A Test Bed for New Seasonal Hydrologic Forecasting Approaches in the Western United States

H ydrologic extremes are costly to the nation. Annual U.S. drought and flood damages over the last decade have averaged between $6–$8 and $2 billion, respectively (FEMA 1995). Losses associated with the four-year 2000s drought in the western United States are likely to be in the tens of billions of dollars. To the extent that floods and droughts can be mitigated by management of water stored in reservoirs, improved streamflow prediction can help to reduce these losses. Potential economic benefits result from accurate hydrologic forecasts in years that are not hydrologically extreme as well. For instance, Yao and Georgakakos (2001) and Hamlet et al. (2002) have shown how hydropower revenues can be increased through incorporation of climate information in hydrologic forecasts, while Brumbelow and Georgakakos (2001) have shown the benefits of improved hydrologic forecasts to management of agricultural water supply. Despite the potential benefits of improved hydrologic forecasts, most operational hydrologic prediction at seasonal lead times and related water and energy management decisions are based on methods and data sources that have been in place for almost half a century. In particular, the primary operational method of seasonal and subseasonal streamflow forecasting in the western United States is regression of seasonal streamflow volume on indicator variables, primarily point observations of snow-water equivalent. This is especially the case for long-lead (e.g., monthly to seasonal) hydrologic forecasts that are the basis for hydropower and water supply management A TEST BED FOR NEW SEASONAL HYDROLOGIC FORECASTING APPROACHES IN THE WESTERN UNITED STATES