1. David Stevens, Nancy Mesner, Terry Glover, Arthur Caplan
Water Quality Modeling in the Bear River Basin to Support Water Quality Trading
Jeff Horsburgh
David Stevens, Nancy Mesner, Terry Glover, Arthur Caplan
Utah State University

2. USEPA Targeted Watersheds Grant
Develop an integrated, Internet-based Watershed Information System (WIS)
Investigate the feasibility of a water quality trading program
Develop a water quality model to support the water quality trading program

3. Why do we need a model to study the potential for trading?
Simulate the physical, chemical, and biological processes that affect pollutant concentrations
Consider spatial and temporal nature of pollutant loading
Calculate delivery ratios to
Determine the environmental equivalence of load reductions and potential trades

4. Background on Trading Trading presumes that a TMDL has been completed
Assumes that pollutant loads and load reductions have already been allocated to the sources in the watershed

5. Why do we need a model? Producer X River Mile 15
Producer Y River Mile 5
Animal Producer Z River Mile 10
Point Source
River Mile 1
Agricultural Diversion (50%)
River Mile 7
Monitoring Location Receptor Point River Mile 0

6. Why do we need a model? Producer X River Mile 15
Total Phosphorus Load = 100 kg
Producer X could only reduce loading at the receptor point by 40.5 kg even though his total loading is 100 kg!
Delivery Ratio = 40.5/100 = 0.405
Loss to uptake/settling prior to diversion = 10% or 10 kg (90 kg left)
Loss to uptake/settling prior to receptor point = 10% or 4.5 kg (40.5 kg left)
Loss to Diversion = 50 % or 45 kg/yr (45 kg left)
Monitoring Location Receptor Point River Mile 0

7. Why do we need a model? Producer Y River Mile 5
Producer Y can reduce loading at the receptor point by 90 kg if he eliminates his entire loading.
Delivery Ratio = 90/100 = 0.9
Total Load = 100 kg
Producer Y River Mile 5
Loss to uptake/settling prior to receptor point = 10% or 10 kg (90 kg left)
Monitoring Location Receptor Point River Mile 0

8. What does this mean?
The point source’s cost per unit phosphorus reduction with Producer X would be higher than Producer Y
Producer X can only get credit for 40% of any load reduction that he creates
Producer Y can get credit for 90%
It would be more economical for the point source to trade with Producer Y
It is critical to have an estimate of the delivery ratios

9. Model and Trading Focus Area
Bear River from Oneida Narrows Reservoir to Cutler Reservoir
Cub River
Little Bear River
Spring Creek
Focus on areas with existing TMDLs

10. 303(d) Listed Water Bodies In or near the Model Focus Area
Water body
Pollutants
Weston Creek
TP, Sediment
Newton Reservoir
TP, DO, Water Temperature
Clarkston Creek TP
Cub River
Porcupine Reservoir
Temperature
Hyrum Reservoir
TP, DO
Spring Creek
TP, DO, Ammonia, Temperature, Fecal Coliform
Little Bear River
TP = Total Phosphorus, DO = Dissolved oxygen

11. TMDL – Existing TP Loads by Source
Total Phosphorus, kg/yr
Phosphorus Source
Spring Creek
Hyrum Reservoir
Little Bear River
Newton Reservoir
Newton Creek
Cub River
Point sources
33050
881.7
1049.9
1660
Agriculture nonpoint
1390
3503
2,872
1,311
Urban nonpoint
1990
167
AFO/CAFO nonpoint
600
1550
556
3218
Other nonpoint
1300
2405
206.7
73.7
18396
Spring Creek
Hyrum Reservoir

12. The Modeling Challenges
Large areas to be modeled
Relatively few data available to populate, calibrate, etc.
High spatial resolution needed to compare sources
Watershed scale models (HSPF, SWAT) cover just about everything, but you don’t get the spatial resolution that you need
Field scale models give you high resolution load estimates, but don’t tell you what happens downstream

13. Modeling Approach
One integrated modeling system made up of the following components:
Hydrologic model component (generates flows)
Watershed loading model component (generates constituent loads)
Stream response model component (simulates concentrations in the streams)
Accounting model component (accounts for diversions, interbasin transfers, local load reductions, etc.)

14. Model Segmentation Subwatersheds Stream Reaches Control Points
Primary unit of modeling
Loads are estimated at the subwatershed scale
Stream Reaches
Loads are routed through the existing stream network (from control point to control point)
One reach per subwatershed
Control Points
Breaks in the stream network at important locations
Calibration points
Diversion points
Major tributary confluences

15. Model Segmentation (cont.)

16. Hydrologic Model Component TOPNET/TOPMODEL
Estimation of the amount of flow generated within each subwatershed
Source: Bandaragoda et al., 2004

17. Watershed Loading Component
Simulate the amount of loading generated within each subwatershed
Simulated loads will be based on:
Amount of flow generated in the subwatershed
Area of each land use within each subwatershed
Number and types of agricultural producers (e.g., number of animals, etc. – whatever info we can get!!!!!)
Point source data
Loads generated by the model will be compared to loads estimated in TMDL documents

18. Stream Response Component QUAL2K
Simulate the changes in TP concentration as water moves from control point to control point in the stream

19. Accounting Model Component
Superimposed over the other two components to account for withdrawals, transfers, etc.

20. Constituents to be Modeled
Total phosphorus is the focus
Other constituents that need to be simulated because they interact
Water temperature
Nitrogen concentrations
Algae
Dissolved Oxygen
BOD/organic matter
Inert/conservative tracer

21. What will the results look like?
River Mile
Delivery Ratio 1
0.93 2
0.86 3
0.79 4
0.71 5
0.64 6
0.57 7
0.50 8
0.43 9
0.36 10
0.29 11
0.21 12
0.14 13
0.07 14
0.00 15
Flows and concentrations at control points
Also along the length of modeled reaches if necessary
Tables of delivery ratios
Scenario specific

22. How will the trading study use the results?
Evaluate Potential Sellers
Seller A –
River Mile 10
Delivery Ratio = 0.29
Total Loading = 100 kg

23. How will the trading study use the results?
TMDL Load Allocation = 50 kg (TMDL Requires a 50 % reduction from nonpoint sources)
Additional Load Reduction via BMP = 25 kg
Total potential load reduction = 75 kg
Issue – there is no regulatory mechanism to force nonpoint sources to reduce loading?
Potential Parallel Paths
Path 1: Credits for Sale = (25)*(0.29) = 7.25 kg
Path 2: Credits for sale = (75)*(0.29) = kg

24. Other Benefits
Spatial
Evaluate potential trading scenarios to make sure that water quality standards are met under a variety of conditions
Temporal

25. Water Quality Model Development Process
By component:
Coding (if required)
Verification of code and testing
Population – finding model inputs
Calibration – adjust model parameters so that results match data
Validation/Corroboration

26. Modeling Challenges Revisited Data Availability
Relatively few data available to populate, calibrate, etc.
Water quality data
Streamflow data
Diversion and water transfer data
Climate data
Stream hydraulic characteristics

27. Consider Total Phosphorus Little Bear River at Mendon Road
Utah DWQ

28. Little Bear River at Mendon Road All Utah DWQ TP Data No Streamflow Gage Available
Total Phosphorus observations
241 observations from 1976 – 2004
(one outlier of 6 mg/L removed for plotting)
Streamflow observations
162 observations from

29. Last 10 Years?
In the past 20 years or so, ~$5 Million has been spent in public cost share funds in this watershed to improve water quality
Data more than 10 years old are not representative of current conditions
Total Phosphorus
99 observations from 1994 – 2004
59 % Reduction in available data
Streamflow
72 observations from 1994 – 2004
56 % Reduction in available data

30. What if I want to characterize seasonal variability?

31. What about monthly variability?

32. What about interannual variability?
Streamflow data from the only active USGS gage in the watershed show HUGE variability in flow from year to year!
The average TP concentration during the dry years is 60 % higher than for the wet years

33. Simplified Conceptual Model Phosphorus Loading
How large are the bumps versus the baseline?

34. Simplified Loading Conceptual Model With Traditional Data

35. Modeling Challenges Revisited Spatial Resolution
High spatial resolution needed to compare sources
Controls the spatial resolution of the model
Requires that we know the locations of sources with some certainty!

36. Why do we need a model? Producer X River Mile 15
Producer Y River Mile 5
Animal Producer Z River Mile 10
Point Source
River Mile 1
Agricultural Diversion (50%)
River Mile 7
Monitoring Location Receptor Point River Mile 0

37. Spatial Resolution What we know: Where we are struggling:
Point source locations
Land use/land cover
Where we are struggling:
Locations of “Producers” or potential trading participants
Animal feeding operations
Actual farm units

38. Water Quality Task Force
Project Partners
USU Project Collaborators:
Nancy Mesner Terry Glover
David Stevens Arthur Caplan
United States EPA
Gary Kleeman
Bear River Commission
Jack Barnett
Water Quality Task Force
Mitch Poulson
Wyoming DEQ
Jack Smith
Idaho DEQ
Lynn Van Every
Utah DEQ
Mike Allred

39. Outreach and Education
Contact Information
Modeling
Jeff Horsburgh
David Stevens
Outreach and Education
Nancy Mesner