Comparing the Distribution of Hail in Model Output to High-Resolution Hail Observations

Recent advances in near-term convective storm-scale numerical weather prediction have led to a significant interest in lengthening warning lead time through warn-on-forecast (WoF), with the goal of creating probabilistic guidance for tornadoes, hail, wind, and flash flooding. A number of studies have focused primarily on the predictability of supercell structure and tornado-like vortices; however, to accurately forecast all hazards associated with severe thunderstorms, the ability of storm-scale numerical models to represent (and forecast) these hazards will also need to be addressed. More specifically, to judge the feasibility of WoF for severe hail (diameter ≥ 2.54 cm), it will be necessary to compare the distribution of hail-related model fields to high-quality, detailed hail datasets that are not generally available in Storm Data due to inherent spatial and temporal limitations. This study compares a set of storm-scale model simulations using WRF-DART to the approximately 300 unique hail reports collected by the field project HailSTONE for the giant-hail producing supercell near Gotebo, Oklahoma on 23 May 2011. The representativeness of vertically integrated graupel and simulated radar reflectivity of hail in these simulations will be discussed, as will the necessity of using observational data with a similar resolution to the data produced by high-resolution convection-allowing models for experimental WoF verification.