Quantifying Runoff Mitigation in STEP 4 Calculations with VFSMOD EU-Scenarios in SWAN 3.0 Bjoern Roepke 6 th European Modelling Workshop, Paris 10 th 12 th June 2012
Existing Vegetated Filter Strip (VFS) Structures & Legislation in the EU EU policies facilitating VFS for risk reduction & ecosystem improvement Nutrient directive Sustainable use directive - pesticides Common Agricultural Policy (CAP) National action plans
Quantifying Runoff Mitigation Established width based factors Mechanistic VFS approach Key drivers hydrologic response (infiltration; sedimentation) & phase distribution of the chemical. VFSMOD-W can significantly improve prediction of pesticide trapping efficiency compared with field slope and buffer width In STEP 4 exposure assessments, SWAN- VFSMOD can be used to quantify VFS mitigation efficiency from: Reichenberger et al. (2006) from: Reichenberger et al. (2006) Average buffer strip efficiency: 50% for 5m 90% for 10m 97.5% for 20m Scatter indicates that other factors than VFS-width drive buffer efficiency from: Sabbagh et al.. (2009)
VFSMOD-W Model Development
VFS Key Drivers Hydrologic Response VFS are complex dynamic systems! Driving Mitigation Infiltration Is governed by soil physical properties; vegetative cover; antecedent moisture content; rainfall intensity and inflow; slope Hydraulic resistance Is a function of vegetation type; Inflow volume Compound Sorption coefficient
VFSMOD-W: Model to Describe Reduction in Pesticide Transfer Across a Vegetated Filter Strip Predicted vs. measured reductions in pesticide transfer across vegetated filter strips (Sabbagh et al., 2009): - development (n=47; left-hand figure) - evaluation (n=120; right-hand figure) datasets
Benchmarking VFS models Mean Error in Buffer Reduction Efficiency Paetzold study / Germany
EU Vegetated Filter Strip Scenarios for STEP 4 FOCUSsw calculations
CORPEN audit EU STEP4 VFSMOD Scenarios STEP 4 VFS Scenario Project Development of European VFS scenarios representative for the FOCUS R landscapes to be used to parameterize the vegetated filter strip model VFSMOD-W in STEP 4 PECsw calculations. Project successfully finished final report available! Project Contractors: University of York & FERA Colin Brown, Wendy van Beinum University of Piacenza Ettore Capri, Marco Trevisan, Matteo Balderacchi
Software Development FOCUS PRZM Edge-of- field runoff VFSMOD-W SWAN Interception in VFS FOCUS TOXSWA Fate in surface water Parameter requirements?
Sensitivity Analysis Sensitive parameters Soil Existing analysis based on field experiments reported in the literature Two soil types and six pesticides with a range of different properties Insensitive parameters Vegetation - saturated hydraulic conductivity - spacing of stems - saturated water content - height Sediment - average diameter of particles - organic carbon content - clay content - hydraulic resistance Muñoz-Carpena et al. (2010). JEQ 39:630-641
Handling of insensitive vs. sensitive parameters Insensitive parameters; Assess likely range across the EU Propose default values relevant to the Step 3 scenarios Appropriate level of conservatism Documentation to justify selection from literature Sensitive parameters; GIS analysis within the framework of FOCUS Step 3 scenarios Generate distributions for each parameter Support selection of conservative values Facilitate higher-tier modelling (e.g. Probabilistic)
Probability distributions for Ksat R1 R2 R3 R4 N for each R scenario small Hydraulic parameters strongly correlated scenario n mean st.dev. R1 348 23.1 20.3 R2 69 98.6 77.5 R3 171 49.5 59.5 R4 222 53.9 56.3 Lognormal distribution k s as variable and area as density
Deriving conservative values for Ksat & θsat Separate simulations for: The four FOCUS R scenarios (n = 69 348) Storm events with 30 mm rain over 1 hour or 8 hours Pesticides with Koc of 100 L/kg or 10,000 L/kg Each run reads Ksat, θsat and θfc for one soil unit θfc used as fixed (and correlated) input for initial water content All other parameters held at constant values relevant to the FOCUS R scenario
Percentile of the frequency distribution Efficiency in removing pesticide varies with properties of soil, pesticide and runoff event 100 90 80 70 60 50 40 30 20 10 0 R3 40 50 60 70 80 90 100 % pesticide removal 8 hour event, Koc=10,000 8 hour event, Koc = 100 1 hour event, Koc = 10,000 1 hour event, Koc = 100
90 th Percentile Worst-Case Values for Hydraulic Properties in the VFS Scenario Parameter R1 R2 R3 R4 K s (m s -1 ) 7.04 x 10-7 2.79 x 10-6 9.25 x 10-7 1.52 x 10-6 θ s (cm 3 cm -3 ) 0.447 0.403 0.472 0.420 θ fc (cm 3 cm -3 ) 0.395 0.312 0.385 0.372
EU VFS Scenario Report Detailed methodology and results are published in Prof. Colin Brown et al s report http://www.york.ac.uk/environment/pesticides/
SWAN VFSMOD Development for Use in FOCUS STEP 4 Exposure Assessment
STEP4 SWAN&EU VFSMOD Scenarios New SWAN 3.0 feature: Conduct STEP 4 calculations with VFSMOD scenarios alternatively to L&M a) STEP4 with fixed 10 & 20m FOCUS L & M values b) Model buffer efficiency with VFSMOD (Prof. Munoz-Carpena) and VFS EU Scenarios for R1 R4 (Prof. Brown)
STEP 4 SWAN VFSMOD - Workflow From STEP3 read PRZM Outputs: ZTS and P2T 0 P2T Stream P2T Pond Apply Swash to: P2T Mitigation Run VFSMOD for each event & generate runoff volume, sediment mass & pesticide mass reduction factors Apply the mitigation factors to the P2T file (by event) generate P2T files Assumptions: 1 ha field with 100 m long buffer along the water body VFS parameterized according to respective EU VFS R scenario (Brown et al) FAO continuous moisture + mass conservation in VFS P2T Stream, Pond are based on SWAN algorithms
STEP 4 runoff mitigation with L&M Red = STEP3 no mitigation Blue line = STEP4 L&M L&M 20m
STEP 4 runoff mitigation with VFSMOD Red = no mitigation Blue = 5m buffer Green = 10m buffer 74% reduction (10m buffer, to max peak)
Testing VFSMOD-W Performance and Impact
Testing Strategy Two stage process; 1. Performance evaluation considering variation in effectiveness taking into account; Range of efate characteristics Range of rainfall / application timing conditions Differences in performance under a range of crop irrigation conditions 2. Technical testing Functionality tests for GUI Fidelity of response relative to pre-implementation datasets developed for performance evaluation in first stage testing
Test Datasets Basic performance test should consider; a range of e-fate characteristics (Koc 10-100) a range of different crops (maize, leafy vegetables, winter cereals) a range of rainfall / application timing conditions Substance 1 DT50 = 150, Koc = 10 Substance 2 DT50 = 150, Koc = 100 Substance 3 DT50 = 150, Koc = 1000 Spring application maize, R1 R4 maize, R1 R4 maize, R1 R4 Summer application vegetables, leafy (early, late), R1 R4, vegetables, leafy (early, late), R1 R4, vegetables, leafy (early, late), R1 R4, Autumn application winter cereals, R1, R3, R4 winter cereals, R1, R3, R4 winter cereals, R1, R3, R4 For each test case the impact of a 5, 10 and 20 m VFS was investigated Test comprises performance assessment for a total of over 6,100 run-off events.
Performance Pesticide Reduction Reduction Effectiveness (%) 5 m VFS 10 m VFS 20 m VFS Minimum 33.46 40.54 51.42 10th Percentile 62.11 77.03 97.23 Median 100.00 100.00 100.00 n = 6,141
Performance Scenario Comparison Illustrative comparison for 10 m VFS based upon simulations featuring all R scenarios; Leafy vegetables (early and late) Substance 1 Increasing vulnerability as follows: R2 < R1 < R4 < R3 Reduction Effectiveness (%) R1 R2 R3 R4 Minimum 66.80 89.45 40.54 52.83 10th Percentile 79.71 100 59.46? Median 100 100 100 100
Performance Crop and Timing Comparison Illustrative comparison for 10 m VFS based upon relatively vulnerable combination; Substance 1 (Kfoc = 10 L/kg) R3 scenario Similar responses between crops/timings Do not over-interpret results Run-off is event driven and weather conditions in each simulation are not directly comparable! Winter cereals (n=30 events) Maize (n=42 events) Leafy vegetables (n=49 events) Winter cereals Reduction Effectiveness (%) Leafy vegetables Maize (early) Leafy vegetables (late) Minimum 47.26 47.26 47.26 40.54 10th Percentile 62.72 71.26 64.54 56.84 Median 100 100 89.67 100
Comparison with Current Regulatory Defaults FOCUS L&M Working Group recommend the following representation for VFS mitigation effectiveness;
Comparison with Current Regulatory Defaults Defaults are based upon 90 th percentile(worstcase) observations in database of run-off studies as follows;
Comparison with Current Regulatory Defaults VFSMod test simulations are not directly comparable to the studies employed as a basis for these defaults Different soils (infiltration and antecedent moisture conditions) Different precipitation / run-on conditions Different vulnerabilities to run-off vs erosion Pragmatic assignment of values from limited database
Comparison Pesticide Reduction (10 m VFS) VFSMod (R1- R4) Reduction Effectiveness (%) FOCUS L&M (aqueous) FOCUS L&M (Sediment) FOCUS Defaults (10 th Percentile) Minimum 40.54 1.90 85.60 10th Percentile 77.03 60.70 86.80 Median 100.00 94.00 99.10
Comparison Pesticide Reduction (20 m VFS) VFSMod (R1- R4) Reduction Effectiveness (%) FOCUS L&M (aqueous) FOCUS L&M (Sediment) FOCUS Defaults (10 th Percentile) Minimum 51.42 14.10 93.20 10th Percentile 97.23 80.60 93.50 Median 100.00 97.00 99.80
Next Step: Linking Risk Management Runoff Mitigation Measures Use in-field buffers Position buffers to break up long/steep slopes Stopping and infiltrating water at source Establish talweg buffers Position buffer in areas of concentrated runoff formation Avoiding runoff in talweg positions Use riparian buffers Position buffers alongside water bodies Stopping runoff before entering ditches/streams/lakes Use edge-of-field buffers Position buffer at downslope field edge Stopping runoff from exiting field Manage field access areas Vegetate field access areas Stopping formation of linear runoff towards roads Establish hedges Position at downslope edge of field Stopping runoff from exiting field; providing wind shield and biodiversity benefits Establish/maintain woodlands Position at downslope/riparian landscape position Stopping runoff in landscape; providing wind shield and biodiversity benefits
Runoff Mitigation Measures Toolbox for Flexible Mitigation Mitigation measure toolbox: Soil management Cropping practices Vegetative buffers Reduce tillage intensity Manage tramlines Prepare rough seedbed Establish in-field bunds Use Crop rotation Do strip cropping Use in-field buffers Establish talweg buffers Use riparian buffers Use edge-of-field buffers Manage surface soil compaction Manage subsoil compaction Do contour tilling/disking Use annual cover crops Use perennial cover crops Manage field access areas Establish hedges Establish/maintain woodlands Retention structures Use edge-of-field bunds Establish Retention ponds/artificial Establish vegetated ditches wetlands Adapted use of pesticides Adapt application timing Optimize seasonal timing Adapt product and rate selection Optimized irrigation Adapt irrigation technique Optimize irrigation timing and rate
Publications: Fox et al. (2010). Influence of flow concentration on input factor importance and uncertainty in predicting pesticide surface runoff reduction by vegetative filter strips. Journal of Hydrology 384:164-173. doi:10.1016/j.jhydrol.2010.01.020. Jones et al. (2010). Modeling the Removal of Pesticides in Runoff by Vegetative Buffer Strips. Paper EC04C-4 presented at the SETAC Europe 20th Annual Meeting 23-27 May 2010, Seville, Spain. Muñoz-Carpena et al. (2010). Parameter importance and uncertainty in predicting runoff pesticide reduction with filter strips. J. Environ. Qual. 39(1):1-12 Poletika et al. (2009). Chlorpyrifos and atrazine removal from runoff by vegetated filter strips: experiments and predictive modeling. Journal of Environmental Quality, 38 (3) 1042-1052. Roepke et al. (2009): Modeling runoff mitigation capability of vegetated filter strips. Poster presentation at the Pesticide Behaviour in Soils, Water and Air Symposium; 14-16 September; York, UK. Sabbagh et al. (2009). Effectiveness of vegetative filter strips in reducing pesticide loading: Quantifying pesticide trapping efficiency. Journal of Environmental Quality, 38 (2) 762-771. Winchell & Estes (2009). A Review of Simulation Models for Evaluating the Effectiveness of Buffers in Reducing Pesticide Exposure. US EPA MRID No. 47773401. Paetzold study / Germany
Resources and SWAN Distribution VFSMOD Documentation http://abe.ufl.edu/carpena/vfsmod/index.shtml VFSMOD Publications http://abe.ufl.edu/carpena/vfsmod/citations.shtml EU VFSMOD Scenario Development Report (Brown et al, 2012) + SWAN Software download (November 2012) http://www.york.ac.uk/environment/pesticides/ Note: This site will be the official source for SWAN distribution! Updates will be posted here instead of the current email distribution.
With thanks to the co-authors Matteo Balderacchi Marco Trevisan Ettore Capri Colin Brown Bjoern Roepke, Horatio Meyer Neil Mackay Beate Erzgraeber Mark Greener Denis Yon Wendy van Beinum Sabine Beulke Rafael Muñoz-Carpena