A Data Sensitivity Study of Hail Prediction Using a High Resolution Ensemble Forecast
In this particular study, an Observation Simulation Experiment (OSE) was used to evaluate the sensitivity of model hail forecasts to the assimilated data. Two long-track, severe hail producing storms were selected. Ensemble forecasts were run for each storm at 500 meter grid spacing using the non-hydrostatic, mesoscale model known as the Advanced Regional Prediction System (ARPS). For each forecast a 40 member ensemble was generated. Data from multiple sources, including the Oklahoma Mesonet, weather radar, soundings, and profilers were assimilated into the model using the Ensemble Kalman Filter (EnKF) method. Different combinations of these observations were added and removed from the forecast to determine the relative impact that each observing network had on the prediction of hail. The Maximum Estimated Size of Hail (MESH) grid and polarimetric radar data are used for verification of the forecasts.
Preliminary results from the project indicate that radar data provides the most significant improvement to forecasting hail. This is expected because the data has both a high spatial and temporal resolution and provides kinematic, microphysical, and thermodynamic information used to resolve precipitation. Data from dense surface networks, such as the Oklahoma Mesonet also had some significant impact on model forecasts due to their ability to provide critical low-level thermodynamic information.