understanding the uncertainties associated with · pdf filewater quality trading (wqt) to...
TRANSCRIPT
![Page 1: Understanding the Uncertainties Associated with · PDF filewater quality trading (WQT) to work, but ... – Point source loads (after satisfying model input format) – Fertilizer](https://reader033.vdocuments.us/reader033/viewer/2022051320/5a9e409b7f8b9a2e688c9234/html5/thumbnails/1.jpg)
Office of Research and DevelopmentNational Risk Management Research Laboratory
Understanding the Uncertainties Associated with Watershed Modeling for Water Quality Trading
Christopher T. Nietch, Matthew T. Heberling, and Amr Safwat
![Page 2: Understanding the Uncertainties Associated with · PDF filewater quality trading (WQT) to work, but ... – Point source loads (after satisfying model input format) – Fertilizer](https://reader033.vdocuments.us/reader033/viewer/2022051320/5a9e409b7f8b9a2e688c9234/html5/thumbnails/2.jpg)
2
• Local, state, and federal agencies want water quality trading (WQT) to work, but there are many barriers. We focus on two:
– Uncertainties associated with modeling the watershed system
– Thin markets (too few participants)
• We use monitoring, modeling, and economic tools to address these barriers.
• Our approach is to use a case-study with traditional actors and potentially alternative participants at a tractable scale.― The study had to be large scale, but not so large that the spatial extent would make
capturing the uncertainties associated with credit generation and WQ attainment untraceable.
― We hypothesized that with a good monitoring program and an appropriate model we cold link all future trades to water quality consequence.
![Page 3: Understanding the Uncertainties Associated with · PDF filewater quality trading (WQT) to work, but ... – Point source loads (after satisfying model input format) – Fertilizer](https://reader033.vdocuments.us/reader033/viewer/2022051320/5a9e409b7f8b9a2e688c9234/html5/thumbnails/3.jpg)
3
Research Conceptual framework:Making WQT more viable
We developed a framework in response to questions:1) In the face watershed
uncertainties and existing structure of point and non-point sources is water quality trading feasible?
2) If not, can we increase participation by considering non-typical participants?
3) Are there unintended consequences and/or additional benefits of these approaches?
We define feasibility as the preliminary assessment for the potential of trading to meet water quality goals.
![Page 4: Understanding the Uncertainties Associated with · PDF filewater quality trading (WQT) to work, but ... – Point source loads (after satisfying model input format) – Fertilizer](https://reader033.vdocuments.us/reader033/viewer/2022051320/5a9e409b7f8b9a2e688c9234/html5/thumbnails/4.jpg)
East Fork Watershed Case Study and Stakeholders-EFWCoop:
4
“Only through adequate testing, refinement, and calibration….And review and buy-in from stakeholders will model related uncertainties be overcome” (Walker and Selman 2014)
1300 km2 or321,236 acres
![Page 5: Understanding the Uncertainties Associated with · PDF filewater quality trading (WQT) to work, but ... – Point source loads (after satisfying model input format) – Fertilizer](https://reader033.vdocuments.us/reader033/viewer/2022051320/5a9e409b7f8b9a2e688c9234/html5/thumbnails/5.jpg)
5
East Fork Watershed System:Good monitoring program design
Spatially and temporally dense monitoring program
Nested monitoring schemes within different land use types (agr & urban)
![Page 6: Understanding the Uncertainties Associated with · PDF filewater quality trading (WQT) to work, but ... – Point source loads (after satisfying model input format) – Fertilizer](https://reader033.vdocuments.us/reader033/viewer/2022051320/5a9e409b7f8b9a2e688c9234/html5/thumbnails/6.jpg)
6
• Different types of uncertainty are encountered in WQT
• Walker and Selman (2014) categorize them as:– Biophysical and Scientific Uncertainty– Extreme Event Uncertainty– Behavioral Uncertainty– Regulatory Uncertainty– Market Uncertainty
• We address the Biophysical and Scientific Uncertainty by conducting research that integrates and evaluates all five aspects that are defined as potential mitigation strategies:
– Improved Science– Direct Measurement– Effectiveness Estimates– Estimation tools and models– Trading Ratios
![Page 7: Understanding the Uncertainties Associated with · PDF filewater quality trading (WQT) to work, but ... – Point source loads (after satisfying model input format) – Fertilizer](https://reader033.vdocuments.us/reader033/viewer/2022051320/5a9e409b7f8b9a2e688c9234/html5/thumbnails/7.jpg)
7
• Most interested in quantify uncertainty inTotal Nitrogen (TN) and Total Phosphorus (TP) loads
– from farm fields– from septic systems– after AgBMPs are implemented– at the watershed outlet– attenuated
• Fixed Inputs (need to rationalize uncertainty as “captured”)
– Weather Data– Land uses– Point source loads (after satisfying model input format)– Fertilizer application rates (??)– Model adjustments to simulate BMPs (??)– BMP implementation costs (both for point and non-points
sources)• Soil and Water Assessment Tool (SWAT) – Key
Modeling PackageSemi-distributed, physically based, divided into land and channel phases.
• Stormwater Management Model (SWMM) used for fine scale modeling of urban area.
Neitsch et al., 2011, http://swat.tamu.edu/media/99192/swat2009-theory.pdf
EPA-SWMM 5.0, Rossman 2010
![Page 8: Understanding the Uncertainties Associated with · PDF filewater quality trading (WQT) to work, but ... – Point source loads (after satisfying model input format) – Fertilizer](https://reader033.vdocuments.us/reader033/viewer/2022051320/5a9e409b7f8b9a2e688c9234/html5/thumbnails/8.jpg)
Ranked TN Loads among Farmer BMP applicants
SWAT Model Output
8
Nitrate Loading at Parcel Scale – HRU
Level
High-resolution SWAT at watershed scale has trade-offs: Long runtimes and high data management requirements.
Nietch and others, 2013 Alternative land use method for spatially-informed watershed management decision-making using SWAT, JEE,139:1413-1423.
![Page 9: Understanding the Uncertainties Associated with · PDF filewater quality trading (WQT) to work, but ... – Point source loads (after satisfying model input format) – Fertilizer](https://reader033.vdocuments.us/reader033/viewer/2022051320/5a9e409b7f8b9a2e688c9234/html5/thumbnails/9.jpg)
9
• Feasibility analysis makes cost comparisons among supply and demand for credits. That is, see if traditional WQT is even an economically viable option
• Analyze data on cost per pound of nutrient reduction for many different BMPs and at different WWTP control levels.
– However, out of 10 WWTPs in the UEFW only 5 had incentive to trade. – Also given the density of NPS relative to the location and size of PS demand, trading would
not result in necessary WQ improvement.
≈$50KAnnual
≈$1KAnnualwithout trading, vs. with tradingOr,
+Urban GI-types
CTIC. 2011.
![Page 10: Understanding the Uncertainties Associated with · PDF filewater quality trading (WQT) to work, but ... – Point source loads (after satisfying model input format) – Fertilizer](https://reader033.vdocuments.us/reader033/viewer/2022051320/5a9e409b7f8b9a2e688c9234/html5/thumbnails/10.jpg)
10
• Determining the significance of BMP scenarios:
• The uncertainty in the load reduction is needed to define credit availability
• If I’m a farmer, and I want to start using Cover Crops, how do I know how many credits I will get?
1) We need to know what effect the Cover Crop BMP type has on N and/or P runoff reduction. I.e., the difference between before and after BMP placement,
2) If one credit = 1lbs TN removed, then TNreduced = number of credits available (not accounting for uncertainty)
Corn growing through decomposing cover of annual ryegrass in Clermont County farmer’s field
![Page 11: Understanding the Uncertainties Associated with · PDF filewater quality trading (WQT) to work, but ... – Point source loads (after satisfying model input format) – Fertilizer](https://reader033.vdocuments.us/reader033/viewer/2022051320/5a9e409b7f8b9a2e688c9234/html5/thumbnails/11.jpg)
11
So, how do we capture uncertainty at Farm Field Scale using SWAT Model while maintaining tractability at watershed scale?
Say there is 11 AgLandUses x 10 soil types x 3 slope classes = 333 unique Field Scale Conditions:333 conditions x 2 scenarios x 1000 runs x 2 constituents = 1,332,000 individual model runs
We can’t manage the data from that many model runs!
As an alternative, we obtain the “field-scale” uncertainty by conducting an analysis at the headwatershed scale in small agricultural drainages
We still have to run the watershed model on the order of:
1000 runs x 2 scenarios x 2 constituents = 4000 times
We have to apply the same fractional uncertainty to individual fields.
![Page 12: Understanding the Uncertainties Associated with · PDF filewater quality trading (WQT) to work, but ... – Point source loads (after satisfying model input format) – Fertilizer](https://reader033.vdocuments.us/reader033/viewer/2022051320/5a9e409b7f8b9a2e688c9234/html5/thumbnails/12.jpg)
12
Watershed Scale: Flow (cms)
TN (kg/month)
TP (kg/month)
p-factor 0.69r-factor 0.83R2 0.82NSE 0.75PBIAS 4.4KGE 0.62
p-factor 0.32r-factor 0.56R2 0.56NSE 0.49PBIAS 19KGE 0.58
p-factor 0.52r-factor 0.84R2 0.53NSE 0.51PBIAS 10.2KGE 0.66
Nitrogen Load Distribution for the Upper EFW – best model
simulation 1000 runs
![Page 13: Understanding the Uncertainties Associated with · PDF filewater quality trading (WQT) to work, but ... – Point source loads (after satisfying model input format) – Fertilizer](https://reader033.vdocuments.us/reader033/viewer/2022051320/5a9e409b7f8b9a2e688c9234/html5/thumbnails/13.jpg)
13
Headwatershed Scale
TN (kg/month) p-factor 0.61
r-factor 1.73
R2 0.56
NSE 0.55
PBIAS 0.6
KGE 0.57
Headwatershed Monitoring
![Page 14: Understanding the Uncertainties Associated with · PDF filewater quality trading (WQT) to work, but ... – Point source loads (after satisfying model input format) – Fertilizer](https://reader033.vdocuments.us/reader033/viewer/2022051320/5a9e409b7f8b9a2e688c9234/html5/thumbnails/14.jpg)
14
Uncertainty at headwatershed scale
Uncertainty at watershed scale
Scaled to Annual w/ 95%CI for
mean estimate
. ±3.5%
.
. ±4.0%
.
![Page 15: Understanding the Uncertainties Associated with · PDF filewater quality trading (WQT) to work, but ... – Point source loads (after satisfying model input format) – Fertilizer](https://reader033.vdocuments.us/reader033/viewer/2022051320/5a9e409b7f8b9a2e688c9234/html5/thumbnails/15.jpg)
15
Scenario gMean L90PPU U90PPU %LE %UE EfficiencyBaseCase‐NoTill 15.0 3.1 90.8BMP‐CoverCrop 7.2 2.1 25.3Difference 8.7 ‐21.9 46.2 352 431 0.58
TN reduced normalized to cover crop area (lbs/yr.acre cc )) at headwatershed scale
Scenario gMean L90PPU U90PPU %LE %UE EfficiencyBaseCase‐NoTill 18.9 4.6 97.7BMP‐CoverCrop 12.3 4.7 36.8Difference 8.4 ‐21.8 47.8 358 467 0.45
TN reduced normalized to cover crop area (lbs/yr.acre cc )) at watershed scale
– Results suggest that a 3.5 to 4.0 Trade Ratio should be adopted for cover crops.
– On average there is about a 10% loss in treatment efficiency at the watershed scale.
![Page 16: Understanding the Uncertainties Associated with · PDF filewater quality trading (WQT) to work, but ... – Point source loads (after satisfying model input format) – Fertilizer](https://reader033.vdocuments.us/reader033/viewer/2022051320/5a9e409b7f8b9a2e688c9234/html5/thumbnails/16.jpg)
16
NPS dominates PS dominates
Under assumed sufficient credit demand
Target still violated
Flow exceedance frequencyUncertainty Changes
• Determining the significance of BMP scenarios II:
• Meeting the WQ Target
WQ Target
TP (m
gL)
WQ Target
TP (m
gL)
Measured Uncertaintye.g., from the Great Miami River
![Page 17: Understanding the Uncertainties Associated with · PDF filewater quality trading (WQT) to work, but ... – Point source loads (after satisfying model input format) – Fertilizer](https://reader033.vdocuments.us/reader033/viewer/2022051320/5a9e409b7f8b9a2e688c9234/html5/thumbnails/17.jpg)
Conclusions
• With a scalable model application we can rationalize credit availability and gauge risk of not meeting WQ target, therefore can help reduce regulatory uncertainty.
• A handful of detailed case studies like this one should be able to provide probability distributions to BMP scenarios that would allow estimating confidence ranges outside of complex modeling environment.
• In our system, uncertainty captured or not, WQT using traditional participants is not feasible for meeting WQ Targets.
17
• BMP uncertainty includes, baseline model uncertainty, BMP-specific effect uncertainty, and how model is simulating BMP effect at large scale. This is “Prediction Uncertainty” and can be quite large.
• In our system a 3.5 Trade Ratio minimum would be needed to account for prediction uncertainty.
• There is currently no guidance for capturing prediction uncertainty!
![Page 18: Understanding the Uncertainties Associated with · PDF filewater quality trading (WQT) to work, but ... – Point source loads (after satisfying model input format) – Fertilizer](https://reader033.vdocuments.us/reader033/viewer/2022051320/5a9e409b7f8b9a2e688c9234/html5/thumbnails/18.jpg)
References
18
– Arabi, M., R.S. Govindaraju, and M.M. Hantush. 2007. A probabilistic approach for analysis of uncertainty in the evaluation of watershed management practices. Journal of Hydrology,333:459-471.
– Arabi, M., J.R. Frankenberger, B.A. Engel, and J.G. Arnold. 2008. Representation of agricultural conservation practices with SWAT. Hydrological Processes. 22, 3042-3055.
– Walker, S. and M.Selman. 2014. Addressing risk and uncertainty in water quality trading markets. World Resources Institute. Issue Brief. 10G St NE, Suite 800, Washington, DC. 20002. www.wri.org.
– Horan, R.D. and J. S. Shortle. 2011. Economic and Ecological Rules for Water Quality Trading. JAWRA, 47(1):59-69.– Woznicki, S. A., and A.P. Nejdhashemi. 2014. Sensitivity analysis of BMPs under climate change scenarios. JAWRA,1-23.,DOI:
10.1111/j.1752-1688.2011.00598.x– Karchar, S.C., J.M. VanBriesen, and C.T. Nietch. 2013. Alternative land use method for spatially informed watershed
management decision making using SWAT. Journal of Environmental Engineering. 139:1413-1423.– Wabash River Watershed Water Quality Trading Feasibility Study – Final Report. 2011. Prepared by the Conservation
Technology Information Center. http://www.ctic.org/media/pdf/Wabash%20WQT%20Feasibility%20Study_091411_final%20report.pdf
– Neitsch, S. L., J.G. Arnold, J.R. Kiniry, and J.R. Williams. 2011, Soil and Water Assessment Tool Theoretical Documentation Version 2009. Texas Water Resources Institute Technical Report No. 406. Texas A&M University System, College Station, Texas 77843-2118. http://swat.tamu.edu/media/99192/swat2009-theory.pdf.
– Rossman, L.A. 2010. Stormwater Management Model User’s Manual Version 5.0. EPA/600/R-05/040.
The data, expressed ideas and opinions, and interpretation herein are those of the primary presenter (Nietch) and do not reflect official EPA position or policy.