Adding Confidence to Seasonal Drought Forecasts using reference evapotranspiraiton anomalies
Download as PPTX, PDF1 like1,063 views
The document discusses using reference evapotranspiration (ETo) anomalies as an additional tool for improving seasonal drought forecasts. It finds that ETo forecasts often have greater predictive skill than precipitation forecasts alone, especially in certain regions and seasons. A case study of the 2002 Southwest U.S. drought showed ETo predictions better captured the spatial extent and severity levels of drought compared to precipitation forecasts. Overall, the use of ETo anomalies could help add confidence to drought outlooks when combined with other drought indicators in seasonal forecasts.
Adding Confidence to Seasonal Drought Forecasts using reference evapotranspiraiton anomalies
- 1. Adding Confidence to Seasonal Drought Forecasts Using Reference Evapotranspiration Anomalies U.S. Drought Monitor Forum April 15th, 2015 Reno, NV Dan McEvoy, Graduate Research Assistant and PhD Candidate Desert Research Institute and University of Nevada, Reno Photo credit: Steve Carver
- 2. What is evaporative demand (E0)? (Figure: http://www.fao.org/docrep/s2022e/s2022e07.htm) • Evapotranspiration rate that could occur given an unlimited water supply • Potential evapotranspiration • Atmospheric demand • Reference evapotranspiration • Often estimated using temperature alone, but a physically based model should be used Physically-based E0 contains valuable information related to drought dynamics “ATMOSPHERIC THIRST”
- 3. Reference Evapotranspiration (ETo) ASCE Standardized Reference ET (Allen et al. 1998, 2005): 𝐸𝑇𝑜 = Δ Δ + 𝛾(1 + 0.34𝑢2) 𝑅 𝑛 − 𝐺 𝜆 + 𝛾 Δ + 𝛾 1 + 0.34𝑢2 900 𝑇 + 273 𝑢2(𝑒𝑠 − 𝑒 𝑎) Advective component Radiative component Rn = net radiation (shortwave + longwave) G = ground heat flux (assumed to be zero) T = mean daily temperature U2 = mean daily wind speed at 2-m es = saturation vapor pressure ((es_tmax+es_tmin)/2) ea = actual vapor pressure (from q and surface pressure) λ = latent heat of vaporization Δ = slope of saturation vapor pressure-temperature curve γ = psychrometric constant Cn = 900 (grass reference) Cd = 0.34 (grass reference) Gridded weather/climate data: • Maximum and temperature (2 m) • Specific or relative humidity (2 m) • Wind speed (10 m) • Downward shortwave radiation at surface
- 4. Evaporative Demand Drought Index (EDDI) (EDDI; Hobbins et al. 2011, McEvoy et al 2014) • EDDI: same standardization procedure as SPI with ETo as only input • Based on the historical probability distribution for a given accumulation window • EDDI identifies current western “snow drought” despite above normal precipitation in some locations. Represents previous 6 months, October – March • 4-km spatial resolution • Daily temporal resolution • METDATA, 1979-present • (Abatzoglou 2011) • http://metdata.northwest knowledge.net/)
- 5. EDDI and Snow Drought Red: <50%, Brown: 50-69%, Yellow: 70-89%
- 6. EDDI as an Early Warning Indicator: 2012 Evaporative Stress Index (1-month ESI) Standardized Soil Moisture Index (1-month SSI) Standardized Precipitation Index (1-month SPI) 1-month EDDI (ESI data provided courtesy Martha Anderson and Chris Hain) USDM
- 7. Motivation: Poor Skill in Precipitation Forecasts From Saha et al. 2014: “Except for the first month (lead 0), which is essentially weather prediction in the first two weeks, there is no skill at all, which is a sobering conclusion” Other studies also highlight this problem: e.g., Lavers et al. 2009, Yuan et al. 2011, Yuan et al. 2013, Wood et al. 2015
- 8. Motivation: Seasonal ETo Prediction Previous research : • Tian, D. and C. J. Martinez, 2014: Seasonal prediction of regional reference evapotranspiration based on Climate Forecast System version 2. J. Hydrometeor., 15, 1166-1188. Moderate skill, cold seasons No skill, warm seasons • Only AL, GA, and FL • No evaluation of humidity • No drought perspective
- 9. (Karl and Koss, 1984) CONUS Scale Analysis: Seasonal ETo Prediction 9 NCDC Climate Regions (Karl and Koss, 1984) • CFSv2 Reforecast • CFSR used as baseline observations • Monthly mean Tmax, Tmin, specific humidity, wind speed, solar radiation, and precipitation • Monthly mean ETo • Deterministic skill: temporal anomaly correlation (ensemble mean) • Probabilistic skill: Heidke Skill score during drought events
- 10. CFSv2 Reforecast Data (CFSRF) (CFSv2: Saha et al., 2014) • Reforecast period of record: 1982-2009 monthly means • Spatial resolution: 1° x 1° • Lead times used: 1 to 9 months
- 11. Season 1 ETo vs. PPT Regional Skill • ETo forecasts almost always contain greater skill than PPT • Certain regions and seasons contain a large skill gap Lead 1-month, Season 1 AC
- 12. Case Study: 2002 Southwest Drought • Forecasts initialized in March, 2002 • Spatial extent better predicted by ETo • Severity poorly forecast by both metrics in southwest
- 13. Case Study: 2002 Southwest Drought • ETo better predicted D0- D4 and D1-D4 percentages • Poor prediction of severe to exceptional drought, both ETo and PPT Red = ETo Blue = PPT Percent area of southwest region in given drought categories: AMJ 2002
- 14. ETo Predictability and ENSO ENSO event: Oceanic Niño Index (ONI) >1 or <-1 Reduce noise from weak ENSO signal 76 events (months) of 336 possible (ONI: http://www.cpc.ncep.noaa.gov/products/analy sis_monitoring/ensostuff/ensoyears.shtml)
- 15. ETo Predictability and ENSO
- 16. Conclusions and Ongoing Efforts • Skill analysis with second set of observations (NARR or ERA-40) • Produce experimental real-time seasonal ETo forecast from CFSv2 • Apply analysis to NMME Phase II output Thank You! Questions? mcevoyd@dri.edu The use of ETo anomalies as a seasonal forecast tool could add confidence to drought predictions when used in addition to PPT and other drought metrics. NMME: http://www.cpc.ncep.noaa.gov/prod ucts/NMME/
- 17. CFSR E0 and Precipitation Anomalies Climate Forecast System Reanalysis (CFSR) (CFSR; Saha et al. 2010)