DOES DOWNSCALING IN SPACE AND TIME DEGRADE THE DEPENDABILITY OF SEASONAL CLIMATE FORECASTS?

Authors: Schneider, Jeanne; Garbrecht, Jurgen

Submitted to: Proceedings of the World Water and Environmental Resources Congress
Publication Type: Proceedings
Publication Acceptance Date: 4/1/2006
Publication Date: 5/22/2006
Citation: Schneider, J.M., Garbrecht, J.D. 2006. Does downscaling in space and time degrade the dependability of seasonal climate forecasts? In: Graham, R., editor. Proceedings of the World Water and Environmental Resources Congress, May 21-25, 2006, Omaha, Nebraska. 2006 CDROM.

Interpretive Summary: The National Oceanic and Atmospheric Administration (NOAA) Climate Prediction Center issues total precipitation and average temperature forecasts for 3-month periods for relatively large areas. In order to include climate forecast information in farm level decision support systems, the forecast information needs to be downscaled to individual sites and daily time steps. Since dependability of seasonal climate forecasts is expected to depend on scale (specifically, more dependable for larger areas and longer time periods), there is concern that downscaling might dilute dependability to the point of non-utility. As an initial exploration, this study will determine any change in the dependability of precipitation and average temperature forecasts downscaled to monthly time steps for several locations in Oklahoma. Downscaling to daily time steps is done in two stages, first to monthly forecasts, and then to collections of daily forecasts. This initial study will examine only the first step, the monthly downscaling. Current forecast dependability for Oklahoma at the larger scales is limited, so any loss in dependability due to downscaling to monthly forecasts and specific locations should be readily apparent.

Technical Abstract: The National Oceanic and Atmospheric Administration (NOAA) Climate Prediction Center issues total precipitation and average temperature forecasts for 3-month periods for relatively large areas (forecast divisions are about 9,000 sq. km.). Incorporation of the climate forecasts into farm level decision support systems necessitates downscaling the forecasts to local and daily time scales. Since any measure of the skill of seasonal climate forecasts is expected to depend on scale (specifically, more skillful for larger areas and longer time periods), there is concern that downscaling will dilute current forecast skill, possibly to the point of non-utility. As an initial exploration, this study will determine any change in the dependability of precipitation and average temperature forecasts downscaled to local and monthly time scales for several locations in Oklahoma. Dependability is a skill measure that assesses only those forecasts that are discernibly different from climatology. Downscaling to daily time scales is done in two steps, first to monthly forecasts, and then to generated ensembles of daily forecasts. This initial study will examine only the first step, the monthly downscaling. In order to assess the relative impact of downscaling in space versus time, forecasts will be downscaled separately in each, and then downscaled in both. Current forecast dependability for Oklahoma at the larger scales is limited compared to regions of the U.S. that experience stronger ENSO impacts, so any degradation from downscaling to monthly forecasts and specific locations should be readily apparent.