Start
April 12, 2019 - 3:00 pm
End
April 12, 2019 - 4:30 pm
Categories
Convective Meteorology (Mesoscale Dynamics)Using Machine Learning Applications and HREFv2 to Enhance Hail Prediction for Operations
Convective Meteorology (Mesoscale Dynamics)
Amanda Burke
April 12, 2019
3:00 pm/ NWC 5600
Abstract: Severe hail results in, on average, more than a billion dollars of damage within the continental United States each year. Thus, timely and accurate operational hail forecasts are vital to allow the public to take action. Machine learning (ML) is capable of synthesizing multiple datasets to obtain optimal predictions for severe weather hazards, including severe hail. In this study, hail is predicted using a random forest ML model with input from the High-Resolution Ensemble Forecast version 2 (HREFv2) and Maximum Estimated Size of Hail (MESH). MESH, a Multi-Radar Multi-Sensor (MRMS) product, was used as an observational dataset to train and verify the ML model.
Results suggest that the ML hail forecasts skillfully predict severe hail over spatially similar regions, as compared to observed SPC hail reports. Also, calibration of the ML predictions towards familiar forecaster output subjectively and objectively maps the predictions towards optimal probabilistic forecast output. Finally, regional analysis indicates there are subjective differences between a regionally trained model and one trained on all storms within the contiguous United States (CONUS).