infiltration assumptions were similar
to the CBM ponds:
• The pond is rectangular with a triangular profile.
• The full pool depth is 10 feet.
• Each pond has a trapezoidal weir to accommodate overflows.
• Infiltration from all ponds is 15 mm/day (based on literature reviews: Payne and Saffer, 2005;
Neff, 1980).
• The length to width ratio of the pond is 4:3.
The upstream drainage area was calculated in a
GIS for a subset of stock ponds. Based on the
subset of data, the relationship between the known
pond total volume and watershed area was
calculated (Figure 3-20). The upstream watershed
area was assumed to be composed of 50 percent
shrubland and 50 percent grassland.
It was assumed that stock pond infiltration raises
the water table and provides subirrigation to plants
downstream of the ponds. This subirrigated area
was assumed to follow the stream channel for
1,000 meters, subirrigating a 30 meter wide area of
land. It was assumed that the 30,000 square meter
area was composed of grassland and emergent
herbaceous wetlands.
-------
LSPC Model Setup
Table 3-13. Total volume of stock ponds per modeling subbasin
Subbasin
Total Stock
Pond Volume
(Acre-feet)
Subbasin
Total Stock
Pond Volume
(Acre-feet)
Subbasin
Total Stock
Pond Volume
(Acre-feet)
Subbasin
Total Stock
Pond Volume
(Acre-feet)
3002
83
3068
1,308
1031
87
1070
238
3003
28
3069
115
1032
85
1071
37
3004
105
3075
55
1033
96
1072
101
3007
245
3077
6
1034
42
1073
15
3008
351
3078
474
1035
142
1074
50
3009
450
3079
0
1037
15
1075
41
3010
391
3080
6
1038
20
1076
16
3011
101
3086
146
1039
24
1077
27
3012
241
3087
46
1040
31
1078
18
3013
110
3088
3
1042
1
1079
16
3015
165
1002
86
1043
176
1081
34
3017
44
1003
57
1044
89
1083
103
3018
26
1005
126
1045
48
1084
310
3019
271
1007
16
1046
1
1085
25
3020
62
1008
15
1047
1
1087
113
3021
23
1009
4
1049
100
1089
1
3022
1,603
1010
46
1050
43
1091
28
3024
58
1011
79
1051
194
1092
1
3026
41
1012
66
1052
27
1094
115
3027
390
1015
46
1053
10
1095
22
3028
253
1016
39
1057
16
1096
107
3031
16
1018
24
1059
20
1097
30
3034
70
1019
26
1060
143
1098
18
3037
57
1020
26
1061
223
1101
25
3038
33
1022
24
1062
103
1102
33
3040
121
1023
26
1063
110
1106
1
3041
10
1024
48
1064
31
1107
36
3043
1
1025
48
1065
47
1108
45
3045
6
1026
62
1066
136
1109
186
3046
4
1027
1
1067
96
1110
126
3066
15
1029
34
1068
167
1111
52
3067
214
1030
40
1069
166
1112
16
1113
1
52
-------
LSPC Model Setup
CUi
Stock Ponds
Streams
! Counties
Tribal Land
r 1 Tongue Rrvei Vteterstied
Powder River
County
WYOMIMO
Rosebud
County
Sheridan
County
Big Horn
County
D 5 10 20 Miles
I i I
MONTANA
Johnson
County
Figure 3-21. Location of stock ponds in the Tongue River watershed.
53
-------
LSPC Model Setup
3.9 High-Altitude Reservoirs and Diversions
Small lakes and reservoirs are prolific throughout the Bighorn Mountains, particularly in the Goose Creek
watershed (southeast portion of the Tongue River watershed in the Bighorn Mountains). Several of the
reservoirs store mountain snowmelt and are regulated to provide irrigation and drinking water to
downstream users. A series of dams, diversions, and ditches exist to store and move irrigation water as
needed. Figure 3-22 provides a simplified schematic of the reservoirs and diversions in the Goose Creek
watershed.
BIG GOOSE CREEK
WATERSHED
Dome Lake
Reservoir
(Subbasin 3063)
Cross Creek
Reservoir
(Subbasin 3055)
Bighorn
Reservoir
(Subbasin
3053)
\°0
*
Pefalto - Mountain
Supply Ditch
Park Reservoir
Diversion twitch
Granger
Reservoir
(Subbasin
3036)
Little Goose Creek
(Subbasin 3036)
-VAfcO "
G0°se
Gtee
Willets
West Fork Little Goose Creek ^
Reservoir
(Subbasin 3035)
(Subbasin 3035)
WATERSHED
Figure 3-22. Simplified schematic of high-altitude reservoirs and diversions in the Goose Creek
watershed (not to scale).
The five major high-altitude reservoirs in the Goose Creek watershed were included in the LSPC model -
Bighorn, Dome Lake, Park, Sawmill and Twin Lakes Reservoirs. The remaining smaller reservoirs (i.e.,
Willets, Last Chance, Martin, Cross Creek, and Granger Reservoirs) were not modeled because
information indicates that the reservoirs are too small to have a substantial impact on watershed
hydrology (i.e., Last Chance and Granger Reservoirs have a 90 and 146 acre-foot capacity, respectively),
reservoirs are not always filled (e.g., Willets, Martin, and Granger Reservoirs), or reservoir operation is
captured in a more downstream reservoir (i.e., Cross Creek and Last Chance Reservoirs) (USFS, 2005)
(SEO, 2001;2002;2003;2004;2005).
The five reservoirs were established in the LSPC model based on the reported reservoir storage, depth,
and outlet information summarized in the Powder/Tongue River Basin Plan, Technical Memoranda J -
Storage Operation and Description (WWDC, 2002g). Model input parameters are summarized in Table
3-14.
54
-------
LSPC Model Setup
Table 3-14. Summary of modeled high-altitude reservoirs and their associated input parameters.
Reservoir Name
LSPC Subbasin
Year Completed
Dam Height (ft)
Modeled Storage (acre-
feet)
Bighorn
CO
LO
O
CO
1909
29
4,629
Dome Lake
3063
1907
23
1,506
Park
3050
1933
71
10,362
Sawmill
3057
1960
31
1,275
Twin Lakes
3059
1937
67
4,042
The high altitude reservoirs are all managed to store mountain snowmelt and supply water to downstream
users when needed. Managed releases were simulated in LSPC based on measured release data, diversion
data, and high altitude USGS gages. The simulation of releases for each reservoir is described below.
• Bighorn Reservoir - Modeled releases from this reservoir were based on the times series of data
for the Mountain Supply Diversion (obtained from the Wyoming SEO and Tongue/Powder Basin
Plan) (WWDC, 2002g). An average of 25 cfs is released from the reservoir (subbasin 3053)
between June and September of each year. Water is diverted into the Little Goose Creek
watershed (subbasin 3036) via the Mountain Supply Ditch. Any overflows were allowed to spill
over and were routed downstream to subbasin 3050.
• Dome Lake - A daily time series of releases from 2001 to 2005 was available from the Wyoming
SEO. This time series was directly used to model releases from Dome Lake (from subbasin 3063
to 3062). The average monthly flow was calculated from the daily time series and applied to the
years 1990-2000. An average of 12 cfs is released between June and September of each year.
Any overflows were allowed to spill over and were routed downstream to subbasin 3062.
• Park Reservoir - The Park Diversion Ditch diverts water from the Park Reservoir (subbasin
3050) into the Little Goose Creek watershed (subbasin 3036). Wyoming SEO data state that an
average of 16 cfs is diverted into this ditch between June and September of each year. The
reservoir operator is also required to maintain a minimum of 4.5 cfs of water in the East Fork Big
Goose Creek during the irrigation season (WWDC, 2002g). This withdrawal was routed out of
the reservoir and into subbasin 3048. Any overflows were allowed to spill over and were routed
downstream to subbasin 3048.
• Sawmill Lake - There were no available monitoring data for releases from the Sawmill
Reservoir. The Tongue/Powder Basin Plan states that 15 cfs is initially released from the
reservoir in late July (from subbasin 3057), and that the flow tapers to 5 cfs by the end of the
season (WWDC, 2002g). Releases were modeled as 15 cfs starting on July 25 of each year and
decreasing to a flow of zero on September 30. Water is released into subbasin 3056. Any
overflows were allowed to spill over and were routed downstream to subbasin 3056.
• Twin Lakes - There were no available daily or monthly monitoring data for releases from the
Twin Lakes Reservoir. The Tongue/Powder Basin Plan states that the reservoir is used to
maintain 1 to 6 cfs of water in the West Fork of Big Goose Creek, with three days of flushing
flows each year, generally in May or June (WWDC, 2002g). A time series of releases was
created for the LSPC model to mimic this description. The 50 cfs of flushing flow was modeled
as occurring on May 30-June 1 of each year. Water is released from the reservoir (subbasin
3059) to subbasin 3062. Any overflows were allowed to spill over and were routed downstream
to subbasin 3062.
Three major high altitude diversions move water from the Big Goose Creek to the Little Goose Creek
watersheds (see Figure 3-22). The Park Diversion diverts flow from Park Reservoir to Willow Creek and
the Mountain Supply Diversion diverts water from the East Fork of Big Goose Creek to the Granger
55
-------
LSPC Model Setup
Reservoir and Willow Creek (WWDC, 2002g). Another diversion moves water from the West Fork of
Big Goose Creek to the Last Chance Reservoir and Willow Creek.
Flow data for the Park and Mountain Supply Diversions were obtained from the Tongue/Powder River
Basin Plan Technical Memoranda A - Irrigation Diversion Operation and Description (WWDC, 2002d).
Continuous flow data for the Park Diversion and Mountain Supply Diversion were also obtained from the
Wyoming State Engineer's Office (SEO, 2006). Diversions were modeled based on the reported flow
data and period of operation for each year. Water is withdrawn and the specified flow and modeled
concentration and input into the appropriate LSPC subbasin. The diversion from the West Fork of Big
Goose Creek to the Last Chance Reservoir was not modeled because data indicate that the Last Chance
Reservoir is only periodically filled (SEO; 2001-2006).
3.10 Other Diversions
Diversions are prolific throughout the Tongue River watershed, and are generally used to supply
irrigation water to cropland near the point of diversion. The amount of diverted flow widely varies, and
can range from less than 1 to more than 100 cfs of water (WWDC, 2002d). In Wyoming alone, 248
points of diversion were identified in the Tongue River watershed by the Powder/Tongue Basin Plan
(WWDC, 2002a). Similar data were not available for Montana.
Little flow data are available for diversions. Some larger diversions have continuous flow meters, and
Wyoming SEO and Montana DNRC personnel periodically obtain spot measurements on smaller
diversions. For the most part, the location, volume of timing of diverted water in the Tongue River
watershed is largely unknown. Furthermore, in the case of larger diversions serving multiple users, it is
not clear which fields are irrigated from the diversion at what time and volume.
Because of the large uncertainty in the diversion data, a simplified approach was chosen for modeling
diversions in LSPC. Only diversions that move a large amount of water (i.e., more than 10 cfs) between
modeling subbasins were included in the LSPC model. The remaining diversions were not explicitly
modeled, but were implicitly captured in the irrigation module (further described in Section 3.11). Water
for irrigated land was withdrawn directly from the modeled stream segment (as opposed to from a
diversion). This method assumes that irrigated land within a modeling subbasin (as defined in Section
3.2) receives water from the stream segment in the same modeling subbasin.
Two of the major high altitude diversions (Mountain Supply and Park Diversions) were previously
discussed in Section 3.9. The other five modeled diversions are discussed individually in Sections 3.10.1,
3.10.2, and 3.10.3.
3.10.1 Mead-Coffeen, Piney-Cruse, and Prairie Dog Diversions
Three ditches (Mead-Coffeen, Piney-Cruse, and Prairie Dog) divert water from the Powder River
watershed (Piney Creek near Story, WY) to the Tongue River watershed (WWDC, 2002d). Water is
delivered to the Prairie Dog Creek and Little Goose Creek subwatersheds, as shown in Figure 3-23. The
diversions were setup in the LSPC model based on flow records obtained from the Tongue/Powder River
Basin Plan and Wyoming SEO (WWDC, 2002d; SEO, 2006b). Average monthly flow data were
available through 2000, and daily data were available from 2001 to 2005. An estimated average monthly
flow was used for 2006. Both the Piney-Cruse and Prairie Dog Ditches divert water solely to the Prairie
Dog Creek watershed. These diversions were routed to modeling subbasin 3009. The Meade-Coffeen
Ditch provides irrigation water to both the Prairie Dog Creek and Little Goose Creek subwatersheds.
Based on the irrigation data provided in the Tongue/Powder River Basin Plan, 40 percent of the flow from
56
-------
LSPC Model Setup
the Meade-Coffeen Ditch was routed to the Little Goose Creek watershed (LSPC modeling subbasin
3027) and 60 percent was routed to the Prairie Dog Creek watershed (LSPC modeling subbasin 3009)
(WWDC. 2002d). Table 3-15 summarizes the ditches and average monthly flows.
C&unti«i
Gftlflft
Mo deled Sir trams.
Taragiw River Riwir.ww
LSPC MMtellng
Tribal Land
Tftmtatfn Divottlsn*
MONTANA
WYOMING
0 25 5 I^MJr-5.
1 I J 1 I I 1 I J
Figure 3-23. Transbasin diversions from the Powder River watershed to the Tongue River
watershed.
Table 3-15. Average monthly flow (cfs) for the three diversions from the Powder River watershed.
Diversion
LSPC
Subbasin
Apr
May
Jun
Jul
Aug
Sep
Oct
Meade-Coffeen (To Prairie Dog
Creek)
3009
1.6
3.2
6.4
8.4
8.2
5.6
0.0
Meade-Coffeen (To Little Goose
Creek)
3027
1.0
2.1
4.3
5.6
5.5
3.7
0.0
Piney-Cruse
3009
3.0
7.1
11.5
14.5
14.5
9.8
0.0
Prairie Dog
3009
15.1
39.6
48.4
53.8
50.0
25.3
0.0
Total Flow Imported from Piney
Creek
20.7
52.0
70.7
82.3
78.2
44.4
0.0
WWDC, 2002d
57
-------
LSPC Model Setup
No water chemistry data were available for the three diversions, and few data were available for Piney
Creek (Personal Communications, Michael Whitaker, Wyoming SEO, December 26, 2006). Water
chemistry data from the Tongue River at Dayton (USGS Gage 06298000) were used in the LSPC model
to represent the water chemistry of the diverted water from Piney Creek. This gage was chosen because it
is geographically similar to Piney Creek near Story, Wyoming. Both locations are at the base of the
Bighorn Mountains and consist primarily of mountain runoff. Both are also located before any major
influences from anthropogenic sources (e.g., irrigation, point sources, etc.). Finally, the Tongue River at
Dayton was chosen because it has a robust set of water chemistry data from 1966 to 2002, and all of the
parameters of concern (i.e., calcium, magnesium, sodium, total nitrogen, and total phosphorus) are
available. Average monthly values of calcium, magnesium, sodium, total nitrogen, and total phosphorus
were obtained from this gage and input into the LSPC transbasin time series.
It is acknowledged that this approach introduces some error and uncertainty into the model. The Tongue
River at Dayton is located more than 25 miles from Story, Wyoming, and undoubtedly has different water
chemistry concentrations than observed in Piney Creek. Second, there is year-to-year variability in water
chemistry data which is not being represented by using a monthly average approach. Future collection of
water chemistry data in Piney Creek and/or the three diversions could help to strengthen model
performance and reduce error.
3.10.2 Tongue and Yellowstone Diversion
The Tongue River Diversion Dam is located on the Tongue River near the confluence of Pumpkin Creek
and approximately 12 miles upstream of Miles City. Water is diverted out of the Tongue River at the dam
for use in both the Tongue River watershed and the Yellowstone River watershed. The diversion was
input into the LSPC model based on daily flow data provided by Montana DNRC (DNRC, 2006b). Data
were only available from 1997 to 2005. Average monthly values were calculated and input into the
model for 1990 to 1996. Water was diverted out of subbasin 1026 and routed out of the watershed. The
average flow for the diversion was 129 cfs, and it was generally operational from May to October of each
year.
3.10.3 Sheridan Water Treatment Plant
The City of Sheridan diverts water out of Big Goose Creek near the canyon for use in the city's public
water supply (WDEQ, 2006d). Flow data for the diversion were obtained from the Powder/Tongue Basin
Plan and the Wyoming SEO. Daily flow data were available from 2001 to 2005, and average monthly
data were available from 1990 to 2000. Diverted flow ranges from 4 to 19 cfs, with an average of 8 cfs
(WWDC, 2002d; SEO, 2006). Flow is diverted year-round and exhibits a seasonal pattern, with more
flow diverted in the summer and less in the winter. In the LSPC model, water was diverted out of
subbasin 3046 and ultimately routed out of the model. Returns from Sheridan's water supply were
captured as part of the Sheridan Wastewater Treatment Plant discharge to Goose Creek (see Section
3.7.1).
58
-------
LSPC Model Setup
3.11 Irrigation
Irrigation and other water withdrawals/diversions are a potential source of pollutants in the Tongue River
watershed. Through irrigation, crops uptake water and concentrate salts in the root zone. Additional
leaching then moves salts out of the root zone, where they mix with alluvial groundwater (MDEQ, 2001).
The net effect to the surface-groundwater system is a loss of water (by plant uptake) and a higher
concentration of salts. However, water withdrawals have another indirect effect on in-stream
concentrations. Any water withdrawal results in less surface water volume, as only some, but not all of
the water is returned to the river via return flows, springs, and groundwater recharge. Because there is
less surface water, evaporation and additional high salinity inputs (from tributaries, point sources, and/or
groundwater) can lead to higher in-stream salinity concentrations. Irrigation practices also alter the timing
of natural hydrology due to diversions and reservoir releases.
Adequately simulating the impacts of irrigation is critical to obtaining a thorough understanding of both
current and "natural" water chemistry issues in the Tongue River watershed. However, there is little
available information about the amount of water applied to irrigated land. Major diversions often have
flow gages or flow grab samples, but they do not indicate how much water is actually applied to the land,
when the water is applied, and which specific parcel of land is irrigated from the diversion. Water rights
information was available from both the Montana DNRC and Wyoming SEO, but rights are often over-
appropriated, and again do not indicate the exact volume of water withdrawn at a specific time. The
irrigated land analysis (as described in Section 3.5.2) defines the location of irrigated land, but even this
information has associated errors. The amount of irrigated land can change per year based on
development, fallowed land, and precipitation. Because it was not in the scope of this analysis to
precisely define the volume and timing of applied irrigation water, a simplified method for simulating
irrigation was chosen for the LSPC model.
Irrigation withdrawal volumes were modeled in LSPC as a function of the irrigation demand (defined as
the difference between effective-crop-potential evapotranspiration [PEVT] and precipitation [PREC]
evaluated over a user-specified window of days). Since local field data were available for consumptive
irrigation requirements (CIR), which have already taken into account demand and transport losses,
irrigation rate (direct withdrawal volume) was used instead of computing demand. A weighted average
withdrawal volume was computed based on the consumptive irrigation requirements (CIR) for alfalfa,
beans, corn, and pasture at Sheridan, Wyoming. Table 3-16 shows the computed weighted average CIR
used in the LSPC model for all irrigated land.
59
-------
LSPC Model Setup
Table 3-16. Crop irrigation requirements for irrigated land in the Tongue River watershed.
Weight:
0.884
0.020
0.003
0.093
Weighted Average
Month
Alfalfa
Beans
Corn
Pasture
Typical CIR (in)
Jan
0.0000
0.0000
0.0000
0.0000
0.0000
Feb
0.0000
0.0000
0.0000
0.0000
0.0000
Mar
0.0000
0.0000
0.0000
0.0000
0.0000
Apr
0.0000
0.0000
0.0000
0.0000
0.0000
May
1.8381
0.0000
0.0000
0.1738
2.0120
Jun
3.5084
0.0560
0.0070
0.3346
3.9058
Jul
5.6912
0.1197
0.0199
0.5532
6.3841
Aug
4.6749
0.0712
0.0196
0.4682
5.2339
Sep
2.1916
0.0000
0.0097
0.2177
2.4191
Oct
0.6186
0.0000
0.0000
0.0579
0.6765
Nov
0.0000
0.0000
0.0000
0.0000
0.0000
Dec
0.0000
0.0000
0.0000
0.0000
0.0000
Total CIR
18.5228
0.2469
0.0562
1.8054
20.6314
Adapted fromTrelease etal., 1970
Based on the irrigated land analysis described in Section 3.5.2, full-supply and partial-supply irrigation
areas were input as independent land use categories in the LSPC model. Areas defined as having full
supply irrigation were assumed to be irrigated at 100 percent of the CIR, while partial supply irrigation
areas were assumed to be irrigated at 50 percent of the CIR. Full-supply irrigation was assumed to be 60
percent sprinkler and 40 percent flood irrigation, while partial supply assumed to be 100 percent flood
irrigation.
The difference between sprinkler and flood irrigation in the model is the storage compartment or pathway
where irrigation water is reintroduced to the land segment. Sprinkler irrigation is subject to interception
storage (trapping by plant leaves or branches) in the same way as direct precipitation. Flood irrigation
bypasses interception and is applied directly to the land surface.
The irrigation rate in the model is computed using a monthly variable multiplier for the PEVT time series
such that the long-term monthly CIRs, as defined in Table 3-16, are maintained. Years with higher PEVT
will require more irrigation, while years with lower PEVT will require less; however, the long-term
monthly CIR relationship is preserved. This method also assumes that irrigated acreage does not change
over the modeling time period, that the same land is irrigated every year, the same crops are grown, and
the same type of irrigation (flood versus sprinkler) is the same from year to year. It was also assumed that
irrigation does not occur if the stage of the stream falls below 0.1 ft, reflecting that fact that irrigation
relies on a minimum volume of water in the stream to be viable.
60
-------
LSPC Model Setup
3.12 Surface and Subsurface Water Chemistry Concentrations
The LSPC model simulates flow to perennial streams by three different pathways: surface runoff,
interflow, and groundwater discharge. Surface runoff, by definition, proceeds across the land surface
when infiltration capacity is exceeded. Interflow represents short-distance lateral subsurface pathways,
typically occurring on hill slopes when capacity for deep infiltration is exceeded. Interflow proceeds
laterally through soils beneath the root zone (typically via the B or upper C soil horizon), and, in a model
at the large scale represented by the Tongue, typically emerges to join surface flow within the physical
dimensions of model subbasins. Groundwater discharge represents slower discharge from the surficial
aquifer. Groundwater flow proceeds through the saturated zone at and below the water table.
Flows from these three pathways encounter different geologic media that have different chemical
compositions. The varying chemical compositions result in varying surface runoff, interflow, and
groundwater concentrations. Concentrations also vary spatially as soils and geology formations change
throughout the Tongue River watershed. The following sections describe the methodology used for
assigning water chemistry concentrations to surface runoff, interflow, and groundwater pathways in the
LSPC model.
3.12.1 Surface Runoff Concentrations
Concentrations of cations in direct surface runoff are generally not of major concern. Except in areas of
high water tables or confined depressions, precipitation tends to leach salt from the surface into the
subsurface, rather than creating high concentrations in runoff, and runoff from intense precipitation events
may not have time to achieve equilibrium with soil salt concentrations (Tuteja et al., 2002). Because of
this, it was assumed that surface runoff concentrations are equal to the concentrations in precipitation.
Average concentrations of calcium, magnesium, and sodium in precipitation were obtained from the
National Atmospheric Deposition Program's (NADP) Little Bighorn Battlefield monitoring gage (located
southeast of Billings, Montana) and average values were initially applied in the LSPC model (NADP,
2006). Average values were 0.182 mg/L Ca, 0.026 mg/L Mg, and 0.059 mg/L Na. Values were then
slightly varied during calibration. Nutrient surface runoff concentrations were modeled using build-up
wash-off routines, as described in Section 4.2.2.
3.12.2 Subsurface Concentrations
Both interflow and groundwater concentrations are difficult to characterize in the Tongue River
watershed because of the wide spatial and vertical variations. The various sources of interflow and
groundwater chemistry data are discussed below (literature review), followed by the rationale and
assumptions used for setting concentrations in the LSPC model.
3.12.2.1 Literature Review
In the arid climate of the Tongue River, the water table can be at considerable depth below the ground
surface - except in alluvial and irrigated areas (WWDC, 2002f). Outside of alluvial areas, groundwater
concentrations should reflect the geochemistry of the surface aquifer - as either determined from wells
screened in the upper aquifer or by geochemical modeling of the uppermost saturated geologic formation.
Within alluvial/irrigated areas with higher water tables, the groundwater discharge concentrations most
likely reflect the concentrations observed or calculated at the bottom of the soil C horizon, which consists
of fractured but unweathered regolith.
61
-------
LSPC Model Setup
Bedrock geology of the Tongue River basin in Montana is primarily occupied by the Fort Union
formation with a small amount of the Wasatch formation (Lewis and Roberts, 1978; WWDC, 2002b).
These two formations have similar Tertiary deposits containing interbedded layers of sandstone, siltstone,
mudstone, limestone, coal, and carbonaceous shale deposited by fluvial and swampy systems (USGS,
1999). In the Wyoming portion of the basin the plains area is primarily Wasatch east of Sheridan. To the
south and west the uplift of the Bighorn Mountains exposes older Fort Union shales along the margins,
with limestone, dolomite, and plutonic rocks, among others, at higher elevations. Groundwater
geochemistry is expected to be generally similar in the Wasatch and Fort Union areas (although subject to
significant local variability), but is likely to be quite different in the Bighorn Mountains.
Recent USGS studies of ground water in the Wasatch and Fort Union formations in the Powder River
basin (Bartos and Ogle, 2002) report median values of 225 mg/L Na, 15.5 mg/L Ca and 15.4 mg/L Mg,
while samples from coalbed aquifers had median values of 210 mg/L Na, 26 mg/L Ca, and 15 mg/L Mg.
The interquartile range for Na is 90 to 480 mg/L. Ca and Mg concentrations in these samples are
apparently lower than expected in discharge because both ions (unlike Na) have a significant negative
correlation with depth. Shallow wells, which are the ones most likely associated with discharging
groundwater, tend to have much higher concentrations of Ca and Mg.
Very few well samples appear to be available from the Bighorn Mountains. However, Larson and
Daddow (1984) do report five samples from the Madison Limestone, which is present across much of the
western portion of the Upper Tongue area. These samples are highly depleted in Na (0-1 mg/L) with 15-
30 mg/L Mg and 40-88 mg/L Ca.
Detailed information on ion concentrations in irrigated alluvial soils at 15 sites in the Tongue basin is
provided in the 2006 Agronomic Monitoring and Protection Program (AMPP) report (Schafer et al.,
2006). The authors conclude that ion concentrations (as meq) at the base of the soil profile are little
affected by irrigation and, as noted above, these saturation paste results can potentially be used to estimate
discharging groundwater concentrations. The results for these soils show considerable variability from
site to site (Figure 3-24), particularly for Na which ranges from 81 to 6250 mg/L. Median concentrations
are shown in Table 3-17.
6
5
> 4
o
O"
0)
I
I
£
rp rtp <£> Qf
Discharge concentration (mg/L)
~ Ca
¦ Mg
~ Na
Figure 3-24. Histogram of Discharge Concentrations from Base of C Horizon, AMPP Alluvial
Soils (calculated from data in Schafer et al., 2006)
62
-------
LSPC Model Setup
Table 3-17. Summary of Discharge Concentrations (mg/L) from Base of C Horizon, AMPP Alluvial
Soils (calculated from data in Schafer et al., 2006)
Parameter
Median
Mean
Geometric Mean
Calcium
112.2
175.3
119.7
Magnesium
82.7
175.4
96.3
Sodium
197.7
1,154.7
386.7
The NRCS soil surveys provide information on salinity (as SC) and SAR for soils throughout the nation.
This provides some insights into the expected spatial variability of subsurface salt concentrations, but the
information turns out to be of somewhat limited value for the following reasons:
• SAR and SC results are reported only as relatively broad ranges (e.g., SC of 1 to 8 (imhos/cm).
The use of broad ranges does not provide sufficient precision to accurately resolve the Na
concentration using the equations given above.
• Upland soil profiles often do not extend to the full extent of the overburden. Further, except in
alluvial areas, the water table position is likely to be in the bedrock, so that soil profile data tells
us little about saturated flow concentrations.
• Given only SAR and SC, there is no firm basis for assumptions about the ratio of Ca to Mg
equivalence or the concentration of minor cations.
Despite these calculations, some qualitative comparisons can be made. Estimated discharge
concentrations from the basal horizon of alluvial soils in the prairie region of the watershed in Montana
(Map Unit MT668) and Wyoming (Map Unit WY055) are compared in Table 3-18, using weighted
percentages of the Map Units ID (MUID) components, and with the additional assumptions that the
concentration of minor cations is 5 mg/L and that the ratio (in meq) of Ca:Mg is 60:40.
Table 3-18. Calculated discharge concentrations from alluvial soils.
Soil MUID
Ca (mg/L)
Mg (mg/L)
Na (mg/L)
MT668
233
256
592
WY055
268
294
121
These results suggest that prairie ground water in the Upper Tongue watershed (i.e., predominantly in
Wyoming) should have a lower concentration of Na, lower SAR, and higher Ca and Mg than ground
water in the Lower Tongue. The difference between the MT668 and WY055 results in Table 3-18 is
primarily due to the presence of three soils with elevated SAR (Map Units MT0106, MT0286, and
MT0181), the last two of which also have elevated salinity, that compose about 17 percent of the MUID.
In contrast, basal horizon SAR values in Wyoming alluvial soils are low, and generally similar to the
remaining soil components in MT668. This suggests that the aerial extent of high-SAR soils (at bottom
of soil profile) could provide a useful basis to vary groundwater concentrations by watershed.
63
-------
LSPC Model Setup
3.12.2.2 Modeled Subsurface Concentrations
Based on the research discussed in Section 3.12.2.1, there is no clear and precise external source of
information available to set subsurface cation concentrations generally throughout the LSPC Tongue
River modeling domain. There are, however, a variety of sources of information that help inform and
constrain potential concentrations. These constraints were used to help calibrate the subsurface
concentrations (described in 4.2). The rationale and assumptions for setting subsurface concentrations in
the LSPC model are described below. Final calibrated values are discussed in Section 4.2.
Ground Water
Groundwater concentrations were assumed to primarily reflect chemistry of the underlying aquifer -
especially the chemical characteristics near the discharge point.
Prairie. Concentrations were set based on local spring data obtained from the USGS NWIS
Database. Median values were used as constant annual values for all land uses. Limited data were
available for most of the region, and this is believed to be a major source of uncertainty in the LSPC
model (see Section 5.7.7)
Mountains. Available evidence suggests that ground water from this area has low sodium
concentrations and moderate Ca and Mg. Concentrations were set as annual average values for all land
uses using the data reported by Larson and Daddow (1984). As with the prairie region, limited data were
available.
Irrigated Land. For groundwater discharges from irrigated lands cation concentrations are
primarily determined by soil characteristics (under current irrigation water chemistry). The best available
evidence for this component appeared to be the median values derived from the AMPP study (Schafer et
al., 2006), as summarized in Table 3-17. These were used as starting values for the model but
subsequently adjusted during calibration.
Alluvial Channels downstream of Ponds. Infiltration from ponds sub-irrigates the ephemeral
channel downstream (see Section 3.7.2.2 and 3.8). Given the assumption that groundwater concentrations
are determined primarily by the underlying bedrock - but influenced by the elevation in water table, these
groundwater concentrations were set equal to those for irrigated land.
Interflow
Interflow represents shorter lateral flow pathways through the subsoil. Concentrations should differ from
those in ground water due to mineralization of Ca and Mg, but Na concentrations are likely to be similar
(Wheaton and Brown, 2005). It should be noted that LSPC (and the HSPF model from which it is
derived) route interflow from surface detention storage prior to entry into the soil profile. Functionally,
however, interflow is used to represent lateral subsurface flow through the soil that occurs more rapidly
than groundwater flow from the water table. This occurs when vertical infiltration capacity is exceeded at
some point in the unsaturated soil profile. In arid climates, this conversion to lateral flow may occur at a
significant depth - despite the way it is represented in LSPC - allowing time for desorption of salts in the
upper soil profile.
Prairie. Sodium concentrations in interflow were set equal to those in groundwater. Calcium and
magnesium concentrations are, however, likely to be enriched due to the shorter flow pathways, whereas
concentrations in ground water are reduced by mineralization. Assuming that mineralization rates of Ca
64
-------
LSPC Model Setup
and Mg are approximately similar, the interflow concentrations were adjusted upward by a single
calibration factor, Kh
Mountains. For the Upper Tongue Mountains, cation concentrations are low and soil profiles are
short. For this region it was appropriate to set equal groundwater and interflow concentrations.
Irrigated Land. Schafer et al. (2006) demonstrate a peak in cation concentrations at an average
depth of 3 feet, likely coincident with the main locus of lateral flow under saturated conditions. (The
result could, however, be an artifact of the drought conditions of recent years.) As with groundwater, this
source appears to be the best information currently available. For irrigation with existing river water,
interflow concentrations were initially set to the average 3-ft peak shown in Figure 5-10 of Shafer et al.
(2006), which has an SAR of 4.0 and SC of 5.4. This translates to discharge cation concentrations of 521
mg/L Ca, 389 mg/L Mg, and 1012 mg/L Na. As with groundwater discharge from irrigated land, these
concentrations were adjusted during calibration.
Alluvial Channels Downstream of Ponds. Water exflltrating from ponds is assumed to be
applied to soil storages and is available for plant ET, but not for interflow. Interflow will occur only in
response to direct precipitation (or snowmelt) on the channel area. The subirrigation will, however, raise
the water table, and any simulated interflow will likely be mixed with ground water. Therefore,
infiltration concentrations were set equal to groundwater concentrations for this area.
65
-------
-------
LSPC Calibration
4.0 LSPC CALIBRATION
Calibration is defined as "the process of adjusting model parameters within physically defensible ranges
until the resulting predictions give the best possible fit to the observed data" (USEPA, 2003). For LSPC,
calibration is required for both hydrology (flow) and water chemistry and is an iterative procedure of
parameter evaluation and refinement as a result of comparing simulated and observed values at specified
locations in a watershed. 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. Because these characteristics vary throughout a watershed, calibration
generally occurs at more than one site. Also, calibration generally covers several years to capture a
variety of climactic conditions. The calibration procedure results in parameter values that produce the
best overall agreement between simulated and observed values throughout the calibration period.
Several different methods were employed to judge the adequacy of the model fit to the observed data.
Hydrologic calibration followed the standard operating procedures for the model described in Donigian et
al. (1984) and Lumb et al. (1994). Daily, monthly, seasonal, and total modeled flows were compared to
observed data, and error statistics were calculated for the percent difference (i.e., [Modeled -
Observed]/[Observed] * 100). The percent errors were then compared to recommended tolerance targets
from Donigian et al. (1984) and Lumb et al. (1994). Tolerance targets for the flow simulation were
modified slightly from the literature values to better assess the Tongue River watershed calibration.
Donigian et al. (1984) and Lumb et al. (1994) recommend seasonal targets (e.g., spring, summer) of ±30
percent. The targets were used to assess monthly results (e.g., January, February) and growing season
results (i.e., growing season versus non-growing season), which are the basis for the Montana salinity and
SAR water quality standards. Targets are show in Table 4-1. Model results were also visually compared
to observed data, and daily and monthly data were plotted as scatter plots with regression analyses.
Table 4-1. Recommended criteria for the Tongue River watershed hydrology calibration.
Category
Recommended Criteria (%)
Error in total volume:
±10
Error in the mean of the 10% lowest flows:
±10
Error in the mean of the 10% highest
flows:
±15
Error in Monthly Volumes:
±30
Error in Growing Season Volumes:
±30
Error in Nongrowing Season Volumes:
±30
Modified from Lumb et al., 1994 and Donigian et al. 1984
67
-------
LSPC Calibration
Fewer water chemistry data were available for most calibration locations. Where daily data were
available, water chemistry calibration followed the standard operating procedures previously described
for the hydrologic calibration. Donigian et al. (1984) suggest targets for the percent error in mean water
chemistry data. The authors specifically note, however, that "tolerance ranges should be applied to mean
values and that individual events or observations may show larger differences, and still be acceptable".
Water chemistry at all locations was also judged with graphs, box plots, and scatter plots.
Table 4-2. Recommended criteria for the Tongue River watershed chemistry calibration.
Parameter
% Difference Between Simulated and Measured Values
Very Good
Good
Fair
Water Temperature
< 7
8-12
13-18
Water Quality/Nutrients
< 15
15-25
25-35
Source: Donigian et al. (1984)
The following sections discuss the methodology for calibrating the Tongue River LSPC model. Section
4.1 discusses the hydrologic calibration process, Section 4.2 discusses the water quality calibration
process, Section 4.3 summarizes the results, and Section 4.4 presents information on the evaluation of the
model. Final calibrated values for all input parameters in the LSPC model are included in the model input
file, which is available upon request. All evaluated calibration tables and plots are included in Appendix
B.
4.1 Hydrologic Calibration Methodology
Hydrologic calibration was performed at 13 USGS gages selected because they had continuous flow
records for multiple years and were located in strategic positions throughout the Tongue River watershed.
Table 4-3 summarizes the 13 USGS gages, and gages are shown in Figure 4-1.
Table 4-3. USGS monitoring sites used for the Tongue River LSPC model hydrology calibration.
Site Name
Station
ID
Altitude
(ft)
Drainage Area
(mi2)
LSPC
Subbasin
Tongue River near Dayton, WY
06298000
4,060
206
3090
Tongue River at the Montana-Wyoming State Line,
MT
06306300
3,429
1,453
3006
Tongue River at Tongue River Dam near Decker,
MT
06307500
3,344
1,770
3112
Hanging Woman Creek near Birney, MT
06307600
3,150
470
1095
Otter Creek at Ashland, MT
06307740
2,917
707
1059
Tongue River near the Brandenberg Bridge, MT
06307830
2,760
3,948
3047
Pumpkin Creek near Miles City, MT
06308400
2,490
697
1007
Tongue River at Miles City, MT
06308500
2,360
5,379
1002
Wolf Creek at Wolf, Wyoming
06299950
4,525
38
3070
Big Goose Creek near Sheridan, Wyoming
06302000
4,505
120
3046
Little Goose Creek near Bighorn, Wyoming
06303500
4,860
52
3029
East Fork Big Goose Creek near Bighorn,
Wyoming
06300500
8,320
20
3051
Coney Creek above Twin Lakes near Bighorn,
Wyoming
06301480
8,690
3.4
3060
68
-------
LSPC Calibration
063077M
heyenne
¦ W3QTW0
O6JO75O0
MONTANA
WYOMING
CSJfJfiJOO
O6W95O0
IKWMTO
iMjrioMO
Calibration Gages
• Hydrology
¦ Hydrology & Water Quality
Streams
Counties
Tribal Land
Tongue River Watershed
Big Horn
County
nhnsoii
Sheridan
County
MONTANA
WYOMING
0 10 20 40 Mites
1 I I I I I I I I
Figure 4-1. LSPC calibration sites for the Tongue River watershed.
69
-------
LSPC Calibration
The hydrologic calibration proceeded from upstream gages to downstream gages (as illustrated in Figure
4-1) and involved a comparison of observed data to modeled in-stream flow and an adjustment of key
parameters. Modeling parameters were varied within generally accepted bounds and in accordance with
observed temporal trends and soil and land cover characteristics. An attempt was made to remain within
the guidelines for parameter values set out in BASINS Technical Note 6 (USEPA, 2000). The hydrology
calibration initially focused on the period between water years 1992 and 1998 and the 1999 through 2006
period served as an informal validation period (it should be noted that the 1999 to 2006 period is not
considered a formal validation period because, during calibration, model results for the entire period of
record were also evaluated).
For the purpose of this report, the hydrologic calibration is further discussed in the two following sections
- High Altitude (i.e., Bighorn Mountain) gages (Section 4.1.1) and Prairie gages (Section 4.1.2).
4.1.1 High-Altitude Hydrologic Calibration Methodology
Hydrologic calibration began in the Bighorn Mountains. Because most of the flow in the Tongue River
(about 63% of all the water in the watershed) originates in the mountains, it was important to obtain an
adequate calibration in this region to achieve calibrated results in the downstream region. Specifically, it
was important to achieve an adequate snow budget (snow accumulation and melt processes). Snow
accumulation and melting were calibrated by comparing model output to a computed snow budget based
on the NRCS SNOTEL data. Key calibration parameters were revised from defaults during optimization
and included the snow catch factor (SNOWCF, ratio that accounts for under-catch of snow in standard
precipitation gages), the field adjustment parameter for heat accumulation in the snow pack (CCFACT),
the maximum rate of snow melt by ground heating (MGMELT), and the difference between the mean
elevation of a subwatershed and the gage elevation (ELDAT, to correct for temperature changes between
the gage elevation and subwatershed elevation). Calibration at two SNOTEL gages (Dome Lake and
Sucker Creek) show that snow processes are well represented in the Bighorn Mountains from 1989
through 2006 (Figure 4-2 and Figure 4-3).
Daily flow records at two USGS gages were used to further refine the high-elevation hydrologic
calibration. Input parameters were varied to achieve calibrated results at USGS gages on Coney Creek
and East Fork Big Goose Creek (USGS gages 06301480 and 06300500, respectively). Calibrated results
are presented in Appendix B.
Downstream of the Coney Creek and East Fork Big Goose Creek gages, reservoirs, lakes, and diversions
significantly alter stream flows. Water is stored and released from reservoirs to meet water supply
demands, while diversions move water among various subbasins (i.e., from Big Goose Creek to Little
Goose Creek) (see Section 3.9). Calibration at gages downstream of the reservoirs was reliant upon an
adequate understanding of volume in the reservoirs and timing of releases. Four USGS Gages (USGS
gages on Big Goose Creek, Little Goose Creek, Tongue River, and Wolf Creek) located at the base of the
Bighorn Mountains were used to assess the final calibration for the high-altitude region.
70
-------
i R ainfall (in)
S n artrfa IIW ate r- E q ui vale nt (in)
Air Temp (D eg F)
"Snowfall Temp (Deg F)
S NO TEL Tern perature fDeg F)
Snoifu Pack Depth (in)
3 now Fa II Water-Equivalent (in)
Snouu Mett (in)
-Water Yield FromSnovuPack £in)
3 NO TEL Sroundepth (in)
20 o
Figure 4-2. Snow budget for the Dome Lake SNOTEL gage.
w
-o
o
o
D"
fl)
r+
o'
-------
Figure 4-3. Snow budget for the Sucker Creek SNOTEL gage.
-------
LSPC Calibration
4.1.2 Prairie Hydroiogic Calibration Methodology
In the Prairie region of the Tongue River watershed, model performance is sensitive to the specification
of the water-holding capacity within the root zone of the soil profile (expressed through LZSN, the
nominal lower-zone storage) and the infiltration rate index (INFILT), which together control the
partitioning of water between surface and subsurface flow. Values for LZSN in the three watersheds
ranged between 5 and 9 inches depending primarily on soil conditions.
INFILT in HSPF is an index of infiltration rate and is not directly interpretable from measured field
infiltration rates. BASINS Technical Note 6 recommends values in the range of 0.1 to 0.4 in/hr for B
soils, 0.05 to 0.1 in/hr for C soils, and 0.01 to 0.05 in/hr for D soils. Values were re-optimized by starting
from the center of the recommended ranges and modifying the value for each soil class proportionately.
Final values ranged between 0.008 and 0.24.
Interception of precipitation in LSPC is represented by a storage capacity (CEPSC) factor, which
accounts for moisture that is retained by vegetation or other ground cover and not available for infiltration
or overland flow. Calibrated values ranged between 0.00 and 0.40 inches, depending on land use and
time of year.
Key parameters for the subsurface flow response include the ground water recession coefficient
(AGWRC), and the interflow inflow and recession parameters (INTFW and IRC). AGWRC was set by
optimizing model performance for baseflow recession, with relative variation among land uses based on
past experience, resulting in values from 0.90 to 0.998. Interflow should be fairly high in this landscape,
and the interflow recession constant parameter was specified as gradually varying between 0.3 and 0.85.
The interflow inflow factor ranged from 1 to 3 depending on soils, topography, and land use.
Deep aquifer infiltration (DEEPFR) represents the fraction of infiltrating water that percolates to deep
aquifers and is therefore "lost" water removed from the system. Within these watersheds, DEEPFR is
usually higher in headwater areas far away from the main stem. The range of DEEPFR values was
optimized accordingly during the calibration process and ranged between 0.0 and 0.2.
Overland flow in LSPC is simulated using the Chezy-Manning equation and an empirical expression that
relates outflow depth to detention storage. Monthly variations of Manning's "n" were specified to
account for changing ground cover conditions and values ranged between 0.10 (barren lands) and 0.2
(forest).
Monthly variability in hydroiogic response was specified by setting monthly values for the upper zone
nominal soil storage and the lower zone ET parameter. In each case, the values specified are consistent
with the range recommended in BASINS Technical Note 6 (USEPA, 2000).
All final calibrated values for the Tongue River LSPC model are included the model input file, which is
available upon request.
73
-------
LSPC Calibration
4.2 Water Chemistry Calibration Methodology
After hydrology was sufficiently calibrated, water chemistry calibration was performed. The water
chemistry calibration consisted of running the watershed model, comparing output to available
observation data, and adjusting pollutant loading and in-stream water quality parameters within a
reasonable range.
Water chemistry calibration occurred at eight USGS gages selected because there were sufficient data.
Table 4-4 summarizes the gages assessed during the calibration process, and the gages are shown in
Figure 4-1.
Table 4-4. USGS monitoring sites used for the Tongue River LSPC model water chemistry
calibration.
Site Name
Station
ID
Altitude
(ft)
Drainaqe Area
(mi2)
LSPC
Subbasin
Tongue River near Dayton, WY
06298000
4,060
206
3090
Tongue River at the Montana-Wyoming State
Line, MT
06306300
3,429
1,453
3006
Tongue River at Tongue River Dam near Decker,
MT
06307500
3,344
1,770
3112
Hanging Woman Creek near Birney, MT
06307600
3,150
470
1095
Otter Creek at Ashland, MT
06307740
2,917
707
1059
Tongue River near the Brandenberg Bridge, MT
06307830
2,760
3,948
1047
Pumpkin Creek near Miles City, MT
06308400
2,490
697
1007
Tongue River at Miles City, MT
06308500
2,360
5,379
1002
Similar to hydrologic calibration, the water chemistry calibration process for each watershed proceeded
from upstream gages to downstream gages and involved a comparison of observed data to modeled in-
stream water quality and an adjustment of key parameters. Modeling parameters were varied within
generally accepted bounds and in accordance with observed temporal trends and soil and land cover
characteristics. The water chemistry calibration relied on data for the entire period of record at each
location (Table 4-4), but focused primarily on data collected since 2004 when continuous data became
available at many monitoring locations.
The following sections discuss the water chemistry calibration for salinity and SAR (as modeled by
calcium, magnesium, and sodium concentrations) (Section 4.2.1) and nutrients (Section 4.2.2). Water
temperature was also calibrated in the Tongue River upstream of the Tongue River Reservoir (Section
4.2.3), as it is needed for the CE-QUAL-W2 reservoir model (further discussed in Section 6.2.3).
4.2.1 Salinity and SAR
As discussed in Section 3.1, salinity (measured as specific conductance [SC]) and SAR are not directly
simulated in the Tongue River application of LSPC. Instead, the model predicts concentrations of the
primary cations that compose the two indicators (calcium, magnesium, and sodium). Cations are modeled
using constant (surface and interflow) and monthly varying (groundwater) concentrations in outflow from
land units. Since little mineralization is expected, a decay rate of zero was assumed.
Calibration points for the Upper Tongue watershed (i.e., upstream of the Tongue River Reservoir) were at
Dayton (USGS gage 06298000) and State Line (USGS gage 06306300). Dayton was calibrated first to
74
-------
LSPC Calibration
isolate the mountain watershed Group 1. Subsequent watershed groups associated with the mountains
(Groups 2 and 3) were assigned concentrations resulting from the Dayton calibration, as there were few
other water chemistry data collected near the Bighorn Mountains. Interflow and groundwater
concentrations in LSPC Group 1 were set to values consistent with data from the Madison limestone
formation and springs data (see Section 3.12.2.1). Initially, values were set as constant parameters by
month, land use, and flow component. During calibration, groundwater and interflow concentrations were
adjusted slightly for all but irrigated land uses. In addition, the observed data suggested decreased
groundwater concentrations in portions of summer. To account for this trend, constituents associated with
groundwater outflow were decreased in June through August (July and August at State Line).
In the Upper Tongue watershed, assignments associated with prairie groups (Groups 4, 5, and 6) were
based on calibration at State Line for Group 4: Prairie. In the Lower Tongue watershed, parameters were
then based on the calibration of Group 4 (Prairie) in the Upper Tongue watershed. Parameters were then
varied to achieve calibrated results for the main stem Tongue River and tributaries.
As described in Section 3.12, initial surface, interflow, and groundwater concentrations for calcium,
magnesium, and sodium were based on a variety of sources. Concentrations in surface outflow from land
units (SOQC) were based on data from the National Atmospheric Deposition Program (as described in
Section 3.12.1). Small adjustments were made during calibration. Higher concentrations were assigned to
the prairie region, where the sedimentary soils have naturally higher salt concentrations than the Bighorns
Mountains.
Soil cation concentrations cited in Schafer et al (2006) provided initial concentrations in interflow and
groundwater concentrations associated with irrigated land uses. While initial values were based on
median values from Schafer et al. (2006), calibration resulted in much lower assignments to outflow,
particularly for sodium. Sodium concentrations in interflow were reduced by an order of magnitude.
Irrigation likely flushes salts below the simulated interflow region into groundwater where it re-
equilibrates with the soil matrix, resulting in lower outflow concentrations. Calibration reduced interflow
concentrations of calcium and magnesium by 70 to 75 percent. Groundwater concentrations were adjusted
to values near those for the other simulated land uses.
4.2.2 Nutrients
Nutrients in the LSPC model were simulated using buildup/washoff functions for the surface runoff
contribution of nutrients, and concentration assignments for interflow and groundwater outflow. General
first order decay processes describing the in-stream loss rate were specified for each constituent, which in
this system likely represents uptake and trapping by periphyton to a large degree. In the Upper Tongue
watershed, the station at Dayton was calibrated first followed by State Line. Nutrients were not modeled
in the Lower Tongue River watershed because none of the 303(d) listed stream segments downstream of
the Tongue River Reservoir were listed as impaired for nutrients.
First-order decay rates were taken from the USGS SPARROW work (Smith et al., 2003) for streams with
flow less than 28 m3/s; TN at 0.455 per day and TP at 0.258 per day. The rate for TN was unchanged
during calibration. First-order decay rates for TP varied spatially. Subbasins 3022 and 3068 (which
receive effluent from the Sheridan WWTP and Ranchester/Dayton Lagoons, respectively), were assigned
a decay rate of 0.8 per day during calibration. It was assumed that there was extra uptake possibly due to
periphyton mats very near to the discharge location of these point sources. All other reaches within
subbasins were set to 0.4 per day. The effects of seasonality were represented with the use of an
Arrhenius temperature correction coefficient on the decay rate, which was set at 1.04 for TN and TP
(USEPA, 1985). Water temperature simulation was performed using a regression equation with air
75
-------
LSPC Calibration
temperature. The equation was developed from observed water and air temperatures at the State Line
station on a monthly basis (further described in Section 4.2.3).
Initial nutrient concentrations of interflow and groundwater were based on observed data during baseflow
periods. Adjustment to the initial concentrations in interflow and groundwater were made during the
calibration process. While phosphorus levels are expected to be low in groundwater discharge, the
groundwater component of the model also represents baseflow transport through small, unmodeled stream
reaches. These reaches are not explicitly represented due to the scale of the Tongue River model. During
the groundwater-dominated receding limb of the late spring and summer period, organic and sorbed
phosphorus continue to be mobilized by flow in these small reaches, resulting in apparent higher
"groundwater" concentrations of TP. To capture this pattern in observed data at Dayton and State Line,
ground water values assigned for TP were set from observed data at Dayton and State Line respectively.
Then the concentrations were adjusted during the calibration process.
Initial values for buildup/washoff parameters were based on several sources. Land uses other than
cropland and urbanized land were assigned initial TN values based on concentration data collected by the
National Atmospheric Deposition Program, converted to corresponding accumulation factors (Butcher,
2003). For cropland and urban land, the initial TN concentration values were based on data summarized
in Haith et al. (1992), with a similar conversion to accumulation rates. Likewise, initial concentration
values of TP for crop land were derived Haith et al. (1992) and converted to accumulation rates. For
urban land, typical accumulation rates were provided by Haith et al., and were assigned an accumulation
of zero. Calibrated concentrations for both TN and TP were slightly higher than the initial values. It is
recognized that in this system, overland runoff, when compared to base flow, is relatively small and
infrequent.
Maximum storage (SQOLIM) equal to five times the accumulation rate was assigned across land uses.
Values typically range from approximately 5 to 8 times the accumulation (Sartor and Boyd, 1972). The
rate of surface runoff that removes 90 percent of stored nutrient (WSQOP) typically ranges from 0.4 to
0.7 in/hr. A value of 0.6 in/hr was assumed for this model application. Since the model was not
particularly sensitive to the SQOLIM and WSQOP, these parameters in the model were held constant
across land uses and parameters during calibration, except for default ID 4 in which three land uses were
assigned a maximum storage at nine times the accumulation rate.
76
-------
LSPC Calibration
4.2.3 Water Temperature
Water temperature in the Upper Tongue watershed was simulated in LSPC by use of regression equations
which make each surface, interflow, and groundwater water temperature the dependent variable and air
temperature the independent variable. An analysis was performed on the observed water and air
temperature data at State Line. The data of each were worked into monthly averages and then plotted in
Excel. A linear regression was performed which set the monthly average air temperature as the
independent variable and monthly average water temperature as the dependent variable (Figure 4-4).
LSPC allows for unique specification of water temperature for the three components: surface, interflow,
and groundwater. However, in this application, and lacking adequate observed data to describe the
forcing, each of these three components were assigned the same regression equation using air temperature
from the meteorological forcing files as the independent variable.
The temperature simulation was further defined by using default assumptions. The model was developed
with an assumption that 80 percent of all reaches are exposed to solar radiation (CFSAEX). The
longwave radiation coefficient (KATRAD) was set to 9.5. The conduction-convection heat transport
coefficient (KCOND) was set to 6.12. And lastly, the evaporation coefficient (KEVAP) was set to 2.24.
Water temperature was not calibrated for the lower Tongue watershed because temperature dependant
parameters (i.e., nutrients) were not modeled.
5 10 15 20 25
Monthly Average Air Temperature (deg C)
Figure 4-4. Water versus air temperature regression at State Line (06306300)
y=0.9375x-2.8933
R2= 0.9845
77
-------
-------
Performance and Evaluation
5.0 PERFORMANCE AND EVALUATION OF THE CALIBRATED
MODEL
The process whereby model performance was optimized (i.e., calibration) was described in Section 4.0.
An evaluation of model performance (i.e., how well the model represents the real world) is presented in
this section.
Typically, the performance of a calibrated model
is evaluated through "validation". Model
validation is defined as "subsequent testing of a
pre-calibrated model to additional field data,
usually under different external conditions, to
further examine the model's ability to predict
future conditions" (USEPA, 1997). Its purpose
is to ensure that the calibrated model properly
assesses all the variables and conditions that can
affect model results, and demonstrate the ability
to predict field observations for periods separate
from the calibration effort (Donigian, 2003).
The Tongue River LSPC model was not strictly
validated in the traditional sense. First, limited
data precluded validation in many cases. With
the exception of discrete periods of time (that
vary from monitoring station to monitoring station) when continuous chemistry data have been collected,
chemistry data is limited in the Tongue River Watershed. As a result, it was not possible to select one
period of time, and one set of data for calibration and another for validation. Also, given the large
watershed area and extreme variability in model forcing functions across the watershed (e.g., geography,
climate, precipitation, hydrogeology, rainfall/runoff relationships, etc.), it was necessary to calibrate over
the entire period of record to ensure that model output was not calibrated to a condition derived from
limited data over a short period of space and time. The calibration period represented a wide range of
conditions/events (e.g., high flows, low flows, variable dam operation practices, variable rates of
permitted discharges, high snow periods, low snow periods, etc.) resulting in an informal form of
validation.
Model performance has been evaluated through qualitative and quantitative comparisons of predicted to
observed data from the calibration time period (i.e., full period of record). A summary of model
performance relative to the prediction of average SC, SAR, and flow at time periods including the full
period of record, the growing season, the non-growing season, and monthly is presented below. Detailed
comparisons of predicted and observed data are presented in Appendix B. The criteria in Table 5-1 were
used to rank model performance in cases where sufficient observed data were available for quantitative
comparisons (Donigian, 2000).
Table 5-1. Model performance criteria (Adapted from Donigian, 2000).
Parameter
% Difference Between Predicted and Observed Values
Very Good
Good
Fair
Hydrology/Flow
<10
10-15
15-25
Water Quality/Nutrients
< 15
15-25
25-35
Model Evaluation Results
The LSPC model results were evaluated at USGS
monitoring gages located throughout the Tongue
River watershed. USGS gages had observed
data for varying time periods, and included both
grab samples and continuous (average daily)
observations. As part of the model evaluation,
over 130 graphs and tables were generated to
compare model results to observed data. Where
sufficient data were available, statistics showing
mean values and percent error were also
generated. A summary of the results is
presented in Section 5.0, and the complete set of
data used to calibrate both hydrology and water
chemistry is included in Appendix B.
79
-------
LSPC Calibration
5.1 Flow
Hydrology is well calibrated for the main stem of the Tongue River upstream of the Brandenburg Bridge.
Simulated flows mimic daily and seasonal patterns. For example, a time series of predicted daily mean
flow and observed daily mean flow is shown in Figure 5-1 (Tongue River at State Line). Similar figures
for all of the evaluated locations are presented in Appendix B. As shown in Table 5-2, with the exception
of March through May, model performance relative to the prediction of average flow is generally good to
very good at the evaluated time scales upstream of the Brandenburg Bridge.
Aug Monthly Rainfall (in) Avg Obserwd Flow (10/1/1992 to 9/30/2006 ) Avg Modeled Flow (Same Period)
Date
Figure 5-1. Time series of daily mean predicted and observed flow for the Tongue River at State
Line near Decker, Montana (USGS gage 06306300).
At Miles City, the observed seasonal patterns are generally mimicked by the predicted results, but May is
over predicted and February through April, June, and August are under predicted. In Hanging Woman
Creek flow predictions were fair to good for the entire period of record, the growing season and during
the 10 percent of the highest flows. Flow predictions were poor for the remaining time periods in Hanging
Woman Creek and for all time periods in the other two tributaries shown in Table 5-2.
80
-------
Performance and Evaluation
Table 5-2. Model performance - average flow at various timescales.
Time
Period
Tongue
River
near
Dayton,
WY
Tongue
River
near
State
Line
Tongue
River
below
TRR
Dam
Tongue River
below
Brandenburg
Bridge
Tongue
River at
Miles
City
Hanging
Woman
Creek
near
Birney
Otter
Creek
near
Ashland
Pumpkin
Creek
near
Miles City
All Data
Very
Good
Very
Good
Very
Good
Very Good
Good
Fair
Poor
Poor
Growing
Season
Very
Good
Very
Good
Very
Good
Very Good
Fair
Good
Poor
Poor
Non-
growing
Season
Very
Good
Very
Good
Very
Good
Very Good
Very
Good
Poor
Poor
Poor
January
Good
Very
Good
Very
Good
Very Good
Very
Good
Poor
Poor
Poor
February
Good
Good
Very
Good
Very Good
Poor
Poor
Poor
Poor
March
Very
Good
Poor
Very
Good
Poor
Poor
Poor
Poor
Poor
April
Poor
Poor
Very
Good
Very Good
Poor
Poor
Poor
Poor
May
Very
Good
Very
Good
Poor
Poor
Poor
Poor
Poor
Poor
June
Very
Good
Very
Good
Fair
Good
Poor
Poor
Poor
Poor
July
Fair
Very
Good
Very
Good
Very Good
Fair
Poor
Poor
Poor
August
Good
Fair
Very
Good
Very Good
Poor
Poor
Poor
Poor
September
Very
Good
Very
Good
Very
Good
Very Good
Very
Good
Poor
Poor
Poor
October
Very
Good
Good
Very
Good
Good
Fair
Poor
Poor
Poor
November
Fair
Very
Good
Very
Good
Very Good
Very
Good
Poor
Poor
Poor
December
Very
Good
Very
Good
Very
Good
Very Good
Very
Good
Poor
Poor
Poor
10%
Highest
Flows
Very
Good
Very
Good
Very
Good
Very Good
Fair
Fair
Poor
Poor
10%
Lowest
Flows
Very
Good
Good
Very
Good
Poor
Poor
Poor
Poor
Poor
81
-------
LSPC Calibration
5.2 Specific Conductance
As shown in Table 5-3, where sufficient data are available for comparison, model performance relative to
the prediction of average SC is generally good to very good at time scales ranging from the entire period
of record to monthly. The exceptions are the months of April, May and June in the Tongue River at State
Line and June and July in Hanging Woman Creek where poor to fair ranks were attained. Insufficient data
were available for a quantitative evaluation of model performance in the Tongue River near Dayton,
Wyoming and for the non-growing season and winter months in the Tongue River at Miles City, and
Hanging Woman, Otter and Pumpkin Creeks. Line graphs comparing predicted to observed values are
presented in Appendix B to facilitate a qualitative evaluation.
Table 5-3. Model performance - average SC at various time scales.
Time
Period
Tongue
River
near
Dayton,
WY
Tongue
River
near
State
Line
Tongue
River
below
TRR
Dam
Tongue
River below
Brandenburg
Bridge
Tongue
River
at
Miles
City
Hanging
Woman
Creek
near
Birney
Otter Creek
near Ashland
Pumpkin
Creek
near
Miles
City
All Data
NA
Very
Good
Very
Good
Very Good
Very
Good
Good
Very Good
Very
Good
Growing
Season
NA
Very
Good
Very
Good
Very Good
Very
Good
Good
Very Good
Very
Good
Non-
growing
Season
NA
Very
Good
Very
Good
Very Good
NA
NA
NA
NA
January
NA
Very
Good
Good
Very Good
NA
NA
NA
NA
February
NA
Very
Good
Very
Good
Very Good
NA
NA
NA
NA
March
NA
Very
Good
Very
Good
Very Good
Very
Good
NA
NA
NA
April
NA
Fair
Very
Good
Very Good
Very
Good
Very
Good
Very Good
Poor
May
NA
Fair
Very
Good
Very Good
Good
Very
Good
Very Good
Poor
June
NA
Poor
Very
Good
Very Good
Good
Fair
Very Good
Very
Good
July
NA
Very
Good
Very
Good
Very Good
Good
Fair
Very Good
Poor
August
NA
Very
Good
Very
Good
Very Good
Very
Good
NA
Good
NA
September
NA
Very
Good
Very
Good
Very Good
Very
Good
NA
Very Good
NA
October
NA
Good
Very
Good
Very Good
Very
Good
Poor
Good
Poor
November
NA
Very
Good
Very
Good
Very Good
NA
NA
NA
NA
December
NA
Very
Good
Very
Good
Good
NA
NA
NA
NA
82
-------
Performance and Evaluation
5.3 SAR
As shown in Table 5-4, where sufficient data are available for comparison, model performance relative to
the prediction of average SAR is generally good to very good at time scales ranging from the entire period
of record to monthly. The exceptions are the months of April through August in the Tongue River at
State Line, May and June in the Tongue River at Miles City, and June and July in Hanging Woman Creek
where poor to fair ranks were attained. Insufficient data were available for a quantitative evaluation of
model performance in the Tongue River near Dayton, Wyoming and below the Tongue River Reservoir
Dam. Insufficient data were also available for the non-growing season and winter months in the Tongue
River at Miles City, and Hanging Woman, Otter and Pumpkin Creeks. Line graphs comparing predicted
to observed values are presented in Appendix B to facilitate a qualitative evaluation.
Table 5-4. Model
)erformance - average SAR at various timescales.
Time Period
Tongue
River
near
Dayton,
WY
Tongue
River
near
State
Line
Tongue
River
below
TRR
Dam
Tongue River
below
Brandenburg
Bridge
Tongue
River at
Miles
City
Hanging
Woman
Creek
near
Birney
Otter
Creek
near
Ashland
Pumpkin
Creek
near
Miles
City
All Data
NA
Very
Good
NA
Very Good
Good
Good
Very
Good
NA
Growing
Season
NA
Very
Good
NA
Very Good
Good
Good
Very
Good
NA
Non-
growing
Season
NA
Very
Good
NA
NA
NA
NA
NA
NA
January
NA
Very
Good
NA
NA
NA
NA
NA
NA
February
NA
Very
Good
NA
NA
NA
NA
NA
NA
March
NA
Very
Good
NA
Very Good
Good
NA
NA
NA
April
NA
Poor
NA
Very Good
Very
Good
Good
Very
Good
NA
May
NA
Poor
NA
Good
Poor
Good
Very
Good
NA
June
NA
Fair
NA
Good
Poor
Fair
Very
Good
NA
July
NA
Poor
NA
Very Good
Good
Poor
Very
Good
NA
August
NA
Fair
NA
Very Good
Very
Good
NA
Very
Good
NA
September
NA
Very
Good
NA
Good
Very
Good
NA
Very
Good
NA
October
NA
Very
Good
NA
Very Good
Very
Good
NA
Very
Good
NA
November
NA
Good
NA
NA
NA
NA
NA
NA
December
NA
Good
NA
NA
NA
NA
NA
NA
83
-------
LSPC Calibration
5.4 Total Nitrogen
Insufficient data were available for a quantitative evaluation of model performance with regards to total
nitrogen. Line graphs comparing predicted to observed values are presented in Appendix B to facilitate a
qualitative evaluation. Figure 5-2 shows an example line graph from the total nitrogen calibration, where
observed grab samples are compared to model output for the Tongue River at the State Line. Calibration
of total nitrogen in the Tongue River Reservoir is discussed in Section 6.3.
• Measured Modeled
Figure 5-2. Time series of TN data for the Tongue River at State Line near Decker, Montana (USGS
gage 06306300).
5.5 Total Phosphorus
Insufficient data were available for a quantitative evaluation of model performance with regards to total
phosphorus. Line graphs comparing predicted to observed values are presented in Appendix B to
facilitate a qualitative evaluation. Figure 5-3 shows an example line graph from the total phosphorus
calibration, where observed grab samples are compared to model output for the Tongue River at the State
Line. Calibration of total phosphorus in the Tongue River Reservoir is discussed in Section 6.3.
84
-------
Performance and Evaluation
0.8
0.7
? 0.6
O)
— 0.5
U)
3
O
£ 0.4
Q.
U)
O
£ 0.3
re
° 0.2
0.1
0.0
Jun-00 Dec-00 Jun-01 Dec-01 Jun-02 Dec-02 Jun-03 Dec-03 Jun-04 Dec-04 Jun-05 Dec-05 Jun-06
Figure 5-3. Time series of TP data for the Tongue River at State Line near Decker, Montana (USGS
gage 06306300).
5.6 Salinity and SAR Calibration in the Tongue River Reservoir
There are few salinity and SAR data available for the Tongue River Reservoir. Because of this, the
salinity and SAR calibration of the reservoir occurred at the USGS gage located just downstream of the
Tongue River Reservoir Dam. The CE-QUAL-W2 model and calibration of nutrients in the reservoir are
discussed in Section 6.0.
Simulated flows mimic daily and seasonal patterns at the USGS gage downstream of the Tongue River
Reservoir, and hydrology calibration targets (as described in Table 4-1) are met for most months and
seasons. The exception is during May, where the model tends to over predict flows (see Appendix B).
Daily observed salinity and SAR data at the gage downstream of the Tongue River Reservoir provided a
robust dataset for calibration. Trends for both parameters were well matched over the available period
(October 1, 2000 to September 20, 2006), and the mean error targets were met for most of the evaluated
periods. April, May, and June were the exception (and July for SAR), where the model tended to under
predict both salinity and SAR values.
5.7 Model Limitations
• Measured Modeled
The Tongue River LSPC model is only capable of representing processes that are captured in the model
input data (see Section 3). Events that are unknown to the models, such as storm events, undocumented
point source discharges, or undocumented flow alterations, cannot be replicated. Therefore, limitations in
the input data drive the limitations, error, and uncertainty in the LSPC models. The following sections
summarize the known limitations in the model input data, and how these data limitations potentially affect
model output.
85
-------
LSPC Calibration
5.7.1 Weather Data
As discussed in Section 3.4, weather data (e.g., temperature, precipitation, cloud cover, PEVT, wind
speed) are critical for running the LSPC model. Precipitation data are the source for all modeled flows,
while other weather data control temperature, snowfall, and evaporation processes. Therefore, the
accuracy of modeled flows (and water quality, indirectly) tends to increase as the number of weather
gages increases. Unfortunately, as described in Section 3.4, weather gages near the Tongue River
watershed are sparse. On average, a single weather gage provides data for an average area of 338 square
miles, and several thousand feet of elevation change (depending on location). The lack of weather gages
was most evident when trying to calibrate storm events and snowmelt events. Nevertheless, the SNOTEL
gages in the mountain region were more densely distributed. This significantly improved the quality of
the high-altitude snowfall/snowmelt representation, which accounts for about 63 percent of all the water
in the Tongue River watershed.
5.7.2 Flow Alterations
Flow alterations (diversions, storage, releases) are pervasive throughout the Tongue River watershed.
The location of the flow alteration, as well as the volume and timing of flow, is required to accurately
model stream flows and water quality. While some diversions and reservoirs had daily flow data, most
had little to no data. The lack of data affected the flow calibration process, particularly during the
summer months when flow alterations are present. The storage and management of water from the
Tongue River Reservoir also proved to be a major limitation to the model performance. Detailed data
about reservoir management and overflows would help to strengthen model performance in the Tongue
River downstream of the Tongue River Reservoir.
5.7.3 Point Sources Discharges
As described in Section 3.7, point source discharges have the potential to affect flow and water quality in
a stream. The LSPC model can account for these sources by using time-series inputs of flow and
concentrations. However, most point sources only report data on a monthly basis (or less frequently), and
data was extrapolated to provide daily model input. In other cases, very little information was available
about the point source, and best professional judgment was used to estimate flow, timing, water quality,
and/or outfall location. Point source uncertainties have the greatest potential to affect model output
during low flow events, when point sources make up a larger percentage of the load.
5.7.4 Physiographic Characteristics
LSPC is driven by the basic physiographic characteristics that make up a watershed - land use, soils,
slopes, and geology (see Section 3.5). Therefore, physiographic data must be accurate and complete for
each watershed. Potential errors were introduced into the model because several of these physiographic
characteristics were simplified to facilitate modeling. Also, physiographic characteristics change over
time, and may or may not be represented by the available data and the chosen calibration period.
However, this process most likely does not introduce much modeling error when compared to the other
potential sources or error.
5.7.5 Observation Data
There is always the possibility of analytical uncertainty in any reported observation that derives from the
inherent imprecision of analytical techniques, and, occasionally, from laboratory analysis and reporting
86
-------
Performance and Evaluation
errors. Perhaps more importantly, grab samples submitted for chemical analyses represent a specific
location and point in time that is not entirely consistent with the spatial and temporal support of the
model. LSPC represents waterbodies as discrete reaches, which are assumed to be fully mixed. Real
waterbodies vary continuously in both longitudinal and lateral dimensions, as well as in time. A sample
taken from a specific location may not be representative of the average concentration across the stream
cross-section, and even less representative of the average across an entire model reach. Further, a sample
taken at a discrete point in time may not be representative of the average concentration that would be
observed across a modeling time step - particularly when the sample is taken near a source of discharge
or during the course of a runoff event. This phenomenon most likely introduces model error during storm
events or during periods with short-term discharges.
5.7.6 Hydrology Calibration Data
A lack of hydrology calibration data were one source of model error. Few flow gages met the LSPC
calibration criteria (i.e., gages had too little data and/or not enough recent data). For example, there was
generally a lack of current data with which to calibrate the tributaries, and no long-term calibration gages
were available for smaller Great Plains subwatersheds. The result of the lack of flow gages is that
varying flow errors are introduced throughout the watersheds. The errors are not quantifiable, simply
because there are no other flow gages with which to validate the hydrologic calibration.
5.7.7 Water Chemistry Calibration Data
While there were over 100 stations with water chemistry data in the Tongue River watershed, most had
few recent data. Stations with the most recent data were used to calibrate water chemistry. The available
data at most stations generally consisted of discrete grab samples collected over a period of several years.
As a result, there was generally insufficient data to calibrate to all potential conditions throughout the
watershed, such as storm events, low flows, high flows, and spring snowmelt. Specifically, the following
water chemistry data gaps were identified:
• Few recent nitrogen and phosphorus monitoring data were available for streams throughout the
Tongue River watershed. Specifically, few data regarding nitrogen and phosphorus species (i.e.,
nitrate, nitrite, organic nitrogen, and orthophosphorus) were available. Because of this, the model
could only be calibrated to total nitrogen and total phosphorus concentrations (which were also
limited throughout the watershed).
• Aside from the 12 stations with continuous sampling, most stations had few recent salinity or
SAR data. On average, one water chemistry station was used to calibrate 450 square miles of
watershed area.
• Monitoring data for all parameters of concern in the Tongue River Reservoir were limited, and
depth profiles were only available from a small number of sampling events in 2001 and 2003.
• Groundwater chemistry data were limited for all parameters of concern.
• No water chemistry data were available for the high altitude reservoirs, high altitude diversions,
and interbasin diversions.
• Limited water chemistry data were available for irrigation return flows and irrigation impacts to
groundwater quality.
87
-------
-------
CE-QUAL-W2
6.0 CE-QUAL-W2 MODEL SETUP AND CALIBRATION
As described in Section 2.5, a CE-QUAL-W2 (W2) model was chosen to simulate nutrient and
eutrophication processes in the Tongue River Reservoir. Two different types of models were necessary to
simulate conditions within the Tongue River watershed. A watershed model (LSPC) was used to address
the generation of loads over the land surface, through groundwater, and to predict the resulting impact on
stream water chemistry upstream of the Reservoir. A separate model was necessary to simulate nutrients
within the Tongue River Reservoir. The U.S. Army Corps of Engineers CE-QUAL-W2 (W2) model was
selected (see Section 2.5 for a description of the W2 model) and the setup and calibration of the Tongue
River Reservoir model is described in the following section.
6.1 CE-QUAL-W2 Model Configuration
Configuration of the W2 model involved setting up a computational grid using available bathymetry data
and setting initial conditions, boundary conditions, and hydraulic and kinetic parameters for the
hydrodynamic and water chemistry simulations. This section describes the configuration and key
components of the model.
89
-------
CE-QUAL-W2
Major Hoods
Shoreline
Contours <4 ft i
Main stem Ton pe River
Figure 6-1. Tongue River Reservoir bathymetry
90
-------
CE-QUAL-W2
6.1.1 CE-QUAL-W2 Segmentation/Computational Grid Setup
The computational grid defines how the Tongue River Reservoir is represented in the W2 model. The
reservoir was represented as a single main branch based on its long and narrow shape. Montana DNRC
provided information on the reservoir's bathymetry following the 1996—1999 rehabilitation project and
this was used to generate the computational grid for the W2 model (DNRC, 2005). The average segment
width, depth, and orientation information were derived from the bathymetry information and bottom
roughness and initial water surface elevation was assigned for each segment. The initial water surface
was set at the maximum of 1028 meters and the default bottom roughness of 70 suggested in the W2
manual was assigned.
It should be noted that a portion of the upstream area defined by the Montana DNRC bathymetry file is
usually dry, which poses a limitation for W2. Since this area contributes a very small portion of the
overall volume of the reservoir, it was not included in the model setup. The difference between the actual
reservoir volume and the volume in the model bathymetric file was less than one percent.
The model was configured with 15 longitudinal segments with lengths ranging from 435 to 930 meters.
The model also contains up to a maximum of twelve 1-meter thick vertical layers. The model
segmentation and longitudinal profile of the reservoir is shown in Figure 6-2 and Figure 6-3. Note that
only the active cells are shown in the figures, and the first and last cells are not shown. The active cells
represent the cells that may contain water during the simulation. W2 requires that the user also specify
boundary cells (having zero widths) as part of the computational grid.
91
-------
CE-QUAL-W2
92
-------
CE-QUAL-W2
XXXXXXXXXX)
XXXXXXXXXX)
XXXXXXXXXX)
XXXXXXXXXX)
XXXXXXXXXX)
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
X X X X X X
> X X X X >
X 1
< X X X X
< X X X X
< X X X X
X !
: x x x x
: x x x
r x x x
8 10
Segment #
12
14
16
18
Figure 6-3. Longitudinal profile of the Tongue River Reservoir
93
-------
CE-QUAL-W2
6.1.2 Initial Conditions
The W2 model requires the user to specify initial temperature and water chemistry conditions at the start
of the model run. A constant initial temperature of 0° C (January 2000) was specified throughout the
reservoir and initial condition values for water chemistry parameters were based on observed in-reservoir
monitoring data from sampling conducted in the year 2000.
The number and location of inflow/outflows also must be specified when defining initial conditions in the
W2 model. For the Tongue River Reservoir, inflows were specified at segments 2 (branch inflow), 4
(tributary inflow), and 7 (tributary inflow). Outflow was specified at segment 16. An initial water
surface elevation was specified as 1028 meters (which is equal to the deepest point in the reservoir).
6.1.3 Boundary Conditions
Boundary conditions are required as inputs for the W2 model and represent external contributions of flow
and pollutants into the reservoir. The upstream boundary for the Tongue River Reservoir was set at the
USGS gage near the Montana-Wyoming state line (gage 06306300), which is approximately 3 miles
upstream of the reservoir (depending on reservoir storage). The downstream boundary condition of the
reservoir was established using the daily flow record for the USGS gage located immediately downstream
of Tongue River Reservoir Dam (gage 06307500).
6.1.4 Reservoir Outflow
Outflows from the model were computed based on a selective withdrawal algorithm in W2 that withdraws
water from each of the different vertical layers. The purpose of using this algorithm is to account for the
mixing and hydrodynamics in the system for each layer.
6.1.5 Point Sources
Three permitted point source discharges were input directly into the reservoir model (permit IDs shown in
parentheses):
• Spring Creek Coal Company (MT0024619)
• Decker Mine West (MT0000892)
• Decker Mine East (MT0024210)
EPA's Permit Compliance System (PCS) database reported no flow data for the Spring Creek Coal
Company for the calibration period; therefore no flow was input for this point source. Flows from outfall
007 of the Decker Mine West were assigned to segments 2 of the reservoir model, and flows from outfall
002 from the Decker Mine East were assigned to segment 4 (Figure 6-4). A distributed tributary with a
very small flow (1 cfs) was also configured for the model to account for nonpoint sources directly
adjacent to the reservoir.
94
-------
CE-QUAL-W2
Figure 6-4. Point source locations in Tongue River Reservoir model.
95
-------
CE-QUAL-W2
6.1.6 Meteorological Data
Meteorological data are an important component of
the W2 model because the surface boundary
conditions are determined by weather conditions.
The meteorological data required by the W2 model
are hourly air temperature, dew point temperature,
wind speed, wind direction, and cloud cover. The
nearest weather station with these data at the hourly
scale is the Sheridan County Airport station located
approximately 25 miles south of the reservoir
(Figure 6-5). Data from this station were also more
complete than data from other potential weather
stations.
SHUWCWN COUWV
~ Surfiv-c iWrfjyT' $rcre«
f I SUM
A/in11 Ftar/iifl fbt]
¦I n i?i
o 125- 175
I
3?S 375
MB V? . 4t>
¦i ^ o
¦I iXJ HO
H so >20
H 160
HI I BO. IUJ
II
+
Figure 6-5. Location of weather station used for Tongue River Reservoir modeling.
6.1.7 CE-QUAL-W2 Calibration Time Period
Calibration of the Tongue River Reservoir was complicated because the reservoir was rehabilitated
between 1996 and 2000, which changed the dam height, reservoir dimensions, and storage volume.
Because CE-QUAL-W2 cannot account for changing dimensions and capacity, the Tongue River
Reservoir was calibrated to post-rehabilitation conditions only (i.e., 2000 through 2006). The model was
calibrated to water chemistry data from the year 2001 and then validated to data from the year 2003.
96
-------
CE-QUAL-W2
6.2 Upstream Conditions
Flow and nutrient output from the Tongue River LSPC model provided the initial input for the Tongue
River Reservoir CE-QUAL-W2 model. Therefore, any uncertainties or error in the Upper Tongue River
LSPC model were carried through to the Tongue River Reservoir during preliminary calibrations. To
improve the Tongue River Reservoir calibration, the reservoir was calibrated as a separate system that
was not directly linked to the Upper Tongue River LSPC model. Rather, flow and water chemistry data
from the Upper Tongue River watershed model were simulated by using measured data from USGS gage
06306300 (Tongue River at the Montana-Wyoming State Line). Daily flows were available at this gage
from 1960 to present, while water chemistry data was collected at varying frequencies (quarterly or
monthly). For days with no water chemistry data, concentrations were interpolated from the two nearest
concentrations. Using this method, the Tongue River Reservoir could be calibrated without the bias from
the Upper Tongue River LSPC model. The following sections describe the methodology for determining
the upstream inputs to the CE-QUAL-W2 model.
6.2.1 Flow
Upstream flow was simply derived from the daily flow records available at USGS gage 06306300. By
using this methodology to calibrate the reservoir, potential flow errors from the Upper Tongue River
LSPC model were removed. However, the disadvantage to this method is that it assumes that there are no
additional flows or withdrawals between the USGS gage and the reservoir boundary. In reality, Badger
Creek flows into this segment, and several industrial outfalls are present.
6.2.2 Water Chemistry
The water chemistry component of the W2 model requires loading of dissolved and particulate organic
material, ammonia, nitrate-nitrite, ortho-phosphorous and dissolved oxygen. Data at USGS gage
06306300 were available from 2000 to 2003 (collected approximately once per month) and linear
interpolation was used to estimate daily concentrations between sampling dates.
Concentrations of organic matter were not available from the USGS gage and thus had to be estimated.
Based on previous modeling applications, dissolved organic material (DOM) loadings were estimated to
be one-half of the organic nitrogen load and particulate organic material (POM) loadings were estimated
to account for the remaining half of the organic nitrogen and the total organic phosphorous. The DOM
and POM form the source of carbon for the model. Table 6-1 presents the annual average water
chemistry concentrations for the various parameters for the period 2000 to 2003.
97
-------
CE-QUAL-W2
Table 6-1. Annual average water chemistry concentrations (mg/L) at the USGS gage along the
Tongue River at Decker, Montana (gage ID 06306300).
Date
Count
TDS
P04
NH4
NOx
LDOM
LPOM
DO
2000
16
393.55
0.007
0.041
0.068
0.172
0.211
10.29
2001
12
508.21
0.032
0.064
0.108
0.271
0.326
9.15
2002
10
410.09
0.009
0.038
0.061
0.216
0.252
9.66
2003
9
399.73
0.010
0.039
0.057
0.179
0.224
10.20
Notes: TDS = Total Dissolved Solids; P04 = Orthophosphorus; NH4 = Ammonia; NOx = Nitrate + Nitrite; LDOM = Labile Dissolved Organic Matter;
LPOM = Labile Particulate Organic Matter; DO = Dissolved Oxygen.
6.2.3 Temperature
Daily temperatures for each inflow into the reservoir were derived from historical temperature data (i.e.,
33 in-stream measurements from 1974 to 2003). A polynomial equation was fitted to the observed
historical data and a temperature time series for each Julian day was calculated using the derived equation
(Figure 6-6).
Figure 6-6. Polynomial fit used to derive temperature time series from observed historical data
98
-------
CE-QUAL-W2
6.3 CE-QUAL-W2 Model Calibration and Validation
After establishing the CE-QUAL-W2 input conditions (Section 6.2), the first step in the model calibration
process was to match observed reservoir water surface elevations and temperatures in the year 2000 to
maintain the water balance and reproduce the thermal structure of the reservoir. This step is known as the
hydrodynamic calibration. Water chemistry was calibrated following the successful hydrodynamic
calibration and then the model was run for the year 2003 (without changing any parameters) as part of the
model validation process.
Daily water surface elevation data for the calibration and validation time periods were not available for
the Tongue River Reservoir. However, monthly water surface elevation data were estimated from
monthly storage volumes provided by Montana DNRC. Daily water surface elevations calculated by the
model were converted to monthly values and evaluated against the estimated monthly water surface
elevation data (Figure 6-7). The model results follow the seasonal trend fairly well and have an estimated
absolute mean error of only 0.75 meters.
1055 n
1050
1035
1030
1025
• Obser\ed Election (m)
—i—Modeled Elevation (m)
J
- *
• • • • #
• •
• r
..•
• /
>CDO'Z.^Dcr~^K>
oooooOoAo
ooooo—
> CO o 2 o
o o o o o
(r7l^>,S(r(r}>(i0O2C>c-7,S>,S(r(r>,c0O
oocSoo°°oo6o6oocuoo00oo6
Figure 6-7. Water Surface Elevation Calibration.
During model calibration it became apparent that the reservoir was significantly affected by highly
dynamic and flashy weather patterns reported at the Sheridan County Airport. For example, wind speeds
were reported as greater than 20 meters/second during some months and diurnal air temperature was also
found to vary by as much as 30 °C on some days in October.
W2 allows for spatially adjusting the wind-sheltering coefficient (WSC) and values ranging from 0.60 to
0.85 were used in the model to address the dynamic weather patterns. The W2 model uses wind speed to
calculate evaporation.
99
-------
CE-QUAL-W2
Figure 6-8. Monitoring station locations.
100
-------
CE-QUAL-W2
Water chemistry and temperature data were available at three locations along the Tongue River Reservoir.
Figure 6-8 shows the monitoring station locations used for calibration1 Temperature and dissolved
oxygen monitoring profile data (at a 1 meter vertical resolution) for the years 2001 and 2003 were
available. Vertical profile data for salinity (EC) were available only in 2003.
Figure 6-9 and Figure 6-10 show the temperature calibration and validation respectively at each of the
three sampling locations. In general the model represents the summer stratification and fall turnover
fairly well and captures the temperature trend in the months with the highest evaporation (July and
August). However, the model does overestimate temperatures in September and October. This may be
due to a combination of errors in the estimated temperature boundary conditions and differences between
the meteorological data reported for Sheridan and those that occurred at the reservoir.
Calibration of the water chemistry model involved minor adjustments to the default recommended rate
coefficients. Figure 6-11 to Figure 6-12 compare the observed versus predicted dissolved oxygen for the
calibration and validation time periods, respectively. Considering the sparse boundary conditions, the
agreement was considered to be reasonable with the model representing the annual trend in the reservoir's
response and capturing the critical summer period (especially in 2003). However, the 2001 dissolved
oxygen trends in July and August do not seem to follow the observed trends as well. Figure 6-11 and
Figure 6-12 indicate that the model predicts hypolimnetic anoxic conditions fairly well except for several
periods in 2001. This is probably due to a variety of factors including the hydrodynamic calibration,
boundary conditions, lateral averaging of the model, etc.
The model was also run to predict nutrient (orthophosphorus) and algae concentrations, although only a
preliminary calibration was possible due to the limited data. Figure 6-13 and Figure 6-14 display the
results and indicate a fair agreement between observed and simulated conditions.
1 The data in 2001 were collected by MDEQ and the data in 2003 were collected by EPA. Due to differences in naming conventions for the stations by EPA and
MDEQ the stations are named Station 1, 2, and 3 in this report.
101
-------
CE-QUAL-W2
Station 1 7/26/2001
Temperature (C)
0 5 10 15 20 25 30
8/16/2001
Temperature (C)
5 10 15 20 25 30
O
O
o
o
o
o
o
9/26/2001
Temperature (C)
5 10 15 20 25 30
10/26/2001
Temperature (C)
5 10 15 20 25 30
O
O
o
o
o
o
o
o
o
o
o
Station 2 7/26/2001
Temperature (C)
0 5 10 15 20 25 30
0 -j 1 1 1 1—£—I
8/16/2001
Temperature (C)
0 5 10 15 20 25 30
9/26/2001
Temperature (C)
10/26/2001
5 10 15 20 25 30
O
o
o
o
o
o
Temperature (C)
0 5 10 15 20 25 30
) j I-qH 1 1 1 1
o
o
o
o
s 8
0
0
1
2 -
3 -
! 4 --
Is
Temperature (C)
5 10 15 20 25 30
0
0.0
1.0 --
2.0 --
3.0 --
1! 4.0 --
I 5-0
o
6.0
7.0
8.0
9.0
10.0
Temperature (C)
5 10 15 20 25 30
0
0.0 —
1.0 --
2.0 --
3.0 --
1! 4.0 --
I 5-0 -
o
6.0 --
7.0 --
8.0 --
9.0 --
10.0 --
Temperature (C)
5 10 15 20 25 30
0.0 —
1.0 -
2.0 -
3.0 -
£ 4.0 -
I 5-0 -
o
6.0 -
7.0 -
8.0 -
9.0 -
10.0 -
Temperature (C)
5 10 15 20 25 30
o
o
Observed
- Predicted
Figure 6-9. Temperature (deg C) calibration (2001).
102
-------
CE-QUAL-W2
Station 1
6/27/2003
Temperature (C)
0 5 10 15 20 25 30
7/29/2003
Temperature (C)
0 5 10 15 20 25 30
8/21/2003
Temperature (C)
0 5 10 15 20 25 30
s-10 -
Q
15 -
20 -
s-10 -
Q
15 -
20 -
O
o
o
o
8-1°-
Q
15 -
20 -
O
o
o
o
o
o
o
o
o
Station 2 6/4/2003
6/27/2003
7/29/2003
8/21/2003
Temperature (C)
0 5 10 15 20 25 30
Temperature (C)
0 5 10 15 20 25 30
Temperature (C)
0 5 10 15 20 25 30
8-10
Q
15 --
20
10 --
15 -L
&5-
Q
10 -
Temperature (C)
0 5 10 15 20 25 30
Station 3 6/4/2003
Temperature (C)
6/27/2003
Temperature (C)
7/29/2003
Temperature (C)
8/21/2003
Temperature (C)
0 5 10 15 20 25 30 0 5 10 15 20 25 30 0 5 10 15 20 25 30 0 5 10 15 20 25 30
o- 5
(D
Q
10
O
o
o
o
&5
Q
10
Q
10
Q
10
O Observed
Predicted
Figure 6-10. Temperature (deg C) validation (2003).
103
-------
CE-QUAL-W2
Station 1 7/26/2001
8/16/2001
9/26/2001
10/26/2001
DO (mg/l)
5 10 15
DO (mg/l)
5 10 15
DO (mg/l)
5 10 15
10
DO (mg/l)
5 10 15
—a
o
o
o
o
o
o
o
o
o
o
Station 2 7/26/2001
9/26/2001
15 -
10/26/2001
DO (mg/l)
0 5 10 15 20
DO (mg/l)
0 5 10 15 20
10 -
o
o
DO (mg/l)
0 5 10 15 20
O
o
o
DO (mg/l)
0 5 10 15 20
Station 3 7/26/2001
8/16/2001
9/26/2001
0
1
2
3
14
-C
£5
° 6
7
8
9
10
DO (mg/l)
5 10 15
20
0
0.0
1.0
2.0
3.0 -
§, 4.0 -
I 5-0 -
O
6.0 -
7.0
8.0
9.0
10.0
DO (mg/l)
5 10 15
O
o
20
0.0 —
1.0 --
2.0 --
3.0 --
£ 4.0 --
I 5-0 -
° 6.0 --
7.0 --
8.0 --
9.0 --
10.0 --
DO (mg/l)
5 10 15 20
O
o
0
0.0 —
1.0 --
2.0 --
3.0 --
£ 4.0 --
I 5-0 -
o
6.0 --
7.0 --
8.0 --
9.0 --
10.0 --
10/26/2001
DO (mg/l)
5 10 15
20
O Observed
Predicted
Figure 6-11. Dissolved oxygen calibration (2001).
104
-------
CE-QUAL-W2
Station 1 6/27/2003 7/29/2003 8/21/2003
DO (mg/l) DO (mg/l) DO (mg/l)
Station 3 6/4/2003
6/27/2003
7/29/2003
8/21/2003
DO (mg/l)
0 5 10 15 20
DO (mg/l)
0 5 10 15 20
DO (mg/l)
0 5 10 15 20
DO (mg/l)
0 5 10 15 20
&5"
o
10 -
o
o
o
o
O- 5 -
CD
Q
10 -
10 -
&5"
O
10 -
O Observed
Predicted
Figure 6-12. Dissolved Oxygen validation (2003).
105
-------
CE-QUAL-W2
Figure 6-13. Chlorophyll-a calibration/validation at Station 1.
o Observed
Predicted
r
05
CD
D
CD
8
05
CD
"O
8
Figure 6-14. P04 calibration/validation at Station 1.
106
-------
CE-QUAL-W2
6.4 CE-QUAL-W2 Modeling Coefficients
Various modeling coefficients are needed to describe the water chemistry reaction rates in the reservoir.
Initial estimates were obtained from W2 default values, general literature values (USEPA, 1985), and
from the W2 User's Manual (Cole, 2003). These coefficients were then refined, as necessary, through
iterative model simulations until the model captured the major processes influencing the reservoir and
reasonably predicted the observed data.
The hydrodynamic calibration mainly involved adjusting the wind sheltering coefficients. W2 allows for
spatially adjusting the wind-sheltering coefficient (WSC). The hydrodynamic calibration was found to be
sensitive to the wind. The Tongue River system has highly dynamic and flashy weather patterns with
wind speeds greater than 20 m/s during some months and diurnal air temperatures were found to vary by
approximately 30 deg C in October. A WSC of 0.6 to 0.85 was used in the model.
The water chemistry calibration coefficients as well as the phytoplankton calibration coefficient data for
the reservoir are presented below in Table 6-2. Default W2 parameters were in general found to be
acceptable. The model was also sensitive to the SOD values. A value of 1.25 g-cm"2day_1 was found to
achieve the most reasonable DO calibration. No sediment oxygen demand measurements have been
made for the Tongue River Reservoir and this value was within the range of that reported in the literature
(Cole, 2003). Minor adjustments were made to the temperature bounds for the algal rates to better mimic
the observed dissolved oxygen concentration observations in the lake.
107
-------
CE-QUAL-W2
Table 6-2. Kinetic Coefficients used in the calibration of Tongue River Reservoir.
Parameter
Description
Units
Value
P04R
Sediment release rate of phosphorous
fraction of SOD
0.015
N03DK
Nitrate decay rate
day"1
0.050
N03T1
Lower temperature for nitrate decay
°C
5
N03T2
Upper temperature for nitrate decay
°C
30
N03K1
Lower temperature rate multiplier for nitrate decay
Unitless
0.10
N03K2
Upper temperature rate multiplier for nitrate decay
Unitless
0.99
NH4DK
Ammonium decay rate
day"1
0.12
NH4R
Sediment release rate of ammonium
Fraction of SOD
0.05
NH4T1
Lower temperature for ammonium decay
°C
5
NH4T2
Upper temperature for ammonium decay
°C
30
NH4K1
Lower temperature rate multiplier for ammonium decay
Unitless
0.10
NH4K2
Upper temperature rate multiplier for ammonium decay
Unitless
0.99
SOD
Sediment Oxygen Demand
g02 cm"2day"1
1.25
AG
Growth rate
day"1
2.0
AR
Dark respiration rate
day"1
0.04
AE
Excretion rate
day"1
0.04
AM
Mortality rate
day"1
0.10
AS
Settling rate
day"1
0.1
AHSP
Phosphorous half-saturation coefficient
g.m"3
0.003
AHSN
Nitrogen half-saturation coefficient
g.m"3
0.014
ASAT
Light saturation
W.m"3
100
AT1
Lower temperature for minimum algal rates
°C
1
AT2
Lower temperature for maximum algal rates
°C
25
AT 3
Upper temperature for minimum algal rates
°C
28
AT4
Upper temperature for maximum algal rates
°C
30
AK1
Lower temperature rate multiplier for minimum algal rates
0.10
AK2
Lower temperature rate multiplier for maximum algal rates
0.99
AK3
Upper temperature rate multiplier for minimum algal rates
0.99
AK4
Upper temperature rate multiplier for maximum algal rates
0.10
108
-------
CE-QUAL-W2
6.5 CE-QUAL-W2 Data Limitations
The Tongue River Reservoir CE-QUAL-W2 model is only capable of representing processes that are
captured in the model input data (see Section 6.1). Events that are unknown to the model, such as storm
events, illicit discharges, or flow alterations, cannot be replicated. Therefore, limitations in the input data
drive the limitations, error, and uncertainty in the CE-QUAL-W2 model. The following sections
summarize the known limitations in the model input data, and how these data limitations potentially affect
model output.
6.5.1 Point Sources Discharges
As described in Section 6.1.5, point source discharges have the potential to affect flow and water
chemistry in the Tongue River Reservoir. The CE-QUAL-w2 model can account for these sources by
using time-series inputs of flow and concentrations. However, most point sources only report data on a
monthly basis (or less frequently), and data was extrapolated to provide daily model input. In other cases,
little information was available about the point source, and best professional judgment was used to
estimate flow, timing, water chemistry, and/or outfall location.
6.5.2 Observation Data
There is always the possibility of analytical uncertainty in any reported observation that derives from the
inherent imprecision of analytical techniques, and, occasionally, from laboratory analysis and reporting
errors. Perhaps more importantly, grab samples submitted for chemical analyses represent a specific
location and point in time that is not entirely consistent with the spatial and temporal support of the
model. Real reservoirs vary continuously in both longitudinal and lateral dimensions, as well as in time.
A sample taken from a specific location may not be representative of the average concentration across the
cross-section, and even less representative of the average across an entire model reach. Further, a sample
taken at a discrete point in time may not be representative of the average concentration that would be
observed across a modeling time step - particularly when the sample is taken near a source of discharge
or during the course of a runoff event.
6.5.3 Water Chemistry Calibration Data
There were few recent water chemistry data for the Tongue River Reservoir. Post-rehabilitation data
were only available for 2001 (4 sampling events) and 2003 (6 sampling events), and samples were only
collected between April and October of each year. Samples were obtained at three sites during each
event. Dissolved oxygen depth profiles were available for both years, but nutrient depth profiles were
only available for 2003. The lack of data for the reservoir proved to be a major complication for
calibration. No data was available for calibrating winter conditions (November through March), and few
data were available to describe water chemistry concentrations at depth.
109
-------
-------
Uncertainty
7.0 UNCERTAINTY
Uncertainties in the scientific sense are a component of all aspects of the modeling process and are due to
(USEPA, 2003):
¦ uncertainty in the underlying science and algorithms of a model (model framework uncertainty)
¦ data uncertainty
¦ uncertainty regarding the appropriate application of a model
Identifying the types of uncertainty that significantly influence model outcomes and communicating their
importance is critical to successfully integrating information from models into the decision-making
process. This report has attempted to identify the key aspects of uncertainty associated with the Tongue
River modeling effort by:
¦ Reporting model uncertainty (discrepancy between observation and prediction) for both the
hydrologic and water quality calibration results (see Appendix B and Section 6.3).
¦ Documenting the known data limitations and uncertainty used for model inputs (see Section 5.7
and 6.5).
¦ Discussing the potential implications of model uncertainty as they relate to the intended use of the
model (see Section 7.1).
7.1 Model Use
To date, the LSPC model's primary purpose has been to support the development of the Tongue River
Assessment Report. Specifically, the purpose of the model is to:
1) Simulate stream flows in Hanging Woman, Otter, and Pumpkin Creek for time periods when no
monitoring data were available to assist in the evaluation/description of drought conditions.
2) Estimate the magnitude of hydrologic change associated with anthropogenic flow alterations
associated with stock ponds and irrigation withdrawals/returns in the Hanging Woman, Otter, and
Pumpkin Creek and the Tongue River watersheds.
3) Estimate daily and average monthly SC and SAR under various scenarios (i.e., existing conditions
and the "natural" condition) for comparison to the Montana numeric instantaneous maximum and
average monthly SC and SAR standards.
4) Estimate nutrient loads from the Upper Tongue River watershed to the Tongue River Reservoir.
Based on the data limitations and uncertainty, the model is better suited to answer some of these questions
than others. Each of the model uses are presented separately below followed by a discussion of model
uncertainly and limitations.
(1) Simulate stream flows in Hanging Woman, Otter, and Pumpkin Creek for time periods when no
monitoring data were available to assist in the evaluation/description of drought conditions.
Over the long term, as shown in Table 5-2, prediction of average flow was fair (-20%) in Hanging
Woman Creek. For the growing season, predicted average flow was good (-12%) in Hanging Woman
Creek. Therefore, the model is reasonably well-suited for long-term predictions of average stream flows
in Hanging Woman Creek. Model performance regarding the prediction of stream flow in Otter and
Pumpkin Creeks, however, was generally poor.
111
-------
Uncertainty
It should be noted that the purpose of these model predictions (see Appendix H of the Assessment Report)
was informational only. So long as the uncertainty is acknowledged, it is felt that use of the model results
is appropriate.
(2) Estimate the magnitude of hydrologic change associated with anthropogenic flow alterations
associated with stock ponds and irrigation withdrawals/returns in the Hanging Woman, Otter,
Pumpkin Creek, and Tongue River watersheds.
The assumptions associated with simulation of irrigation and stock ponds are presented in Sections 3.8
and 3.11 and in Appendix J of the Assessment Report. In the absence of field data describing their
hydrologic characteristics, it was not possible to specifically calibrate hydrology associated with these
two sources of human-caused flow alteration. The hydrologic processes and fate and transport of
chemical constituents were characterized by the LSPC model using a literature-based understanding
combined with best professional judgment and consideration in the overall hydrologic calibration. The
inability to specifically calibrate precludes an assessment of how well the model has simulated these
features.
Donigian (2000) points out that factors such as the availability of alternative assessment procedures and
the purpose of the model application need to be considered when allowing for uncertainty. In this case, at
the watershed scale, there is no alternative assessment procedure for estimating the potential magnitude of
hydrologic change associated with anthropogenic flow alteration. While the simulations of these two
sources are not calibrated, it is felt that it provides the best means for assessing the hydrologic effects of
irrigation and stock ponds in the absence of monitoring data.
Given the un-quantified uncertainty, model predictions associated with potential impacts of anthropogenic
flow alteration should be used with caution.
(3) Estimate daily and average monthly SC and SAR under two scenarios (i.e., existing conditions
and the "natural" condition) for comparison to the Montana numeric instantaneous maximum and
average monthly standards (i.e., relative comparisons between two modeled scenarios).
The performance of the model for the prediction of SC and SAR is summarized in Section 5.2 and 5.3 and
in Appendix B. The "existing condition" and "natural condition" scenarios are described in Appendix J of
the Assessment Report.
Existing Condition SC
For the "existing condition", performance of the model relative to prediction of average monthly and
instantaneous maximum SC can be determined directly through examination of the calibration results.
The calibration results vary by water body and site. As shown in Table 5-3, for most months and time
periods greater than one month, model performance for the prediction of average SC is very good (i.e.,
plus or minus 15 percent or less) in the Tongue River near the State Line, below the Tongue River
Reservoir, and at Miles City. Insufficient data were available to quantitatively evaluate model
performance in the Tongue River at Dayton, Wyoming. The time series chart showing predicted and
observed SC in the Tongue River at Dayton (Figure B-7 in Appendix B), however, qualitatively suggests
that the LSPC predictions of SC are good at this location.
Insufficient data were available for a quantitative analysis of model performance during the non-growing
season and associated months in the Tongue River at Miles City and Hanging Woman Creek, Otter Creek,
and Pumpkin Creek. At the long-term (annual or greater) and growing season time scales at these four
112
-------
Uncertainty
locations, the LSPC predictions of average SC were good (i.e., plus or minus 15 to 25 percent) and very
good (i.e., plus or minus 15 percent or less).
Model performance at the daily time step was evaluated by comparing observed average daily values to
predicted average daily values using regression analysis and box plots (see Appendix B). R-squared
values ranged from 0.01 in Pumpkin Creek to 0.51 in the Tongue River at State Line indicating a general
poor fit of observed to predicted average daily values. The poor correlations between observed and
predicted daily average values likely reflects the difficulties in exactly duplicating the timing of flows,
given the uncertainties in the timing of model inputs, mainly precipitation.
The box plot analysis, among other things, allows for ready examination of the model performance
relative to the prediction of maximum and minimum values. In most cases, the maximum and minimum
values are over or under predicted (see Appendix B for magnitude) suggesting that model results should
be used with caution for daily maximum or minimum values (Table 7-1).
Table 7-1. Qualitative comparison of predicted and observed minimum and maximum SC values
(Symbols indicate under or over prediction)
Location
Minimum
Maximum
Tongue River at Dayton, WY
T
T
Tongue River at State Line
T
T
Tongue River below the TRR
~
~
Tongue River below Brandenberg
Bridge
T
~
Tongue River at Miles City
T
-
Hanging Woman Creek near Birney
T
-
Otter Creek at Ashland
T
~
Pumpkin Creek near Miles City
T
T
~ = Over prediction
T = Under prediction
— = Approximately equal
In summary, use of the LSPC model developed for the Tongue River watershed for the prediction of
average daily SC for a given day would likely be unreliable. On the other hand, predictions at the
monthly or greater time step are generally good to very good.
Existing Condition SAR
As with SC, performance of the model relative to prediction of average monthly and instantaneous
maximum SAR under the "existing condition" scenario can be determined directly through examination
of the calibration results. The calibration results vary by water body and site. As shown in Table 5-4,
where sufficient data were available for a quantitative evaluation, for most months and time periods
greater than one month, model performance for the prediction of average SAR is good to very good (i.e.,
plus or minus 25 percent or less). The exceptions are the Tongue River at State Line and Miles City and
Hanging Woman Creek during some of the spring and summer months were model performance was poor
to fair.
Insufficient data were available for a quantitative evaluation of model performance for SAR at the
Tongue River near Dayton, below the Tongue River Reservoir Dam, Pumpkin Creek, and during many of
the non-growing season months at the other site shown in Table 5-4.
Model performance at the daily time step was evaluated by comparing observed average daily values to
predicted average daily values using regression analysis and box plots (Appendix B). Sufficient data were
113
-------
Uncertainty
only available for regression analysis in Otter and Hanging Woman Creeks and the Tongue River at State
Line, below Brandenberg Bridge, and Miles City where R-squared values ranged from 0.04 in Hanging
Woman Creek to 0.45 in the Tongue River at State Line indicating a general poor fit of observed to
predicted average daily values. The poor correlations between observed and predicted daily average
values likely reflects the difficulties in exactly duplicating the timing of flows, given the uncertainties in
the timing of model inputs, mainly precipitation.
The box plot analysis, among other things, allows for ready examination of the model performance
relative to the prediction of maximum and minimum values. In most cases, the maximum and minimum
values are over or under predicted (see Appendix B for magnitude) suggesting that model results should
be used with caution for daily maximum or minimum values (Table 7-2).
Table 7-2. Qualitative comparison of predicted and observed minimum and maximum SAR values
(Symbols indicate under or over prediction)
Location
Minimum
Maximum
Tongue River at Dayton, WY
T
T
Tongue River at State Line
—
~
Tongue River below the TRR
—
T
Tongue River below Brandenberg
Bridge
—
~
Tongue River at Miles City
—
—
Hanging Woman Creek near Birney
—
~
Otter Creek at Ashland
~
—
Pumpkin Creek near Miles City
-
T
~ = Over prediction
T = Under prediction
— = Approximately equal
In summary, use of the LSPC model developed for the Tongue River watershed for the prediction of
average daily SAR for a given day would likely be unreliable. On the other hand, predictions at the
monthly or greater time step are generally good to very good.
Natural Condition SC and SAR
As described in Appendix J of the Assessment Report, a number of anthropogenic factors such as
irrigation, agriculture, CBM discharge, wastewater treatment discharge, mining, etc. were removed from
the model to estimate the potential magnitude of human affect.
In the absence of field data describing the hydrologic and pollutant fate/transport characteristics
associated with many of these factors, it was not possible to specifically calibrate SC and SAR loading
from these sources. These sources were addressed in the model using a literature-based understanding of
their characteristics.
However, the potential magnitude of change between the existing and natural conditions was based on a
relative comparison of two model scenarios, thereby minimizing the error/uncertainty associated with
model fit to the observed data. As a result, the uncertainty associated with comparisons of SC and SAR
between the two scenarios is largely a function of the model's ability to simulate each of the various
anthropogenic factors. While it is not possible to evaluate the model's ability to simulate each of these
sources, no other assessment methodology is currently available to estimate what water quality conditions
might have been like in the absence of man's influence. As a result, the model provides one of the only
means for evaluating the impact of human's actions at the watershed scale.
114
-------
Uncertainty
(4) Estimate nutrient loads from the Upper Tongue River watershed to the Tongue River Reservoir.
Two model scenarios were developed; one for the existing condition and one for the natural condition
(i.e., the absence of human actions). For the existing condition, performance of the LSPC model relative
to prediction of nutrient loads to the Tongue River Reservoir can be evaluated by examination of the flow
and water quality calibration results for the Tongue River at the State Line near Decker, Montana. As
summarized in Table 5-2, model performance with regards to flow at the State Line was very good (± <
10%) for time periods greater than monthly (i.e., growing season, non-growing season, full period of
record). However, as shown in Appendix B (Section B.3.2.3 and B3.2.4), the LSPC model appears to
under predict TN and over predicts peak TP.
Similar to SC and SAR (Number 3 above), a number of anthropogenic nutrient sources were then
removed from the model to estimate the potential magnitude of human affect (i.e., define the "natural
condition).
In the absence of field data describing the hydrologic and pollutant fate/transport characteristics
associated with many of the nutrient sources, it was not possible to specifically calibrate TN and TP
loading from these sources. These sources were addressed in the model using a literature-based
understanding of their characteristics.
However, the potential magnitude of change between the existing and natural conditions was based on a
relative comparison of two model scenarios, thereby minimizing the error/uncertainty associated with
model fit to the observed data. As a result, the uncertainty associated with comparisons of nutrient
loading between two model scenarios is largely a function of the model's ability to simulate each of the
various anthropogenic factors. While it is not possible to evaluate the model's ability to simulate each of
these sources, no other assessment methodology is currently available to estimate what water quality
conditions might have been like in the absence of man's influence. As a result, the model provides one of
the only means for evaluating the impact of human's actions at the watershed scale.
7.2 Conclusions
As described above, the model has been used primarily to; 1) fill-in spatial and temporal gaps where no
monitoring data are available for SC, SAR, and flow and 2) to make relative comparisons between two or
more model scenarios considering SC, SAR, nutrients, and flow. The former use has been largely for
descriptive purposes. While acknowledging uncertainty is still important, the consequences of prediction
error are relatively insignificant. The later use has been to answer "what if' questions such as what would
SC, SAR, nutrient, and/or temperature levels be in the absence of human influence? Since answers to
these types of questions may be used to inform future decisions, uncertainty and an understanding of
prediction error may be important.
However, is should be noted that the purpose of the model has been to answer questions that can only be
answered through simulation since sufficient monitoring data do not exist to answer the questions by
other means. For example, no data exist from a period in time prior to the onset of human influence. The
only way to estimate what water quality conditions may have been like at that time, in the absence of
monitoring data from that period in time and/or data from a suitable reference stream, is through model
simulation. Thus, the models described in this document have been developed and used because there is
no alternative method for obtaining answers to the questions that need to be asked to ultimately interpret
Montana's water quality standards (i.e. estimate "natural conditions") and/or assess the potential
magnitude of impact associated with man's past, current, and future influences on water quality.
115
-------
Uncertainty
In general, the fit between observed and predicted SC and SAR is good to very good at time scales greater
than one month at most locations and there is greater uncertainty associated with the non-growing season
months than the growing season months. Uncertainty associated with daily predictions and for prediction
of instantaneous maximum values is relatively high and the model should only be used with caution at
these time scales.
Qualitatively, it appears that a reasonable fit between predicted and observed TN and TP values was
obtained in the Upper Tongue River. So long as the model results are used with caution and in
combination with other data and information the model is adequately suited for the intended purpose
relative to nutrients.
In summary, the current quality of fit is sufficiently good that the model is judged ready for supporting
the development of the Tongue River Assessment Report. If a higher level of "proof' (validation or
corroboration) is required for regulatory application, it is recommended that appropriate data quality
objectives be pre-defined and tested on new sampling data obtained after the period used for model
calibration.
116
-------
Steps to Improve Performance
8.0 STEPS TO IMPROVE MODEL PERFORMANCE
The Tongue River LSPC and CE-QUAL-W2 models were set up and calibrated primarily to support the
specific purposes associated with the Tongue River water quality assessment. Despite model
uncertainties, the models are believed to be appropriate for water quality assessment purposes, especially
since model output was only one of several types of information used to make the assessment decisions
(i.e., a weight-of-evidence approach was employed that also relied on a variety of other data). The
models also have the potential to support a variety of future watershed management needs, such as TMDL
development, water quality standards compliance, and watershed-based permitting activities. However, a
detailed re-analysis of the modeling quality objectives should be undertaken to determine the suitability of
the existing modeling framework to support these potential decision needs. Additional model refinements
are likely necessary to achieve the new modeling quality objectives. For example, several potential
model refinements were identified during the calibration process including the following:
• The modeling of salt transport might be improved by incorporating a geochemical representation
of salt leaching and transport into the LSPC model. In this way the modeling of groundwater
discharges could be improved by taking into account geochemistry, water table elevation, and soil
characteristics.
• The modeling of nutrients in the Upper Tongue River could possibly be improved with more
information on point source discharges and stream characteristics that might affect nutrient
transport to the reservoir (e.g., location and extent of large periphyton beds).
A variety of data limitations also limited model performance and the following data needs were
identified:
• Weather data is believed to be the primary limitation in the Tongue River LSPC model
(particularly in the prairie region). Long-term use of the models would greatly benefit from
additional weather stations throughout the Tongue River watershed.
• Information on the location, timing, and volume of water diverted and returned from the main
stem Tongue River was limited and inconsistent. Studies to better understand this important
characteristic of the watershed would help the modeling process.
• Information on point source discharge characteristics was also limited and inconsistent. Better
data on key point sources would help to determine the magnitude of their impact.
• Additional nutrient and salinity sampling should be performed to update the LSPC and CE-
QUAL-W2 models. For example, sampling of Tongue River in-stream concentrations for P04
and NOx would help to determine the characteristics of incoming nutrient loads.
117
-------
-------
References
9.0 REFERENCES
Anderson , E.A., 1968: Development and Testing of Snow Pack Energy Balance Equations. Water
Resources Research, 4, 19-37.
Anderson, E.A. and N.H. Crawford. 1964. The Synthesis of Continuous Snowmelt Runoff Hydrographs
on a Digital Computer. Technical Report No. 36, Department of Civil Engineering, Stanford University,
Stanford, CA.
APHA. 1992. Standard Methods for the Examination of Water and Wastewater. 18th ed. American Public
Health Association, Washington, DC.
Bartos, T.T. and K.M. Ogle. 2002. Water Quality and Environmental Isotopic Analyses of Ground-Water
Samples Collected from the Wasatch and Fort Union Formations in Areas of Coalbed Methane
Development - Implications to Recharge and Ground-Water Flow, Eastern Powder River Basin,
Wyoming. Water-Resources Investigations Report 02-4045. U.S. Geological Survey, Cheyenne, WY.
Bedient, P. B., and Huber, W. C., 1992, Hydrology and Floodplain Analysis: Addison-Wesley Publishing
Company, Reading, Massachusetts, 692 p.
Bicknell, B.R., J.C. Imhoff, J. Kittle, A.S. Donigian, and R.C. Johansen. 1996. Hydrological Simulation
Program -FORTRAN, User's Manual for Release H. U.S. Environmental Protection Agency,
Environmental Research Laboratory, Athens, GA.
BLM. 2003. Final Environmental Impact Statement and Proposed Plan Amendment for the Powder
River Basin Oil and Gas Project. U.S Department of Interior Bureau of Land Management, Wyoming
State Office, Buffalo Field Office. Document WY-070-02-065. Buffalo, Wyoming. Available online at
http://www.wy.blm.gov/nepa/prb-feis/.
Butcher, J.B. 2003. Buildup, Washoff, and Event Mean Concentrations. Journal of the American Water
Resources Association. 39(6): 1521-1528.
Cole, T.M., and S. A. Wells. 2003. CE-QUAL-W2: A two-dimensional, laterally averaged,
Hydrodynamic and Water Quality Model, Version 3.1. Instruction Report EL-03-1. US Army
Engineering and Research Development Center, Vicksburg, MS.
Crawford, N.H. and R.K. Linsley. 1966. Digital Simulation in Hydrology: The Stanford Watershed
Model IV. Palo Alto, CA: Dept. of Civil Engineering, Stanford University; Tech. Rep. No. 39.
DNRC. 2003. Pond data for Southeast Montana [Computer File]. Montana Department of Natural
Resources and Conservation [Producer and Distributor], Helena, Montana.
DNRC. 2004. DNRC Water Resources Division Annual Report - 2004. Montana Department of
Natural Resources and Conservation. Helena, Montana. Available online at http://dnrc.mt.gov/About_Us/
publications/2004/wrd.pdf (accessed February 26, 2007).
DNRC. 2005. CAD files for the Tongue River Reservoir [Computer File]. Montana Department of
Natural Resources and Conservation [Producer and Distributor]. Helena, Montana.
119
-------
References
DNRC. 2006a. Draft Montana Watershed Boundaries 6th Code HUCs (Dated November 2006)
[Computer File]. Montana Department of Natural Resources and Conservation [Producer and
Distributor], Helena, Montana. Available online at http://nris.state.mt.us/nsdi/watershed/index.html
(Accessed December 12, 2006).
DNRC. 2006b. Daily Flow Data for the T&Y Diversion near Miles City, Montana [Computer File].
Montana Department of Natural Resources and Conservation [Producer and Distributor], Helena,
Montana.
Donigian, A. S., Jr., J. C. Imhoff, B.R. Bicknell, and J. L. Kittle, Jr. 1984. Application Guide for the
Hydrological Simulation Program - FORTRAN. EPA 600/3-84-066. U. S. EPA, Environmental Research
Laboratory, Athens, Georgia.
Donigian, Jr., A.S., 2000. HSPF Training Workshop Handbook and CD. Lecture #19. Calibration and
Verification Issues. Presented and prepared for U.S. EPA, Office of Water, Office of Science and
Technology, Washington, D.C.
Donigian. A.S. 2003. Watershed Model Calibration and Validation: The HSPF Experience.
Dunn, T. and Leopold, L.B. 1978. Water in Environmental Planning Freeman Press, San Francisco, 818
P-
Fidelity Exploration & Production Company. 2004. DMR Data for CBM Outfalls near Decker, Montana
[Computer File]. Fidelity Exploration & Production Company [Producer and Distributor]. Sheridan,
Wyoming.
Haith, D.A., R. Mandel, and R.S. Wu. 1992. GWLF, Generalized Watershed Loading Functions, Version
2.0, User's Manual. Dept. of Agricultural & Biological Engineering, Cornell University, Ithaca, NY.
Homer, C., C. Huang, L. Yang, B. Wylie, and M. Coan. 2004. Development of a 2001 National Land-
Cover Database for the United States. Photogrammetric Engineering & Remote Sensing, 70(7): 829-840.
Jensen, M.E. and Haise, H.R. 1963. Estimating Evapotranspiration from Solar Radiation. Journal of the
Irrigation and Drainage Division of the American Society of Civil Engineers. 89(IR4): 15-41.
Keith, Kristen, J. Bauder, and J. Wheaton. 2003. Frequently Asked Questions - Coal Bed Methane.
Montana State University. Available online at http://waterquality.montana.edu/docs/
methane/cbmfaq.shtml (accessed March 8, 2007).
Larson, L.R. and R. L. Daddow. 1984. Ground-Water-Quality Data from the Powder River Structural
Basin and Adjacent Areas, Northeastern Wyoming. Open-File Report 83-939. U.S. Geological Survey,
Cheyenne, WY.
Lewis, B.D., and Roberts, R.S. 1978. Geology and Water-yielding Characteristics of Rocks of the
Northern Powder River Basin, Southeastern Montana: U.S. Geological Survey Miscellaneous
Investigations Map I-847-D, scale 1:250,000, 2 sheets.
Lumb, A.M., R.B. McCammon, and J.L. Kittle, Jr. 1994. Users Manual for an Expert System (HSPEXP)
for Calibration of the Hydrological Simulation Program-Fortran. Water-Resources Investigations Report
94-4168. U.S. Geological Survey. Reston, VA.
120
-------
References
Mao, K. How To Select A Computer Model For Storm Water Management. Pollution Engineering, Oct.
1, 1992, pp. 60-64.
MASS. 2002. Montana Agricultural Statistics Volume XXXIX October 2002. Montana Agricultural
Statistics Service. Helena, Montana.
MBOG. 2006. Online Oil & Gas Information System - Coal Bed Methane [Online]. Montana Board of
Oil and Gas Conservation. Available online at http://bogc.dnrc.state.mt.us/default.asp (Accessed
December 8, 2006).
MDEQ. 1996. Montana 303(d) List - Streams. Montana Department of Environmental Quality;
Planning, Prevention, and Assistance Division, Monitoring and Data Management Bureau, Helena,
Montana.
MDEQ. 2000. Authorization to Discharge Under the Montana Pollutant Discharge Elimination System -
Redstone Gas Partners, LLC [Permit #MT0030457], Montana Department of Environmental Quality.
Helena, Montana.
MDEQ. 2001. Water Quality Impacts from Coal Bed Methane Development in the Powder River
Basin, Wyoming and Montana. Montana Department of Environmental Quality Water Quality Technical
Report. Helena, Montana.
MDEQ. 2003. Total Maximum Daily Load (TMDL) Status Report Tongue River TMDL Planning Area.
Montana Department of Environmental Quality. Helena, Montana. Available online at
http://deq.mt.gov/wqinfo/TMDL/TonguePowderRosebudTMDL.asp.
MDEQ. 2006a. 2006 Integrated 303(d)/305(b) Water Quality Report for Montana. Water Quality
Planning Bureau, Montana Department of Environmental Quality. Available online at
http://www.deq.state.mt.us/CWAIC/default.aspx (Accessed March 5, 2007).
MDEQ. 2006b. Administrative Rules of Montana - Chapter 30 Subchapter 6 - Surface Water Quality
Standards and Procedures. Helena, Montana. Available online at http://arm.sos.state.mt.us/.
MDEQ. 2006c. Fidelity Exploration and Production Company MPDES Permit MT0030457 DMR Data
[Computer File]. Montana Department of Environmental Quality [Producer and Distributor]. Helena,
Montana.
Montana State Engineer. 1947. Water Resources Survey - Big Horn County, Montana. Helena,
Montana.
MRLC. 2007. National Land Cover Database 2001 [Online]. Multi-Resolution Land Characteristics
Consortium. Sioux Falls, South Dakota. Available online at http://www.mrlc.gov/ (Accessed December
2, 2006).
NADP. 2006. Annual Data Summaries for Site MT00 - Little Bighorn Battlefield National Monument.
NADP Program Office, Illinois State Water Survey, Champaign, IL. Available online at
http://nadp.sws.uiuc.edu/ (Accessed December 2, 2006).
Neff, Earl L. 1980. Using Sodium Carbonate to Seal Leaky Stock Ponds in Eastern Montana. Journal of
Range Management. Volume 33(4). p. 292-295.
121
-------
References
NRCS. 1986. Urban Hydrology for Small Watersheds Technical Release 55. June 1986.
NRCS. 2005. SNOTEL Data Collection Network Fact Sheet [Online]. Natural Resources Conservation
Service. Portland, Oregon. Available at http://www.wcc.nrcs.usda.gov/snow/ (Accessed April 22, 2005).
Payne, A.A. and D. M. Saffer. 2005. Surface water hydrology and shallow groundwater effects of
coalbed natural gas development, upper Beaver Creek drainage, Powder River Basin, Wyoming. Chapter
1 in Zoback, M.D., ed., Western Resources Project Final Report - Produced Groundwater Associated
with Coalbed Natural Gas Production in the Powder River Basin. Report of Investigations No. 55.
Wyoming State Geological Survey, Laramie, WY.
Penman, H.L. 1948. Natural Evaporation from Open Water, Bare and Grass. Proceedings from the Royal
Society of London, Series A. Volume 193. pp. 120-145.
Sartor J. and G. Boyd. 1972. Water Pollution Aspects of Street Surface Contaminants. U.S.
Environmental Protection Agency EPA-R2-72-08. Washington, D.C.
Schafer, W.M., N. Fehringer, and K. Harvey. 2006. 2006 Progress Report, Tongue River Agronomic
Monitoring & Protection Program.
Schwab, G.O., D.D. Fangmeier, W.J. Elliott, and R.K. Frevert. 1993. Soil and Water Conservation
Engineering. Fourth Edition.
SEO. 2001. Division II Annual Hydrographer's Report-2001. Wyoming State Engineer's Office.
Sheridan, Wyoming.
SEO. 2002. Division II Annual Hydrographer's Report-2002. Wyoming State Engineer's Office.
Sheridan, Wyoming.
SEO. 2003. Division II Annual Hydrographer's Report-2003. Wyoming State Engineer's Office.
Sheridan, Wyoming.
SEO. 2004. Division II Annual Hydrographer's Report - 2004. Wyoming State Engineer's Office.
Sheridan, Wyoming.
SEO. 2005. Division II Annual Hydrographer's Report-2005. Wyoming State Engineer's Office.
Sheridan, Wyoming.
SEO. 2006a. Daily Flow Data for the Park Diversion [Computer File]. Wyoming State Engineer's
Office (Producer and Distributor). Cheyenne, Wyoming. Obtained December 26, 2006.
SEO. 2006b. Daily Flow Data for the Meade-Coffeen, Piney-Cruse, and Prairie Dog Transbasin
Diversions [Computer File]. Wyoming State Engineer's Office (Producer and Distributor). Cheyenne,
Wyoming. Obtained December 27, 2006.
Smith, R.A., Alexander, R.B., and Schwarz, G.E., 2003. Estimating the natural background
concentrations of nutrients in streams and rivers of the conterminous United States. Environmental
Science and Technology. Vol. 37, 3039-3047.
Theissen, A.H., 1911, Precipitation Averages for Large Areas. Monthly Weather Review, v. 39, p. 1082-
1084.
122
-------
References
Trelease, Frank J., T. Swartz, P. Rechard, and R. Burman. 1970. Consumptive Use of Irrigation Water in
Wyoming. Wyoming Water Planning Report No. 5. Water Resources Series No. 19. Cheyenne,
Wyoming. Available online at http://library.wrds.uwyo.edu/wrs/wrs-19/ (accessed December 29, 2006).
Tuteja, N.K., G.T.H. Beale, G. Summerell, and W.H. Johnston. 2002. Development and Validation of the
Catchment Scale Salt Balance Model - CATSALT (Version 1). New South Wales Dept. of Land and
Water Conservation, Parramatta, New South Wales, Australia.
United States Army Corps of Engineers. 1956. Snow Hydrology. Portland, OR.
U.S. Bureau of Reclamation. 2006. Great Plains Cooperative Agricultural Weather Network - AgriMet
Data [Online]. Available at http://www.usbr.gov/gp/agrimet/ (Accessed February 14, 2006)
USDI. 2003. Final Statewide Oil and Gas EIS and Proposed Amendment of the Powder River and
Billings RMPs. United State Department of the Interior - Bureau of Land Management. Miles City and
Billings, Montana Field Offices. Billings, Montana.
USEPA. 1985. Rates, Constants, and Kinetics Formulations in Surface Water Quality Modeling (Second
Edition). U.S. Environmental Protection Agency Document EPA/600/3-85/040. Office of Research and
Development. Athens, Georgia.
USEPA. 1997. Technical Guidance Manual for Developing Total Maximum Daily Loads: Book 2, Rivers
and Streams; Part 1 - Biochemical Oxygen Demand/Dissolved Oxygen & Nutrient Eutrophication EPA
823/B-97-002.
USEPA. 2000. BASINS Technical Note 6. Estimating Hydrology and Hydraulic Parameters for HSPF.
EPA-823-R00-012. U.S. Environmental Protection Agency. Office of Water. July 2000.
USEPA. 2003. Draft Guidance on the Development, Evaluation, and Application of Regulatory
Environmental Models. Prepared by The Council for Regulatory Environmental Modeling. U.S.
Environmental Protection Agency, Office of Science Policy, Office of Research and Development.
Washington, D.C.
USEPA. 2006. Permit Compliance System Database [Online]. U.S. Environmental Protection Agency.
Washington D.C. Available online at http://www.epa.gov/enviro/html/pcs/pcs_overview.html (Accessed
October 18, 2006).
USEPA. 2007. Water Quality Assessment for the Tongue River Watershed, Montana. U.S.
Environmental Protection Agency, Montana Operation Office. Helena, Montana.
USGS. 2002. 30-Meter National Elevation Dataset for Montana [Computer File]. U.S. Geological
Survey EROS Data Center [Producer] and Montana State Library [Distributor], Helena, Montana.
Available online at http://nris.state.mt.us/gi.asp.
USGS. 2004. Channel-Morphology Data for the Tongue River and Selected Tributaries, Southeastern
Montana, 2001-02. Open-File Report 2004-1260. U.S. Geological Survey. Reston, Virginia. Available
online at http://pubs.usgs.gov/of/2004/1260/ (Accessed December 2, 2006).
123
-------
References
WDEQ. 2003a. Authorization to Discharge Under the National Pollutant Discharge Elimination System
- Dayton Wastewater Treatment Lagoon [Permit #WY0020435], Wyoming Department of
Environmental Quality. Cheyenne, Wyoming.
WDEQ. 2003b. Authorization to Discharge Under the National Pollutant Discharge Elimination System
- Bighorn Mountain KOA Sewage Plant [Permit #WY0026441], Wyoming Department of
Environmental Quality. Cheyenne, Wyoming.
WDEQ. 2004a. Authorization to Discharge Under the National Pollutant Discharge Elimination System
- Burgess Junction Dump Station [Permit # WY0020931], Wyoming Department of Environmental
Quality. Cheyenne, Wyoming.
WDEQ. 2004b. Authorization to Discharge Under the National Pollutant Discharge Elimination System
- AC Ranch N and E, Option IB Facility [Permit #WY0052043], Wyoming Department of
Environmental Quality. Cheyenne, Wyoming.
WDEQ. 2005a. Authorization to Discharge Under the National Pollutant Discharge Elimination System
- Sheridan Wastewater Treatment Plant [Permit #WY0020010], Wyoming Department of Environmental
Quality. Cheyenne, Wyoming.
WDEQ. 2005b. Authorization to Discharge Under the National Pollutant Discharge Elimination System
- Ranchester Wastewater Treatment Lagoon [Permit #WY0022161], Wyoming Department of
Environmental Quality. Cheyenne, Wyoming.
WDEQ. 2005c. Authorization to Discharge Under the National Pollutant Discharge Elimination System
- Wrench Ranch Goose Creek On-Channel [Permit #WY0038628], Wyoming Department of
Environmental Quality. Cheyenne, Wyoming.
WDEQ. 2005d. Authorization to Discharge Under the National Pollutant Discharge Elimination System
-Hape 14 Option2 [Permit #WY0051811], Wyoming Department of Environmental Quality.
Cheyenne, Wyoming.
WDEQ. 2005e. Authorization to Discharge Under the National Pollutant Discharge Elimination System
- Pipeline Ridge [Permit #WY0052345], Wyoming Department of Environmental Quality. Cheyenne,
Wyoming.
WDEQ. 2005f. Authorization to Discharge Under the National Pollutant Discharge Elimination System
- Dunning Facility [Permit #WY0046540], Wyoming Department of Environmental Quality. Cheyenne,
Wyoming.
WDEQ. 2006a. Authorization to Discharge Under the National Pollutant Discharge Elimination System
- Powder Horn Ranch LLC [Permit #WY0036251]. Wyoming Department of Environmental Quality.
Cheyenne, Wyoming.
WDEQ. 2006b. Authorization to Discharge Under the National Pollutant Discharge Elimination System
- Sheridan Water Treatment Plant [Permit #WY0001392], Wyoming Department of Environmental
Quality. Cheyenne, Wyoming.
WDEQ. 2007. Microsoft Access Database of Point Sources in the Tongue River Watershed [Computer
File]. Wyoming Department of Environmental Quality [Producer and Distributor], January 26, 2007.
124
-------
References
Wheaton, J.W. and T.H. Brown. Predicting Changes in Groundwater Quality Associated with Coalbed
Natural Gas Infiltration Ponds. Chapter 2 in Zoback, M.D., ed., Western Resources Project Final Report
- Produced Groundwater Associated with Coalbed Natural Gas Production in the Powder River Basin.
Report of Investigations No. 55. Wyoming State Geological Survey, Laramie, WY.
WWDC. 2002a. 1:24,000-Scale Irrigation Points of Diversion (Service Areas), in Decimal Degrees -
NAD 1927 [Computer File]. Wyoming Water Development Commission [Distributor]. Cheyenne,
Wyoming. Available online at http://waterplan.state.wy.us/.
WWDC. 2002b. Bedrock Geology in Decimal Degrees - NAD 1927 [Computer File]. Wyoming Water
Development Commission [Distributor], Cheyenne, Wyoming. Available online at
http://waterplan.state.wy.us/.
WWDC. 2002c. Wyoming Stock Pond Permit Locations through December 2000, in Decimal Degrees -
NAD 1927 First Edition [Computer File]. Wyoming Water Development Commission [Distributor],
Cheyenne, Wyoming. Available online at http://waterplan.state.wy.us/.
WWDC. 2002d. Powder/Tongue River Basin Water Plan Technical Memoranda A - Irrigation
Diversion Operation and Description. Wyoming Water Development Commission. Cheyenne, Wyoming.
Available online at http://waterplan.state.wy.us/.
WWDC. 2002e. Powder/Tongue River Basin Water Plan Technical Memoranda C - Irrigated Lands
Mapping and Water Rights Data. Wyoming Water Development Commission. Cheyenne, Wyoming.
Available online at http://waterplan.state.wy.us/.
WWDC. 2002f. Powder/Tongue River Basin Water Plan Technical Memoranda D - Agricultural Water
Use. Wyoming Water Development Commission. Cheyenne, Wyoming. Available online at
http://waterplan.state.wy.us/.
WWDC. 2002g. Powder/Tongue River Basin Water Plan Technical Memoranda J - Storage Operation
and Description. Wyoming Water Development Commission. Cheyenne, Wyoming. Available online at
http://waterplan.state.wy.us/.
125
------- |