vvEPA EPA/600/R-14/233 I October 2016 I www.epa.gov
United States
Environmental Protection
Agency
Quantitative Assessment of
Temperature Sensitivity of the
South Fork Nooksack River under
Future Climates using QUAL2Kw
-------
-------
EPA/600/R-14/233
October 2016
Quantitative Assessment of Temperature Sensitivity of the
South Fork Nooksack River under Future Climates using
QUAL2Kw
Prepared by
Jonathan Butcher, Ph.D., P.H.
Mustafa Faizullabhoy
Hillary Nicholas
Peter Cada
J. Todd Kennedy
Preferred Citation:
Butcher, J.B., M. Faizullabhoy, H. Nicholas, P. Cada, and J.T. Kennedy. 2016. Quantitative Assessment
of Temperature Sensitivity of the South Fork Nooksack River Nooksack River under Future Climates
using QUAL2Kw. EPA/600/R-14/233. Western Ecology Division, National Health and Environmental
Effects Research Laboratory, Corvallis, OR.
-------
SFNR Quantitative Assessment - Final
October 2016
Disclaimer
This document has been reviewed in accordance with U.S. Environmental Protection Agency policy and
approved for publication. Mention of trade names or commercial products does not constitute
endorsement or recommendation for use. The information in this document has been funded wholly by
the U.S. Environmental Protection Agency under Contract Number EP-C-08-004 and EP-C-12-060 to
Tetra Tech, Inc.
-------
SFNR Quantitative Assessment - Final
October 2016
Abstract
The Total Maximum Daily Load (TMDL) program, established by the Clean Water Act, is used to
establish limits on loading of pollutants from point and nonpoint sources necessary to achieve water
quality standards. One important use of a temperature TMDL is to allocate thermal loads to achieve water
temperature criteria established for the protection of cold water fisheries. The pollutant in this case is
thermal load and allocations to reduce the load often involve restoration of stream shading, which reduces
the solar input. While many temperature TMDLs have been established, the supporting analyses have
generally assumed a stationary climate under which historical data on flow and air temperature can serve
as an adequate guide to future conditions. Projected changes in climate over the 21st century contradict
this assumption. Air temperature is expected to increase in most parts of the US, accompanied in many
areas by seasonal shifts in the timing and amount of precipitation, which in turn will alter stream flow.
This study evaluates the implications of climate change for the water temperature TMDL developed for
the South Fork Nooksack River in northwest Washington by the Department of Ecology, where multiple
water body segments exceed temperature criteria established for the protection of cold water salmonid
populations (Ecology, 2016). The purpose of this report is to provide a "companion technical methods
manual" as documentation for the draft SFNR temperature TMDL developed by Ecology. The TMDL
analyses use Ecology's QUAL2Kw stream simulation model in conjunction with an analysis of shading
to predict the temperature profile during critical conditions of summer low flow and elevated air
temperatures. The modeling shows that restoration of system potential vegetation shading would
significantly mitigate increasing water temperature. We reran the QUAL2Kw model for future climate
conditions (multiple climate models for the 2020s, 2040s, and 2080s) using gridded downscaled climate
data and hydrologic model runoff predictions developed by the Climate Impacts Group at the University
of Washington to modify the critical conditions inputs using a change factor approach. Results show that
the risk of higher water temperature will accelerate over time. Projected increases in heat inputs and lower
summer flows associated with a reduction in the storage of winter snowpack combine to exacerbate
summer water temperature extremes and may begin to overwhelm the mitigating impact of increased
shading by the 2040s, with a high probability of exceeding cited lethal temperature thresholds under low
flow critical conditions. We note, however, that the TMDL focuses on extreme conditions (e.g., 7-day
low flow with 10-year recurrence), and predictions are more favorable for less extreme flow conditions -
for instance, at a 2-year low flow recurrence water temperatures through the 2080s are generally predicted
to remain below lethal thresholds with enhanced shading in place. Indeed, the importance of system
potential shading is of even greater importance under future climate conditions. Establishing a mature
riparian forest canopy can take 100 years, so it is important to begin planting riparian buffers now to
reduce the anticipated climate change impacts on water temperature. Protection and restoration of local
cold water refuges is another important adaptation strategy to mitigate the effects of climate change on
aquatic life during high temperature events.
-------
SFNR Quantitative Assessment - Final
October 2016
Foreword
The U.S. Environmental Protection Agency (EPA) Region 10 and EPA's Office ofWater (OW) and
Office of Research and Development (ORD) have launched a Pilot Research Project to explore how
projected climate change impacts could be considered in the implementation of a Clean Water Act
(CWA) 303(d) temperature Total Maximum Daily Load (TMDL,) and influence restoration actions in an
Endangered Species Act (ESA) Salmonid Recovery Plan. The Pilot Research Project uses a temperature
TMDL being developed by Washington's Department of Ecology (Ecology) for the South Fork Nooksack
River (SFNR) in Washington, as the pilot TMDL for climate change vulnerability analysis. An
overarching objective of the Pilot Research Project is to support the goals and priorities of EPA's climate
adaptation plans.
A range of projected climate change impacts from the Intergovernmental Panel on Climate Change
(IPCC) Scenarios were evaluated as a risk assessment to thoroughly consider plausible futures of potential
impacts to salmonids.
The project consists of two separate research assessments:
The quantitative assessment (this report) provides a comparison of QUAL2Kw modeled stream
temperatures, including riparian shading, with and without climate change for the 2020s, 2040s and
2080s. A range of projected climate change impacts from a high, medium and low impact scenario are
analyzed for each time period. This assessment discusses and considers the relevant CWA water quality
standards developed to protect beneficial uses, including cold water fisheries.
The qualitative assessment (in review) is a comprehensive analysis of freshwater habitat for ESA salmon
restoration in the SFNR under climate change. The objective of the qualitative assessment is to identify
and prioritize climate change adaptation strategies or recovery actions for the SFNR that explicitly
include climate change as a risk.
Together, these two assessments identify comprehensive actions to protect CWA beneficial uses (salmon
habitat) and ESA recovery goals under potential climate change.
Stakeholder outreach and Tribal engagement is considered a critical element of this project. Workshops,
webinars and working Interdisciplinary Teams have been utilized throughout the life of this project. The
result is actionable science and, with the participation of scientists, environmental practitioners and
decision makers, supports the co-production of knowledge for climate change adaptation.
Foreword by
One EPA Team:
EPA Region 10
EPA Office ofWater
EPA Office of Research and Development
iv
-------
SFNR Quantitative Assessment - Final
October 2016
Contents
List of Abbreviations xi
Acknowledgements xiii
Executive Summary xv
1 Introduction 1-1
2 The Water Temperature Model for the South Fork Nooksack River 2-1
2.1 Summary of Water Temperature Model Development 2-4
2.1.1 Shade Modeling 2-5
2.1.2 QUAL2Kw 2-6
2.2 TMDL Modeling Scenarios 2-9
2.2.1 Scenario Flows and Air Temperature 2-10
2.2.2 Modeling Results 2-12
2.2.3 Evaluation of Historical Conditions 2-13
2.2.4 Draft TMDL Results 2-13
3 Climate Change Modeling 3-1
3.1 Climate Model Applications for the Pacific Northwest 3-1
3.1.1 CIG Climate Projections 3-1
3.1.2 Time Horizons for Analysis 3-3
3.1.3 Climate Scenarios Selected for the South Fork Nooksack River TMDL 3-4
3.2 Future Climate Estimates for the SFNR 3-7
3.2.1 Air T emperature 3-7
3.2.2 Precipitation 3-9
3.2.3 Relative Humidity, Dew Point, and Vapor Pressure Deficit 3-10
3.2.4 Solar Radiation 3-12
3.3 VIC Hydrologic Model 3-13
3.3.1 Water Yield 3-15
3.3.2 Baseflow 3-16
4 Future Climate Boundary Conditions 4-1
4.1 Mainstem and Tributary Flow Boundary Conditions 4-1
4.2 Mainstem and Tributary Water Temperature Boundary Conditions 4-5
4.2.1 Fitting Data 4-6
4.2.2 Regression Analysis 4-7
4.3 Meteorological Forcing 4-12
4.3.1 Air Temperature 4-12
4.3.2 Dew Point Temperature 4-12
4.3.3 Cloud Cover 4-13
4.3.4 Wind 4-13
4.4 Other Boundary Conditions 4-13
v
-------
SFNR Quantitative Assessment - Final
October 2016
4.4.1 Ground Water Inflow Temperatures 4-13
4.4.2 Physical Boundary Conditions 4-13
5 Temperature Model Results 5-1
5.1 Critical Conditions Scenario 5-1
5.2 Climate Change Scenarios 5-1
5.2.1 Climate Change Scenarios: System Potential Shade 5-4
5.2.2 Climate Change Scenarios: Current Shade 5-7
5.2.3 Analysis of Results 5-9
5.3 Sensitivity to Boundary Conditions 5-15
5.4 7Q2 Analysis 5-16
5.5 Potential Changes in Extreme High Flows 5-19
6 Discussion 6-1
6.1 Implications for the TMDL 6-2
6.2 Implications for Restoration and Management Strategies 6-3
7 References 7-1
vi
-------
SFNR Quantitative Assessment - Final
October 2016
List of Tables
Table 2-1. Washington State Temperature Criteria for the South Fork Nooksack River Watershed 2-1
Table 2-2. Calculated 7Q2 and 7Q10 Flows Compared to Calibration Conditions 2-10
Table 2-3. Air Temperature Statistics for the South Fork Nooksack Region 2-11
Table 2-4. Modeling Scenario Results for Typical Low-Flow and Critical Low-Flow and Air
Temperature Conditions 2-12
Table 3-1. Time Horizons for the Climate Analysis 3-3
Table 3-2. Comparison of Selected SFNR Climate Scenarios to 10-model Ensemble for the North
Cascadia Region 3-6
Table 4-1. SFNR Tributary Catchments Included in the QUAL2Kw Model 4-2
Table 4-2. SFNR Tributary Stream Temperature Monitoring Data 4-6
Table 4-3. Coefficients ofWater Temperature Prediction Model 4-10
Table 4-4. Average of Projected Daily Maximum Tributary Temperatures (°C) 4-10
Table 4-5. Average Additive Deltas for August Air Temperature and Dew Point (°C) 4-12
Table 4-6. Additive Deltas on Average Annual Air Temperature (°C) 4-13
Table 5-1. Critical Period Maximum Water Temperature Results Summary 5-2
Table 5-2. Critical Period Daily Average Water Temperature Results Summary 5-3
Table 5-3. Projected Percent Change in Magnitude of Floods of Various Recurrence Intervals in
the SFNR under Different Climate Change Scenarios 5-20
vii
-------
SFNR Quantitative Assessment - Final
October 2016
List of Figures
Figure 2-1. The South Fork Nooksack River and Associated Water Temperature Criteria 2-2
Figure 2-2. Elevation in the South Fork Nooksack River Watershed 2-2
Figure 2-3. Land Use in the South Fork Nooksack River Watershed 2-3
Figure 2-4. Modeled Effective Shade for 2007 Existing Vegetation and 100-year System Potential
Vegetation 2-6
Figure 2-5. South Fork Nooksack River Model Segmentation by 1-kilometer Reach 2-7
Figure 2-6. Comparison of Observed and Simulated Flows for the Calibration Period 2-8
Figure 2-7. Temperature Calibration for QUAL2Kw Model of South Fork Nooksack River 2-9
Figure 2-8. Flow Gages and Air Temperature Monitoring Stations used for Boundary Condition
Development 2-10
Figure 3-1. Schematic of Model and Climate Data Integration for the Quantitative Assessment 3-1
Figure 3-2. Climate Scenario Changes in Precipitation and Temperature Compared to North
Cascadia (NCAP) Ensemble Mean 3-7
Figure 3-3. Climate Model Projections of Monthly Average Air Temperature in the SFNR
Watershed for Low-, Medium-, and High-Impact Climate Scenarios 3-8
Figure 3-4. Climate Model Projections of Monthly Maximum of Daily Average Air Temperature
in the SFNR Watershed for Low-, Medium-, and High-Impact Scenarios 3-9
Figure 3-5. 2080s Climate Model Projections of Average Air Temperature in the SFNR Watershed
at the Headwaters (left) and Outlet (right) 3-9
Figure 3-6. Climate Model Projections of Monthly Median Precipitation in the SFNR Watershed
for Low-, Medium-, and High-Impact Scenarios 3-10
Figure 3-7. Climate Model Projections of 2080s Average Precipitation in the SFNR Watershed at
the Headwaters (left) and Outlet (right) 3-10
Figure 3-8. Climate Model Projections of Monthly Average Relative Humidity in the SFNR
Watershed for Low-, Medium-, and High-Impact Scenarios 3-11
Figure 3-9. Climate Model Projections of 2080s Average Vapor Pressure Deficit in the SFNR
Watershed at the Headwaters (left) and Outlet (right) 3-12
Figure 3-10. Climate Model Projections of Monthly Solar Radiation in the SFNR Watershed for
Low-, Medium-, and High-Impact Scenarios 3-12
Figure 3-11. Comparison of Monthly Average VIC Model Flows to USGS Gaged Flows for the
South Fork Nooksack River at Wickersham (12209000) for 1934-2006 3-14
Figure 3-12. Time Series Comparison of Daily Average VIC Model Flows to USGS Gaged Flows
for the South Fork Nooksack River at Wickersham (12209000) for 2003-2006 3-14
Figure 3-13. Monthly Average Water Yield in the SFNR Watershed Predicted by the VIC Model
for Low-, Medium-, and High-Impact Climate Scenarios 3-15
Figure 3-14. 2080s Average Water Yield in the SFNR Watershed Predicted by the VIC Model for
Headwaters (left) and Outlet (right) Grid Cells 3-16
viii
-------
SFNR Quantitative Assessment - Final
October 2016
Figure 3-15. 2080s Average Baseflow Discharge Rate in the SFNR Watershed Predicted by the
VIC Model for Headwaters (left) and Outlet (right) Grid Cells 3-16
Figure 4-1. 7Q10 Flow Multiplicative Change Factors (Unitless) for Tributary Flow from VIC
Model for Low-, Medium-, and High-Impact Scenarios (Columns) and 2020s, 2040s,
and 2080s Time Horizons (Rows) 4-4
Figure 4-2. Simulated 7Q10 Flows by River Mile under Historic and 2080s Climate in the SFNR 4-5
Figure 4-3. Prediction of 7-day Average Minimum and Maximum Water Temperature 4-9
Figure 4-4. Predicted Maximum Hourly Tributary Temperatures under 7Q10 Flow Conditions for
TMDL Scenario (top) and 2080s High-Impact Climate Scenarios with Current Shade
(bottom) 4-11
Figure 5-1. Maximum Stream Temperature by Reach for Scenario 5 (7Q10 flows, 90th Percentile
Meteorology, System Potential Shade, and Tributaries/Headwaters at Water Quality
Criteria) and Scenarios 12, 13, and 14 (2020 High GCM, 2020 Medium GCM, 2020
Low GCM) 5-5
Figure 5-2. Maximum Stream Temperature by Reach for Scenario 5 (7Q10 flows, 90th Percentile
Meteorology, System Potential Shade, and Tributaries/Headwaters at Water Quality
Criteria) and Scenarios 18, 19, and 20 (2040 High GCM, 2040 Medium GCM, 2040
Low GCM) 5-5
Figure 5-3. Maximum Stream Temperature by Reach for Scenario 5 (7Q10 flows, 90th Percentile
Meteorology, System Potential Shade, and Tributaries/Headwaters at Water Quality
Criteria) and Scenarios 24, 25, and 26 (2080 High GCM, 2080 Medium GCM, 2080
Low GCM) 5-6
Figure 5-4. Comparison of Maximum Temperature Profiles at 7Q10 Flow and 90th Percentile
Meteorology for Existing TMDL Conditions and Medium GCM Scenarios for 2020,
2040, and 2080 (with System Potential Shade) 5-6
Figure 5-5. Maximum Stream Temperature by Reach for Scenario 3 (7Q10 flows, 90th Percentile
Meteorology, Current Shade) and Scenarios 9, 10, and 11 (2020 High GCM, 2020
Medium GCM, 2020 Low GCM) 5-7
Figure 5-6. Maximum Stream Temperature by Reach for Scenario 3 (7Q10 flows, 90th Percentile
Meteorology, Current Shade) and Scenarios 15, 16, and 17 (2040 High GCM, 2040
Medium GCM, 2040 Low GCM) 5-8
Figure 5-7. Maximum Stream Temperature by Reach for Scenario 3 (7Q10 flows, 90th Percentile
Meteorology, Current Shade) and Scenarios 21, 22, and 23 (2080 High GCM, 2080
Medium GCM, 2080 Low GCM) 5-8
Figure 5-8. Spatially Averaged Maximum Water Temperature in the South Fork Nooksack River
Mainstem at Critical Conditions for Existing TMDL Conditions and Future Climate
with Existing and System Potential Vegetation Shade 5-9
Figure 5-9. Change in Spatially Averaged Maximum Water Temperature in the South Fork
Nooksack River Mainstem at Critical Conditions for Future Climate with System
Potential Vegetation Compared to Existing TMDL Conditions and Vegetation 5-10
Figure 5-10. Spatial Distribution of Ratio to Lethal Temperatures at Critical Conditions, Existing
Climate with Existing (top) and System Potential Shade (bottom) 5-11
Figure 5-11. Spatial Distribution of Ratio to Lethal Temperatures at Critical Conditions for all
Climate Scenarios with System Potential Shade 5-13
ix
-------
SFNR Quantitative Assessment - Final
October 2016
Figure 5-12. Percent of Stream Miles in the South Fork Nooksack River where the Daily
Maximum Temperature Exceeds the 1-day Average Maximum Lethality Temperature
of 23 °C under Existing TMDL Conditions and Future Climate Critical Conditions 5-15
Figure 5-13. Sensitivity of 7Q10 TMDL Model Maximum Water Temperature Predictions to
Variations in Boundary Temperature and Flow 5-16
Figure 5-14. Maximum Stream Temperature by Reach for 7Q2 flows, and 50th Percentile
Maximum Air Temperature with System Potential Shade for 2080 High GCM, 2080
Medium GCM, and 2080 Low GCM 5-17
Figure 5-15. Maximum Stream Temperature by Reach for 7Q2 flows, and 50th Percentile
Maximum Air Temperature with Current Shade for 2080 High GCM, 2080 Medium
GCM, and 2080 Low GCM 5-18
Figure 5-16. Comparison of Maximum Stream Temperature by Reach for 2080s Medium GCM
with System Potential Shade at 7Q2 and 7Q10 Flows 5-18
Figure 6-1. Climate Change Ecological Risk Assessment (adapted from USEPA, 1992) 6-1
Figure 6-2. Climate Change Adaptation and Iterative Risk Management (Yohe, 2011) 6-2
x
-------
SFNR Quantitative Assessment - Final
October 2016
List of Abbreviations
7-DADMax
Highest 7-day average of the daily maximum temperatures
7-DADMin
Highest 7-day average of the daily minimum temperature
7Q10 flow
7-day average flow with a 10-year recurrence frequency
7Q2 flow
7-day average flow with a 2-year recurrence frequency
AR4
IPCC Fourth Assessment Report
AR5
IPCC Fifth Assessment Report
CCAP
Coastal Change Analysis Program
CCSM
Community Climate System Model
CFR
Code of Federal Regulations
cfs
cubic feet per second
CGCM
Coupled Global Climate Model
CIG
Climate Impacts Group
DOY
Day of Year
DRTT
Dominant River-tracing-based Streamflow and Temperature model
Ecology
Washington State Department of Ecology
EPA
U.S. Environmental Protection Agency
ESA
Endangered Species Act
GCM
Global Climate Model or General Circulation Model
GIS
Geographic Information System
HADGEM
Hadley Centre Global Environmental Model
HUC
Hydrologic Unit Code
IPCC
Intergovernmental Panel on Climate Change
LiDAR
Light Detecting and Ranging
MIROC
Model for Interdisciplinary Research on Climate
MWMT
Maximum Weekly Maximum Temperature
NAIP
National Agricultural Imagery Program
NCAP
North Cascadia Adaptation Partnership
NCDC
National Climatic Data Center
NO A A
National Oceanic and Atmospheric Administration
NPDES
National Pollutant Discharge Elimination System
NSDZ
Near-stream Disturbance Zone
ODEQ
Oregon Department of Environmental Quality
ORD
EPA Office of Research and Development
xi
-------
SFNR Quantitative Assessment - Final
October 2016
ow
EPA Office of Water
PCM
Parallel Climate Model
PNW
Pacific Northwest
QUAL2Kw
The Washington version of a river and stream water quality model (QUAL2K)
that is in turn a modernized version of EPA's older QUAL2E model
RMSE
Root Mean Square Error
RPD
Relative Percent Difference
SFNR
South Fork Nooksack River
SNOTEL
"Snow telemetry": Automated system of snowpack and related climate sensors
SOD
Summary of the Day weather station
SPV
System Potential Vegetation
TMDL
Total Maximum Daily Load
USDA
U.S. Department of Agriculture
USFS
U.S. Forest Service
USGS
U.S. Geological Survey
VIC
Variable Infiltration Capacity hydrologic model
WAC
Washington Administrative Code
WDNR
Washington State Department of Natural Resources
WRIA
Water Resource Inventory Area
xii
-------
SFNR Quantitative Assessment - Final
October 2016
Acknowledgements
We acknowledge and thank the EPA Region 10 Climate Change and TMDL Pilot Team (Steve Klein,
EPA ORD Project Leader; Laurie Mann and Bruce Duncan, EPA Region 10 and Dwight Atkinson, EPA
OW) and the South Fork Nooksack River Temperature TMDL Team (Laurie Mann, EPA Region 10;
Teizeen Mohamedali and Steve Hood, Washington Department of Ecology; Oliver Grah and Treva Coe,
Nooksack Indian Tribe and Jeremy Freimund, Lummi Nation) for their support contributing to the
development of this report. We also thank the University of Washington (UW), Climate Impacts Group
(CIG) for the Coupled Model Intercomparison Project Phase 3 (CMIP3) and Variable Infiltration
Capacity model (VIC) downscaled Climate Change data for use in the QUAL2Kw modeling.
We also want to acknowledge and thank the Peer Reviewers (Heather Golden, EPA ORD; Dan Isaak,
USFS; Guillaume Mauger, UW-CIG) of this report for their many suggestions and comments that
resulted in a much improved product. Lastly, we want to acknowledge and thank Paul Pickett,
Washington Department of Ecology for his continued involvement and helpful comments during the life
of this project.
-------
SFNR Quantitative Assessment - Final October 2016
xiv
-------
SFNR Quantitative Assessment - Final
October 2016
Executive Summary
Global climate change has the potential for significant impacts on freshwater ecosystems through changes
in both the hydrological and thermal regime. Stream temperatures are projected to increase in most rivers,
resulting in increased stress on cold water fish species including salmon. Changes in hydrology, such as
reduction in summer baseflows, could potentially exacerbate these impacts.
The South Fork Nooksack River (SFNR) is in an area considered typical of the mountainous, remote,
forested landscape of northwest Washington, with minor urban and agricultural land uses along with
extensive timber harvest and associated activities. The SFNR and its tributaries provide migration routes
and spawning and rearing habitat for several salmon species throughout the year.
The Total Maximum Daily Load (TMDL) program, established by the Clean Water Act, has frequently
been used to develop management plans intended to achieve temperature criteria and protect cold water
fisheries. Portions of the SFNR and its tributaries are identified as being impaired by elevated temperature
on Washington's 2010 Clean Water Act 303(d) list. These segments exceed the temperature criteria
established to protect aquatic life uses for the support of cold water salmonid populations. The draft
temperature TMDL developed by the Washington Department of Ecology (Ecology) in conjunction with
EPA Region 10, the Lummi Nation, and the Nooksack Indian Tribe identifies the issues and outlines the
solutions needed to improve river temperatures (Ecology, 2016).
The Quantitative Assessment (this report) provides a comparison of modeled stream temperatures, with
and without proposed TMDL allocations such as increased riparian shading, for critical conditions under
projected climate conditions for the 2020s, 2040s, and 2080s.
The SFNR temperature TMDL analyses use the QUAL2Kw stream simulation model (Pelletier and
Chapra, 2008) in conjunction with an analysis of shading to predict the temperature profile throughout the
SFNR during critical conditions of summer low flow and elevated air temperatures. The analyses predict
that the SFNR will not fully attain the numeric temperature criteria under current climate during summer
low-flow, high-temperature conditions even under full riparian canopy; however, restoration of the
"system potential" riparian shade will help mitigate elevated temperatures deleterious to salmon
populations.
For this pilot research project, the calibrated QUAL2Kw stream temperature model developed for the
TMDL study has been used to estimate the impacts of potential future climate changes on the stream
temperature with and without the restoration of riparian vegetation. To evaluate climate change in the
SFNR, a new set of climate-modified boundary conditions was developed for the QUAL2Kw simulations.
The Climate Impacts Group (CIG) at the University of Washington has assembled output from multiple
Global Climate Models (GCMs) and used statistical downscaling to translate these global model
projections to a finer spatial scale over the Pacific Northwest (PNW). CIG has also used the downscaled
climate data to predict future hydrology using a grid-based macro-scale hydrologic model known as the
Variable Infiltration Capacity (VIC) model. We selected three climate model products that span a range of
low impact, medium impact, and high impact for the 2020s, 2040s, and 2080s. By the 2080s, summer
average air temperatures are predicted to increase by as much as 6 °C or more over the SFNR watershed,
depending on the climate scenario. Presentation of a range of potential impacts provides important
information to an iterative risk assessment approach that can support an adaptive management strategy
relative to future climate change, as recommended by the National Climate Assessment (Melillo et al.,
2014).
The downscaled climate model projections and VIC hydrologic application are still at too coarse a spatial
scale to directly drive a local, site-specific model such as the SFNR QUAL2Kw model. Therefore, we
xv
-------
SFNR Quantitative Assessment - Final
October 2016
applied a delta or change factor method in which the observed historical climate data for the SFNR are
modified by the amount of change predicted by the downscaled climate models to obtain a projection of
future conditions specific to the SFNR. The climate inputs for which change factors are calculated include
several which are analyzed directly from VIC model outputs, such as air temperature and flow, as well as
inputs for parameters that are indirectly calculated from VIC output such as dew point, headwater,
tributary, and ground water temperatures.
The future climate projections are consistent in predicting an increase in air temperature. For the SFNR,
the downscaled climate data, in conjunction with the VIC model, also suggest that summer low flows will
decrease significantly. The QUAL2Kw model predicts that this decrease in flow will also contribute to
increased water temperatures as the thermal mass of the stream is reduced, amplifying the effects of
increased air temperature. The decreases in low flow occur in large part because of a shift from snow to
rain during the winter months, which results in less snow melt during the warm season and reduced
storage of ground water to support base flows. The net result is less water availability during the summer
critical period.
The QUALK2w model simulations suggest that, without restoration of riparian shade, water temperatures
during critical summer low-flow conditions could increase by amounts ranging from 3.5 to almost 6 °C
by the 2080s. Restoration of full system potential riparian shading can help buffer against temperature
increases and mitigate from 30 to 60 percent of the critical period increase; however, even with system
potential shade, the critical condition maximum 7-day average stream water temperatures are projected to
increase by 1.1 to 3.6 °C by the 2080s. In conjunction with this increase, the percent of stream miles in
which critical condition water temperatures are potentially lethal to salmon is predicted by the model
simulations to increase dramatically—from about 18 percent at present to a between 60 and 94 percent in
the 2080s depending on the climate model analyzed.
The TMDL program intentionally focuses on infrequent, worst-case, or "critical" conditions for the
analysis of thermal impairments, using 7-day average low flows that are expected to occur, on average,
once in 10 years (7Q10 flows) and the 90th percentile of projected annual 7-day maximum air
temperatures, as a way of ensuring that standards are met at all times. To estimate more typical summer
periods of maximum stress, additional simulations evaluated responses to the 7-day average low flows
that occur, on average, once in two years (7Q2 flow) and the median projected annual 7-day maximum air
temperature. Under these less stringent conditions, water temperatures through the 2080s are projected to
generally remain below lethal thresholds, with the possible exception of the most downstream reaches of
the SFNR.
Because the QUAL2Kw model of the SFNR predicts spatially averaged water temperatures within stream
segments of 1-kilometer length, it cannot resolve temperature differences at finer scales or evaluate the
availability of cold water refuges. The impact of occasional high-temperature events on salmonids is in
large part determined by whether the fish can find sufficient refuges that are cooler than the reach average
and within their physiological tolerance ranges. Thus, habitat management at a scale smaller than the
spatial scale of the QUAL2Kw model may have an important role in protecting the resource.
In sum, projected changes in future climate in the PNW will result in increased risk of temperature stress
on salmon populations in the SFNR. Because projections of the future are uncertain, the resulting
analyses should not be treated as forecasts; rather, they are designed to assist managers in defining the
potential range of conditions to which adaptation may be needed. Practical implications of the analyses
include the following: First, establishing a mature riparian forest canopy can take 100 years, so it is
important to begin planting riparian buffers now to mitigate anticipated climate change impacts, as well as
to consider the resilience of restored vegetation under future climate conditions. It is also important to
recognize that protection and restoration of local cold water refuges may be a key adaptation strategy to
mitigate the effects of climate change on aquatic life during high temperature events.
xvi
-------
SFNR Quantitative Assessment - Final
October 2016
1 Introduction
The Total Maximum Daily Load (TMDL) program is one of the primary frameworks for the nation to
maintain and achieve healthy water bodies, implemented pursuant to section 303(d) of the Clean Water
Act. The majority of TMDL analyses have been conducted using assumptions of a stationary climate
under which historical data on flow and temperature can be assumed to be an adequate guide to future
conditions (Johnson et al., 2012).
U.S. Environmental Protection Agency (EPA) Region 10 and EPA's Office of Research and
Development (ORD) and Office of Water (OW) are interested in looking beyond these assumptions of
stationary climate to identify the range of potential climate change impacts on water temperature, using a
temperature TMDL as a pilot study. Knowing that multiple technological approaches are available to
assess watershed loading, and that available data can vary dramatically among watersheds, EPA
understands that a pilot study in one watershed might have limited applicability in other watersheds.
Nonetheless, EPA sees benefits to researching the technological and resource issues that arise during
integration of available climate change data into a TMDL project. Already prioritized for TMDL
development by the Washington Department of Ecology (Ecology), the South Fork of the Nooksack
River (SFNR) is impaired by excess temperature and was chosen as the pilot TMDL for the research
project (Ecology, 2016).
Climate models used in the Intergovernmental Panel on Climate Change (IPCC) Fourth and Fifth
Assessment Reports (IPCC, 2007, 2013) confirm observations of increasing temperatures in the Pacific
Northwest (PNW) over the 20th century and consistently project accelerated warming in the 21st century.
Across the PNW, the overall average across all analyzed climate models project increases in average air
temperature of 2.0 °F (1.1 °C) by the 2020s, 3.2 °F (1.8 °C) by the 2040s, and 5.3 °F (2.9 °C) by the
2080s compared to a baseline of 1970 to 1999 (Mote and Salathe, 2010). Precipitation changes are less
certain, but most climate model simulations project changes toward wetter falls and winters and drier
summers, with lower flow further increasing summer water temperature maxima. Together these factors
could significantly increase temperature stress on salmonid populations. Over the entire PNW,
simulations with the DRTT (Dominant River-tracing-based Streamflow and Temperature) model (Wu et
al., 2012) project average increases in summer stream temperatures of 1.37 °C by the 2040s and 2.10 °C
by the 2080s.
Changes in air temperature and precipitation interact synergistically to shape future stream habitat
conditions. Indeed, large reductions in summer flows could have a greater impact on maximum water
temperature in some PNW streams than the direct impact of projected increases in 21st century air
temperature (Cristea and Burges, 2010; Wu et al., 2012). Of particular importance is the potential shift in
overall runoff regime as changes in temperature affect the balance between rain and snow. Hydrologists
characterize the PNW as having three runoff regimes: (1) snowmelt dominant, (2) rain dominant, and
(3) transient or mixed rain and snow systems (Hamlet and Lettenmaier, 2007). In snow-dominant
regimes, much of the winter precipitation is stored in the snowpack, which melts in the spring and early
summer, resulting in peak flows in early spring and a continued supply of cold melt water through
summer. In contrast, rain-dominant systems have little snowpack, resulting in peak flows in late fall and
early winter, lower flows in summer, and greater risk of stream warming (Eisner et al., 2010; Mote and
Salathe, 2010). Transient systems receive a mixture of snow and rain and can have large winter-spring
flow peaks due to rain-on-snow events. Small changes in winter air temperature can result in a shift from
one regime to another. In general, the hydrology of systems currently classified as transient has the
greatest sensitivity to changes in climate forcing because these systems are near the current snowline;
therefore, small changes in temperature can substantially affect snow accumulation (Mote and Salathe,
2010).
1-1
-------
SFNR Quantitative Assessment - Final
October 2016
Salmonid populations are at risk from climate change because of a number of factors. Increasing
summertime stream temperatures, the focus of the temperature TMDL, are likely to be a key pressure
point for many salmon populations, and could be exacerbated if summer flows decrease and limit the
extent of cold water refuges (Mantua et al., 2009). In addition, increases in extreme high flows can have a
strong negative impact on reproductive success due to washout of redds and fry, as was found in the
analysis of climate impacts on ocean-type Chinook in the Snohomish Basin by Battin et al. (2007).
The average changes in climate expected over the PNW can vary substantially at a local scale because of
elevation and aspect, as well as local slope, soil, and land cover conditions. The TMDL is a reach-specific
estimate of allowable loadings and conditions to protect beneficial uses of a water body. Evaluating the
potential impacts of climate change on the TMDL also requires a reach-specific analysis.
This report is one part of a larger research plan that is described in the EPA Region 10 Climate Change
and TMDL Pilot Research Plan (Klein et al. 2013). An important component of the Pilot Research Plan is
a Quantitative Assessment (this document), in which the findings of modeling tools used to develop the
temperature TMDL are reevaluated under a range of potential future climate conditions. In the
Quantitative Assessment, the calibrated QUAL2Kw stream temperature model developed for the TMDL
study is used to estimate the impacts of potential future climate change on critical condition stream
temperature with and without enhancement of existing riparian shading to 100-year system potential
conditions. The approach first calculates altered boundary conditions for the QUAL2Kw TMDL
modeling under the IPCC A IB greenhouse gas emissions storyline (which speculates a balance of fossil
and non-fossil fuel energy sources) and for three time horizons (2020s, 2040s, and 2080s). These
boundary conditions are then used to conduct additional QUAL2Kw modeling analyses of system
response under potential future climate conditions.
Model results provide important information on the potential future response of the system to future
climate, with and without the implementation actions called for in the TMDL. The detailed description of
the approach used in the SFNR pilot analysis provides insight on how place-based analysis of risk
associated with climate change could be incorporated into studies of other watersheds.
1-2
-------
SFNR Quantitative Assessment - Final
October 2016
2 The Water Temperature Model for the South
Fork Nooksack River
The SFNR watershed is located in Whatcom and Skagit counties, northwest Washington, in water
resource inventory area (WRIA) 1 and hydrologic unit code (HUC) 17110004 (Figure 2-1). The SFNR
watershed covers approximately 186 square miles. It originates east of the Twin Sisters Mountain in the
Cascade Mountain Range (Figure 2-2), discharging to the North Fork Nooksack River several miles south
of the Middle Fork confluence. The confluence of all forks becomes the Nooksack River mainstem about
36 miles upstream from where the river discharges into Bellingham Bay.
Numerous tributaries feed the SFNR as it flows down from the Twin Sisters. Major tributaries are
Wanlick Creek, Howard Creek, Cavanaugh Creek, Skookum Creek, Hutchinson Creek, and Black
Slough. The river has an average annual discharge of 1,032 cubic feet per second (cfs) based on Ecology
data at gaging station 01F070 (water years 2004-2010) on the left bank of the SFNR at the Potter Road
Bridge crossing near the town of Van Zandt.
The SFNR is in an area considered typical of the mountainous, remote, forested landscape of northwest
Washington, with minor amounts of urban and agricultural land uses. Forest and shrub land dominate
land use in the watershed, with small amounts of agriculture and development in the lower portion
(Figure 2-3), including portions of the municipalities of Van Zandt and Acme, WA. The Lummi Nation
operates a salmon hatchery and established the Arlecho Creek Preserve in the watershed. The Nooksack
Indian Tribe also owns land and other facilities in the watershed. The headwaters are lands managed by
the U.S. Forest Service (USFS). A portion of the watershed is dominated by alpine tundra and bare rock
of the Twin Sisters summit where vestigial ice fields are present.
The SFNR and its tributaries provide migration routes as well as spawning and rearing habitat for several
salmon species throughout the year. To protect these uses, Washington water quality standards establish
three temperature criteria for the SFNR, expressed as the highest 7-day average of the daily maximum
temperatures (7-DADMax) occurring in a water body. The temperature criteria applicable to the SFNR, as
given in the Washington Administrative Code [WAC 173-201A-200; 2003 edition], are listed in Table
2-1 (also shown on Figure 2-1) and are consistent with USEPA (2003):
Table 2-1. Washington State Temperature Criteria for the South Fork Nooksack River Watershed
Use Classification
Numeric Temperature Criteria
Core summer salmonid habitat, spawning, rearing, and migration
< 16 °C 7-DADMaxa b
Char spawning and rearing
< 12 °C 7-DADMaxa b
Supplemental salmonid spawning and incubation
< 13 °C 7-DADMaxab (Sept 1-Jul 1)
Notes'.
a 7-DADMax means the highest annual running 7-day average of daily maximum temperatures.
b A human-caused variation within the above range of less than 0.3 °C for temperature is acceptable.
The SFNR (Figure 2-1) has 14 mainstem segments and 9 tributary segments identified as impaired by
elevated water temperature on Washington's 2010 303(d) list. These segments are documented to exceed
the temperature criteria established by Ecology to protect aquatic life use categories (salmonid habitat)
and life-stage conditions (spawning and rearing). The temperature TMDL is intended to address these
conditions and identify the solutions needed to improve river temperatures and support designated uses.
2-1
-------
SFNR Quantitative Assessment - Final
October 2016
Todd Creek
[Van Zandt/ r
Black
[Sygitowicz Creek
Hard Scrabble
Falls Creek
|McCarty Creek)
I Skookum Creek
Acme
Bell Creek
Wanlick Creek
i Edfro Creek
iCavanaugh
Creek
Wickers ham
I Deer Creek
Plumbago Creek
Fobes Creek
South Fork Nooksack River: Temperature Standards
TETRA TECH
Whatcom County
Sudden
Valley
J Hutchinson Creek
Legend
I | Watershed Boundary
~ County Boundary
B Water
Major Road
Aquatic Life Use Designations (Temp. Standard)
Core Summer Salmonid Habitat (16°C)
Char Spawning and Rearing (12°C)
Supplemental Spawning/Incubation (13°C)
Applicable from September to July
Howard Creek,
Skagit County
Figure 2-1. The South Fork Nooksack River and Associated Water Temperature Criteria
[Van Zandtj
Sudden
Valley
[Acme
Wickersham
South Fork Nooksack River: Digital Elevation Model
TETRA TECH
Legend
I | Watershed Boundary
Major Road
- South Fork Nooksack River
ation
2,200 meters
60 meters
Figure 2-2. Elevation in the South Fork Nooksack River Watershed
2-2
-------
SFNR Quantitative Assessment - Final
October 2016
Van Zandt
<£fSudden
Valley
y
¦A w. Acme
v Wickersh'am
kL4
Legend
j ™| Watershed Boundary ~ Barren Land
Major Road Deciduous Forest
South Fork Nooksack River Evergreen Forest
NLCD 2006 Land Cover I I Mixed Forest
Value Shrub/Scrub
| Open Water | | Grassland/Herbaceous
| Perennial Ice/Snow | [ Pasture/Hay
| Developed, Open Space Cultivated Crops
~ Developed, Low Intensity | [ Woody Wetlands
| Developed, Medium Intensity [i! [ Emergent Herbaceous Wetlands
H Developed High Intensity
4- JV.IC*
- ¦ *
r*fe.
South Fork Nooksack River: NLCD 2006 Land Cover
NAD_1983_HARN_StatePlane_Washington_South_FIPS_4602_Feet
Map produced 04-10-2012 - S. Job
A
TETRATECH
Figure 2-3. Land Use in the South Fork Nooksack River Watershed
Ecology has published the draft South Fork Nooksack River Temperature Total Maximum Daily Load,
Water Quality Improvement Report, and Implementation Plan (Ecology, 2016). Modeling analyses for the
draft TMDL have been completed and those draft results are discussed here for the purpose of evaluating
potential climate impacts. The draft Ecology documentation shows all available monitoring data,
boundary assumptions, the calibration procedures used, and an evaluation of model predictions versus
observed temperatures. The modeling linkage analysis used to estimate the temperature TMDL consists of
a Shade model (Ecology, 2003b) linked to the QUAL2Kw water quality model. QUAL2Kw (Chapra and
Pelletier, 2003; Ecology, 2003a; Pelletier et al., 2006) is used to simulate in-stream water temperature.
The models were developed for 2007 and 2010 summer conditions.
The Shade model was selected to evaluate solar radiation along the streams using watershed-specific
geographic information system (GlS)-based data derived with the TTools ArcView extension, developed
by Oregon Department of Environmental Quality (ODEQ). TTools uses spatial data to estimate
vegetation and topography perpendicular to the stream channel and samples longitudinal stream channel
characteristics such as the near-stream disturbance zone (NSDZ), riparian vegetation, and elevation.
Ecology's Shade model (Ecology, 2003b) was adapted from a program that ODEQ developed as part of
its HeatSource model version 6.
2-3
-------
SFNR Quantitative Assessment - Final
October 2016
The Shade model quantifies the potential daily solar load and generates the percent effective shade.
Effective shade is the fraction of shortwave solar radiation that does not reach the stream surface because
vegetative cover and topography intercept it. Effective shade is influenced by latitude and longitude; time
of year; stream geometry; topography; and vegetative buffer characteristics, such as height, width,
overhang, and density. Vegetation status was determined from a variety of data sources, including Light
Detection and Ranging (LiDAR) analyses of vegetation height completed in 2009 (downstream portion of
the watershed) and 2005 (upstream portion) by the Puget Sound LiDAR program; classification of
riparian vegetation type from the Riparian Function Assessment created by Duck Creek Associates based
on aerial photography from 1991 and 1995 and provided by the Nooksack Indian Tribe Natural Resources
Department; vegetation type information from the National Oceanic and Atmospheric Administration's
(NOAA) Coastal Change Analysis Program (CCAP) based on 2006 satellite imagery; and additional
digital ortho-imagery for Whatcom and Skagit counties flow in 2006 and 2009 and obtained from the
U.S. Department of Agriculture's (USDA) National Agricultural Imagery Program (NAIP). TTools
output serves as input for the Shade model, which is then used to generate longitudinal effective shade
profiles. Reach-averaged integrated hourly effective shade (i.e., the fraction of potential solar radiation
blocked by topography and vegetation) in turn serves as input into the QUAL2Kw model.
QUAL2Kw is a quasi-steady state model (it assumes steady-state hydraulics but represents the diel heat
budget and water quality kinetics) that is Ecology's preferred tool for estimating temperature TMDLs.
Model selection for the temperature TMDL is documented in Kennedy and Butcher (2012). The model
has also been previously used to evaluate stream temperature response to future climate in the Wenatchee
River basin (Cristea and Burges, 2010).
QUAL2Kw is a modernized version of EPA's standard river water quality model, QUAL2E (Brown and
Barnwell, 1987) and retains the diel heat budget framework of QUAL2E. The modernized version was
first developed as QUAL2K (Chapra et al., 2008), and further updated for use by Ecology as QUAL2Kw.
The algorithms used in QUAL2Kw are documented in Pelletier and Chapra (2008) and Pelletier et al.
(2006). The model simulates daily temperature and the heat budget with hourly variations in input
parameters and boundary conditions. Meteorological conditions have strong influences on water
temperature. Parameters included in QUAL2Kw input that affect stream temperature are effective shade,
solar radiation, air temperature, cloud cover, relative humidity, headwater and tributary temperature, and
hyporheic flow. These parameters are calculated (e.g., effective shade from Shade model), obtained from
weather station information, or interpreted from other sources.
For the TMDL, QUAL2Kw was applied to conduct focused analyses of critical conditions (e.g., late
summer low flow, clear sky, and high air temperature conditions) that exacerbate temperature
impairments, from which TMDL targets can be determined directly. Model inputs for the TMDL
simulations include flow, meteorological, and water temperature boundary conditions developed from
available data. Section 4 describes these assumptions in the context of developing future climate
boundary conditions. The TMDL documentation includes complete details.
2.1 Summary of Water Temperature Model Development
This section summarizes the development of the water temperature model for the SFNR TMDL.
Ecology's draft South Fork Nooksack River Temperature Total Maximum Daily Load, Water Quality
Improvement Report, and Implementation Plan (Ecology, 2016) provides more exhaustive
documentation. The model parameters described below are believed to be in final form, but are subject to
change until the TMDL is approved.
2-4
-------
SFNR Quantitative Assessment - Final
October 2016
2.1.1 Shade Modeling
Many factors contribute to warmer in-stream temperatures, including reduced shading from riparian
vegetation. Ecology typically evaluates the impacts of restoring system potential vegetation (SPV) and
associated shade in stream temperature TMDLs, where SPV refers to the climax tree community expected
to be attained on a given soil type. Increased shading typically reduces daily maximum water
temperatures, but has lesser impacts on minimum and daily average temperature (Johnson, 2004).
Both existing shade and 100-year system potential shade were estimated using the Shade model,
following Ecology protocols (Mohamedali and Stohr, 2011). Key data sources used include digital ortho-
imagery from the USDA's NAIP, LiDAR elevation data, land use and land cover data from NOAA's
CCAP, and local riparian condition data provided by the Nooksack Indian Tribe.
The riparian corridor was defined for shade analysis as a 150-foot buffer outside of the NSDZ on either
side of the river channel. Current vegetation type, height, density, and overhang are measured or
estimated within the riparian buffer, which serve as key inputs for the Shade model. In addition to the
vegetation information, TTools was also used to sample each 100-meter interval for channel wetted
width, NSDZ width, stream aspect, stream elevation, and topographic shade angles in all directions.
Where the NSDZ channel width was artificially high because of large tree islands, manual changes were
made to the widths before running the Shade model. (Outliers were checked if they were more than two
standard deviations away from the mean NSDZ width for the entire reach.) Using this information,
effective shade under current conditions was calculated using channel geometry, vegetation, and solar
position.
Next, system potential shade was developed. SPV is defined as the 100-year mature riparian vegetation
that would naturally occur if the riparian corridor was left undisturbed. The 100-year system potential
along the SFNR was determined spatially using a soil GIS coverage provided by the Washington State
Department of Natural Resources (WDNR) and Whatcom and Skagit County Soil Survey Reports
(Goldin, 1992; Klungland and McArthur, 1989). The WDNR data consists of digitized soil delineations
that designate which type of tree would dominate on each soil type in a system potential scenario. Using
the dominant soil types in the riparian area, 100-year site indices (i.e., the height of the dominant tree
species for a soil type at 100 years of age) were selected for those soil types.
On the basis of these data sources, western hemlock and Douglas-fir are the most common native tree
stands that would dominate the soil types in the SFNR. Therefore, the model for 100-year SPV focuses on
these two species and their 100-year system potential site index height as representative of the potential
for shading in the SFNR watershed. (Note that other species such as western red cedar, red alder, and
bigleaf maple dominate in some topographic positions and site-specific prescriptions will be needed in
specific restoration projects.) The 90th percentile height for Douglas-fir and western hemlock in Whatcom
and Skagit counties from the 100-year site indices for the major riparian soil types is 50.66 meters. This
height is used in the 100-year system potential scenario.
The resulting shade from existing vegetation (approximately 2007 conditions) and system potential shade
for the SFNR watershed is shown in Figure 2-4 shows the resulting shade from existing vegetation
(approximately 2007 conditions) and system potential shade for the SFNR watershed. Note that shade
was modeled along the SFNR mainstem only and not for tributaries, which are not explicitly simulated in
the QUAL2Kw model. Model scenarios that use SPV are based on the red line below for which the
riparian area is allowed to achieve SPV except where roads and houses are currently present.
2-5
-------
SFNR Quantitative Assessment - Final
October 2016
C
O
"3 —
nj •—•
* c
O
TO
o TO
^ IS
is ff
c >
OJ TJ
"K c
£ >
® Q.
C nj
O *-
¦— on
¦m Sr
u o
H3 Q.
,v o
>»
.£2
TJ
OJ
U
O
GQ
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
¦ Existing Shade
• System Potential: All Vegetated
•System Potential: All Vegetated Except Roads/Developed
NSDZ Width
-fr- —A—
J v\__^ Wl
1
-
HI
1 4^
4-
-W
100
50
0
150
10,000 20,000 30,000 40,000
Distance downstream from headwaters (m)
50,000
60,000
TJ
-------
SFNR Quantitative Assessment - Final
October 2016
Hutchinson Creek)
Skookum Creek
South Fork Nooksack River:
Model Reaches Segmentation
NAD_1983_HARN_StatePlane_Washington_South_FIPS_4602_Feet
Map produced 01-21-2013 • H Nicholas
Kilometers
it
TETRA TECH
V les
Wanlick Creek
Skagit
County
Legend
Tributary
Water
~~ County Boundary
Mainstem Model Segmentation
Symbol 1
Symbol 2
Figure 2-5. South Fork Nooksack River Model Segmentation by 1-kilometer Reach
Daily flow boundary conditions for the headwater, tributary, and ground water inputs were based on
steady-state flows during low-flow conditions m the calibration and corroboration periods. Observed flow
data in the watershed included a limited amount of data from U.S. Geological Survey (USGS) and
Ecology gages; much of the mainstem and tributaries are not gaged. Flow values, therefore, were
estimated throughout the watershed using regression equations from Curran and Olsen (2009), with
adjustments to account for differences between observed flow at the gages during the simulation days and
the low-flow statistic assumed to represent those days. Curran and Olsen performed an in-depth analysis
of low-flow hydrology and statistics for 25 gaging sites in the Nooksack River basin and developed
regional regression equations for estimating 12 critical low-flow statistics at ungaged locations. Modeled
flows match available in-stream flow observations well (Figure 2-6).
2-7
-------
SFNR Quantitative Assessment - Final
October 2016
South Fork Nooksack River, Aug 2, 2007
300
250
200
| 150
100
50
¦ Observed Flow
¦ — Model Flow
_ ¦ - -
r
I U
35
—i—
30
—i—
25
—i—
20
—i—
15
—i—
10
River Mile (from confluence)
Note'. The figure was adapted from the draft of Ecology's South Fork Nooksack River Temperature Total Maximum
Daily Load, Water Quality Improvement Report, and Implementation Plan (Ecology, 2016).
Figure 2-6. Comparison of Observed and Simulated Flows for the Calibration Period
In addition to hourly shade provided by the Shade model and flow boundary inputs, other major inputs to
the QUAL2Kw model included reach hydraulic parameters, hourly meteorology time series, and
boundary stream temperatures. Hydraulics inputs were developed using LiDAR and aerial imagery used
in the Shade work and data from a late 1990s USGS seepage study. Meteorology input data were based
on a variety of weather stations that were chosen on the basis of proximity to the watershed, data
availability for both calibration and validation dates, and reported quality assurance. Tributary and
headwater temperatures were based on the available monitoring data. Temperature for direct ground water
inputs were derived based on average annual air temperature with some adjustment during calibration.
2.1.2.2 Calibration, Validation, and Sensitivity
To conduct the model calibration process, a visual comparison of temperature along with a set of
statistical measures was used to compare model predictions and observations along the mainstem of the
SFNR (Figure 2-7). Two primary statistical measures were used: root mean square error (RMSE), a
commonly used measure of model variability, and relative percent difference (RPD) as a measure of bias.
The average RMSE for model calibration (2007) and validation (2010) of maximum temperatures was
less than 0.5 °C. Model bias evaluation showed no evidence of systematic over- or under prediction of
temperature. The TMDL report provides additional details.
2-8
-------
SFNR Quantitative Assessment - Final
October 2016
South Fork Nooksack River (Washington) (8/2/2007)
25 T
5 --
50 40
30 20
10 0
distance upstream (Km)
Temp(C) Average Modeled
¦ Mean Temp-data
Temp(C) Minimum Modeled
Temp(C) Maximum Modeled
~ Minimum Temp-data
~ Maximum Temp-data
Notes: Solid and dashed lines show model results; open and closed squares show observed temperature data. The
figure was taken from the draft of Ecology's South Fork Nooksack River Temperature Total Maximum Daily Load,
Water Quality Improvement Report, and Implementation Plan (Ecology, 2016). The chart shows model segment
numbers for observations.
Figure 2-7. Temperature Calibration for QUAL2Kw Model of South Fork Nooksack River
A sensitivity analysis was conducted to test the calibrated model's response to key inputs. Changes in
boundary temperature inputs (e.g., headwater, tributaries, and ground water) had the greatest impact on
water temperature. Air temperature had the second-largest impact on the model predictions for water
temperature. The second- and third-largest impacts on maximum predicted water temperature were from
boundary flow and bottom width. Changes in Manning's n and shade had the least impact on the model
outputs.
2.2 TMDL Modeling Scenarios
Using the 2007 calibration model as a foundation, a series of modeling scenarios were developed to
evaluate stream temperatures on the mainstem of the SFNR realized under various typical and critical
summer conditions. TMDL modeling scenarios were constructed using Ecology's standard assumptions
and protocols for temperature TMDLs. These include typical low-flow conditions (7Q2 flow - the 7-day
average flow with a 2-year recurrence frequency) and critical low-flow conditions (7Q10 flow - the 7-day
average flow with a 10-year recurrence frequency) coupled with typical and critical 7-day average
maximum temperatures, as explained below. These conditions are selected to correspond to the water
quality criterion for temperature expressed as 7-DADMax. For each of these flow and temperature
regimes, scenarios using shade provided by existing vegetation and 100-year SPV were developed. Those
scenarios that incorporate system potential shade also included a microclimate effect, which is discussed
further below.
2-9
-------
SFNR Quantitative Assessment - Final
October 2016
2.2.1 Scenario Flows and Air Temperature
Flow inputs for TMDL critical condition scenarios were based on calculated 7Q2 and 7Q10 statistics
from Curran and Olsen (2009) for three of the gaged sites in the SFNR watershed using various standard
USGS methods, as shown in Table 2-2. Figure 2-8 shows flow gages and air temperature monitoring
stations used for boundary condition development.
Table 2-2. Calculated 7Q2 and 7Q10 Flows Compared to Calibration Conditions
Location
Gage ID
Flow (cfs)
Calibration
7Q2
7Q10
Skookum Creek
12209490
44
20.6
15.3
South Fork Nooksack at Wickersham
12209000
190
102
75.8
Hutchinson Creek
01C070
7.9
4.92
4.37
Legend
© Meteorological Station
A Flow Station
SF Nooksack River
Tributary
Water
WA Ecology Station
"SF Nooksack at Potter Rd
01F070
WA Ecology Station
"Hutchinson Creek nr Acme
01C070
SNOTEL Station
"Elbow Lake"
~ County Boundary
Whatco
Skagit
County
USGS Station /
"Skookum Creek"'
12209490
USGS Station
"SF Nooksack at Wickersham
12209000
NOAA SOD Station
"Sedro Woolley"
WA457507
South Fork Nooksack River:
Meteorological and Flow Stations
NAD_1983_HARN_StatePlane_Washington_South_FIPS_4602_Feet
Map produced 04-04-2014 - H. Nicholas
Kilometers
TETRA TECH
Figure 2-8. Flow Gages and Air Temperature Monitoring Stations used for Boundary Condition
Development
The 50th percentile air temperatures are calculated as the median of the series of annual maxima based on
annual rolling 7-day average maximum temperatures. The 90th percentile temperature is similarly
calculated from the series of annual maxima. The air temperature analysis used historical records from
2-10
-------
SFNR Quantitative Assessment - Final
October 2016
two meteorological stations, representing lower and upper elevations. Elbow Lake Station 910 (a Snow
Telemetry or ""SNOTEL" station) with available data from August 1995 through 2012 was used as the
upper elevation station. Ecology station 01F070 was used to represent lower elevations during model
calibration, but had a much shorter period of record beginning in 2003. Therefore, an alternate station,
Sedro Woolley Station WA457507 (NOAA Summary of the Day [SOD] Program), farther downstream in
the town of Nooksack, was selected to represent lower elevations in the TMDL analysis. Linear
interpolation on elevation between the high and low stations was used to adjust the daily maximum and
minimum air temperatures for each model reach. Using the period of record at SNOTEL 910 and data
from 1959 to 2009 at WA457507, 50th and 90th percentiles were calculated from the series of hottest
annual 7-day average maximum temperatures (Table 2-3). The WA457507 values were then scaled by
0.47 °C to adjust for the difference relative to station 01F070 (determined using the average difference
between air temperatures from July and August of paired years between the two stations). The same
procedure was used to establish the appropriate percentile of the maximum 7-day average daily minimum
temperature, while the hourly distribution of temperatures within the day was scaled to follow the same
pattern between the maximum and minimum as observed in the calibration data set for August 2, 2007.
Table 2-3. Air Temperature Statistics for the South Fork Nooksack Region
Condition
SNOTEL Station 910
(1996-2012)
Adjusted SOD Station
WA457507a
(1959-2009)
Max 7-
DADMin °C
Max 7-
DADMax °C
Max 7-
DADMin °C
Max 7-
DADMax°C
90th percentile of annual max 7-day series'5
17.56
28.69
15.39
30.07
50th percentile of annual max 7-day series'5
15.21
26.77
13.64
28.01
Notes'.
a Observations at WA457507 are adjusted to 01F070 by adding 0.47 °C.
b Results are shown for the maximum 7-day average of daily temperature minima (7-DADMin) and the maximum
7-day average of daily temperature maxima (7-DADMax).
Ecology protocols for temperature TMDLs consider a microclimate effect. The microclimate effect,
incorporated into all the SPV scenarios, reflects a cooling of air temperature near the stream channel as a
result of the presence of mature riparian vegetation. Brosofske et al. (1997) reported that a buffer width of
at least 150 feet was required to maintain natural riparian microclimate environments in small forest
streams in western Washington in Douglas-fir and western hemlock forests. The average impact of clear
cutting on ambient mean daily air temperature was an increase of 2 °C according to a literature review
provided in Bartholow (2000). Using the aggregated results of this study, Ecology concluded that an
opposite impact could be expected from reforesting an area, such that the full microclimate impact of a
mature riparian forest would decrease ambient air temperature by 2 °C. (Because much of the SFNR is
wider than the streams studied by Brosofske et al. (1997), the validity of this assumption is uncertain.) As
specified by Ecology, this microclimate effect is simulated by a drop in air temperature of 2 °C at every
hour of the day for scenarios in which SPV is present. The presence of a mature forest canopy could
actually lead to less radiative cooling and warmer nighttime air temperatures; however, a fixed decrease
in air temperature at all hours was applied to be consistent with Ecology's TMDL modeling practices
(e.g., Stohr, et al., 2011). As was done in the Lower Skagit temperature TMDL (Lawrence, 2008), dew
point was held at 90th percentile conditions during microclimate runs for 7Q10 flows (unless it exceeded
air temperature) to preserve increased relative humidity expected during the microclimate scenarios
(Zalewsky and Bilhimer, 2004).
2-11
-------
SFNR Quantitative Assessment - Final
October 2016
2.2.2 Modeling Results
Table 2-4 summarizes the results for the draft modeling scenarios for typical low-flow and weather
conditions and critical low-flow and weather conditions. Figures showing the simulations are not
reproduced here; however, the results of the modeling scenarios for existing climate are shown in
conjunction with the climate scenario results in Section 5.
Table 2-4. Modeling Scenario Results for Typical Low-Flow and Critical Low-Flow and Air
Temperature Conditions
Scenario
Condition
Maximum Stream Temperature (°C)
(averaged across select reaches)
All Reaches
Headwaters
to Reach 28a
Reach 28a to
Outlet
Typical Low-Flow Conditions (7Q2 flows; 50th percentile air temperature)
1
Current Shade Conditions: 7Q2
19.0
18.4
19.6
2
System Potential except where developed: 7Q2b
16.9
16.2
17.5
Critical Low-Flow Conditions (7Q10 flows; 90th percentile air temperature)
3
Current Shade Conditions: 7Q10
21.0
20.1
21.8
4
Current Shade Conditions with cooler tributaries'5:
7Q10
20.7
19.6
21.6
5
System Potential Vegetation with microclimate effect
except where developed, cooler tributaries'5: 7Q10
18.7
17.8
19.6
6
System Potential Vegetation with microclimate effect
everywhere, cooler tributaries'5: 7Q10
18.7
17.8
19.6
Notes:
a From the headwaters to reach 28 the summer water quality criterion is 12 °C. For reach 28 to the mouth the water
quality criterion is 16 °C.
b Tributaries are set at current temperatures or water quality criteria temperatures, whichever is cooler.
During both typical low-flow (Scenario 1) and critical low-flow (Scenario 3) conditions, and
corresponding meteorological conditions in the summer, the model estimates that the SFNR exceeds the
numeric water quality criteria of 12 °C (headwaters to reach 28) and 16 °C (reach 28 to outlet) in nearly
all mainstem river segments. Maximum temperatures estimated for Scenario 1 averaged 18.4 °C and
19.6 °C for the upstream and downstream reaches, respectively, and exceed the numeric criteria by 3.6 °C
to 6.4 °C.
To estimate the stream temperature profile under conditions of maximum potential shade, the models
were run with 100-year SPV, associated microclimate effects, and with tributaries and headwaters at or
below the numeric water quality criteria. Under both typical and critical 100-year system potential
scenarios (Scenarios 2 and 5), the model predicts that the stream will continue to exceed the numeric
water quality criteria. Averaged over reaches upstream of Fobes Creek, the predicted temperatures are
4.2 °C and 5.8 °C above the criterion for the 7Q2 and 7Q10 scenarios. Below the point where the numeric
criterion changes to 16 °C (for the simulation dates) the predicted water temperatures are 1.5 °C to 3.6 °C
above the criterion. On average, the impact of system potential shade with the microclimate effect on
critical condition stream temperatures was a cooling of about 2.21 °C relative to stream temperatures
under the current shade scenarios. The addition of 100-year SPV in the developed areas under critical
conditions (Scenario 6) did not have a noticeable effect because the change applied to a relatively small
area (1.6 percent) of the riparian buffer.
2-12
-------
SFNR Quantitative Assessment - Final
October 2016
Additional model runs investigated response to critical conditions for the September through July time
period, when the supplemental spawning criterion of 13 °C applies to the entire mainstem of the SFNR,
using September 11, 2007 meteorology. Lower thermal inputs for September result in lower simulated
water temperatures than are predicted for the summer critical conditions run. The scenario with shading
and tributary temperatures equal to those for TMDL Scenario 5 yield an average maximum stream
temperature of 13.3 °C, slightly greater than the criterion. The summer critical condition analysis is thus
more limiting on thermal load allocations.
Scenario 5 uses shade levels provided by 100-year SPV to estimate water temperatures occurring under
system potential shade at critical low-flow and high air temperature conditions. This is interpreted as a
natural limit on what can be attained under existing climate. This limiting condition then becomes the
applicable temperature water quality criterion under the natural conditions provision for the water quality
criteria, WAC 173-201A-070(2), which states, "Whenever the natural conditions of said waters are of a
lower quality than the criteria assigned, the natural conditions shall constitute the water quality criteria."
2.2.3 Evaluation of Historical Conditions
Several supplemental modeling scenarios undertaken for the TMDL analysis investigated possible stream
temperature responses during critical conditions with inferred natural, historical conditions for the
watershed land cover and stream channel geometry. These scenarios evaluated (1) a 20 percent reduction
in headwater and tributary temperatures, (2) decreased channel width representative of conditions prior to
land use disturbance, (3) shade associated with full climate vegetation (assumed height of 290 feet)
attained everywhere within a wider, 218-foot stream buffer, (4) greater hyporheic exchange, and
(5) combined impacts of all four individual factors. These analyses suggest that, under historical
conditions, stream temperatures in response to 7Q10 critical conditions could be up to about 16 percent
lower than predicted for Scenario 5, with the predicted average maximum stream temperature over all
reaches dropping from 18.7 to 15.8 °C.
2.2.4 Draft TMDL Results
The general goal of the TMDL, however, is to develop allocations of thermal load such that the sum of
these allocations does not exceed the water body's loading capacity. The loading capacity provides a
reference for calculating the amount of pollutant reduction needed to bring water bodies into compliance
with water quality standards. EPA defines loading capacity as, "the greatest amount of loading that a
water can receive without violating water quality standards" [Title 40 of the Code of Federal Regulations
(CFR) 130.2(f)], The loading capacity for the mainstem of the SFNR can be approximated using the
100-year system potential shade scenario (Scenario 5) under critical low-flow (7Q10) and air temperature
conditions (90th percentile of the series of hottest rolling 7-day average of daily maximum temperatures
from each year of record). This scenario also assumes water temperatures within the tributaries and
headwaters are at or below the numeric water quality criteria. Thermal loading under these conditions can
provide the basis for TMDL load allocations.
2-13
-------
SFNR Quantitative Assessment - Final October 2016
2-14
-------
SFNR Quantitative Assessment - Final
October 2016
3 Climate Change Modeling
The climate-altered boundary conditions for the QUAL2Kw modeling of the SFNR are derived from the
work conducted by the Climate Impacts Group (CIG) at the University of Washington (Hamlet et al..
2010; Hamlet et al., 2013; Mauger and Mantua, 2011; Mote and Salathe, 2010). CIG focuses on the
consequences of a warming climate in the PNW. Among their key products is the Washington Climate
Change Impacts Assessment (Miles et al., 2010), a comprehensive assessment of the impacts of climate
change on the state of Washington, developed under mandate of the Washington State legislature. Figure
3-1 provides a general schematic of the relationships between CIG climate products and the TMDL
model, with details in the following sections.
CIG Downscaled Future Climate
Precipitation, Air Temp,
Evaporation
VIC Regional Hydrologic
Model
Change
Factors
Air Temp Solar Radiation,
Vapor Pressure Deficit
Tributary Water
Temperature Prediction
Model
Solar Radiation
Shade Model
\
Existing
Hydrology
Climate-altered
Boundary Flows
QUAL2Kw Future Climate Instream
Temperature Model
Parameters
I I
Calibrated QUAL2Kw Instream Temperature Model
Figure 3-1. Schematic of Model and Climate Data Integration for the Quantitative Assessment
3.1 Climate Model Applications for the Pacific Northwest
3.1.1 CIG Climate Projections
The basis of the CIG climate change assessment is a common set of simulations using 21 Global Climate
Models (GCMs) coordinated through the IPCC as part of the Fourth Assessment Report (AR4) and
described in the IPCC 2007 report (Randall et al, 2007). These models are global in scope, and therefore
the spatial resolution across the state of Washington is coarse (i.e., only a few grid cells cover the state).
Overall, the suite of models from AR4 provide a credible representation of observed climate in the PNW,
reproducing both the observed seasonal cycle and 20th century warming trend of 0.8 °C (Randall et al.,
2007). The large-scale GCMs, however, provide a coarse representation of local climate because the scale
3-1
-------
SFNR Quantitative Assessment - Final
October 2016
of these model results precludes fine-scale resolution of local elevation effects and the details of land-sea
interactions. For this reason it is important to use the models to inform the potential for relative changes
at the local scale, rather than use climate model output directly. Mote and Salathe (2010) evaluated GCM
performance in simulating historical climate over the PNW as a whole by comparing the regional average
model output over the PNW to 0.5° gridded data for 1970-1999 obtained by assimilating observed data
into the Climate Research Unit weather prediction model (Mitchell et al., 2004). This adjusts for the
uneven distribution of weather stations, especially in high terrain, and yields a direct basis for
comparison. The GCMs are in a general sense consistent with observed spatial and temporal patterns of
climate in the PNW. On detailed comparison for 1970-1999, Mote and Salathe found that most of the
GCMs have a cold bias (mean and median of -1.8 °C) on average over the whole PNW region. The
models also all have a wet bias, with a mean bias of +0.6 centimeters per month.
The spatial scale of the global models is too coarse for evaluating the hydrologic and water quality
response of even moderate-size watersheds because it does not resolve local elevation and local land-
water interactions well. For more local analysis it is necessary to downscale the global predictions to
finer-scale local predictions. Downscaling can be accomplished in two general ways: by running a more
detailed regional climate model that is forced by the global model (dynamical downscaling) or through
using statistical relationships between GCM output and local climate variations (statistical downscaling).
The two approaches can yield different results and both have limitations. Statistical downscaling assumes
that spatial patterns of weather remain unchanged under future climate, which might not be true, while
regional climate models used in dynamical downscaling are themselves subject to high levels of
uncertainty and might not fully resolve local topographic effects. CIG used a statistical downscaling
approach and has taken projected time series from GCMs and downscaled the meteorological output to a
1/16th degree (approximately 26. 7 km2 or 6,600 acres at this latitude) resolution for the PNW (see Hamlet
et al., 2013; Polebitski et al., 2007). This has been done for 10 GCMs and multiple emission scenarios for
the period through 2099. Downscaling is also done in two different ways: using a composite delta method
in which there is a single average change (delta) for each month calculated from a time slice of the GCM
for the region that is applied to every day in that month, and a hybrid delta approach that uses statistical
bias correction to maintain the probability distribution.
CIG has also produced gridded estimates of surface runoff and baseflow at the 1/16 degree scale, using
the Variable Infiltration Capacity or VIC model (Eisner et al., 2010; Gao et al., 2010; Liang et al., 1994).
Both the downscaled GCMs and VIC output are available for calendar years 1915 through 2006 as well as
future climate projections based on the hybrid delta method modifications of the 1915-2006 historical
climate series.
Seventy-nine climate scenario products from AR4 are available from CIG covering the SFNR watershed.
These products incorporate a range of greenhouse gas emissions scenarios as interpreted through multiple
GCMs. This project focuses on a limited subset of model results for the IPCC A1B emissions scenario.
The A IB scenario is considered to provide a robust data set for a moderate emission scenario. Other IPCC
emission scenarios predict greater warming, but there are only small deviations among scenarios until the
latter half of the 21st century, and the range of models using the A IB scenario covers most of the A2 (high
emission) scenario range through the 2080s. The upper bound of the suite of GCMs using the A1B
scenario projects an average annual air temperature increase for 2080 that is only 0.6 °C lower than the
upper bound from the A2 scenario.
The IPCC recently began releasing updated GCM results for the Fifth Assessment Report (AR5). The
climate models perform similarly to the AR4 projections for the PNW and have the same climate
sensitivities. Differences in projected results are primarily due to differences in greenhouse gas emission
scenarios. Work funded by the Northwest Climate Science Center is ongoing to develop downscaled
hydrologic projections for the PNW using AR5, but results are not yet available at fine resolution
comparable to the AR4 results.
3-2
-------
SFNR Quantitative Assessment - Final
October 2016
3.1.2 Time Horizons for Analysis
Climate models are not of sufficient spatial resolution to resolve watershed-scale processes and are
furthermore subject to biases in their simulation of climate. It is therefore standard practice to use a
"delta" or "change factor" approach in which future climate time series are represented by perturbing
historical climate time series in accordance with the differences between future and historic climate
indicated by the downscaled GCMs. This approach preserves realistic temporal sequencing and
minimizes bias present in the GCM simulations (Eisner et al., 2010). The GCMs themselves predict
continuous changes in climate over time as a result of the changing concentrations of greenhouse gases
and natural variations in climate. To minimize the impact of random and cyclic fluctuations in climate,
the change analysis is typically based on the average change assessed over 30-year windows.
Downscaled climate projections produced by CIG are the result of bias correction of global models,
spatial downscaling, and translation to a future time frame (Hamlet et al., 2013). The first step is
implemented by comparing the historical simulations of the GCMs with observations, and then applying
monthly differences, or "change factors" to the full GCM record to correct for spatial bias. In other words,
the GCM output is adjusted to match observed data (CIG used data from 1950-1999 for the spatial bias
analysis), and these same adjustments are applied to the remainder of the GCM time series. These
adjustments are applied separately for each calendar month.
The bias correction and statistical downscaling steps are first applied to model output for 1915-2006,
resulting in a 92-year time series that represents historical conditions and their variability.
Future climate products from CIG are also provided as 92-year time series, although relying on the
average differences between two 30-year time windows. The future climate series are based on
modification of the historic, 1915-2006 time series using a hybrid delta approach. The hybrid delta
approach uses cumulative distribution functions to adjust the historical time series based on the change in
the probability distribution of temperature and precipitation projected by the bias-corrected GCMs (see
Hamlet et al., 2010 and Hamlet et al., 2013, Chapter 4). This step relies on the comparison of cumulative
distribution functions between the future period of interest (e.g., 2070-2099) and a chosen historical
window (specified as 1970-1999, which matches the historical time window chosen by Mote and Salathe
(2010)). These change factors are then applied to the 92-year historical time series (1915-2006) to
produce a perturbed time series that has the same time series characteristics of the historical record but a
probability distribution that is shifted to match the change projected by GCMs. Using the 92-year historic
time series has two principle advantages: (1) it provides a more accurate representation of local spatial
and temporal weather variability than is provided by the climate models and (2) using a longer time series
allows for better sampling of cyclic natural variability in the climate system, such as El Nino and the
Pacific Decadal Oscillation. Despite being constructed based on specific 30-year windows, the
downscaled projections are referred to using the shorthand "time horizon" names listed below (refer to
Table 3-1):
Table 3-1. Time Horizons for the Climate Analysis
Time Horizon
Cumulative
Distribution Basis
Resulting Time Series
Historic (1980s)
1970-1999
1915-2006, bias corrected and statistically downscaled
2020s
2010-2039
1915-2006 shifted by relationship of 2010-2039 to 1970-1999
2040s
2030-2059
1915-2006 shifted by relationship of 2030-2059 to 1970-1999
2080s
2070-2099
1915-2006 shifted by relationship of 2070-2099 to 1970-1999
From Table 3-1 it will be noted that a climate time series referred to as the "2020s time horizon" is
actually a 92-year time series that is created by modifying the 1915-2006 historical time series based on
3-3
-------
SFNR Quantitative Assessment - Final
October 2016
the difference between the 2010-2039 projected results (centered at 2025) and those for 1970-1999
conditions (centered at 1985). In addition to the climate shift between these two periods, the resulting
series also incorporates the variability in climate (whether of natural or anthropogenic origin) experienced
over the 1915-2006 historical data. Therefore, it is more appropriate, for instance, to refer to the "2080s
time horizon" than to the specific year 2085 for future scenarios.
The shift expected under future climate is calculated by CIG based on comparison to historic conditions
centered at 1985, not 2013. Because the TMDL is based on calibration to conditions observed in 2007
with subsequent adjustment to the 90th percentile of 7-day maximum air temperatures observed from
1995-2012 there is likely some additional margin of safety in the analysis due to the difference between
the 30-year distribution function of temperatures centered at 1985 and the underlying distribution
statistics that would be appropriate to conditions centered at about 2003.
3.1.3 Climate Scenarios Selected for the South Fork Nooksack River TMDL
As discussed above, Mote and Salathe (2010) evaluated biases in the global-scale climate model
predictions for the PNW. No single GCM fell into the best five of the GCMs for prediction of both
temperature and precipitation; likewise, no GCM fell into the worst five for both temperature and
precipitation. It is thus not appropriate to select a specific GCM based on its perceived prediction skill for
the area; instead, the suite of GCMs is more appropriate for analyzing the potential ensemble range of
future climates (Mote et al., 2011). This is consistent with findings of Knutti et al. (2010) and Pierce et al.
(2009) that attempts to cull the best GCMs yields little difference in representing likely future change
relative to a randomly selected subset of GCMs.
We evaluated three time horizons (2020s, 2040s, and 2080s) using the hybrid delta results from GCMs
under the A1B scenario for the South Fork Nooksack watershed. We selected three GCMs for the analysis
that are anticipated to produce the least warming of air temperature (model low-impact scenario), medium
warming (medium-impact scenario), and highest warming (high-impact scenario), resulting in 3 climate
models x 3 time horizons = 9 runs. This addresses the project objective of evaluating the ensemble range
of outcomes from one IPCC emission scenario for the climate change risk assessment.
Selection of specific GCMs also considered factors other than average increase in annual air temperature.
Maximum risk would be expected to coincide with increases in summer temperatures accompanied by
decreases in summer baseflow. The USFS North Cascadia Adaptation Partnership (NCAP) has suggested
a general storyline for impacts on aquatic habitat that includes the following components:
• Longer duration, higher stream temperatures, and lower summer baseflow.
• Transitions among the three basic PNW streamflow patterns (snowmelt-dominated, transient
(mixed snow and rain runoff), and rain-dominated hydrographs). The lower SFNR is now
classified as transient, but is expected to transition to rain-dominated by the 2020s under A IB.
The upper reaches of SFNR are currently snowmelt-dominated, but are expected to transition to
transient as early as the 2020s.
• Precipitation is expected to increase on an annual basis due primarily to increased winter rainfall
and earlier, more intense fall rainfall. This is likely to increase winter flooding in sensitive
transient river basins such as the SFNR.
Selection of specific GCMs is complicated by the fact that rankings switch for different time periods.
Nonetheless, we have identified three GCMs that meet the general criteria for low, medium, and high
warming while also demonstrating consistency with the NCAP storyline. These are as follows:
3-4
-------
SFNR Quantitative Assessment - Final
October 2016
Low-Impact Scenario: This scenario uses the CGCM3.1_t47 or Third Generation Coupled Global Climate
Model (CGCM) from the Canadian Centre for Climate Modelling and Analysis (www.ec.gc.ca/ccmac-
cccma/default.asp?n=1299529F-l'). Across the entire Columbia Basin this scenario shows the least
warming by the 2040s and remains below the average through the 2080s. Temperature results for the grid
cells intersecting the SFNR watershed are consistent with low warming. This scenario also shows a strong
increase in total annual precipitation and a small decrease in summer precipitation compared to other
scenarios, and is consistent with low impact in the NCAP storyline. CGCM3.1_t47 was also found to
have the best performance across the 1970-1999 historical data set on evaluation of combined
performance in predicting air temperature, precipitation, and sea level pressure (Mote and Salathe, 2010).
High-Impact Scenario: This scenario uses HADGEM1 (the Hadley Centre Global Environmental Model)
from the Hadley Centre in the UK (Johns et al., 2006). Across the entire Columbia Basin this scenario has
the second greatest increase in air temperature for the 2040s (and the highest increase of those available in
the VIC model output) and the greatest increase in air temperature for the 2080s (+5.39 °C). Results for
the SFNR are similar to findings for the Columbia Basin. This GCM also predicts a large decrease in
summer precipitation and is consistent with high impact under the NCAP storyline. HADGEM1 exhibits
average biases across the PNW that are near the ensemble mean for both air temperature and
precipitation.
Medium-Impact Scenario: We chose to use a single GCM rather than a composite of multiple GCMs to
represent a medium-impact condition because of concerns that a composite might not correctly reproduce
correlations between air temperature, humidity, and other variables that jointly influence water
temperature extremes. No GCM exactly follows the central trend of the ensemble of models on both
temperature and precipitation; however, CCSM3 (the Community Climate System Model) provides a
good compromise and is used for this purpose. CCSM3 is supported by the National Center for
Atmospheric Research in Boulder, CO (Collins et al. 2006). This model falls near the average on
temperature change for the PNW in both the 2040s and 2080s. There is a relatively large predicted
decrease in summer precipitation, although not as extreme as the high warming scenario. Using this
model could help distinguish between impacts due to air temperature and those due to precipitation
changes. CCSM3 also exhibits biases on the historical 1970-1999 data sets that are near the ensemble
mean (Mote and Salathe, 2010).
Table 3-2 compares predicted temperature and precipitation for the three selected GCMs for the SFNR to
the minimum, mean, and maximum predictions from a 10-model ensemble identified in NCAP's draft
(May 2013) North Cascadia Vulnerability Assessment (http://www.northcascadia.org/). Figure 3-2
provides a graphical comparison of the selected GCMs and the NCAP ensemble mean.
3-5
-------
Table 3-2. Comparison of Selected SFNR Climate Scenarios to 10-model Ensemble for the North Cascadia Region
Period
Month
PCM1 (NCAP
least warming)
SFNR Model Low-Impact
CGCM3.1_t47 (least
warming)
NCAP 10-model
ensemble mean
(moderate
warming)
SFNR Model
Medium-Impact
CCSM3
MIROC 3.2
(NCAP most
warming)
SFNR Model
High-Impact
HADGEM1
(most warming)
Temperature change relative to 1970-1999 baseline (° C)
2040s
DJF
2.0
2.2
1.8
1.9
2.7
1.7
MAM
1.3
1.4
1.7
1.7
3.0
3.0
JJA
1.9
1.5
2.7
3.3
2.8
3.8
SON
2.0
1.4
2.2
1.7
2.4
3.0
Annual
1.8
1.6
2.1
2.2
2.7
2.9
2080s
DJF
3.2
3.0
3.4
2.4
4.6
4.3
MAM
2.0
2.0
3.2
2.9
4.8
5.6
JJA
3.3
2.7
4.9
4.0
4.9
6.1
SON
2.4
2.2
3.9
2.5
4.3
4.8
Annual
2.7
2.5
3.8
3.0
4.6
5.2
Precipitation change relative to 1970-1999 baseline (percent)
2040s
DJF
-8
21
4
0
6
-7
MAM
10
9
4
8
1
5
JJA
-3
-15
-10
-39
-8
-41
SON
-6
12
3
6
17
4
Annual
-2
13
0
1
4
-4
2080s
DJF
9
17
9
-1
9
-3
MAM
5
18
5
12
9
10
JJA
-19
-12
-15
-25
-30
-33
SON
-13
16
9
3
14
26
Annual
-5
14
2
1
0
6
Note'.
PCM: Parallel Climate Model; MIROC: Model for Interdisciplinary Research on Climate; DJF: December, January, February; MAM: March, April, May; JJA: June,
July, August; SON: September, October, November. Results are weighted averages across model grid cells covering the SFNR watershed.
-------
SFNR Quantitative Assessment - Final
October 2016
2040s
¥ 6.0
J 2.0
I DJF
IMAM
IJJA
¦ SON
2040s
rr
NCAP Mean CGCM3.1 CCSM3 HADGEM1
NCAP Mean CGCM3.1
2080s
2080s
¦ MAM
¦ MAM
NCAP Mean CGCM3.1
CCSM3
HADGEM1
NCAP Mean CGCM3.1
CCSM3
HADGEM1
Note: Results are weighted averages across all climate model grid cells intersecting the SFNR watershed. DJF =
December, January, February; MAM = March, April, May; JJA = June, July, August; SON = September, October,
November.
Figure 3-2. Climate Scenario Changes in Precipitation and Temperature Compared to North
Cascadia (NCAP) Ensemble Mean
3.2 Future Climate Estimates for the SFNR
Model predictions for future air temperature, precipitation, and other climate variables from the low,
medium, and high-impact GCMs are summarized below for the grid cells intersecting the SFNR
watershed. Projections reported here for the SFNR are generally consistent with those reported in draft
form by Mauger and Mantua (2011) for the North Cascades area (Omernik ecoregion 77), which contains
the SFNR watershed. Their analysis of CIG GCM modeling products reports monthly average air
temperature in the 2080s rising by about 2.8 °C in the winter and 4.6 °C in the summer, with precipitation
varying widely among models, but tending on average to increase by about 19 percent in the winter and
decrease by about the same percentage in the summer. As a result of temperature increases, potential and
actual evapotranspiration is also predicted to increase in the summer, further reducing summer flows.
3.2.1 Air Temperature
Figure 3-3 shows air temperature projections area-weighted and spatially averaged over the downscaled
grid cells intersecting the SFNR watershed and temporally averaged over the 91-year time series
associated with each climate time horizon. (Note that in this and similar subsequent figures, the monthly
averages are connected by smoothed lines to improve legibility.) The results are generally consistent with
the average changes for the PNW reported by Mantua et al. (2010): By the 2080s the average summer
(July-September) air temperature is projected to rise by amounts ranging from 2.81 °C (CGCM3) to
3-7
-------
SFNR Quantitative Assessment - Final
October 2016
6.31 °C (HADGEM1), while the average winter (December-February) air temperature is projected to rise
by amounts ranging from 2.44 (CCSM3) to 4.28 °C (HADGEM1).
His orica
critical Period
Average Temperature (deg C)
Scenario: cgcm3.1_t47^A1B (low impact scenario)
BP|
^5
Average Temperature (deg C)
Scenario: ccsm3_A1B (medium impact scenario)
Wmk
w
Us
¦k
N
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
3
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Average Temperature (deg C)
Scenario: hadgem1_A1B (high impact scenario)
Historical
Critical Period
Note'. Results shown are medians across all the CIG grid cells intersected by the SFNR watershed. August is
highlighted as a critical period because it typically combines the highest air temperatures and lowest precipitation.
Figure 3-3. Climate Model Projections of Monthly Average Air Temperature in the SFNR Watershed
for Low-, Medium-, and High-Impact Climate Scenarios
The increase in the maximum daily average temperature might be even greater than the increase in the
monthly average, as shown in Figure 3-4. The spatial median maximum air temperature for the SFNR,
which occurs in August, is projected to increase by 3.32 °C (CGCM3) to 7.74 °C (HADGEM1).
3-8
-------
SFNR Quantitative Assessment - Final
October 2016
Maximum Temperature (deg C)
Scenario: ccsm3_A1B (medium impact scenario)
Maximum Temperature (deg C)
Scenario: cgcm3.1 _t47_A1B (low impact scenario)
Historical
Historical
28 -
28 -
Critical Period
Critical Period
2020s
2020s
2040s
2040s
20 ¦
20 ¦
2080s
2080s
Jul
Jul
Maximum Temperature (deg C)
Scenario: hadgem1_A1B (high impact scenario)
Historical
28 -
Critical Period
2020s
2040s
20 -
2080s
Jul
Note'. Results shown are medians across all the CIG grid cells intersected by the SFNR watershed.
Figure 3-4. Climate Model Projections of Monthly Maximum of Daily Average Air Temperature in
the SFNR Watershed for Low-, Medium-, and High-Impact Scenarios
The pattern of projected air temperature increases is similar across the SFNR watershed; however, the
absolute magnitude of the temperature decreases at higher elevations (Figure 3-5).
Historical
Critical Period
20 -
'HADGEM1
u
-CCSM3
ai
'CGCM3.
+-<
TO
ai
a.
E
ai
H
<
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
25
Historical
Critical Period
20
•HADGEM1
u
•CCSM3
I 15
TO
01
Q.
! 10
~CGCM3.1
'<
5
0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Figure 3-5. 2080s Climate Model Projections of Average Air Temperature in the SFNR Watershed
at the Headwaters (left) and Outlet (right)
3.2.2 Precipitation
Projected changes in future precipitation (Figure 3-6) are less dramatic than changes in air temperature.
All three GCMs predict an increase in fall precipitation by the 2080s accompanied by a small decrease in
3-9
-------
SFNR Quantitative Assessment - Final
October 2016
summer precipitation. The low-impact scenario (CGCM3) projects an increase in winter-spring
precipitation, while the high-impact scenario (HADGEM1) suggests a decrease in January-February
precipitation. The spatial distribution of 2080s precipitation is shown in Figure 3-7.
Precipitation (mm)
Scenario: ccsm3_A1B (medium impact scenario)
Precipitation (mm)
Scenario: cgcm3.1 _t47_A1B (low impact scenario)
Historical
Historical
Critical Period
Critical Period
Precipitation (mm)
Scenario: hadgem1_A1B (high impact scenario)
Historical
Critical Period
Note'. Results shown are medians across all the CIG grid cells intersected by the SFNR watershed.
Figure 3-6. Climate Model Projections of Monthly Median Precipitation in the SFNR Watershed for
Low-, Medium-, and High-Impact Scenarios
Historical
Critical Period
HADGEM1
•CCSM3
-CGCM3.1
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Historical
60 -
Critical Period
•HADGEM1
- 50 -
E
E
T 40 -
o
>CCSM3
•CGCM3.1
Q.
QJ
20 -
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Figure 3-7. Climate Model Projections of 2080s Average Precipitation in the SFNR Watershed at
the Headwaters (left) and Outlet (right)
3.2.3 Relative Humidity, Dew Point, and Vapor Pressure Deficit
Evaporative cooling can be an important component of the stream heat balance under low-flow
conditions. The evaporation rate depends on the moisture content above the air-water interface and wind
3-10
-------
SFNR Quantitative Assessment - Final
October 2016
mixing. Ground-level atmospheric moisture is summarized via the relative humidity (percent of the
saturation vapor concentration), dew point temperature (temperature of saturation), and vapor pressure
deficit (the absolute difference between saturation and ambient vapor pressure, in Pascals). The VIC
output from CIG includes estimates of relative humidity and vapor pressure deficit; however, these
estimates do not come directly from the GCMs but rather were calculated in a preprocessing step within
the VIC model. Both are based on empirical relationships to the average and daily range of air
temperatures and so could omit important responses to changes in forcing conditions (Guillaume Mauger,
Climate Impacts Group, personal communication, March 28, 2014). The relative humidity results project
only a small change, with a potential decrease in summer relative humidity under the high-impact
scenario (Figure 3-8).
Relative Humidity (%)
Scenario: cgcm3.1_t47_A1B (low impact scenario)
Historical
Critical Period
2020s
2040s
2080s
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Relative Humidity (%)
Scenario: ccsm3_A1B (medium impact scenario)
Historical
Critical Period
2020s
2040s
2080s
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Relative Humidity (%)
Scenario: hadgem1_A1B (high impact scenario)
Historical
Critical Period
2020s
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Note: Results shown are medians across all the CIG grid cells intersected by the SFNR watershed.
Figure 3-8. Climate Model Projections of Monthly Average Relative Humidity in the SFNR
Watershed for Low-, Medium-, and High-Impact Scenarios
While changes in relative humidity are small, calculated estimates of dew point temperature show a large
increase as a result of increases in air temperature. This is significant for thermal modeling because the
dew point temperature tends to control the daily minimum air temperature and thus affects the amount of
nighttime cooling of water temperature that can occur.
VIC model results for all three of the GCMs project increases in vapor pressure deficit by the 2080s, with
larger increases at lower elevations (Figure 3-9). This might in turn drive greater evaporation and
increased evaporative cooling, providing some buffering against the projected air temperature increases.
3-11
-------
SFNR Quantitative Assessment - Final
October 2016
Historical
Critical Period
-~-HADGEM1
-B-CCSM3
-A-CGCM3.1
Historical
Critical Period
-~-HADGEM1
-B-CCSM3
-A-CGCM3.1
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Figure 3-9. Climate Model Projections of 2080s Average Vapor Pressure Deficit in the SFNR
Watershed at the Headwaters (left) and Outlet (right)
3.2.4 Solar Radiation
Figure 3-10 summarizes projected incident solar radiation from the VIC model output. As with humidity,
the VIC pre-processor calculates solar radiation based on an empirical relationship to temperature. Little
change is forecast, although small decreases in summer solar radiation are predicted as a result of
increased cloud cover.
Solar Radiation (W/m2)
Scenario: ccsm3_A1B (medium impact scenario)
Solar Radiation (W/m2)
Scenario: cgcm3.1 _t47_A1B (low impact scenario)
Historical
Historical
Critical Period
Critical Period
Solar Radiation (W/m2)
Scenario: hadgem1_A1B (high impact scenario)
Historical
Critical Period
Note'. Results shown are medians across all the CIG grid cells intersected by the SFNR watershed.
Figure 3-10. Climate Model Projections of Monthly Solar Radiation in the SFNR Watershed for
Low-, Medium-, and High-Impact Scenarios
3-12
-------
SFNR Quantitative Assessment - Final
October 2016
3.3 VIC Hydrologic Model
CIG estimated hydrologic response to climate scenarios using the VIC model (Gao et al., 2009; Liang et
al., 1994; Liang et al., 1996). The VIC model is a macro-scale hydrologic model that simulates watershed
hydrology using estimates of vegetation, soil properties, topography, and daily weather variations. The
VIC application for the state of Washington (Eisner et al., 2010) uses the Penman-Monteith energy
balance approach to estimate potential evapotranspiration and then estimates surface and subsurface
runoff rates at the daily time scale for model grid cells (l/16th degree spatial scale or about 26.7 km2).
Results can be routed and aggregated to a desired streamflow location or spatial scale.
For the Washington State application, VIC was calibrated and validated to runoff in the Columbia and
Yakima rivers on streamflow at specific gage locations, and gridded temperature and precipitation data
adjusted for orographic effects using the PRISM climatology (Daly et al., 2002). The model is not
calibrated to individual small watersheds such as the SFNR.
The VIC output matches observed summer flows well in the SFNR, but does not seem to provide a
particularly close fit to observed winter flows as seen by comparison to the Wickersham gage in Figure
3-11. Notably, VIC appears to not fully represent the role of snow in supporting May-June runoff. (VIC
does predict a transient regime for higher elevations in the SFNR under historical conditions, but
apparently weights the total watershed runoff too much toward rain-dominated conditions, as can be seen
below in Figure 3-14 and Figure 3-15.) Similar results are found for comparison to different 30-year
periods within the gage record. Comparisons of daily time series (e.g., Figure 3-12) show that VIC does a
reasonable job of tracking daily variations in runoff and reproducing critical summer flows in the SFNR,
but tends to over predict large winter runoff events. Wu et al. (2012) report similar results for validation
tests of VIC at 12 PNW gages and obtained median daily and monthly Nash-Sutcliffe coefficients (Nash
and Sutcliffe, 1970) of 0.56 and 0.70, respectively, and median daily correlation coefficient of 0.90 across
the 12 sites. For these reasons it is best to use the VIC results to compute seasonal relative change factors
for future climate rather than applying the model output directly in the temperature model.
3-13
-------
SFNR Quantitative Assessment - Final
October 2016
Runoff (mm/day)
Cells area-weighted to Wickersham Gage
18
Critical Period
16
USGS 12209000
14
VIC Model
12
10
8
6
4
2
0
Jan
Feb
Mar
Apr May
Jun
Jul
Aug Sep
Oct
Nov
Dec
Figure 3-11. Comparison of Monthly Average VIC Model Flows to USGS Gaged Flows for the
South Fork Nooksack River at Wickersham (12209000) for 1934-2006
18,000
16,000
14,000
12,000
10,000
8,000
6,000
4,000
2,000
0 F1 i — 1 1 l' ™ T 1 I 1 P* ^ 1 |
Oct-03 Dec-03 Feb-04 Apr-04 Jun-04 Aug-04 Oct-04 Dec-04 Feb-05 Apr-05 Jun-05 Aug-05 Oct-05 Dec-05 Feb-06 Apr-06 Jun-06 Aug-06
Figure 3-12. Time Series Comparison of Daily Average VIC Model Flows to USGS Gaged Flows for
the South Fork Nooksack River at Wickersham (12209000) for 2003-2006
Over the entire state of Washington, the VIC applications suggest that future climate will result in a
significant decrease in April 1 snow water equivalent storage, with accompanying shifts from snow-
dominant to transient and from transient to rain-dominant regimes (Eisner et al., 2010; Mantua et al.,
2010). Annual runoff across the state is projected to increase by 0-2 percent by the 2020s, 2-3 percent by
the 2040s, and 4-6 percent in the 2080s, primarily due to projected increases in winter precipitation. In
contrast, warm season (April-September) flow is projected to decrease by 33-43 percent by the 2080s.
3-14
-------
SFNR Quantitative Assessment - Final
October 2016
3.3.1 Water Yield
As was shown above, the VIC model does a reasonable job of estimating summer low flows that are
critical for temperature impacts in the SFNR, but does not provide a particularly accurate fit to historical
winter-spring hydrology in the watershed. VIC's ability to predict responses of SFNR hydrology to
significant changes in future climate is untested. More generally, the VIC model code has been widely
applied and the theoretical construct of the model shown to provide a satisfactory representation of the
water balance. We therefore assume that VIC simulations under projected future climate conditions
provide at least a reasonable representation of the relative change that can be expected in watershed
response. Thus, change factors, based on comparison of VIC output for a future scenario and historical
conditions, applied to the observed historical flow series provide a reasonable estimate of the flow regime
that can be expected under future climate conditions.
In general, VIC simulations of future watershed responses to all three GCMs project an increase in winter
flows coupled with a decline in summer flows (Figure 3-13). The pattern varies considerably by
elevation. At the headwaters, VIC correctly predicts the historical transient regime with the important role
of snow in delaying runoff. This regime is predicted to change drastically, with a shift to a rain-dominated
regime consistent with regional results (e.g., Mantua et al., 2010; Wu et al., 2012). On the other hand,
predicted changes in hydrology at low elevations are generally small (Figure 3-14).
Scenario: ccsm3_A1B (medium impact scenario)
Scenario: cgcm3.1J47J\1B (low impact scenario)
Historical
Historical
Critical Period
Critical Period
Scenario: hadgenTLAlB (high impact scenario)
Historical
Critical Period
Figure 3-13. Monthly Average Water Yield in the SFNR Watershed Predicted by the VIC Model for
Low-, Medium-, and High-Impact Climate Scenarios
3-15
-------
SFNR Quantitative Assessment - Final
October 2016
25
Historical
Historical
Critical Period
Critical Period
20
HADGEM1
¦HADGEM1
~u
E
E, 15
2
0)
>¦
•CCSM3
¦CCSM3
•CGCM3.1
'CGCM3.1
0)
§
5
0
2
3
4
5
6
7
8
9 10 11 12
Figure 3-14. 2080s Average Water Yield in the SFNR Watershed Predicted by the VIC Model for
Headwaters (left) and Outlet (right) Grid Cells
3.3.2 Baseflow
Thermal processes in streams are strongly dependent on the ratio of the surface area exposed to the
atmosphere to the volume and thermal mass of the stream (Wu et al., 2012). As a result, stream
temperature is often observed to co-vary with streamflow (Gu and Li, 2002). Summer baseflow is
therefore important in determining the water temperature extremes in the SFNR, with lower baseflow
resulting in increased heat gain during summer critical conditions.
The baseflow component of runoff emphasizes strong elevation differences within the SFNR. Under
historical conditions, baseflow at high elevations represents a transient regime, with winter snow
supporting spring and summer baseflow, with a peak in April and May, while the lower elevations have
rain-dominated regime with peak baseflow in February. By the 2080s the model predicts little change in
summer baseflow at low elevations; however, the baseflow pattern at high elevations shifts from a
transient to a rain-dominated regime (Figure 3-15), showing this transition more clearly than the total
flow plots presented above. This results in increased winter baseflow, but strong decreases in baseflow
generated in the upper part of the watershed during the spring and summer.
10
Historical
Historical
4.5
9
Critical Period
Critical Period
8
'HADGEM1
'HADGEM1
e 7
I 6
5 5
o
4
-CCSM3
-CCSM3
-CGCM3.
•CGCM3.
4-
O)
(/>
ra
DQ
3
2
0.5
0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Figure 3-15. 2080s Average Baseflow Discharge Rate in the SFNR Watershed Predicted by the VIC
Model for Headwaters (left) and Outlet (right) Grid Cells
3-16
-------
SFNR Quantitative Assessment - Final
October 2016
4 Future Climate Boundary Conditions
The draft TMDL analysis was developed using a steady-state QUAL2Kw model (Ecology, 2003a)
application to critical conditions within the SFNR. For the purposes of this assessment, and to assist with
current and future salmon recovery planning, each of the critical conditions inputs is reevaluated under
estimated future climate conditions. This section describes how each of the boundary conditions in the
draft TMDL model was modified for the future climate scenarios.
4.1 Mainstem and Tributary Flow Boundary Conditions
The QUAL2Kw model represents the mainstem SFNR up to the confluence with Wanlick Creek and
consists of 58 segments (see Figure 2-5 above). The model requires flow boundary conditions for the
headwater reach and for each of the 37 tributary inflows, as well as for the direct ground water inputs to
the mainstem (see Table 4-1).
The critical conditions run for the draft TMDL is based on 7Q10 flow, representing a critical low-flow
condition combined with air temperatures of a similar recurrence (the 90th percentile 7-day annual
maximum). Some model simulations were also conducted using the 7Q2 flow combined with the median
summer maximum temperature to represent the temperature stress on salmon populations during an
average or "typical" year. Flow boundary conditions under future climates are based on an estimate of the
effect of climate on flow during low-flow periods. To do this, the climate change application incorporates
predicted changes in summer baseflow from assessments conducted by CIG with the VIC hydrologic
model. VIC simulations suggest that total annual streamflow volume could increase slightly throughout
much of the PNW in the 2040s and 2080s (Eisner et al., 2010). However, summer flows are projected to
decrease across most of the region, with a mean decrease of more than 30 percent by the 2080s (Wu et al.
2012).
While the VIC model predictions of the water balance differ from historic gage records for the SFNR, the
fit is generally good for the critical late summer period. Therefore, mapping and extrapolation of CIG
estimates to the QUAL2Kw domain is necessary. Specifically, the CIG output is applied using a change
method in which the TMDL 7Q10 flow at the model headwaters and for all tributary and diffuse inflows
is modified by the ratio of CIG estimates of low flows under historic and future climate conditions.
4-1
-------
SFNR Quantitative Assessment - Final
October 2016
Table 4-1. SFNR Tributary Catchments Included in the QUAL2Kw Model
Index
Name
Drainage Area
(km2)
Mean Elevation
(m)
Confluence
Distance from
Headwaters (km)
Forested Area
(%)
1
Wanlick
25.183
645.310
0
81.93%
2
SFN Headwaters
53.933
815.590
0
69.80%
3
Unnamed Trib 22
5.019
598.566
2.1
68.66%
4
Unnamed Trib 21
3.261
629.726
2.7
50.36%
5
Unnamed Trib 20
1.691
1036.890
2.8
74.27%
6
Unnamed Trib 19
5.474
552.084
4.4
76.36%
7
Unnamed Trib 18
2.353
691.332
6.4
89.78%
8
Unnamed Trib 17
1.937
624.672
8.1
80.78%
9
Unnamed Trib 16
1.290
596.201
10.5
57.97%
10
McGinnes
2.817
460.355
11.1
63.36%
11
Howard
20.013
429.725
11.7
67.57%
12
Unnamed Trib 15
1.359
503.633
13.9
91.63%
13
Unnamed Trib 14
1.980
512.113
16.9
84.10%
14
Canyon
3.671
452.229
19
80.44%
15
Unnamed Trib 12
2.028
418.746
21.2
61.99%
16
Unnamed Trib 11
1.472
464.442
22.9
52.14%
17
Unnamed Trib 10
2.133
443.301
24.6
77.52%
18
Plumbago
17.835
262.050
25.1
62.19%
19
Fobes
5.781
219.222
27.1
65.45%
20
Unnamed Trib 8
5.905
232.520
29
64.05%
21
Cavanaugh
24.147
389.141
30.8
76.64%
22
Unnamed Trib 7
2.071
280.670
31.2
76.91%
23
Edfro
7.622
278.374
32.9
73.48%
24
Skookum
57.677
259.506
34.5
66.91%
25
Unnamed Trib 6
4.163
235.014
38.1
73.76%
26
Pond
3.731
155.147
39.6
79.38%
27
Unnamed Trib 5
1.109
101.906
40.9
19.81%
28
Hutchinson
46.037
175.113
41.1
66.76%
29
Unnamed Trib 4
8.998
148.683
43.5
47.09%
30
Jones
7.130
225.005
44.5
57.45%
31
McCarty
6.232
164.868
46.8
57.37%
32
Standard
3.802
230.584
48.3
78.76%
33
Hard Scrabble Falls
2.686
240.694
50.4
71.93%
34
Sygitowicz
6.666
171.917
51.6
77.48%
35
Unnamed Trib 2
5.933
175.004
52.7
56.94%
36
Black Slough
26.003
124.464
54.2
51.06%
37
Unnamed Trib 1
4.777
137.119
55.3
80.70%
Note'. Tributaries are un-nested and each discharges directly to the SFNR mainstem.
4-2
-------
SFNR Quantitative Assessment - Final
October 2016
The VIC model historical conditions run is based on meteorology from January 1, 1915, to December 31,
2006, with current land use (from the 2006 digitization). The VIC model output includes the following
climate variables:
• Precipitation
• Average, minimum, and maximum air temperature
• Solar radiation
• Relative humidity
• Vapor pressure deficit
• Evapotranspiration
• Potential evapotranspiration
• Runoff (surface flow depth per unit area)
• Baseflow (subsurface flow depth per unit area)
• Soil moisture
• Snow depth
• Snow water equivalent
The VIC output (for each 92-year time series associated with the distribution statistics of a 30-year time
horizon, as shown in Table 3-1) was converted to unit-area 7-day flows, and the 7-day flows analyzed to
determine the empirical 7-day flow with 2- and 10-year recurrence intervals. The same procedure was
followed for each of the future climate scenarios and the ratio between future climate and historic
conditions for 7Q10 flows was calculated for each VIC grid cell (and area-weighted if necessary) to
estimate a ratio for each tributary watershed. These ratios were applied to the boundary inflow to yield the
climate-modified inflow estimate for the critical condition. The flows are thus adjusted using a
multiplicative change factor approach in which a multiplicative factor of 1 results in no change and values
less than 1 result in future 7Q10 estimates that are less than historic estimates. The 7Q10 flow change
factors are summarized graphically in Figure 4-1. The least change is associated with the low-impact
GCM, which predicts relatively wetter summer conditions.
The resulting ratios for tributary flow vary strongly by elevation, with greater reductions at higher
elevations (which have the highest water yield under historic conditions) and larger fractional reductions
for the 7Q2 flow than for the 7Q10 flow. For example, in the 2080s under the high (HADGEM1)
scenario, the predicted 7Q2 at the eastern ridgeline of the watershed is only 26 percent of the existing 7Q2
flow, while the baseflow of tributaries near the mouth of the SFNR remains at 90 percent of their existing
7Q2 flow. This reflects a shift away from snow-dominated runoff at high elevations and is broadly
consistent with the VIC model output for the mainstem Nooksack River reported by Mantua et al. (2010),
who found that 7Q2 flow there was predicted to be between 85 and 95 percent of existing 7Q2 flows
through the 2040s and between 75 and 85 percent of existing flows by the 2080s. For the 7Q10 flows, the
corresponding ratios are 53 and 94 percent.
Flows for direct ground water inputs are analyzed in the same manner as the tributary flows, except that
the multiplicative change factors are based on VIC output for subsurface flow only. For the portion of the
lower Nooksack that is a losing stream under low-flow conditions the rates of water loss (which sum to
about 6 percent of the total flow) are left unchanged.
4-3
-------
2020 Medium GCM
2040 Medium GCM
2080 Medium GCM
7Q10 Flow Deltas
0 30 - 0 47
0.47 - 0.66
I I 0.66 - 0.83
0.83-0.90
I 0.90-1.00
South Fork Nooksack River:
7Q10 Climate Change Scenario Flow Deltas:
All GCM Scenarios
NAD_1983_HARN_StatePlane_Washington_South_FlPS_4602_Feet
Map produced 04-16-2013 - H. Nicholas
Kilometers
TETRA TECH
C/)
1
o
c
3
§
C/3
CO
a
CO
3
CD
3
I
~n
3'
a-
O
§
Cr
Figure 4-1. 7Q10 Flow Multiplicative Change Factors (Unitless) for Tributary Flow from VIC Model for Low-, Medium-, arid High-Impact 3?
Scenarios (Columns) and 2020s, 2040s, and 2080s Time Horizons (Rows) ^
O)
-------
SFNR Quantitative Assessment - Final
October 2016
Figure 4-1 shows the spatial distribution of changes in tributary inflows. Total flow within the SFNR is
calculated by the QUAL2Kw model as the integrated result of all headwater and tributary inflows. The
2080s predictions for 7Q10 low-flow conditions within the SFNR are compared to the historic 7Q10
flows by river mile in Figure 4-2. The absolute decrease in 7Q10 flows is greatest at the mouth (River
Mile 0); however, the relative decrease is greatest at the headwaters (31 percent) and steadily shrinks to
an approximately 20 percent decline near the mouth, reflecting the smaller change in tributary inflows at
lower elevations.
120
Historical
100
— • 2080s High Impact
2080s Medium Impact
2080s Low Impact
80
60
40
20
0
35
30
25
20 15
River Mile (from confluence)
Figure 4-2. Simulated 7Q10 Flows by River Mile under Historic and 2080s Climate in the SFNR
4.2 Mainstem and Tributary Water Temperature Boundary
Conditions
The QUAL2Kw model application provides a process-based simulation of the energy balance and
temperature conditions in the mainstem SFNR; however, it requires specification of water temperatures
for all influent boundary conditions. The tributary inflows influence resulting water temperatures in the
SFNR, although active heat exchange across the air-water interface damps out the influence on
downstream water (Mohseni and Stefan, 1999). The modeling scenarios assume tributary temperatures
associated with the summer low-flow calibration run, which had tributary temperatures that might be
slightly lower than 7Q10 flow conditions. Sensitivity analyses to temperature forcing in the tributaries is
presented below in Section 5.3.
QUAL2Kw inputs headwater temperatures as a diurnal (24-hour) time series, developed from monitored
data, while tributary temperatures are characterized by a daily mean and range. For future climate
conditions the diurnal pattern or shape of the curve between the daily maximum and daily minimum
temperature is assumed to remain unchanged; however, the daily maximum and daily minimum
temperature could change. These changes are represented by an additive delta approach. That is, an
4-5
-------
SFNR Quantitative Assessment - Final
October 2016
absolute estimated change (as °C) is applied to both the maximum and minimum tributary water
temperature.
The CIG climate analysis provides daily minimum and maximum air temperature; however, the VIC
model does not predict stream water temperature. Therefore, an approach was developed in which the
water temperature deltas were assessed using a regression approach that uses the VIC model output as
explanatory variables. The Research Plan anticipated that the best approach would be to develop
predictive multiple linear regressions based on VIC model output for daily minimum and maximum air
temperature and daily surface and subsurface flow. However, subsequent testing showed that this
approach did not yield acceptable results. Therefore, additional nonlinear approaches were developed and
tested based on peer-reviewed literature for the prediction of stream temperature from landscape and
climate variables (e.g., Flint and Flint, 2011; Isaak et al., 2010; Mohseni et al., 1998). The tributary
temperature prediction method (the combined regression method described in Section 4.2.2) is able to
predict observed stream temperature from climate variables, but does not use the VIC flow output as a
predictor of water temperature.
4.2.1 Fitting Data
The major obstacle to the regression approach proposed in the Research Plan is that good spatial coverage
of temperature monitoring data from the SFNR tributaries is not contemporaneous with the VIC historical
run, which covers 1915-2006. In contrast, the tributary stream temperature monitoring data used for the
TMDL model was collected in July and August of 2007 through 2012 (Table 4-2). Most of the earlier
temperature data in the system are from the mainstem only.
Table 4-2. SFNR Tributary Stream Temperature Monitoring Data
Tributary Name
Station Name (Years Monitored)
Wanlick Creek
SF0210 (2007), WanlicklO (2010)
Plumbago Creek
411 (2010, 2012)
Deer Creek (tributary to Plumbago Creek)
SF0135 (2007), 412 (2010, 2012)
Cavanaugh Creek
SFT016 (2007), 410 (2010, 2012)
Edfro Creek
SFT015 (2007), 409 (2010, 2012)
Skookum Creek
SF0130 (2007), 413 (2010, 2012), USGS Station 12209490
(2009-2010)
Hutchinson Creek
408 (2010, 2012), Ecology Station 01C070 (2003-2011)
Upper Tributary to Hutchinson Creek
604 (2012)
Second Upper Tributary to Hutchinson Creek
605 (2012)
McCarty Creek
SF0033(2007)
Jones Creek
608 (2012)
Upper Tributary to Jones Creek
609 (2012)
Hard Scrabble Falls Creek
610 (2012)
Sygitowicz Creek
611 (2012)
Black Slough Creek
407 (2012)
Upper Tributary to Black Slough Creek
601 (2012)
Tinling Creek (tributary to Black Slough Creek)
602 (2012)
Upper Tributary to Tinling Creek (tributary to Black
Slough Creek)
603 (2012)
Note'. Monitoring by Nooksack Indian Tribe, unless otherwise indicated.
4-6
-------
SFNR Quantitative Assessment - Final
October 2016
Given the lack of temporal overlap, it was not possible to predict daily variations in stream temperature
from contemporaneous VIC output. Instead, the focus was shifted to a more generalized representation of
critical period stream temperatures from which appropriate deltas could be calculated. Specifically, the
revised analysis focused on the 7-DADMax and the corresponding maximum of the 7-day average
minimum temperature in the monitoring series. Using the 7-day average is consistent with the temporal
basis of the temperature water quality criterion and also dampens the effects of any individual anomalous
observations. Examination of the changes in both the daily minimum and daily maximum of the critical
temperature conditions is needed to define the proper diurnal cycle.
4.2.2 Regression Analysis
A variety of approaches were evaluated for predicting the observed 7-day average critical temperatures.
Multiple linear regression has been highly successful in predicting thermal regime in mountain streams.
For example, Isaak et al. (2010) developed multiple regression models of both mean stream temperature
and maximum weekly maximum temperature (MWMT) using a large data set for the Boise River
watershed. Isaak et al. (2010) developed models both with and without spatial covariance structures based
on the degree of flow connection between different sites. The predictor variables in Isaak's final models
are elevation, solar radiation, mean flow, and either mean or maximum weekly maximum air temperature.
The models with spatial covariance structure yield R2 values of 0.857 and 0.925 for the MWMT and
mean stream temperature models, respectively; however, the nonspatial models yielded much lower R2
values of 0.543 and 0.679.
We constructed similar models for the SFNR, but using VIC-predicted climate conditions rather than
observations for the tributaries and without defining a spatial correlation structure as all the SFNR
tributary sites are on individual small tributaries to a single mainstem and are thus not flow connected
with one another. The resulting models obtained adjusted multiple R2 values ranging from 0.575 to 0.649,
which is consistent with the quality of fit obtained by Isaak et al. (2010) for the nonspatial models —
despite the fact that we were constrained to use noncontemporaneous data—but still relatively weak as a
predictive tool. We then examined regression models of the type developed by Flint and Flint (2008,
2011) to predict daily average temperature in the Klamath River basin of northern California that
incorporate a nonlinear day-of-year (DOY) function and modified the approach to predict 7-day averages
of maximum and minimum stream temperature. Stepwise regression fits to individual tributary data sets
were promising, with R2 values often around 0.8, but a good fit was not obtained to the entire data set
simultaneously.
We also fit nonlinear logistic models of the type developed by Mohseni et al. (1998) to predict weekly
stream temperatures observed in 2007-2012 using average air temperature from paired weeks in the
2000-2006 VIC model output. Mantua et al. (2009, 2010) fit models of this form to stations throughout
the PNW including the mainstem Nooksack River at North Cedarville and Nooksack River above Middle
Fork. Our results for many individual streams yielded multiple adjusted R2 values around 0.75; however,
there was significant variability in parameter values between individual sites, and fit was poor for
Wanlick Creek, near the headwaters, and Hardscrabble Creek. Further, the Mohseni models require
minimum and maximum temperature parameters and it is unclear if these parameters would be stable and
appropriate for application under future climates. This model form, therefore, was deemed suboptimal for
prediction under changed climate. However, the application did suggest the potential utility of using data
from matched calendar weeks from different years.
Based on the various approaches discussed above, a nonlinear regression approach was undertaken that
combines the Flint approach with findings of the general stepwise regression to predict the entire set of
7-day average minimum temperature and 7-day average maximum temperature (all observations) based
on the Flint variables plus three potential additional terms for elevation (ELV, meters), drainage area
(DRN, km2), and fraction of area forested (FOR) in the following nonlinear form:
4-7
-------
SFNR Quantitative Assessment - Final
October 2016
Ts = a + b(Rn) + c(VDD) + d(Tmean) + e(DA) + f(DA2) + g(ELV) + h(FOR) + i(DRN).
Here, Ts is the predicted stream water temperature in Celsius, Rn is the solar radiation (W/m2), VDD is
the vapor density deficit (Pa), Tmean is the average air temperature, ELVis elevation (m), FOR is the
fraction of area forested, and DRN is the drainage area (km2). DA is a day-of-year (DOY) function that
includes an internal fitting parameter, I:
Both forward and backward stepwise regressions converged on the same model equation in which all but
one parameter in the equation for the average maximum temperature were significant. The model fit is
shown in Figure 4-3; the coefficients and statistics are summarized in Table 4-3 and attain adjusted
multiple R2 values across all sites of 0.77 for the 7-DADMax and 0.80 for the corresponding maximum of
the 7-day average minimum temperature. In general, the model is unbiased but somewhat imprecise due
to the necessity of fitting noncontemporaneous stream water temperature and air temperature data. For the
summer period of July 15-August 26 the average errors are -0.32 °C and -0.37 °C for the minimum and
maximum, respectively, while the average absolute errors are 1.20 °C and 1.80 °C. The noise in the
model fit is likely due to site-specific conditions in individual tributaries. The five largest under
predictions of summer 7-day average maximum temperature all occur at station 410 on Cavanaugh Creek,
while the 23 largest over predictions are all at Ecology station 01C070 on Hutchinson Creek.
The equations presented in Table 4-3 are used to calculate maximum 7-day average minima and maxima
for each tributary during the model critical period under existing and future climates. These in turn are
used to calculate additive deltas with which to modify the daily maximum and daily minimum
temperatures specified in the TMDL model. (Using the delta method helps to minimize the potential
impacts of bias in predictions for individual stations in the regression model.) Finally, the hourly
temperatures are fit to the existing shape between the maximum and minimum over the course of the day.
Note that this approach preserves a difference between the future runs without TMDL implementation
and runs with SPV for the TMDL. The TMDL implementation runs, in addition to system potential shade,
assume that under current conditions the tributary and headwater temperatures are at water quality criteria
if the current 7-DADMax is greater than the applicable criterion. Thus, the system potential runs and the
runs with current shade apply the same additive deltas to tributary temperatures, but the system potential
run applies the deltas to a lower baseline in tributaries that currently exceed temperature criteria.
Table 4-4 and Figure 4-4 show examples of the predicted daily maximum tributary water temperatures for
TMDL critical (7Q10) conditions for the 2080s, along with the current conditions daily maximum
temperature used for the 7Q10 TMDL run.
4-8
-------
SFNR Quantitative Assessment - Final
October 2016
20
7-day Average Minimum
15
1 10
-Q
o
10
Predicted (°C)
15
20
25
7-day Average Maximum
20
q_ . 15
-a
CD
>
-a 10
O
10 15
Predicted (°C)
20
25
Note'. Data from Nooksack Indian Tribe stations are from June-October. The Ecology and
USGS stations have year-round data. Results from July 15-August 26 are highlighted in red.
Figure 4-3. Prediction of 7-day Average Minimum and Maximum Water Temperature
4-9
-------
SFNR Quantitative Assessment - Final
October 2016
Table 4-3. Coefficients of Water Temperature Prediction Model
Coefficient
7-day Average Minimum
Temperature (°C)
7-day Average Maximum
Temperature (°C)
a (constant)
1.509
1.249
b (Rn: Solar Radiation, W/m2)
-0.010
-0.006
c (VDD: Vapor Density Deficit, Pa)
266.622
432.452
d (Tmean: Average Air Temp, °C)
0.519
0.518
e (DA: Day of Year function)
-0.128
0.000
f (DA2)
-0.666
0.575
g (ELV: Elevation, m)
0.000
0.001
h (FOR: Forested Fraction)
3.066
4.491
i (DRN: Drainage Area, km2)
-0.010
-0.027
t (Day of year parameter, degrees)
1.000
-0.607
Summary Statistics
Count
1052
1052
Multiple R
0.896
0.879
Adjusted Multiple R2
0.801
0.772
Standard Error of Estimate
1.290
1.761
F-ratio
605.9
509.0
P value
< 0.001
<0.001
Average Error (7/15-8/26)
-0.31
-0.34
Average Absolute Error (7/15-8/26)
1.24
1.81
Table 4-4. Average of Projected Daily Maximum Tributary Temperatures (°C)
Existing Vegetation
System Potential Vegetation
Low
Medium
High
Low
Medium
High
Baseline
14.45
14.45
14.45
13.21
13.21
13.21
2020s
15.47
15.86
15.62
14.23
14.62
14.39
2040s
15.87
16.14
15.90
14.63
14.90
14.66
2080s
16.13
16.60
16.64
14.89
15.36
15.40
Note'. Average of the projected diel maximum for all tributaries is shown.
4-10
-------
SFNR Quantitative Assessment - Final
October 2016
South Fork Nooksack River:
Climate Change Scenario Tributary Max Temperatures:
Current Conditions
N
0 1.5 3
6
I
0 1.5
3 6
It
TETRA TECH
NAD 1983 HARN SlalePlana \M»shinglon Soulh FIPS 4602 F«ol
Map produced 09-13-2013 - H. Nicholas
Legend
SF Nooksack Mainstem
Tributary
Max Stream Temp (C)
11.1
12.0
12.1
13.0
13.1
13.0
13.1
14.0
14.1
15.0
15.1
16.0
16.1
17.0
17.1
18.0
18.1
19.1
Whatcom County
Skagit County
South Fork Nooksack River:
Climate Change Scenario Tributary Max Temperatures:
2080 High - Current Shade
TETRA TECH
Legend
SF Nooksack Mainstem
Tributary
Max Stream Temp (C)
11.1
12.0
12.1
13.0
13.1
13.0
13.1
14.0
14.1
15.0
15.1
16.0
16.1
17.0
17.1
18.0
18.1
19.1
Whatcom County
Skagit County
Figure 4-4. Predicted Maximum Hourly Tributary Temperatures under 7Q10 Flow Conditions for
TMDL Scenario (top) and 2080s High-Impact Climate Scenarios with Current Shade (bottom)
4-11
-------
SFNR Quantitative Assessment - Final
October 2016
4.3 Meteorological Forcing
The QUAL2Kw model requires direct input of several meteorological variables, by reach, as described
below.
4.3.1 Air Temperature
Air temperature under future climate is represented through an additive delta approach. Monthly average
VIC model output for the historical run (1915-2006) was subtracted from the VIC model output for the
climate scenario runs (2020s, 2040s, 2080s) to determine the difference in minimum and maximum air
temperatures at each grid cell for the month represented in the TMDL runs. The delta change values for
each climate scenario were added to the baseline condition to calculate a new daily maximum and daily
minimum air temperature for each model segment. Weighted interpolation between the old and new
maximum and minimum temperatures was used to calculate new air temperature based on the baseline
condition on an hourly basis for each reach. Table 4-5 shows averages of the air temperature delta values
for August critical conditions. Note that the changes in August temperature are greater in the medium-
impact scenario than in the high-impact scenario for the 2020s and 2040s, even though the average annual
change in temperature is greater in the high-impact scenario.
Table 4-5. Average Additive Deltas for August Air Temperature and Dew Point (°C)
Scenario
Low Impact
Medium Impact
High Impact
Air Temperature
Dew Point
Air Temperature
Dew Point
Air Temperature
Dew Point
2020s
1.35
0.69
2.71
1.71
2.04
0.98
2040s
2.32
2.02
3.96
2.49
3.67
1.41
2080s
3.13
2.75
4.76
4.16
6.69
3.76
Note'. Results shown are average of change in maximum and minimum August temperatures averaged across all VIC
grid cells intersecting the SFNR mainstem.
4.3.2 Dew Point Temperature
The VIC model does not directly output dew point temperature, but does provide relative humidity (RH)
and air temperature (T, °C). From these, dew point temperature (Td) is readily calculated via the Magnus-
Tetens formula (Barenbrug, 1974) as:
T _ b-g{T,RH)
J-a-a(T,RH) with
a(T,RH) = T- + \n(RH) a = 17 27, and b = 237.7 °C
b+T
We used RH rather than vapor pressure deficit to calculate Td to maintain consistency with the approach
used in the TMDL. After calculating dew point temperatures, the same additive delta method is applied as
was used for air temperature. This involves establishing a delta-modified maximum and minimum daily
dew point temperature with weighted interpolation between the minimum and maximum. Table 4-5 above
shows average delta values. In all cases, the increase in dew point is smaller than the increase in air
temperature.
4-12
-------
SFNR Quantitative Assessment - Final
October 2016
4.3.3 Cloud Cover
The TMDL application assumes zero cloud cover (clear sky) as a critical condition. No adjustments are
made to this assumption under future climates.
4.3.4 Wind
Wind speed influences conductive and convective heat exchanges at the water surface. Although wind is
liable to change under future climate conditions, downscaled analysis of this variable is not available from
CIG. Wind stress, therefore, is kept constant at TMDL conditions for all climate change scenarios. Hourly
wind speeds for the critical condition runs range from 0 to 1.73 meters per second.
4.4 Other Boundary Conditions
4.4.1 Ground Water Inflow Temperatures
The QUAL2Kw model includes direct ground water discharge to the SFNR and the temperature of this
discharge has an important effect on the heat balance in the river. Under future climate conditions, the
ground water inflow temperature is modified using an additive delta. To derive the delta it is assumed that
ground water temperatures are ultimately proportional to annual average air temperatures with the
superposition of an annual cycle that results in ground water temperatures that are a few degrees warmer
than the annual average air temperature during the late summer critical period. Therefore, the delta in the
annual average air temperature is used to modify the ground water inflow temperature.
Because the annual average air temperature deltas have a very small range between VIC grid cells, the
median delta was chosen from each climate scenario condition to be applied to all ground water inputs
(Table 4-6).
Table 4-6. Additive Deltas on Average Annual Air Temperature (°C)
Scenario
Low Impact
Medium Impact
High Impact
2020s
0.977
1.212
1.464
2040s
1.630
2.154
2.878
2080s
2.472
2.948
5.193
4.4.2 Physical Boundary Conditions
Physical boundary conditions are generally assumed to be unchanged under future climates. This
simplifying assumption introduces uncertainties into the prediction.
Riparian Shade: The QUAL2Kw core model uses estimates of existing shade on the mainstem of the
river based on observations (e.g., LiDAR, aerial imagery) and the Shade model. The TMDL application
also evaluates the natural condition of full potential shade ("system potential" conditions). Even though
there is a likelihood that riparian vegetation composition (e.g., diversity and abundance) could be altered
as a result of climate change, we do not have the tools to predict these changes and thus do not evaluate
climate-induced changes in riparian shade outside the range of conditions to be evaluated for the TMDL.
Channel Structure: The channel structure is set to existing conditions for all model setups, including
climate change scenarios. It is acknowledged that alterations in climate, such as increased winter high-
flow events, could alter channel geometry. For example, powerful high-flow events could result in a
widening of the channel with lessened shade and increased solar input. We do not, however, have the
4-13
-------
SFNR Quantitative Assessment - Final
October 2016
tools to predict these changes and thus do not evaluate potential climate-induced changes in channel
structure.
Hyporheic Exchange: In rivers with coarse (sand, gravel, cobble) bed sediments, a portion of the flow
(hyporheic flow) occurs within the bed sediment. Heat flux between the water and sediment during
hyporheic flow can help stabilize stream water temperatures and provide cooling during summer months.
Changes to flow and temperature under future climates could alter the effectiveness of hyporheic
exchange in maintaining stable water temperatures.
The user inputs to QUAL2Kw that control the effect of hyporheic exchange on water temperature are the
thermal conductivity, thermal diffusivity, the effective thickness of the sediment layer, and the bulk
hyporheic exchange flow. These parameters are determined primarily through calibration. The primary
uncertainty for addressing climate impacts through hyporheic flow in the QUAL2Kw model is that the
bulk hyporheic exchange flow is a user-specified input. The fraction of hyporheic flow adopted for the
TMDL model is not altered for the future climate model runs, resulting in specification of a linear change
in hyporheic flow as total flow in the reach changes. Therefore, no additional changes to boundary
conditions are needed to evaluate how the effects of hyporheic exchange could change under future
climates.
4-14
-------
SFNR Quantitative Assessment - Final
October 2016
5 Temperature Model Results
The QUAL2Kw model of TMDL critical conditions was reapplied with future climate forcings as
described in Section 4. We report the resulting maximum water temperature predictions for the 2020s,
2040s, and 2080s under low-, medium-, and high-impact climate scenarios. Note that because QUAL2Kw
is a steady-state model (with diel variability) the daily maximum water temperature predicted by the
model under critical conditions, as defined below, is assumed to be equivalent to the 7-DADMax
temperature defined in the water quality standards.
5.1 Critical Conditions Scenario
The SFNR temperature TMDL scenarios are based on a steady-state analysis of critical conditions, which
combine high air temperatures with low flows, leading to increased heating. Specifically, the following
assumptions were used to estimate critical conditions:
• 7Q10 flows are assumed. The 7-day average low flow that occurs, on average, once every 10
years.
• Headwater and tributary temperatures are at water quality criteria (or observed current
temperature if cooler). This is applied for scenarios with SPV (not current vegetation).
• Air temperature, dew point, and relative humidity are at critical conditions (90th percentile air
temperature defined as the 90th percentile of the series of the annual maxima of 7-day average air
temperatures for each week of record, and corresponding atmospheric moisture content),
modified to account for a microclimate effect associated with SPV. As specified by Ecology, the
microclimate effect is simulated by a drop in temperature of 2 °C at every hour of the day.
• Direct ground water discharge rates based on 1990s seepage study and a method synthesized from
Curran and Olsen (2009) as incorporated into the model calibration to conditions of August 2,
2007, and adjusted to 7Q10 flows.
• Temperature of direct ground water discharge initially based on annual average air temperature
and adjusted during model calibration.
• Cloudless sky, consistent with maximum temperature impact in later summer.
As discussed in Section 2.1, modeling scenarios were developed to estimate natural conditions by
assuming SPV. Modeling scenarios were also developed assuming current shade, recognizing that
attaining SPV can take decades. Runs that included current shade omitted the microclimate effect
associated with SPV, and also omitted the decreased headwater and tributary temperatures.
5.2 Climate Change Scenarios
Ecology and EPA requested 18 future climate simulations. These runs (identified as Scenarios 9 through
26 in the draft TMDL document) address 2020s, 2040s, and 2080s climate for the high-impact, medium-
impact, and low-impact climate scenarios, with current or system potential shade. We constructed each
scenario by perturbing the results (Scenario 5 or Scenario 3) by the climate deltas or change factors
discussed in Section 4. The future scenarios, along with the two comparable current climate TMDL runs,
are summarized in Table 5-1, along with the resulting maximum predicted stream temperatures. Results
are discussed in more detail below.
5-1
-------
SFNR Quantitative Assessment - Final
October 2016
Table 5-1. Critical Period Maximum Water Temperature Results Summary
Scenario
(Scenario numbers are as given in the
draft TMDL document.)
Maximum Stream Temperature (°C)
(averaged across reaches)
All Reaches
Headwaters to
Reach 281
Reach 281 to Outlet
Critical Condition TMDL Runs
3 (Current climate and shade, 7Q10)
21.04
20.15
21.84
5 (Current climate, system potential, 7Q10)
18.73
17.78
19.60
2020 GCM Comparisons
9 (2020 High, current shade)
23.40
22.75
23.98
12 (2020 High, system potential)
21.02
20.26
21.71
10 (2020 Medium, current shade)
23.65
22.94
24.30
13 (2020 Medium, system potential)
21.30
20.48
22.05
11 (2020 Low, current shade)
22.78
22.10
23.39
14 (2020 Low, system potential)
20.42
19.64
21.13
2040 GCM Comparisons
15 (2040 High, current shade)
24.61
24.07
25.10
18 (2040 High, system potential)
22.15
21.45
22.78
16 (2040 Medium, current shade)
24.68
24.04
25.26
19 (2040 Medium, system potential)
22.31
21.54
23.02
17 (2040 Low .current shade)
23.55
22.84
24.19
20 (2040 Low, system potential)
21.22
20.39
21.96
2080 GCM Comparisons
21 (2080 High, current shade)
26.97
26.54
27.36
24 (2080 High, system potential)
24.61
24.00
25.15
22 (2080 Medium, current shade)
25.70
25.07
26.27
25 (2080 Medium, system potential)
23.35
22.56
24.06
23 (2080 Low, current shade)
24.46
23.81
25.04
26 (2080 Low, system potential)
22.10
21.33
22.80
Notes:
1 From headwaters to 27.5 km downstream, the water quality criterion is 12 °C 7-DADMax; below this point it is 16 °C.
The increases in maximum water temperatures shown in Table 5-1 arise from the combination of
increased air temperature, increased ground water and tributary water temperatures (both of which are in
turn dependent on air temperature, among other factors), and decreased flows. For the 2080s the median
ratio of the change in maximum water temperature to change in maximum air temperature is 0.96.
Table 5-1 reports the simulated daily maximum temperature under critical conditions because it is the
surrogate for 7-DADMax temperature, which is the focus of the TMDL. Projected changes in daily
average temperature for the August critical day (Table 5-2) are smaller than changes in the daily
maximum temperature. The results are generally consistent with the findings of Wu et al. (2012), who
projected mean summer temperature changes for transient systems in the PNW of up to 3.97 °C for the
2040s and up to 5.71 °C for the 2080s.
5-2
-------
SFNR Quantitative Assessment - Final
October 2016
Table 5-2. Critical Period Daily Average Water Temperature Results Summary
Scenario
(Scenario numbers are as given in the
draft TMDL document.)
Average Water
Temperature (°C)
Change Relative to
Scenario 3 or
Scenario 5 Baseline
(°C)*
Ratio to Average
Air Temperature
Change
Critical Condition TMDL Runs
3 (Current climate and shade, 7Q10)
17.21
5 (Current climate, system potential, 7Q10)
15.35
2020 GCM Comparisons
9 (2020 High, current shade)
19.12
1.90
0.93
12 (2020 High, system potential)
17.22
1.87
0.92
10 (2020 Medium, current shade)
19.41
2.20
0.81
13 (2020 Medium, system potential)
17.54
2.19
0.81
11 (2020 Low, current shade)
18.62
1.40
1.04
14 (2020 Low, system potential)
16.73
1.38
1.02
2040 GCM Comparisons
15 (2040 High, current shade)
20.16
2.95
0.80
18 (2040 High, system potential)
18.22
2.87
0.78
16 (2040 Medium, current shade)
20.32
3.11
0.79
19 (2040 Medium, system potential)
18.44
3.09
0.78
17 (2040 Low, current shade)
19.40
2.18
0.94
20 (2040 Low, system potential)
17.53
2.18
0.94
2080 GCM Comparisons
21 (2080 High, current shade)
22.33
5.11
0.76
24 (2080 High, system potential)
20.46
5.11
0.76
22 (2080 Medium, current shade)
21.28
4.07
0.86
25 (2080 Medium, system potential)
19.41
4.06
0.85
23 (2080 Low, current shade)
20.15
2.94
1.07
26 (2080 Low, system potential)
18.28
2.93
1.07
*Note that the Baseline Scenario for all "current shade" runs is Scenario 3, and the Baseline Scenario for all "system
potential" runs is Scenario 5.
The ratios of change in daily average water temperature to change in air temperature for the 2080s (0.76-
0.94, median 0.85) are somewhat greater than the typical ratio of changes in summer average water
temperature to change in air temperature of 0.67 reported by Isaak et al. (2011). However, the researchers
also note that "Comparison of trends in annual temperature extremes... suggested more rapid warming of
the year's highest and lowest weekly temperatures." Specifically, they report an increase in 7-day average
maximum water temperature of about 0.28 °C per decade corresponding to air temperature increases of
0.36 °C (as an average across seven unregulated sites), for a ratio of 0.78. The somewhat higher ratios
predicted for the SFNR are likely due to the additional impact of large reductions in 7Q10 flows projected
for the higher elevation portions of the SFNR watershed as the climate pattern shifts from transient snow
to rain-dominated conditions. Cristea and Burges (2009) applied the QUAL2Kw model to predict future
water temperatures in the Wenatchee River watershed. Their Table 3, reporting effects of future climate
with current shade at 7Q10 flows, shows a ratio of predicted increase in daily average water temperature
to change in July-August air temperature of 0.52 to 0.62 in the mainstem Wenatchee (which is influenced
by Lake Wenatchee at the headwaters) and 0.61 to 0.83 in Nason Creek; however, their results also show
ratios of change in maximum water temperature to change in air temperature from 0.87 to 1.1 in Nason
5-3
-------
SFNR Quantitative Assessment - Final
October 2016
Creek (median 1.0). The larger ratio in Nason Creek, consistent with the results presented here for the
SFNR, appears to be due to the strong influence of reduced 7Q10 flows.
5.2.1 Climate Change Scenarios: System Potential Shade
The critical conditions scenario (Scenario 5) represents the 7Q10 critical low-flow condition and
accompanying 7-day average maximum air temperature with SPV and associated microclimate climate
effects (see Section 2.2). Climate change scenarios focus on examining the effects of future climate on the
TMDL critical conditions run both with and without SPV and the accompanying microclimate effect.
Figures compare longitudinal profiles of maximum stream temperature to the numeric water quality
criteria (the criteria applicable in August) and the comparable system potential modeling run (Scenario 5).
In addition, approximate thresholds for salmonid lethality are provided based on WAC 173-201A-
200(l)(c)(vii)(A) (the natural conditions provision for water quality standards) and an Ecology study
(Hicks, 2002) that evaluated lethal temperatures for cold water fish. The approximate lethality thresholds
from these studies are 23 °C, the within-day temperature maximum threshold, and 22 °C, a critical
temperature for the 7-DADMax. It should be noted that these are summary values and reported lethal
temperatures differ among salmon species as well as by life stage. (See Richter and Kolmes (2005) for a
review of data.) Given that the steady-state critical conditions run is intended to represent the
7-DADMax temperature, comparison to the 22 °C threshold is most relevant for evaluating risk.
The two comparison levels (22 °C and 23 °C) are estimates of temperatures at which 50 percent of
individuals in a sample are at risk of dying in laboratory studies (for the given exposure period). These are
presented for comparison purposes only and are not necessarily predictive of outcomes in natural systems.
First, the QUAL2Kw model simulates reach-average temperatures and does not account for the possible
presence of small-scale cold water refuges that can provide shelter during high-temperature events.
Second, while extensive research has been undertaken, the temperature tolerance levels of salmonids are
complex and depend on the rate of temperature change and the temperatures at which fish have been
previously acclimatized. Many of the laboratory tests on which lethal temperatures are based used
constant temperature exposures and did not replicate the diel variability of temperatures found in nature.
In addition, salmonid populations in streams with warmer waters can show adaptive capacity (or
phenotypic plasticity) that increases their ability to survive warmer temperatures (Beechie et al., 2006;
Waples et al., 2004). Nonetheless, it is clear that water temperatures in excess of 22 °C are sub-optimal
for salmonid survival and impose significant stress on salmonid populations. Increased water
temperatures, especially 7-DADMax temperatures that exceed this level, are likely to diminish the health
of fish populations, so management measures to minimize the occurrence of such conditions are essential.
The TMDL simulation (Scenario 5), even with SPV, exceeds the numeric temperature criteria throughout
the river and approaches the temperature levels identified as potentially lethal for 1-day and 7-day
exposures (22 °C and 23 °C, respectively) in the downstream reaches. Figure 5-1, Figure 5-2, and Figure
5-3 show the projected temperature profiles for system potential shade conditions in the 2020s, 2040s,
and 2080s, respectively. In the 2020s the QUAL2Kw model projects an increase in the instantaneous
maximum water temperature of approximately 2 °C (relative to the TMDL scenario with existing climate
conditions plus system potential shade) under all scenarios (note that the medium-impact scenario
produces slightly higher water temperature projections in 2020 than the high-impact scenario), while a 3-
4 degree increase is predicted for the 2040s. By the 2080s the different climate scenarios begin to diverge
further, but all show a strong projected increase in stream water temperature across the length of the river.
Figure 5-4 compares the maximum temperature profiles under critical conditions for existing climate
(without system potential shade) and 2020, 2040, and 2080 climate (with system potential shade). The
addition of system potential shade results in little change through 2020, but a gradual increase in
temperature in all reaches in later decades.
5-4
-------
SFNR Quantitative Assessment - Final
October 2016
30
25
n. 15
10
5
0
0
10
20
30
Longitudinal Distance from Upstream (km)
40
50
1-Day Maximum lethality (23 C) — — 7-DADMax Lethality (22 C)
Water Quality Standards — — Scenario 5: 7Q10 TMDL
Scenario 12: 2020 High Scenario 13: 2020 Medium
Scenario 14: 2020 Low
Figure 5-1. Maximum Stream Temperature by Reach for Scenario 5 (7Q10 flows, 90th Percentile
Meteorology, System Potential Shade, and Tributaries/Headwaters at Water Quality Criteria) and
Scenarios 12,13, and 14 (2020 High GCM, 2020 Medium GCM, 2020 Low GCM)
30
25
S 20
°r 15
.i 10
5
0
0
) 30
Longitudinal Distance from Upstream (km)
40
1-Day Maximum lethality (23 C) — — 7-DADMax Lethality (22 C)
Water Quality Standards — — Scenario 5: 7Q10 TMDL
Scenario 18: 2040 High Scenario 19: 2040 Medium
Scenario 20: 2040 Low
Figure 5-2. Maximum Stream Temperature by Reach for Scenario 5 (7Q10 flows, 90th Percentile
Meteorology, System Potential Shade, and Tributaries/Headwaters at Water Quality Criteria) and
Scenarios 18, 19, and 20 (2040 High GCM, 2040 Medium GCM, 2040 Low GCM)
5-5
-------
SFNR Quantitative Assessment - Final
October 2016
2. 20
Q- 15
0
10
20
30
40
50
Longitudinal Distance from Upstream (km)
1-Day Maximum lethality (23 C) — — 7-DADMax Lethality (22 C)
Water Quality Standards — — Scenario 5: 7Q10 TMDL
Scenario 24: 2080 High Scenario 25: 2080 Medium
Scenario 26: 2080 Low
Figure 5-3. Maximum Stream Temperature by Reach for Scenario 5 (7Q10 flows, 90th Percentile
Meteorology, System Potential Shade, and Tributaries/Headwaters at Water Quality Criteria) and
Scenarios 24, 25, and 26 (2080 High GCM, 2080 Medium GCM, 2080 Low GCM)
30
°r 15
£ 10
0
10
20
30
40
50
Longitudinal Distance from Upstream (km)
1-Day Maximum lethality (23 C) — — 7-DADMax Lethality (22 C)
Water Quality Standards — — Scenario 3:7Q10 Base
Scenario 13: 2020 Medium Scenario 19: 2040 Medium
Scenario 25:2080 Medium
Figure 5-4. Comparison of Maximum Temperature Profiles at 7Q10 Flow and 90th Percentile
Meteorology for Existing TMDL Conditions and Medium GCM Scenarios for 2020, 2040, and 2080
(with System Potential Shade)
5-6
-------
SFNR Quantitative Assessment - Final
October 2016
5.2.2 Climate Change Scenarios: Current Shade
We also estimated the impact of climate change on Scenario 3 (as described in Section 2.2), which also
uses 7Q10 flow but has a baseline condition of current shade, no microclimate effect, and existing
tributary temperatures on which each climate scenario is overlain. Longitudinal profile results are shown
in Figure 5-5, Figure 5-6, and Figure 5-7 for the 2020s, 2040s, and 2080s, respectively. Not surprisingly,
predicted water temperatures are noticeably warmer than under the SPV scenarios, and they are projected
to exceed the 1-day maximum lethality threshold of 23 °C over much of the river, even by the 2020s. This
is of practical concern because it could take considerably longer than a decade to achieve SPV. However,
it should be recalled that TMDLs are based on extreme critical conditions, and more typical conditions
will not be as adverse, as shown in Section 5.4. In addition, the QUAL2Kw model is predicting reach-
average conditions and the presence of micro-scale cooler refuges could reduce lethality.
30
25
u
bo
O)
3 20
OJ
3
ro
u
E
D
E
X
n
2
0
10
20
30
40
50
Longitudinal Distance from Upstream (km)
1-Day Maximum lethality (23 C)
Water Quality Standards
Scenario 9: 2020 High
Scenario 11: 2020 Low
7-DADMax Lethality (22 C)
— — Scenario 3: 7Q10 Base
Scenario 10: 2020 Medium
Figure 5-5. Maximum Stream Temperature by Reach for Scenario 3 (7Q10 flows, 90th Percentile
Meteorology, Current Shade) and Scenarios 9,10, and 11 (2020 High GCM, 2020 Medium GCM,
2020 Low GCM)
5-7
-------
SFNR Quantitative Assessment - Final
October 2016
30
°r 15
£ 10
0
10
20
30
40
50
Longitudinal Distance from Upstream (km)
1-Day Maximum lethality (23 C) — — 7-DADMax Lethality (22 C)
Water Quality Standards — — Scenario 3:7Q10 Base
Scenario 15: 2040 High Scenario 16: 2040 Medium
Scenario 17: 2040 Low
Figure 5-6. Maximum Stream Temperature by Reach for Scenario 3 (7Q10 flows, 90th Percentile
Meteorology, Current Shade) and Scenarios 15,16, and 17 (2040 High GCM, 2040 Medium GCM,
2040 Low GCM)
30
25
°r 15
£ 10
5
0
0
) 30
Longitudinal Distance from Upstream (km)
40
50
1-Day Maximum lethality (23 C) — — 7-DADMax Lethality (22 C)
Water Quality Standards — — Scenario 3:7Q10 Base
Scenario 21: 2080 High Scenario 22: 2080 Medium
Scenario 23: 2080 Low
Figure 5-7. Maximum Stream Temperature by Reach for Scenario 3 (7Q10 flows, 90th Percentile
Meteorology, Current Shade) and Scenarios 21, 22, and 23 (2080 High GCM, 2080 Medium GCM,
2080 Low GCM)
5-8
-------
SFNR Quantitative Assessment - Final
October 2016
5.2.3 Analysis of Results
Comparison between scenarios is facilitated by displaying average 7Q10 water temperatures throughout
the modeled reaches of the river on the same plot.
Figure 5-8 summarizes all 18 climate scenarios and the expected steady increase in water temperature
over time. It also shows that SPV reduces average water temperature by about 2 °C. This is sufficient to
mitigate water temperature increases in the 2020s, but not for later periods.
30
u
O
a;
3
+¦»
fU
a;
Q.
E
i£
E
3
£
"x
ro
25
20
15
•High-Existing Shade
Med-Existing Shade
-A- Low-Existing Shade
-•-High-SPV
Med-SPV
Low-SPV
Current
2020s
2040s
2080s
Note'. Results shown are the average of the maximum water temperatures from all model segments. High, Med, and
Low refer to the high-impact, medium-impact, and low-impact future climate scenarios. SPV = System Potential
Vegetation.
Figure 5-8. Spatially Averaged Maximum Water Temperature in the South Fork Nooksack River
Mainstem at Critical Conditions for Existing TMDL Conditions and Future Climate with Existing
and System Potential Vegetation Shade
Figure 5-9 further explores the time trends in projections with SPV, confirming that the additional shade
is able to buffer the temperature effects due to climate change through the 2020s, but that a steady
increase in water temperature is predicted to occur in future decades even with SPV. This figure also
shows that the ranking of climate scenarios as high, medium, and low impact is appropriate for the 2080s,
but that in the 2020s the impacts are slightly greater under the "medium" scenario.
5-9
-------
SFNR Quantitative Assessment - Final
October 2016
u
O
a;
3
¦M
05
a;
Q.
E
i£
E
3
£
"x
ro
ai
60
£
ns
3
2
1
-•-High-SPV
Med-SPV
Low-SPV
-1
Current
2020s
2040s
2080s
Note'. Results shown are the average of the maximum water temperatures from all model segments. High, Med, and
Low refer to the high-impact, medium-impact, and low-impact future climate scenarios. SPV = System Potential
Vegetation.
Figure 5-9. Change in Spatially Averaged Maximum Water Temperature in the South Fork
Nooksack River Mainstem at Critical Conditions for Future Climate with System Potential
Vegetation Compared to Existing TMDL Conditions and Vegetation
The spatial distribution of the results is summarized graphically in the following figures. Figure 5-10
shows the results for existing climate, with and without system potential shade. In these figures each
modeled stream segment is shaded according to the ratio of the predicted daily maximum water
temperature under critical 7Q10 flow conditions to the lethality threshold. The left side of the river
(facing downstream) is colored according to the ratio to the 1-day lethality threshold of 23 °C. The right
side is colored according to the ratio to the 7-day lethality threshold of 22 °C. A ratio of 1 is shown in
yellow. The green to blue range represents conditions below the target, while the orange to red range
represents conditions above the target temperature. Figure 5-11 shows the spatial distribution for the
2040s and 2080s climate scenarios, with system potential shade.
5-10
-------
SFNR Quantitative Assessment - Final
October 2016
Hutchinson Creeki
Skookum Creek
7-day Lethal
South Fork Nooksack River - Temperature Distribution
Current Conditions - Existing Shade
TETRA TECH
1-day Lethal
Wanlick Creek
Legend
Tributary
Ratio to Target
0.63
Hutchinson Creeks
Skookum Creek
7-day Lethal
South Fork Nooksack River - Temperature Distribution
Current Condition - System Potential
TETRA TECH
1-day Lethal
Wanlick Creek
Legend
Tributary
Ratio to Target
0.63
Figure 5-10. Spatial Distribution of Ratio to Lethal Temperatures at Critical Conditions, Existing
Climate with Existing (top) and System Potential Shade (bottom)
5-11
-------
SFNR Quantitative Assessment - Final October 2016
5-12
-------
SFNR Quantitative Assessment - Final
October 2016
Hutchinson CreeM
Skookum Creek
7-day Lethal
South Fork Nooksack River - Temperature Distribution
2040 - Low System Potential
NAD 1983_MARN_S!atePlane_V\feshinaton_South FIPS 4602_Fee<
Map produced 04-25-2014 - P. Cada
1-day Lethal
Wanlick Creek
Legend
Tributary
Ratio to Target
0.63
Hutchinson Creeks
Skookum Creek
7-day Lethal
South Fork Nooksack River - Temperature Distribution
2040 - Medium System Potential
MD_1983_HARN_SSaiePlane_V&shington_South_RPS_4602_Feet
Map produced 04-25-2014 - P. Cada
*
TETRA TECH
1-day Lethal
Legend
Tributary
Ratio to Target
0.63
Wanlick Creek
Hutchinson Creeks
Skookum Creek
7-day Lethal
South Fork Nooksack River-Temperature Distribution
2040 - High System Potential
TETRA TECH
1-day Lethal
Wanlick Creek
Legend
Tributary
Ratio to Target
0.63
Legend
Tributary
Ratio to Target
0.63
1.00
1.22
Hutchinson CreeM
Skookum Creek
1-day Lethal
V
7-day Lethal
Wanlick Creek
South Fork Nooksack River - Temperature Distribution
2080 - Low System Potential
-|^ TETRA TECH
Hutchinson Creeks
Skookum Creek
7-day Lethal
South Fork Nooksack River - Temperature Distribution
2080 - Medium System Potential
TETRA TECH
1-day Lethal
Legend
Tributary
Ratio to Target
0.63
Wanlick Creek
Hutchinson Creeks
Skookum Creek
7-day Lethal
South Fork Nooksack River - Temperature Distribution
2080 - High System Potential
TETRA TECH
Wanlick Creek
Legend
Tributary
Ratio to Target
1 0.63
-1.00
1 1.22
1-day Lethal
Note order of images: 2040 low-, medium-, and high-impact scenarios along the top; 2080 low-, medium-, and high-impact scenarios along the bottom.
Figure 5-11. Spatial Distribution of Ratio to Lethal Temperatures at Critical Conditions for all Climate Scenarios with System Potential Shade
-------
SFNR Quantitative Assessment - Final October 2016
-------
SFNR Quantitative Assessment - Final
October 2016
The results for individual segments are also summarized as percent of stream miles projected to exceed
the 1-day average maximum lethality temperature of 23 °C (Figure 5-12). In this figure, the range from
the low-impact to high-impact scenario is shown as a colored area, while the medium-impact scenario
results are shown as a line (even though the medium scenario exceeds the high scenario for the 2020s).
The figure shows that with current shade potentially lethal temperatures could occur in most reaches of
the SFNR under critical low flow/high air temperature conditions by the 2020s. SPV could significantly
delay such impacts, but the risk of lethal temperatures is projected to increase significantly in the 2080s
even with full shade.
100%
= 80%
>
15
JZ
OJ
_i
"O
I
rH
A
W)
aj
c
aj
60%
40%
£ 20%
0%
Current Vegetation
High
Med
Low
System Potential Vegetation
1
Current
2020s
2040s
2080s
Note'. The shaded bands represent the range from the climate scenarios selected as minimum and maximum impact
in terms of air temperature. The lower, red band represents results with system potential shade, while the upper,
blue band represents results with current shade. The lines with markers represent the medium-impact climate
scenario (in terms of atmospheric temperature) for both shade scenarios. Current conditions represent existing
shade only because restoration of system potential vegetation will take decades to accomplish.
Figure 5-12. Percent of Stream Miles in the South Fork Nooksack River where the Daily Maximum
Temperature Exceeds the 1-day Average Maximum Lethality Temperature of 23 °C under Existing
TMDL Conditions and Future Climate Critical Conditions
5.3 Sensitivity to Boundary Conditions
Stream water temperature predictions are the net result of the interaction of shade, flow, and various
sources of thermal energy. The 7Q10 TMDL model with SPV and medium-impact GCM (Scenario 25)
was used to conduct sensitivity analyses relative to direct forcing temperatures in the headwaters and
tributaries, in ground water discharge, and in the air, as well as sensitivity to dew point temperature and
tributary and headwater flow. The first four, temperature-based variables were each perturbed by ± 1 °C
and the results displayed in a tornado diagram (Figure 5-13). Flows were varied by ±5 percent. Responses
(as the spatially averaged maximum stream temperature) to perturbations of temperatures in tributary,
headwater, and ground water inflow temperatures are relatively small (about 0.25 degree per degree for
the first two and 0.18 degree per degree for the ground water inflow temperatures). The magnitude of
direct response to air temperature is similar. The greatest response is to dew point temperature
(0.32 degrees per degree), showing the importance of evaporative cooling in maintaining stream
temperatures. This effect could limit the ability of full SPV to control temperatures because mature
riparian canopy will also limit wind and tend to maintain higher water vapor pressure and dew point
temperature.
5-15
-------
SFNR Quantitative Assessment - Final
October 2016
As expected, increasing flow leads to decreases in the maximum stream temperature. The scale of this
metric is not directly comparable to the temperature changes, but the response is -0.025 °C per percent
increase in headwater tributary flow. As the 2080s scenarios show reductions of more than 50 percent in
7Q10 flows in the upper watershed, the reductions in flow are an important contributor to the predicted
increase in maximum temperature, consistent with the findings of Cristea and Burges (2010) for the
Wenatchee River watershed.
22.5
22.7
22.9
Spatially Averaged Maximum Stream Temperature (°C)
23.1 23.3 23.5 23.7 23.9
24.1
24.3
24.5
Dew Point Temperature (±1 °C)
Air Temperature (±1 °C)
Tributary + Headwater Temperature (±1 °C)
Groundwater Temperature (±1 °C)
Decrease
Increase
Tributary and Headwater Flow (± 5%)
Figure 5-13. Sensitivity of 7Q10 TMDL Model Maximum Water Temperature Predictions to
Variations in Boundary Temperature and Flow
5.4 7Q2 Analysis
Analysis of the critical condition 7Q10 flows and 90th percentile meteorology presented in the previous
section a high risk of exceeding lethal thresholds for salmon on a reach average (although small, cold
water refuges might persist and allow for fish survival). The TMDL analysis of critical conditions,
however, purposefully represents relatively extreme worst-case conditions that will not occur every year.
Although the future climate models predict significant impacts under these conditions, it should also be
recognized that the ability of climate:watershed model pairs to predict rare events at the extreme tails of
the annual distribution of flow and air temperature is limited relative to their ability to predict changes in
more common events (Tebaldi et al., 2006). Model confidence is generally higher when analyzing less
extreme (more frequent recurrence) conditions.
The modeling scenarios undertaken for the TMDL included an average annual condition, based on 7Q2
flows accompanied by the median of the annual series of 7-day average maximum air temperatures and
associated predicted water temperature to simulate the maximum stream temperatures that salmon are
expected to encounter during a typical year. We also updated this scenario for 2080s climate conditions,
5-16
-------
SFNR Quantitative Assessment - Final
October 2016
using the same techniques described in the previous sections. The estimated 7Q2 flow was updated with
the same multiplicative change factor applied to the 7Q10 flow1.
QUAL2Kw results for 2080s climate coupled with SPV and 7Q2 flows and meteorology are shown in
Figure 5-14. The low-impact scenario remains at or below the 7-day lethality temperature, while even the
high-impact climate change scenario remains between the 1-day and the 7-day lethality temperature over
most of the length of the mainstem SFNR. However, the high-impact scenario does predict temperatures
greater than 23 °C for the lower 7 km of the SFNR even under these less extreme, more typical flow
conditions. This could present a migration barrier because thermal blockages to migration for salmon are
reported to consistently occur in the range of 19-23 °C (Mantua et al., 2010; McCullough et al., 2001;
Richter and Kolmes, 2005). Results with current shade (Figure 5-15) are about 2 °C warmer.
Selection of the critical flow has a considerable impact on the results, with maximum projected stream
temperatures for 2080 about 2-3 degrees cooler at 7Q2 flows than at 7Q10 flows. (See Figure 5-16 for the
medium-impact scenario.) At 7Q2 flows, the 7-DADMax lethality temperature of 22 °C is projected to be
exceeded in only the most downstream reaches of the river under this scenario.
30
25
2. 20
f
10
20 30 40
Longitudinal Distance from Upstream (km)
-1-Day Maximum lethality (23 C)
Water Quality Standards
-2080 Low GCM
2080 High GCM
7-DADMax Lethality (22 C)
7Q2 System Potential Baseline
2080 Medium GCM
Figure 5-14. Maximum Stream Temperature by Reach for 7Q2 flows, and 50th Percentile Maximum
Air Temperature with System Potential Shade for 2080 High GCM, 2080 Medium GCM, and 2080
Low GCM
1 This is necessary because the VIC model output is not consistent with the observed relationship between 7Q10
flows and 7Q2 flows at the Wickersham gage. VIC predicts a much greater relative difference between 7Q2 and
7Q10 flows under historical climate than is seen in the gaged flow data. VIC also predicts that the difference
between 7Q10 and 7Q2 flows will narrow under future climate conditions, implying a greater rate of change in 7Q2
than in 7Q10. As a result, applying the VIC model change factors (calculated for 7Q2 flows) to the observed flows
can result in predicted future condition 7Q2 flows that are less than predicted 7Q10 flows. Therefore, the relative
change calculated for VIC model 7Q10 flows is used to provide a consistent basis of comparison that agrees with the
observed 7Q2:7Q10 relationship and ensures that future 7Q2 remains greater than 7Q10.
5-17
-------
SFNR Quantitative Assessment - Final
October 2016
2. 20
£ 10
20 30 40
Longitudinal Distance from Upstream (km)
•1-Day Maximum lethality (23 C)
Water Quality Standards
• 2080 Low GCM
2080 High GCM
7-DADMax Lethality (22 C)
7Q2 Current Shade Baseline
2080 Medium GCM
Figure 5-15. Maximum Stream Temperature by Reach for 7Q2 flows, and 50th Percentile Maximum
Air Temperature with Current Shade for 2080 High GCM, 2080 Medium GCM, and 2080 Low GCM
S 20
£ 10
7-DADMax Lethality (22 C)
2080 Medium GCM, 7Q2
20 30
Longitudinal Distance from Upstream (km)
Water Quality Standards —
^—2080 Medium GCM, 7Q10
7Q2 Current Shade Baseline
Figure 5-16. Comparison of Maximum Stream Temperature by Reach for 2080s Medium GCM with
System Potential Shade at 7Q2 and 7Q10 Flows
5-18
-------
SFNR Quantitative Assessment - Final
October 2016
5.5 Potential Changes in Extreme High Flows
The draft temperature TMDL focuses on critical conditions for water temperature, which occur during
summer low flow. We have seen that flows might decrease and water temperatures might increase during
these critical periods under future climate. However, climate change can also lead to other changes in the
hydrology of the SFNR. Most notably, as shown in Section 3.3, higher elevation runoff is expected to
shift from a transient regime, with a mix of rain and snow, to a rain-dominant regime, with flows
occurring earlier in the year. A possible result in this regime shift is an increase in extreme high flows,
which can cause egg scour and loss. Indeed Battin et al. (2007), using the Shiraz salmon population model
(Scheuerell et al., 2006) in conjunction with climate change and hydrologic response models, predicted
that changes in winter high flows would have a greater negative impact on salmon populations than
changes in water temperature in the Snohomish basin (a nearby watershed south of the SFNR basin, also
draining to the Puget Sound). Seiler et al. (2003) also demonstrated that flood magnitude was a significant
predictor for survival rates for Chinook salmon in Skagit River. Effects of changes in high flows should
therefore also be evaluated in salmon recovery planning for the SFNR.
Previous analyses with the VIC model for the PNW indicate that the magnitude and frequency of flooding
are likely to increase dramatically in the winter months in currently transient runoff watersheds, with
substantial increases in the magnitude of the 20-year flood in Puget Sound drainages by the 2080s
(Mantua et al., 2010).
The VIC model is not specifically calibrated for the SFNR and cannot be expected to provide exact
estimates of flood magnitude. The model appears to have some weaknesses in its representation of
historical spring high flows. However, as in the analysis of boundary conditions, it is appropriate to look
at the projected relative changes in high flows to inform the potential magnitude of future changes. To do
this we calculated average flow depth (mm/d) for the weighted average of VIC grid cells intersecting the
SFNR watershed and extracted the series of annual maxima for historic and future climate conditions.
Estimates of peak runoff for various return periods were then calculated using the Gumbel Type I extreme
value distribution, as described in Maidment (1993). (The Gumbel distribution is used rather than log
Pearson III because the latter distribution requires estimates of skew and the ability of the VIC model to
reproduce the skew of the series of annual maxima is untested.)
Results shown in Table 5-3 suggest that the magnitude of floods of various recurrence intervals could
increase by amounts ranging from 4 to 39 percent. Interestingly, the low-impact (for temperature)
scenario, CGCM3.1, produces some of the largest projected increases in flood magnitude, with a 39
percent increase in the 2-year event and a 35-year projected increase in the 25-year event by the 2080s. In
contrast, the medium-impact scenario (CCSM2) produces increases in flood magnitude of 10 to 17
percent, with no acceleration overtime. HADGEM1 (the high-impact scenario for temperature) predicts
only small changes in flood magnitude through the 2040s, but increases of 26 to 31 percent by the 2080s.
5-19
-------
SFNR Quantitative Assessment - Final
October 2016
Table 5-3. Projected Percent Change in Magnitude of Floods of Various
Recurrence Intervals in the SFNR under Different Climate Change Scenarios
GCM
Recurrence
2020s
2040s
2080s
CGCM3.1
2-year
+23 %
+39 %
+39 %
10-year
+ 19 %
+32 %
+35 %
25-year
+ 19 %
+30 %
+34 %
CCSM2
2-year
+ 17 %
+ 16 %
+ 15 %
10-year
+ 12 %
+ 12 %
+ 12 %
25-year
+ 10 %
+ 12 %
+ 11 %
HADGEM1
2-year
+5 %
+7 %
+31 %
10-year
+5 %
+5 %
+27 %
25-year
+5 %
+4 %
+26 %
As an example, the empirical 10-year recurrence daily flow peak at the Wickersham gage, based on
1935-2008 monitoring, is 11,500 cfs. The range of projections for the 10-year flow in the 2080s is from
about 12,900 to 15,500 cfs.
The VIC modeling suggests there is a risk of increased magnitude of flood flows, which in turn could
affect salmon populations. The work done on channels by flood events is controlled, in part, by the
boundary shear stress, which is itself a linear function of slope and flow depth. Hydraulic equations
suggest that flow depth in the SFNR changes approximately in relation to the ratio of flows raised to the
0.45 power. Therefore, boundary shear stress may be predicted to increase by up to 1.39"45 = 1.16 times
relative to historic conditions for the 2-year flow under CGCM3.1 projections for the 2080s.
5-20
-------
SFNR Quantitative Assessment - Final
October 2016
6 Discussion
This paper presents a quantitative assessment of the potential range of impacts of changing climate on the
water temperature in the SFNR. Climate models and hydrologic models based on projections of future
climate are subject to wide ranges of uncertainty, with different models and scenarios pointing to a range
of possible futures. We have confronted these uncertainties by evaluating a range of GCMs (from low to
high anticipated impact on air temperature) and by estimating predicted hydrologic responses via a delta
change factor method applied to observed and calibrated time series (rather than using model output
directly). The resulting analyses should not be treated as forecasts of the future; rather, they are designed
to assist managers in defining the potential range of climates to which adaptation might be needed.
The Quantitative Assessment can be thought of as an embedded ecological risk assessment that evaluates
climate change vulnerability associated with the SFNR temperature TMDL (Figure 6-1). It is intended to
help inform the development of implementation plans for the TMDL that take into account needs for
climate change adaptation. Discussions of the implications of the climate change ecological risk
assessment to the TMDL process and to restoration and management strategies for the SFNR are provided
in the following sections.
Climate Change Ecological Risk Assessment
Climate Change Adaptation
(Risk Management)
Communication: Risk Assessor
and Risk Manager (Results)
Communication:
Risk Assessor and
Risk Manager
(Planning)
-------
SFNR Quantitative Assessment - Final
October 2016
6.1 Implications for the TMDL
The climate change analysis was designed to provide risk managers with an understanding of potential
climate change impacts (magnitude and timing) on stream temperature and streamflow. TMDLs have
typically been constructed with the assumption of a stationary climate under which historical data on flow
and temperature can be assumed to be an adequate guide to future conditions. TMDLs based on
assumptions of critical conditions derived from historic climate might not accurately represent risks under
potential future climate regimes. TMDLs have by and large not explicitly considered climate change to
date. In addition, data for estimating future impacts of climate change have not been readily available to
state agencies that develop TMDLs.
The TMDL regulations in 40 CFR require that the TMDL be based on an analysis of loading capacity that
is sufficient to attain water quality standards and beneficial uses.
A TMDL is a regulatory tool for addressing specific identified water quality impairments, but TMDL
allocations do not necessarily constitute a complete strategy for preserving and enhancing valued
resources. The TMDL program's focus on attaining standards under critical conditions may not address
the need for practical management strategies that enhance and protect the resource under less extreme, but
more frequently encountered conditions. In general, the TMDL should be viewed not as an end to itself
but as one tool within a more holistic resource management strategy (NRC, 2001; Serveiss et al., 2005).
Evaluation of climate change vulnerability could help inform the implementation of the SFNR TMDL.
Climate change is time dependent. Managers need first to understand the extent to which designated uses
are likely to be attainable at different time horizons and what adaptation strategies should be
implemented, on what schedule, to protect those uses. The pace (timing/rate) and priorities of restoration
actions for TMDL implementation to ameliorate potential impacts of climate change is a key component
of an iterative risk management strategy, as recommended by the National Climate Assessment (Figure
6-2). A key finding of the quantitative analysis for the SFNR is that system potential shade can likely
provide substantial resiliency into the future that will help protect beneficial uses, especially if combined
with other actions that provide cold water refuges during high-temperature events. To approach the level
of protection provided by 100-year SPV (at least 60- to 70-year-old trees) by the 2080s will require
planting and protection of all riparian areas along the mainstem SFNR by 2020.
Iterative Risk Assessment
6. Monitor and | (
reevaluate implemented
adaptation options
5. Implement
adaptation options
2. Assess the
vulnerabilities and
risk to the system
1. Identify current and
future climate changes
relevant to the system
¦/ 3. Develop an adaptation
strategy using risk-based
prioritization schemes
4. Identify opportunities
for co-benefits and
synergies across sectors
Adaptive Management
Framework
OtltrmkM Mixthn
Figure 6-2. Climate Change Adaptation and Iterative Risk Management (Yohe, 2011)
6-2
-------
SFNR Quantitative Assessment - Final
October 2016
6.2 Implications for Restoration and Management Strategies
The modeling analyses of water temperature maxima associated with projected future climate show that
restoring system potential shade will have a strong beneficial impact on the summer temperature regime
in the SFNR, while controlling the temperature of tributary inflow also has a small, but important impact.
Achieving system potential shade will take time, and consideration must be given to the resilience of
restored vegetation under future climate conditions. Although climate change is expected to have a
negative impact on forest growth due to increased summer drought, the key risk to planted riparian forests
is tree mortality caused by summer drought during early stand establishment (0-10 years after
planting). Species selection, seed source, planting density and early stand establishment culturing
methods can be evaluated to provide riparian forest vegetation that will be more resilient to increased
summer drought effects due to climate change. In addition, potential upland forest management strategies
that could help maintain summer flows by delaying the melt of winter snowpack could be evaluated as a
climate change adaptation strategy.
The modeling analyses do suggest that climate change will have a significant effect on temperature in the
river, and could substantially reduce preferred salmon habitat. However, it must be remembered that the
TMDL modeling analysis is purposefully based on an analysis of reasonable worst-case conditions (7Q10
flow combined with 90th percentile annual air temperature maximum) that may occur at a sufficiently low
frequency so as to allow recovery or adaptation of the population. Analyses of more typical 7Q2
conditions still suggest significant stress on the salmon population; however, the projected level of risk
under 7Q2 flows are not nearly as high as the 7Q10 QUAL2Kw-modeled predictions. Notably, many
rivers within the current salmon range, including the Snake and Willamette River basins, have monitored
temperatures above published lethal or protective thresholds, yet salmon currently occupy the majority of
these rivers (Beechie et al., 2012).
As noted above, the TMDL can be viewed as one component in a holistic strategy for managing and
protecting the resource. It can serve and motivate broader scale salmon management strategies that are
designed to optimize the resource. Several implications for restoration and management strategies are
suggested by the results presented in this report:
Despite the benefits of increased shade, future climate scenarios indicate water temperature regimes that
increasingly deviate from preferred habitat for salmon. The model, however, predicts reach-average
temperature on an approximately 1-kilometer scale. The impact of occasional high-temperature events is in
large part determined by whether the fish can find sufficient cold water refuges that are cooler than the reach
average and within their physiological tolerance ranges. Thus, habitat management at a scale smaller than
the spatial scale of the model will have an important role in protecting the resource and the implementation
plan should combine system potential shade with other options that provide localized cooler habitat. In
addition, watershed management that increased stream stability (and thus resulted in a narrowing of the
treeless riparian zone) would further increase effective shade on the river and mitigate warming.
Beechie et al. (2012) addressed the question of protecting salmon habitat in the face of anticipated climate
change for the 2080s, considering conditions similar to those projected for the SFNR—a decrease in
summer low flows, an increase in maximum monthly flows, and stream temperature increases of between
2 °C and 6 °C. They concluded that restoring floodplain connectivity, restoring streamflow regimes, and
re-aggrading incised channels are most likely to ameliorate streamflow and temperature changes and
increase habitat diversity and population resilience.
In sum, restoration of SPV on both the mainstem and tributaries of the SFNR has the potential to mitigate
some of the impacts of climate change on water temperature, but only through about 2020. This finding
highlights the importance of combining the implementation of system potential shading with other measures
that provide cold water refuges during high-temperature events to protect the resource beyond 2020.
6-3
-------
SFNR Quantitative Assessment - Final October 2016
6-4
-------
SFNR Quantitative Assessment - Final
October 2016
7 References
Barenbrug, A.W.T. 1974. Psychrometry andPsychrometric Charts, 3rd ed. Cape and Transvaal Printers,
Cape Town, SA.
Bartholow, J.M. 2000. Estimating cumulative effects of clearcutting on stream temperatures. Rivers, 7(4),
284-297.
Battin, J., M.W. Wiley, M.H. Ruckelshaus, R.N. Palmer, K.K. Bartz, H. Imaki, and E. Korb. 2007.
Projected impacts of climate change on salmon habitat restoration. Proceedings of the National
Academy of Sciences 104:6720-6725.
Beechie, T., E. Buhle, M. Ruckelshaus, A. Fullerton, and L. Holsinger. 2006. Hydrologic regime and the
conservation of salmon life history diversity. Biological Conservation, 130:560-572.
Beechie, T., H. Imaki, J. Greene, A. Wade, H. Wu, G. Pess, P. Roni, J. Kimball, J. Stanford, P. Kiffney,
and N. Mantua. 2012. Restoring salmon habitat for a changing climate. River Research and
Applications. doi: 10.1002/rra.2590.
Brosofske, K.D., J. Chen, R.J. Naiman, and J.F. Franklin. 1997. Harvesting effects on microclimate
gradients from small streams to uplands in western Washington. Ecological Applications,
7(4): 1188-1200.
Brown, L.C., and T.O. Barnwell. 1987. The Enhanced Stream Water Quality Models QUAL2E and
QUAL2E-UNCAS. EPA/600/3-87-007. U.S. Environmental Protection Agency, Athens, GA.
Chapra, S.C., and G.J. Pelletier. 2003. QUAL2K: A Modeling Framework for Simulating River and
Stream Water Quality: Documentation (Version 2.11). Civil and Environmental Engineering
Department, Tufts University, Medford, MA.
Chapra, S., G. Pelletier, and H. Tao. 2008. QUAL2K: A Modeling Framework for Simulating River and
Stream Water Quality: Documentation and User's Manual. Civil and Environmental Engineering
Department, Tufts University, Medford, MA.
Collins, W.D., C. M. Bitz, M.L. Blackmon, G.B. Bonan, C.S. Bretherton, J.A. Carton, P. Chang, S.C.
Doney, J.J. Hack, T.B. Henderson, J.T. Kiehl, W.G. Large, D.S. McKenna, B.D. Santer, and R.D.
Smith. 2006. The Community Climate System Model: CCSM3. Journal of Climatology,
19( 11):2122—2143.
Cristea, N.C., and S. J. Burges. 2010. An assessment of the current and future thermal regimes of three
streams located in the Wenatchee River basin, Washington State: Some implications for regional
river basin systems. Climatic Change, 102:493-520.
Curran, C.A., and T.D. Olsen. 2009. Estimating Low-Flow Frequency Statistics and Hydrologic Analysis
of Selected Streamflow-Gaging Stations, Nooksack River Basin, Northwestern Washington and
Canada. Scientific Investigations Report 2009-5170. U.S. Geological Survey, Reston, VA.
http://pubs.usgs.gov/sir/2009/517Q.
Daly, C., W.P. Gibson, G. Taylor, G.L. Johnson, and P. Pasteris. 2002. A knowledge-based approach to
the statistical mapping of climate. Climate Research, 22:99-113.
7-1
-------
SFNR Quantitative Assessment - Final
October 2016
Ecology (Washington Department of Ecology). 2003a. QUAL2Kw.xls - A Diurnal Model of Water
Quality for Steady Flow Conditions. Washington Department of Ecology, Olympia, WA.
www.ecv.wa.gov/programs/eap/models.html.
Ecology (Washington Department of Ecology). 2003b. Shade.xls - A Tool for Estimating Shade from
Riparian Vegetation. Washington State Department of Ecology, Olympia, WA.
www.ecv.wa.gov/programs/eap/models.html.
Ecology (Washington Department of Ecology). 2016. Preliminary Draft South Fork Nooksack River
Temperature Total Maximum Daily Load, Water Quality Improvement Report, and
Implementation Plan. October 2016.
Eisner, M.M., L. Cuo, N. Voisin, J.S. Deems, A.F. Hamlet, J. Vano, K.E.B. Mickelson, S.Y. Lee, and
D.P. Lettenmaier. 2010. Implications of 21st century climate change for the hydrology of
Washington State. Climatic Change. doi:10.1007/sl0584-010-9855-0.
Flint, L.E., and A.L. Flint. 2008. A basin-scale approach to estimating stream temperatures of tributaries
to the lower Klamath River, California. Journal of Environmental Quality, 37:57-68.
Flint, L.E., and A.L. Flint. 2011. Estimation of Stream Temperature in Support of Fish Production
Modeling under Future Climates in the Klamath River Basin. Scientific Investigations Report
2011-5171. U.S. Geological Survey, Reston, VA.
Gao, H., Q. Tang, X. Shi, C. Zhu, T.J. Bohn, F. Su, J. Sheffield, M. Pan, D. Lettenmaier, and E.F. Wood.
2010. Water-Budget Record from Variable Infiltration Capacity (VIC) Model. Chapter 6 in
Algorithm Theoretical Basis Document for Terrestrial Water Cycle Models.
http://gridl.cos.gmu.edu:8090/QPeNDAPClient/ATBD Chapter6 doc.pdf.
Goldin, A. 1992. Soil Survey of Whatcom County Area, Washington. U.S. Department of Agriculture,
Soil Conservation Service, in cooperation with Washington State Department of Natural Resources
and Washington State University, Agriculture Research Center.
Gu, R.R., and Y. Li. 2002. River temperature sensitivity to hydraulic and meteorological parameters.
Journal of Environmental Management, 66(l):45-56.
Hamlet, A.F., and D.P. Lettenmaier. 2007. Effects of 20th century warming and climate variability on
flood risk in the western US. Water Resources Research, 43:W06427.
Hamlet, A.F., P. Carrasco, J. Deems, M.M. Eisner, T. Kamstra, C. Lee, S.-Y. Lee, G.S. Mauger, E. P.
Salathe, I. Tohver, and L.W. Binder. 2010. Final Report for the Columbia Basin Climate Change
Scenarios Project. University of Washington, Climate Impacts Group, Seattle, WA.
Hamlet, A.F., M.M. Eisner, G.S. Mauger, S.-Y. Lee, I. Tohver, and R.A. Norheim. 2013. An overview of
the Columbia basin climate change scenarios project: Approach, methods, and summary of key
results. Atmosphere-Ocean, 51 (4):392—415.
Hicks, M. 2002. Evaluating Standards for Protecting Aquatic Life in Washington's Surface Water Quality
Standards - Temperature Criteria - Draft Discussion Paper and Literature Summary. Publication
No. 00-10-070. Washington State Department of Ecology, Olympia, WA.
7-2
-------
SFNR Quantitative Assessment - Final
October 2016
IPCC (Intergovernmental Panel on Climate Change). 2007. Climate Change 2007: Synthesis Report-
Summary for Policymakers. Intergovernmental Panel on Climate Change, Geneva.
http://www.ipcc.ch/pdf/assessment-report/ar4/svr/ar4 svr spm.pdf.
IPCC. 2013. Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the
Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Stocker, T.F., D. Qin,
G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex and P.M. Midgley
(eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 1535
pp, doi: 10.1017/CBO9781107415324.
Isaak, D.J., C.H. Luce, B.R. Rieman, D.E. Nagel, E.E. Peterson, D.L. Horan, S. Parkes, and G.L.
Chandler. 2010. Effects of climate change and wildfire on stream temperature and salmonid
thermal habitat in amountain river network. Ecological Applications, 20(5): 1350-1371.
Isaak, D.J., S. Wollrab, D. Horan, and G. Chandler. 2011. Climate change effects on stream and river
temperatures across the northwest U.S. from 1980-2009 and implications for salmonid fishes.
Climatic Change, doi: 10:1007/s 10584-011-0326-z.
Johns, T.C., C. F. Durman, H. T. Banks, M. J. Roberts, A. J. McLaren, J. K. Ridley, C. A. Senior, K. D.
Williams, A. Jones, G. J. Rickard, S. Cusack, W. J. Ingram, M. Crucifix, D. M. H. Sexton, M. M.
Joshi, B.-W. Dong, H. Spencer, R. S. R. Hill, J. M. Gregory, A. B. Keen, A. K. Pardaens, J. A.
Lowe, A. Bodas-Salcedo, S. Stark, and Y. Searle. 2006. The new Hadley Centre climate model
HadGEMl: Evaluation of coupled simulations. Journal of Climatology, 19(7): 1327-1353.
Johnson, S.L. 2004. Factors influencing stream temperatures in small streams: substrate effects and a
shading experiment. Canadian Journal of Fisheries and Aquatic Sciences, 61(6): 913-923.
Johnson, T.E., J. Butcher, A. Parker, and C.P. Weaver. 2012. Investigating the implications of climate
change for U.S. stream water quality: The EPA Global Change Research Program's "20
Watersheds" project. Journal of Water Resources Planning and Management, 138(5):453-464.
Kennedy, J.T., and J. Butcher. 2012. South Fork Nooksack River Temperature Total Maximum Daily
Load Water Quality Study Design (Quality Assurance Project Plan). Publication No. 12-03-126.
Washington Department of Ecology, Olympia, WA.
Klein, S., J. Butcher, B. Duncan, and H. Herron. 2013. EPA Region 10 Climate Change and TMDL Pilot,
Research Plan. EPA/600/R/13/028. U.S. Environmental Protection Agency, Office of Research and
Development, Ecological Effects Branch, Western Ecology Division, Corvallis, OR.
Klungland, M.W., and M. McArthur. 1989. Soil Survey of Skagit County Area, Washington. U.S.
Department of Agriculture, Soil Conservation Service, in cooperation with Washington State
Department of Natural Resources and Washington State University, Agriculture Research Center.
Knutti, R., R. Furrer, C. Tebaldi, J. Cermak, and G.A. Meehl. 2010. Challenges in combining projections
from multiple climate models. Journal of Climatology, 23(10):2739-2758.
Lawrence, S. 2008. Lower Skagit River Tributaries Temperature Total Maximum Daily Load, Water
Quality Improvement Report. Publication No. 08-10-020. Washington State Department of
Ecology, Northwest Regional Office, Bellevue, WA.
https://fortress.wa.gov/ecv/publications/publications/0810020.pdf.
7-3
-------
SFNR Quantitative Assessment - Final
October 2016
Liang, X., D.P. Lettenmaier, E.F. Wood, and S.J. Burges. 1994. A simple hydrologically based model of
land surface water and energy fluxes for GSMs. Journal of Geophysical Research, (99) D7:14415—
14428.
Liang, X., E.F. Wood, and D.P. Lettenmaier. 1996. Surface soil moisture parameterization of the VIC-2L
model: Evaluation and modifications. Global Planetary Change, 19:137-159.
Maidment, D.R. (ed.). 1993. Handbook of Hydrology. McGraw-Hill, New York.
Mantua, N., I. Tohver, and A. Hamlet. 2009. Impacts of Climate change on Key Aspects of Freshwater
Salmon Habitat in Washington State. Chapter 6 in The Washington Climate Change Impacts
Assessment: Evaluating Washington's Future in a Changing Climate. Climate Impacts Group,
University of Washington, Seattle, WA.
Mantua, N., I. Tohver, and A. Hamlet. 2010. Climate change impacts on streamflow extremes and
summertime stream temperature and their possible consequences for freshwater salmon habitat in
Washington State. Climatic Change, 102:187-223.
Mauger, G.S., and N. Mantua. 2011. Climate Change Projections for USFS Lands in Oregon and
Washington (draft). University of Washington, Climate Impacts Group, Seattle, WA. Accessed
September 9, 2013.
http://cses.washington.edu/picea/USFS ORWA/pub/USFS ImpactsReport DRAFT GMauger.pdf.
McCullough, D.A., S. Spalding, D. Sturtevant, and M. Hicks. 2001. Issue Paper 5: Summary of Technical
Literature Examining the Physiological Effects of Temperature on Salmonids. EPA-910-D-01-005.
U.S. Environmental Protection Agency, Region 10, Seattle, WA.
Melillo, J.M., T.C. Richmond, and G.W. Yohe, Eds. 2014. Climate Change Impacts in the United States:
The Third National Climate Assessment. U.S. Global Change Research Program.
doi:10.7930/J0Z31WJ2. http://nca2014.globalchangc.gov/report/our-changing-climate.
Miles, E.L., M.M. Eisner, J.S. Littell, L.W. Binder, D.P. Lettenmaier. 2010. Assessing regional impacts
and adaptation strategies for climate change: The Washington Climate Change Impacts
Assessment. Climatic Change, 102(l-2):9-27.
Mitchell, T.D., T.R. Carter, P.D. Jones, M. Hulme, and M. New. 2004. A comprehensive set of climate
scenarios for Europe and the globe: the observed record (1901-2000) and 16 scenarios (2001-
2100). Tyndall Centre Working Paper No. 55. University of East Anglia, Norwich, UK.
Mohamedali, T., and A. Stohr. 2011. Determining System Potential Mature Riparian Vegetation and
Shade, Guidance Document. Washington State Department of Ecology, Olympia, WA.
Mohseni, O., H.G. Stefan, and T.R. Erickson. 1998. A nonlinear regression model for weekly stream
temperatures. Water Resources Research, 34(10):2685-2692.
Mohseni, O., and H.G. Stefan. 1999. Stream temperature air temperature relationship: A physical
interpretation. Journal of Hydrology, 218(3-4): 128-141.
Mote, P.W., and E.P. Salathe. 2010. Future climate in the Pacific Northwest. Climatic Change, 102(1-
2): 29-5 0. doi: 10.1007/s 105 84-010-9848 -z.
Mote, P., L. Brekke, P.B. Duffy, and E. Maurer. 2011. Guidelines for constructing climate scenarios. Eos,
Transactions of the American Geophysical Union, 92(31):257-258.
7-4
-------
SFNR Quantitative Assessment - Final
October 2016
Nash, J.E., and J.V. Sutcliffe. 1970. River flow forecasting through conceptual models, part 1 - A
discussion of principles. Journal of Hydrology, 10(3):282-290.
NRC (National Research Council). 2001. Assessing the TMDL Approach to Water Quality Management.
National Academy Press, Washington, DC.
Pelletier, G.J., S.C. Chapra, and H. Tao. 2006. QUAL2Kw - a framework for modeling water quality in
streams and rivers using a genetic algorithm for calibration. Environmental Modelling and
Software, 21:419-425.
Pelletier, G. and S. Chapra. 2008. QUAL2Kw Theory and Documentation (Version 5): A Modeling
Framework for Simulating River and Stream Water Quality. Washington State Department of
Ecology, Olympia, WA.
Pierce, D.W., T.P. Barnett, B.D. Santer, and P.J. Gleckler. 2009. Selecting global climate models for
regional climate change studies. Proceedings of the National Academy of Sciences, 106(21): 8441—
8446.
Polebitski, A., L. Traynham, and R.N. Palmer. 2007. Technical Memorandum #4: Approach for
Developing Climate Impacted Meteorological Data and its Quality Assurance/Quality Control.
Climate Change Technical Subcommittee of the Regional Water Supply Planning Process, Seattle,
WA.
Randall, D.A., R.A. Wood, S. Bony, R. Colman, T. Fichefet, J. Fyfe, V. Kattsov, A. Pitman, J. Shukla, J.
Srinivasan, R.J. Stouffer, A. Sumi, and K.E. Taylor. 2007. Climate Models and Their Evaluation. In
Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth
Assessment Report of the Intergovernmental Panel on Climate Change [S. Solomon et al., Editors]
Cambridge University Press, Cambridge, UK.
Richter, A., and S.A. Kolmes. 2005. Maximum temperature limits for Chinook, Coho, and chum salmon,
and steelhead trout in the Pacific Northwest. Reviews in Fisheries Science, 13:23-49.
Scheuerell, M.D., R. Hilborn, M.H. Ruckelshaus, K.K. Bartz, K.M. Lagueux, A.D. Haas, and K. Rawson.
2006. The Shiraz model: A tool for incorporating anthropogenic effects and fish-habitat
relationships in conservation planning. Canadian Journal of Fisheries and Aquatic Sciences,
63(7): 1596-1607.
Seiler, D., S. Neuhauser, and L. Kishimoto. 2003. Skagit River Wild 01 Chinook Production Evaluation
Annual Report. Washington Department of Fish and Wildlife, Olympia, WA.
Serveiss, V.B., J.B. Butcher, J. Diamond, and K.C. Jones. 2005. Improving the TMDL process using
watershed risk assessment principles. Environmental Management, 36(1): 143-151.
Stohr, A., J. Kardouni, and R. Svrjcek. 2011. Snoqualmie River Basin Temperature Total Maximum
Daily Load: Water Quality Improvement Report and Implementation Plan. Publication No. 11-10-
041. Washington State Department of Ecology, Olympia, WA, and Washington State Department
of Ecology Water Quality Program, Bellevue, WA.
Tebaldi, C., K. Hayhoe, J.M. Arblaster, and G.A. Meehl. 2006. Going to the extremes: An
intercomparison of model-simulated historical and future changes in extreme events. Climatic
Change, 79:185-211.
7-5
-------
SFNR Quantitative Assessment - Final
October 2016
USEPA (U.S. Environmental Protection Agency). 1992. Framework for Ecological Risk Assessment.
EPA/630/R-92/0001. U.S. Environmental Protection Agency, Risk Assessment Forum,
Washington, DC.
USEPA (U.S. Environmental Protection Agency). 2003. EPA Region 10 Guidance for Pacific Northwest
State and Tribal Temperature Water Quality Standards. EPA 910-B-03-002. U.S. Environmental
Protection Agency, Region 10, Office of Water, Seattle, WA.
Waples, R.S., D.J. Teel, J.M. Myers, and A.R. Marshall. 2004. Life-history divergence in Chinook
salmon: Historic contingency and parallel evolution. Evolution, 58(2):386-403.
Wu, H., J.S. Kimball, M.M. Eisner, N. Mantua, R.F. Adler, and J. Stanford. 2012. Projected climate
change impacts on the hydrology and temperature of Pacific Northwest rivers. Water Resources
Research, 48:W11530.
Yohe, G. 2011. Incorporating (Iterative) Risk Management into the National Climate Assessment,
presentation by Gary Yohe, Vice-Chair of the NCADAC, July 12, 2011, Regional Working Group
Background Document, Climate Assessment.
Zalewsky, B., and D. Bilhimer. 2004. Lower Skagit River Tributaries Temperature Total Maximum Daily
Load. Washington State Department of Ecology, Olympia, WA.
7-6
-------
-------
&EPA
United States
Environmental Protection
Agency
Western Ecology Division
National Health and Environmental Effects Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Corvallis, OR 97333
W
Printed on recycled paper
-------
|