Exploring the use of physically based evaporative demand anomalies to improve seasonal drought prediction

Wednesday, 17 December 2014: 10:35 AM
Daniel Mcevoy1, Mike Hobbins2, Justin Lee Huntington1, John Mejia3, Christopher Hain4, Martha C. Anderson5 and James P Verdin6, (1)Desert Research Institute Reno, Reno, NV, United States, (2)National Integrated Drought Information System, Boulder, CO, United States, (3)Desert Research Institute, Reno, NV, United States, (4)Earth System Science Interdisciplinary Center, COLLEGE PARK, MD, United States, (5)USDA ARS, Pendleton, OR, United States, (6)USGS/EROS, Boulder, CO, United States
Providing reliable seasonal drought forecasts continues to pose a major challenge for scientists, end-users, and the water resources community. Precipitation is the most commonly used variable to assess future drought and water supply outlooks. However, studies have shown that precipitation forecasts at seasonal lead times are highly uncertain, and skill drops off dramatically past one month. Recent research has shown that satellite based evapotranspiration (ET) and land surface model based evaporative demand (Eo) anomalies accurately represent drought at different time scales, often capturing the onset of drought several weeks before precipitation-based drought metrics. Therefore, forecasts of accumulated Eo anomalies could provide additional early warning to water managers and the agricultural community. This study has two main objectives: 1) provide evidence that Eo anomalies can capture historical drought episodes, and 2) evaluate the predictive skill of a physically based Eo and individual forcings. Objective 1 is accomplished by comparing the Evaporative Demand Drought Index (EDDI) to the satellite energy balance based Evaporative Stress Index (ESI), and the United States Drought Monitor. EDDI is derived using physically based Eo estimates from the ASCE Standardized Reference ET equation, and forced with the North American Land Data Assimilation System /Parameter Regression on Independent Slopes Model hybrid data set at 4-km spatial resolution over the contiguous United States (CONUS). For objective 2, the National Center for Environmental Prediction Climate Forecast System (CFS) data products are used. A 28-year (1982-2009) reforecast of the CFS version 2 (CFSv2) and the CFS reanalysis is used to establish the climatology and historical skill of CFSv2 in predicting 2-m air temperature (maximum and minimum), 2-m specific humidity, 10-m wind speed, downwelling shortwave radiation at the surface, and ASCE Standardized Reference ET over the CONUS. The biases found in the historical analysis are used as correction factor for real-time CFSv2 runs. An archive of real-time CFSv2 seasonal forecasts initialized during the Fall of 2013, and with lead times of 0-9 months is used to forecast EDDI and evaluate CFSv2 skill during the drought-stricken winter of 2013-2014 in the western United States.