Evaluation of WRF Model Performance for Hailstorms in Southern France Using DPR-GPM Data

Abstract

Understanding and better predicting severe meteorological phenomena such as hailstorms is increasingly important because of their capacity to cause major socioeconomic losses. We analyzed seven case studies with confirmed hail on the ground. To do so, we used four microphysical parameterizations of the Weather Research and Forecasting (WRF) model (horizontal resolution 3 km) that incorporate hail formation. Predictive variables for hailstorm severity were extracted from the Dual-Frequency Precipitation Radar-Global Precipitation Measurement (DPR-GPM) sensor, a novel and effective tool for monitoring extreme events. The main objective was to evaluate WRF model data by comparing them with DPR-GPM observations during hailstorm days in southern France. For this, the SAL score was used, an object-oriented method that robustly evaluates the simulations. To deepen the analysis, other metrics were calculated. The selection of microphysical parameterizations for simulating hail is challenging because of their specific assumptions and dependence on other model parameterizations. However, for southern France, the best parameterization for simulating maximum reflectivity in the vertical profile and vertically integrated precipitation water in the solid phase was the Milbrandt-Yau Double-Moment 7-class. For vertically integrated precipitation water in the liquid phase, it was the WRF Single-Moment 7-class (WSM7), with WRF Double-Moment 7-class (WDM7) standing out in the localization of all three variables. In contrast, the NSSL 2–Moment 7-class scheme generally showed poorer performance in hail prediction and simulation of the analyzed variables. Identifying optimal parameterizations for hailstorm prediction in southern France will enhance future simulations of extreme weather in the Mediterranean.