Hydrologic application of global emsemble precipitation forecasts

A promising source of information about future precipitation is ensemble weather forecasts. Such forecasts have been made operationally out to 16 days by NWS global numerical weather prediction models. To use these forecasts as input to hydrologic forecast models requires that the climatology of ensemble forecasts match the climatology of actual precipitation events. An archive of these products is available on a 2.5 degree grid, globally, beginning in early 1997. The conterminous U. S. portion of these data was abstracted for this study. Observed daily precipitation data from both the NCDC climatological network and the NWS operational network have also been used to estimate actual mean areal precipitation for each 2.5 degree grid element. Climatological monthly precipitation estimates from the PRISM project were used in the mean areal precipitation estimates to account for effects of orography and sharp climate gradients within each grid element. This paper presents preliminary comparisons of the forecast and observed climatological statistics important from a hydrological perspective. Also, approaches to adjusting ensemble products to make them hydrologically more useful will be discussed. Finally, the question of how to downscale the ensemble products to scales required by hydrologic forecast models will be considered.