NowPrecip 2: Precipitation nowcasting in the complex terrain of Switzerland. Part II: Verification

Abstract NowPrecip is a real‐time nowcasting application, operational at MeteoSwiss since 2020. A new version has introduced a novel ensemble‐based approach designed to generate real‐time multi‐member precipitation forecasts and improve rainfall‐related severe weather warnings in Switzerland. This study provides a comprehensive verification of the NowPrecip ensemble by comparing its performance with precipitation observations from five C‐band dual‐polarization radars operated by MeteoSwiss. In addition, the Python‐based Short‐Term Ensemble Prediction System (PySTEPS), a widely used open‐source nowcasting framework, is employed to generate ensemble nowcasts, providing a valuable benchmark for NowPrecip performance. Both PySTEPS and NowPrecip nowcasting simulations are blended with members of the numerical weather prediction (NWP) ensemble derived from the Icosahedral Nonhydrostatic Model (ICON). To this end, we compute widely used deterministic and probabilistic verification metrics across a 3.5‐year dataset over the period 2020–2023. The results show that NowPrecip systems consistently outperform PySTEPS across skill scores, while deterministic verification demonstrates that NowPrecip methods exhibit superior performance compared with the current operational NWP ensemble for lead times up to four hours.