Presentation on theme: "Advancing the Science of Modeling: Industry Perspectives Dave Gustafson 29 March 2011."— Presentation transcript:
Advancing the Science of Modeling: Industry Perspectives Dave Gustafson 29 March 2011
Outline Importance of high quality input data Use best available modeling technology Follow Good Modeling Practices (“GMPs”) Increasing importance of buffers Multiple ecosystem services provided Agreed methods for quantifying benefits Closing comments about the future 2
Acknowledgements 3 Dave Archer – USDA-ARS Nancy Baker – USGS Jeff Frey – USGS Jerry Hatfield – USDA-ARS Doug Karlen – USDA-ARS Cristina Negri – DOE-ANL John Prueger – USDA-ARS Al Barefoot, DuPont Paul Hendley, Syngenta Scott Jackson, BASF Russell Jones, Bayer Iain Kelly, Bayer Mike Legget, CropLife Nick Poletika, Dow Industry Federal Agencies
Importance of High Quality Input Data 4 “GIGO”: a cliché, but still very true USDA-NASS data collection must be supported and should be greatly expanded More frequent collection of more extensive nutrient input data (including timing info, etc.) New collection of data on tillage practices Standardized, enhanced hydrology (NHDPlus) New, higher resolution NEXRAD data should be utilized whenever possible and appropriate
Use Best Available Modeling Technology Pesticide screening tools – OK for Tier 1 only Ensure underlying mathematics of the simulation model is actually correct Pesticide dissipation Dispersion (leaching and in rivers) Modeling for landscape management HIT (Jon Bartholic, Michigan State University) SWAT & APEX (Claire Baffaut, USDA-ARS) 5
GUS: Example Tier 1 Screening Tool Initially proposed as a joke to colleagues at Monsanto Ended up getting published and “going viral” in the early 1990s (pre-Internet) Not appropriate for exposure analysis Only useful for the purpose of determining when higher tier modeling techniques are needed 6 “Groundwater Ubiquity Score: A Simple Method for Assessing Pesticide Leachability,” J. Environ. Toxic. & Chem., 8:339-357 (1989).
Pesticide dissipation is nearly always nonlinear, yet many models still assume linear, 1 st -order dissipation kinetics Dispersion coefficient increases linearly with mean distance traveled, yet nearly all models assume constant D L Getting the Underlying Mathematics Right 7 “Nonlinear Pesticide Dissipation in Soil: A New Model Based on Spatial Variability,” Environ. Sci. & Technol., 24:1032-1038 (1990). “Modeling Root Zone Dispersion: A Comedy of Error Functions,” Chem. Eng. Comm., 73:77-94 (1988). “Fractal-Based Scaling and Scale-Invariant Dispersion of Peak Concentrations of Crop Protection Chemicals in Rivers,” Environ. Sci. & Technol., 38:2995-3003 (2004).
Modeling Challenge: Predicting Peak Concentrations in Surface Water A key regulatory question is the following: What is the “peak” pesticide concentration to which humans and aquatic organisms are exposed via surface water? The answer depends largely on scale Need a proper model for scale effects Exploit scaling properties of fractals to provide such a model 8
One Possible Modeling Approach 9 Determine daily edge-of-field concentrations and flows using an existing regulatory model Feed these into a simple analytical model to simulate scale effects A Fractal-Based, Scale Dependent Analytical Solution to Convective-Dispersion Eq. PRZM or MACRO, etc.
Method Validated Using Heidelberg College (WQL) Monitoring Data 10
Temporal Intensity of Heidelberg Pesticide Monitoring Data 11 Surface water monitoring results from the Water Quality Laboratory. Each plot shows daily streamflow per unit area (Q/A) and concentrations of four herbicides: acetochlor (AC), alachlor (AL), atrazine (AT), and metolachlor (ME) during 1996, a high runoff year.
Excellent Fits Achieved to Shape of Hydrograph and Chemograph 12 Hydrograph following large upstream runoff event in June 1996 Atrazine chemograph following the same runoff event
Additional Modeling Science Issues Challenges of modeling water and contaminant transport at edge-of-field water exit points Agree appropriate scales for watershed modeling, particularly in Regulatory contexts Alternatives to Nash-Sutcliffe (accuracy metric for hydrological models), such as Ehret & Zehe † Data needed for parameterization of buffer performance (more on this later in the talk) 13 † Hydrol. Earth Syst. Sci., 15, 877–896, 2011 www.hydrol-earth-systsci.net/15/877/2011/doi:10.5194/hess-15-877-2011www.hydrol-earth-systsci.net/15/877/2011/doi:10.5194/hess-15-877-2011
Good Modeling Practices (“GMPs”) Modeling results should be reproducible and able to be compared with alternative models All assumptions and methods clearly stated Input data and model source code available Guidance concerning applicability of results Clearly state any limits on valid extrapolation of results (in space or time, especially the future) What weaknesses of the model or modeling report should be known by the user/reader? 14
Buffers: Increasingly Important, & Increasing Challenged ($7 corn) Conservation buffers are areas or strips of land maintained in permanent vegetation to help control pollutants and manage other environmental problems (USDA definition) Used for many years to reduce transport of eroded soil Also provide other benefits, such as reduction of runoff and nutrient entry into surface waters, wildlife habitat improvement, streambank protection, and mitigation of drift (if placed around entire field) 16
VFSMOD: Mechanistic Modeling of Vegetative Filter Strips 17 VFSMOD developed for regulatory modeling of buffer effectiveness Improved understanding of pesticide retention processes Nonlinear, complex relationship, relating pesticide retention to: –Rainfall/run-on event size –VFS length Availability of this new, useful model drives new data needs
Plant a nonfood perennial bioenergy crop (switchgrass, Miscanthus, etc.) as a buffer strip around all sides of all row crop fields Width is negotiable, but probably try to fit 1 or 2 passes of harvest equipment (~15-30’) 7.5M acres for all US corn and soybean fields Assuming 20’ width and 80 acre average field size (40’ for adjoining fields) New Concept: “Bioenergy Buffers” 18
Bioenergy Buffers Provide Multiple Ecosystem Services Improved water quality Additional wildlife habitat Enhanced “C-questration” Sustainable energy source Endangered species protection Mitigation of spray drift source: Jeff Volenec (Purdue) source: Doug Karlen (USDA-ARS) source: DEFRA Switchgrass Elephant Grass Miscanthus giganteus Reed Warbler nest in Miscanthus (UK) 19
Bioenergy Buffer Collaborations Minnesota: Don Wyse (Univ MN), Xcel Energy White Paper on pesticide drift mitigation USDA-ARS (Jerry Hatfield, et al.) Ceres, Dow, DuPont/Danisco, Mendel, Monsanto Field study demonstration Location: Indian Creek watershed near Fairbury IL Key collaborators: CTIC, DOE Argonne 20
New CropLife America Initiative on Buffers and Pesticide Mitigation Buffers now required on many pesticide labels to reduce potential impacts on aquatic organisms Need for agreed modeling methods on quantifying the degree of mitigation provided by buffers Need to further develop and refine practical solutions for positioning, introducing and maintaining buffers Success will require a broad collaboration among Grower Groups, EPA, USDA, State Agencies, etc. Utilize appropriate, standardized label language 21
Closing Comments about the Future Bioenergy Buffers likely to become widespread Either through BCAP-type incentives or by modifying existing conservation programs Continued increases in Nitrogen Use Efficiency Step-changes coming through new Biotech Traits Better input data through Remote Sensing GMPs essential if good science is to prevail 22