Quantile-based short-range QPF evaluation over Switzerland

摘要: Quantitative precipitation forecasts (QPFs) are often verified using categorical statistics. The traditionally used 2×2 contingency table is modified here by applying sample quantiles instead of fixed amplitude thresholds. This calibration based on the underlying distribution and has beneficial implications for quantile difference debiased Peirce skill score split total error into complementary components bias pixel overlap. It shown that they provide a complete verification set with ability to assess full range rainfall intensities. technique enables potential in calibrated forecast be estimated without spurious influences from marginal totals problem hedging therefore avoided. To exemplify feasibility quantile-based contingencies, method applied 6.5 years operational Swiss Federal Office Meteorology Climatology (MeteoSwiss). Daily accumulations COSMO model at 7 km grid size compared high-quality gridded observational record spatially interpolated rain gauge data. scores single points predefined regions. A high-resolution climatology then built up reviewed terms typical characteristics model. seasonal QPF performance exhibits most severe overestimation over Northern Alps during winter, indicative impact ice phase. related updates, such as introduction prognostic scheme, also evaluated. demonstrated continuously increases subsequent versions Over entire time period, strong gradient evident Alps, meaning much higher Alpine south side than north side.