Stochastic modeling of radar-derived maximum estimated size of hail for scenario-based hail loss estimation

Abstract

This study develops a stochastic method for simulating the maximum estimated size of hail (MESH) values at locations within a hail swath and conducts a scenario-based hail loss estimation at regional scale. The method is based on MESHmax (which is obtained by taking the maximum value of MESH data over a hailstorm per location) and hail insurance loss data for the June 13th^, 2020, July 2nd^, 2021, and August 5th^, 2024 Calgary hailstorms. The stochastic MESHmax modeling identifies the hail swath (centerline and surrounding points) using MESH data and characterizes MESHmax along the centerline of the hail swath and at off-centerline locations. In the proposed method, spatial correlations of the MESHmax values along the centerline and off-centerline locations are considered. In addition, an empirical vulnerability curve is developed by relating MESHmax to insurance losses for residential properties. The scenario-based hail loss estimation generates numerous realizations of regional MESHmax maps and integrates them with a hail vulnerability curve for residential properties that is derived from the insurance loss data of three recent hailstorms in Calgary. An illustrative hail loss estimation is performed by considering a hypothetical event similar to the June 13th^, 2020 Calgary hailstorm. Using the developed scenario-based hail loss estimation tool, the probability distribution of the regional hail loss can be obtained. The stochastic simulation of the 2020 hailstorm is capable of reasonably hindcasting actual loss in that event. The sensitivity analysis results highlight significant influences of spatial variability of MESHmax values and uncertainty of the insurance loss generations.