THE ROLE OF CLIMATE VARIABILITY IN OPERATIONAL WATER SUPPLY FORECASTING FOR THE WESTERN UNITED STATES

Abstract

The single greatest uncertainty in seasonal water supply forecasts is the amount of precipitation falling after the forecast issue date. There has been a long history of attempting to incorporate seasonal climate forecasts into operational water supply forecasts. The skill of these precipitation forecasts remains low especially compared to highly confident snow-based streamflow forecasts. Early in the season (e.g., September-December), however, large-scale climate indices are the best available predictors of future water supplies. This dissertation suggests practical methods for issuing climate-based operational streamflow forecasts.This study also documents the existence of strong decadal trends in water supply forecast skill. Across the Western US, 1 April forecast skill peaked in the 1960-1970s and has been on the decline more recently. The high skill period was a very calm period in the Western US, with a near absence of extreme (wet or dry) spring precipitation events. In contrast, the period after 1980 has had the most variable, persistent, and skewed spring and summer streamflows in the modern record. Spring precipitation is also now more variable than it has been since at least the 1930s. This rise in spring precipitation variability in the Colorado/Rio Grande Basins and the Pacific Northwest is the likely cause behind the recent decline in water supply forecast skill.