Rapid hail stone characterization: a 3d computer vision shape analysis model.

Hail stone size distributions are an important predictor for damage intensity. However, accurately modeling hail swaths and simulating hail strikes remains challenging due to both a lack of information regarding the shape of the stone distribution and reporting bias in hail sizes. This study incorporates an analysis of nearly 100 hail pad measurements, each co-located with Understory’s network of ground-based weather sensors. We present a 3d computer vision model that scans the surface of each hail pad to quickly and accurately resolve stone shape, size, anisotropy, density, and ultimately the kinetic energy of impact from each stone imprint. The resulting distribution of hailstones empirically validates the accuracy of Understory’s hail sensors and allows us to optimally characterize hailstorm types, with important implications for improved reporting accuracy and damage likelihood assessment.