Detection of hail signatures from single-polarization C-band radar reflectivity

Five different criteria that estimate hail signatures from single-polarization radar data are statistically evaluated over a 15-year period by categorical verification against loss data provided by a building insurance company. The criteria consider different levels or thresholds of radar reflectivity, some of them complemented by estimates of the 0 °C level or cloud top temperature. Applied to reflectivity data from a single C-band radar in southwest Germany, it is found that all criteria are able to reproduce most of the past damage-causing hail events. However, the criteria substantially overestimate hail occurrence by up to 80%, mainly due to the verification process using damage data. Best results in terms of highest Heidke Skill Score HSS or Critical Success Index CSI are obtained for the Hail Detection Algorithm (HDA) and the Probability of Severe Hail (POSH). Radar-derived hail probability shows a high spatial variability with a maximum on the lee side of the Black Forest mountains and a minimum in the broad Rhine valley.