Analysis of Precipitation Forecasts From The NCEP Global Forecast System

Abstract

Satellite observations extend precipitation measurements from limited-area land surface analyses to a nearly global view of precipitation. As a result, we are able to verify quantitative precipitation forecasts (QPF) from global forecasts. In this study, QPF from the NCEP (National Centers for Environmental Prediction) global forecast system (GFS) have been analyzed with different lead times from 1 to 7 days during the period of October 2005 – September 2006. QPFs over major continents and oceans were composited to verify regional precipitation at various thresholds. The PERSIANN (Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks, Sorooshian et al. 2000) precipitation estimates were used as the observations. Using the PERSIANN products, forecast skill [e.g., root mean square error (RMSE) and equitable threat score (ETS)] and other verification metrics for the GFS QPF were compared for different seasons (winter/summer), regions (land/ocean), and zonal districts (tropical, subtropical, and midlatitudes). Daily QPF for the year demonstrated widely varying forecast skill over different subregions. Since upstream weather over the Pacific Ocean is critical for predicting weather over the continental United States (CONUS), it is important to know both the predictability of QPF and what forecast skill precipitation forecasts process over the ocean. Ebert et al. (2007) compared different nearreal- time satellite precipitation estimates and used the satellite data to verify short-range precipitation forecasts from mesoscale models over several continents. As an alternative observation data, the NCEP CMORPH (CPC MORPHing technique, Joyce et al.2004) satellite precipitation estimates will be used to study the GFS QPF in the future. Uncertainties in the observations are also needed to be considered, since uncertainties in observation data greatly affect forecast skill in verification. We discuss application of this study to future research for projects related to THORPEX (The Observing-System Research and Predictability Experiment), a long-term research program organized under the World Meteorological Organization's World Weather Research Program, and the next-generation global models, such as Finite-volume Icosahedral Model (FIM) at NOAA/ESRL/GSD.