SCALE-DEPENDENT Uncertainties In The Global QPF and Qpe From NWP Model and Satellite Fields

Abstract

Global precipitation forecasts from numerical weather prediction (NWP) models can be verified using the near-global coverage of satellite precipitation retrievals. However, inaccuracies in satellite precipitation analyses complicate the interpretation of forecast errors that result from verification of an NWP model against satellite observations. In this study, assessments of both a global quantitative precipitation estimate (QPE) from a satellite precipitation product and corresponding global quantitative precipitation forecast (QPF) from a global NWP model are conducted using available global land-based gauge data. A scale de- composition technique is devised, coupled with seasonal and spatial classifications, to evaluate these inac- curacies. The results are then analyzed in context with various physical precipitation systems, including heavy monsoonal rains, light Mediterranean winter rains, and North American convective-related and midlatitude cyclone–related precipitation. In general, global model results tend to consistently overforecast rainfall, whereas satellite measurements present a mixed pattern, underestimating many large-scale precipitation systems while overestimating many convective-scale precipitation systems. Both global model QPF and satellite-retrieved QPE showed better correlation scores in large-scale precipitation systems when verified with gauge measurements. In this case, model-based QPF tends to outperform satellite-retrieved QPE. At convective scales, there are significant drops in both model QPF and satellite QPE correlation scores, but satellite QPE performs slightly better than model QPF. These general results also showed regional and seasonal variation. For example, in tropical monsoon systems, satellite QPE tended to outperform model-based QPF at both scales. Overall, the results suggest potential improvements for both satellite estimates and weather forecast systems, in particular as applied to global precipitation forecasts.