High-Resolution Ensemble Forecasts of Hail using One-, Two-, and Three- Moment Microphysics Schemes with Verification Against Single- and Dual- Polarization Radar Data

Severe hail is a major weather hazard that results in injuries and more than a billion dollars in damages annually. Though modern convective-scale NWP forecast ensembles have shown skill in predicting convective storms, relatively few studies have focused on explicit short-term prediction of hail. The Severe Hail Analysis Representation and Prediction Project (SHARP) seeks to investigate and improve the capability of NWP ensemble analyses/forecasts to represent and predict hail in 0-2 hour forecasts using optimally-configured ensemble Kalman filter (EnKF) data assimilation (DA) with sub-kilometer grid spacing.

In this study, ensemble forecast experiments are performed using one-, two-, and three- moment microphysical schemes to investigate the sensitivity of hail forecasts to the choice and complexity of the microphysical parameterization scheme. The case study used a left-splitting supercell which occurred over Oklahoma on 10 May 2010, is chosen because it produced severe hail during an extended period of time. Ensemble forecasts are run at 500 meter grid spacing using the Advanced Regional Prediction System (ARPS); ARPS is a non-hydrostatic, mesoscale NWP model. Each forecast ensemble uses 40 members, and data from multiple sources (including WSR-88D radars, the Oklahoma Mesonet, soundings, and wind profilers) are assimilated into the model using the CAPS EnKF system (Wang et al. 2013). Two ensemble DA and forecasts were run using the Lin et al. (1983) single moment (SM) and the Milbrandt and Yau (2005, hereafter MY) double moment (DM) microphysics schemes, respectively. Ensemble analyses using the MY DM microphysics scheme were used to initialize both DM and triple moment (TM) ensemble forecasts. Gridded Maximum Estimated Size of Hail (MESH) and Hydrometeor Classification Algorithm (HCA) based on polarimetric radar observations are used to verify the ensemble forecasts of maximum hail size, and surface distribution.

Preliminary results suggest that the use of a multi-moment microphysics scheme improves the skill of 0-2 hour hail forecasts. The SM microphysics scheme ensemble over predicted both the maximum hail size and surface distribution, while the multi-moment microphysics scheme ensembles more closely represented observed maximum hail size and spatial distribution.