3 OBJECTIVE EVALUATION OF GLOBAL PRECIPITATION FORECAST

Numerical Weather Prediction (NWP) model quantitative precipitation forecasts (QPF) have been one of interest to forecasters (and public users) as they influence daily life. The rapid increases in model resolution are proportional to the increases in computer resources. The development of precipitation observation networks and the improvement in measuring techniques are greatly promoting the accuracy of observed precipitation. It is possible to build a model precipitation verification system based on high quality observation from the Continental United States (CONUS) area. The model precipitation forecast skills are more varied (sensitive) to different verification techniques, different spatial and temporal resolutions, especially model biases. Somehow, the forecast skills are misleading when the model forecasts are strongly biased. The subject of model bias and bias-free verification will be discussed in the separated paper. The past 4-year NCEP/GFS QPFs have been evaluated by different measures. The skills are improved gradually for past years when model increases the resolution, improves analysis system and physical processes. The winter seasons have more skills than summers. The lower spatial, temporal resolutions have more skills than the higher spatial, temporal resolutions