Runoff Risk Advisory Forecast (RRAF) for Wisconsin Development & Production of a Real-Time Decision Support System for Wisconsin Manure Producers Dustin Goering North Central River Forecast Center (NCRFC) 2013 AWRA Spring Specialty Conference Agricultural Hydrology & Water Quality II St. Louis, MO
What is the NCRFC? and Water Quality? North Central River Forecast Center is 1 of 13 RFCs Covers 341,000 mi 2 over 9 states Staff of 19 responsible for 426 forecast points Real-time modeling of 1,173 sub-watersheds ranging from 6 3,061 mi 2 Primary focus is streamflow and river stage forecasts
What is the NCRFC? and Water Quality? North Central River Forecast Center is 1 of 13 RFCs Covers 341,000 mi 2 over 9 states Staff of 19 responsible for 426 forecast points Real-time modeling of 1,173 sub-watersheds ranging from 6 3,061 mi 2 Primary focus is streamflow and river stage forecasts NWS Strategic Plan Focus on helping decision makers where weather & water forecasts can have effect Address water quality not just quantity Leverage existing capabilities in new ways
RRAF Project Motivation Problem :: Excess contaminated field runoff can degrade water quality Well contamination, fish kills, Gulf of Mexico hypoxic zone, Great Lakes water/beach quality Important Issue 1.25 million Wisconsin dairy cows 34 million tons of waste annually Best Management Practices / Nutrient Management Plans focus on how & where Wisconsin Dept. Agriculture, Trade, and Consumer Protection (DATCP) Tasked to create advisory & notification system for state
RRAF Project Motivation Problem :: Excess contaminated field runoff can degrade water quality Well contamination, fish kills, Gulf of Mexico hypoxic zone, Great Lakes water/beach quality Important Issue 1.25 million Wisconsin dairy cows 34 million tons of waste annually Best Management Practices / Nutrient Management Plans focus on how & where Wisconsin Dept. Agriculture, Trade, and Consumer Protection (DATCP) Tasked to create advisory & notification system for state Goal :: Reduce nutrient loss in field runoff entering water bodies Utilize existing NWS weather & watershed modeling in water quality application Collaboration opportunity with federal, state, and university partners Complement BMPs/NMPs & producer knowledge Doesn t replace farmers responsibility
RRAF Project Motivation Problem :: Excess contaminated field runoff can degrade water quality Well contamination, fish kills, Gulf of Mexico hypoxic zone, Great Lakes water/beach quality Important Issue 1.25 million Wisconsin dairy cows 34 million tons of waste annually Best Management Practices / Nutrient Management Plans focus on how & where Wisconsin Dept. Agriculture, Trade, and Consumer Protection (DATCP) Tasked to create advisory & notification system for state Goal :: Reduce nutrient loss in field runoff entering water bodies Utilize existing NWS weather & watershed modeling in water quality application Collaboration opportunity with federal, state, and university partners Complement BMPs/NMPs & producer knowledge Doesn t replace farmers responsibility Action :: Focus on largest unknown When (when not) to Spread Warn when future runoff risk is high delay application reduce contaminated runoff Integrate land-atmosphere system for 10 day forecast more than is it going to rain tomorrow? Before this, no real-time guidance currently existed to aid decision process for producers
What is the RRAF? Derived from NCRFC operational models Lumped Snow-17 & Sacramento Soil Moisture Accounting Model (SAC-SMA) Incorporates 5 days future precip, 10 days future temps Issued 3x a day with forecast window out 10 days Event Based :: 3 conditions must be met SAC-SMA Interflow runoff > 0 SAC-SMA Upper Zone Tension Water Deficit = 0 Snow-17 Rain + snowmelt > 0 Forecast Precip & Temps NCRFC Soil & Snow Models Simulated Runoff Events DATCP Processing
What is the RRAF? Derived from NCRFC operational models Lumped Snow-17 & Sacramento Soil Moisture Accounting Model (SAC-SMA) Incorporates 5 days future precip, 10 days future temps Issued 3x a day with forecast window out 10 days Event Based :: 3 conditions must be met SAC-SMA Interflow runoff > 0 SAC-SMA Upper Zone Tension Water Deficit = 0 Snow-17 Rain + snowmelt > 0 Event list sent to DATCP They process events with thresholds Display risk on webpage they built and maintain Forecast Precip & Temps NCRFC Soil & Snow Models Simulated Runoff Events DATCP Processing
What is the RRAF? Derived from NCRFC operational models Lumped Snow-17 & Sacramento Soil Moisture Accounting Model (SAC-SMA) Incorporates 5 days future precip, 10 days future temps Issued 3x a day with forecast window out 10 days Event Based :: 3 conditions must be met SAC-SMA Interflow runoff > 0 SAC-SMA Upper Zone Tension Water Deficit = 0 Snow-17 Rain + snowmelt > 0 Event list sent to DATCP They process events with thresholds Display risk on webpage they built and maintain End Product Webpage indicating basin Low Med High runoff risk Available for 216 NWS watersheds in Wisconsin Forecast Precip & Temps NCRFC Soil & Snow Models Simulated Runoff Events DATCP Processing
What is the RRAF? Derived from NCRFC operational models Lumped Snow-17 & Sacramento Soil Moisture Accounting Model (SAC-SMA) Incorporates 5 days future precip, 10 days future temps Issued 3x a day with forecast window out 10 days Event Based :: 3 conditions must be met SAC-SMA Interflow runoff > 0 SAC-SMA Upper Zone Tension Water Deficit = 0 Snow-17 Rain + snowmelt > 0 Event list sent to DATCP They process events with thresholds Display risk on webpage they built and maintain End Product Webpage indicating basin Low Med High runoff risk Available for 216 NWS watersheds in Wisconsin Forecast Precip & Temps NCRFC Soil & Snow Models Simulated Runoff Events DATCP Processing Decision Support Regulatory Tool
Project Challenges Obvious Scale Difference Between Fields & NWS models Average NWS basin in Wisconsin = 301 mi 2 (9 1,800 mi 2 ) Using conceptual, lumped models calibrated for peak streamflow
Project Challenges Obvious Scale Difference Between Fields & NWS models Average NWS basin in Wisconsin = 301 mi 2 (9 1,800 mi 2 ) Using conceptual, lumped models calibrated for peak streamflow Communicate Assumptions to Users Not only information available for basing decisions This tool will never produce perfect prediction everywhere Combine local condition knowledge with forecast One farm may have runoff, the next one may not Rainfall and snowpack distribution, field aspect, etc.
Project Challenges Obvious Scale Difference Between Fields & NWS models Average NWS basin in Wisconsin = 301 mi 2 (9 1,800 mi 2 ) Using conceptual, lumped models calibrated for peak streamflow Communicate Assumptions to Users Not only information available for basing decisions This tool will never produce perfect prediction everywhere Combine local condition knowledge with forecast One farm may have runoff, the next one may not Rainfall and snowpack distribution, field aspect, etc. Long-term Success Dependent on: Will RRAF accurately predict average field scale conditions and runoff risk? Users build trust in the RRAF fewer contamination incidents?
RRAF Validation Compare observed & simulated historical runoff events 4 Edge-of-Field (EOF) sites 0.03 mi 2 vs. 230 mi 2 [0.01% NWS basin area] 7 Small USGS gauged Watersheds 15.9 mi 2 vs. 294 mi 2 [5.41% NWS basin area] Basin specific thresholds differentiate moderate and high risk
RRAF Validation Compare observed & simulated historical runoff events 4 Edge-of-Field (EOF) sites 0.03 mi 2 vs. 230 mi 2 [0.01% NWS basin area] 7 Small USGS gauged Watersheds 15.9 mi 2 vs. 294 mi 2 [5.41% NWS basin area] Basin specific thresholds differentiate moderate and high risk EOF Results Moderate & High Risk :: Hit = 80% Miss = 20% False Alarm = 71% High Risk :: Hit = 64% Miss = 36% False Alarm = 48%
RRAF Validation Compare observed & simulated historical runoff events 4 Edge-of-Field (EOF) sites 0.03 mi 2 vs. 230 mi 2 [0.01% NWS basin area] 7 Small USGS gauged Watersheds 15.9 mi 2 vs. 294 mi 2 [5.41% NWS basin area] Basin specific thresholds differentiate moderate and high risk EOF Results Moderate & High Risk :: Hit = 80% Miss = 20% False Alarm = 71% USGS Results Moderate & High Risk :: Hit = 62% Miss = 38% False Alarm = 45% High Risk :: High Risk :: Hit = 64% Hit = 41% Miss = 36% Miss = 59% False Alarm = 48% False Alarm = 19%
RRAF Validation Continued Average Both Scales Moderate & High Risk :: Hit = 71% Miss = 29% False Alarm = 58% High Risk :: Hit = 53% Miss = 47% False Alarm = 34%
RRAF Validation Continued Average Both Scales Moderate & High Risk :: Hit = 71% Miss = 29% False Alarm = 58% High Risk :: Hit = 53% Miss = 47% False Alarm = 34% Encouraging results overall However, for High Risk: hits and misses while false alarms What is being missed? Should increased misses cause concern? RRAF captures largest events while misses are much smaller in magnitude High Risk Hit/Miss Event Ratio = 9.6 Moderate & High Risk Hit/Miss Event Ratio = 7.7
RRAF Validation Continued Average Both Scales Moderate & High Risk :: Hit = 71% Miss = 29% False Alarm = 58% High Risk :: Hit = 53% Miss = 47% False Alarm = 34% Encouraging results overall However, for High Risk: hits and misses while false alarms What is being missed? Should increased misses cause concern? RRAF captures largest events while misses are much smaller in magnitude High Risk Hit/Miss Event Ratio = 9.6 Moderate & High Risk Hit/Miss Event Ratio = 7.7 Long-term behavior for all 216 basins over 50 year simulation :: Percent Time No Event simulated = 90% Percent Time Moderate Risk = 4% Percent Time High Risk = 6%
RRAF 2011 Verification Compare Edge-of-Field (EOF) runoff vs. Forecasts for 2011 9 Edge-of-Field sites in 4 NWS basins 0.03 mi 2 vs. 433 mi 2 [0.007% NWS basin area!] September December was unusually warm with much below snowpack in southern ⅔ of WI No observed runoff events past early September
RRAF 2011 Verification Compare Edge-of-Field (EOF) runoff vs. Forecasts for 2011 9 Edge-of-Field sites in 4 NWS basins 0.03 mi 2 vs. 433 mi 2 [0.007% NWS basin area!] September December was unusually warm with much below snowpack in southern ⅔ of WI No observed runoff events past early September Verification Results Moderate & High Risk :: Hit = 77% Miss = 23% False Alarm = 66% High Risk :: Hit = 74% Miss = 26% False Alarm = 54% Hit/Miss Ratio remains high = 10.2
RRAF Webpage Clickable basin interface overlaid on Google map Side bar with additional information How-to documentation, links to soil temps, state 590 maps Site News Working Group notes passed along to RRAF users
RRAF in Action Normal Mode = (3) 72-hour risk windows Only 1 event required in that time-frame to activate Highest risk wins (High > Moderate)
RRAF in Action Normal Mode = (3) 72-hour risk windows Only 1 event required in that time-frame to activate Highest risk wins (High > Moderate) Winter Mode uses entire 10-day forecast period Only uses High Risk or Winter Mode Snowmelt generated runoff indicated by hash-marks
RRAF in Action Normal Mode = (3) 72-hour risk windows Only 1 event required in that time-frame to activate Highest risk wins (High > Moderate) Winter Mode uses entire 10-day forecast period Only uses High Risk or Winter Mode Snowmelt generated runoff indicated by hash-marks Clickable Basins provide more forecast detail More forecast details to be provided to users in future
RRAF in Action Normal Mode = (3) 72-hour risk windows Only 1 event required in that time-frame to activate Highest risk wins (High > Moderate) Winter Mode uses entire 10-day forecast period Only uses High Risk or Winter Mode Snowmelt generated runoff indicated by hash-marks Clickable Basins provide more forecast detail More forecast details to be provided to users in future Positive reviews from surveys so far (0-10) Technical quality of product = 7.5 Ease of use = 7.9 Appropriate for NWS to provide this service = 100%
Next Steps Activating smaller secondary threshold to further reduce false alarms Analysis indicates between 6-20% decrease of smallest simulated events Minimal impact on percent of simulated hit events
Next Steps Activating smaller secondary threshold to further reduce false alarms Analysis indicates between 6-20% decrease of smallest simulated events Minimal impact on percent of simulated hit events Product Expansion Collaboration ongoing with Minnesota Department of Agriculture to replicate product in MN
Next Steps Activating smaller secondary threshold to further reduce false alarms Analysis indicates between 6-20% decrease of smallest simulated events Minimal impact on percent of simulated hit events Product Expansion Collaboration ongoing with Minnesota Department of Agriculture to replicate product in MN Continually briefing other agencies on the RRAF Very positive feedback from varied audiences on the product concept and application NCRFC eager to work with other states and universities in the future Open to have other partners expand and improve product different models?
Next Steps Activating smaller secondary threshold to further reduce false alarms Analysis indicates between 6-20% decrease of smallest simulated events Minimal impact on percent of simulated hit events Product Expansion Collaboration ongoing with Minnesota Department of Agriculture to replicate product in MN Continually briefing other agencies on the RRAF Very positive feedback from varied audiences on the product concept and application NCRFC eager to work with other states and universities in the future Open to have other partners expand and improve product different models? Summer 2013 Begin investigating capability of new model SAC-HTET (Heat Transfer & Evapotranspiration) Distributed model on 4-km grid
Questions? Dustin Goering dustin.goering@noaa.gov Brian Connelly Steve Buan brian.connelly@noaa.gov steve.buan@noaa.gov RRAF : http://www.manureadvisorysystem.wi.gov/app/runoffrisk NCRFC: http://www.crh.noaa.gov/ncrfc/