Fine Scale Hail Hazard Prediction using the WRF Model

Real time hail hazard prediction significantly helps property insurers in their mitigation and recovery efforts. A hail prediction system composed of data assimilation and a modified WRF model is presented. Several severe storms occurred during the spring of 2012 and 2015 and caused significant hail damage to property over Oklahoma and Texas. The storms occurred in a weak synoptic scale environment. The study investigates the feasibility and use of a 12-24 h fine horizontal resolution hail swath forecast to estimate the impact of hail to property.

Hindcast of the hail storms are performed at 250-500 m horizontal resolution with a triply nested Weather and Research Forecasting (WRF) model with the National Severe Storms Laboratory (NSSL) double moment microphysics scheme. To mimic an operational forecasting environment, the lateral boundary and initial conditions for the WRF model are provided by the 12-km North American Mesoscale (NAM) and 13-km Rapid Refresh (RAP) model forecast data and are enhanced with a three-dimensional variational (3-DVar) assimilation procedure.

The accumulated surface precipitation is realistically captured compared to observations. The probability of detection (POD) of hail computed using the hindcast hail swath data for runs with 3-DVar outperforms runs without 3-DVar assimilation. The POD scores decrease as the horizontal resolution increases due to a sensitivity of the POD to the spatial and structural errors of the predicted storms. Overall, the higher horizontal resolution coupled with a lead time of about 12-24 h provides a robust hail prediction system that produces hail hazard information for property damage models.