Severe hail, particularly in urban areas, can cause significant injury and millions of dollars in property damage. These impacts, however, could be significantly mitigated with more accurate and precise hail predictions based on 0-2 hour numerical weather prediction (NWP) forecasts. To provide such NWP forecasts requires advances in data assimilation (DA), the development of innovative ensemble forecast methods, and the application of novel data mining techniques to represent and predict hail within state-of-the-art NWP models. The Severe Hail Analysis, Representation, and Prediction project (SHARP) will focus on answering two scientific hypotheses: (1) assimilation of data from new and multiple data sources (including single- and dual-polarization Doppler radars, wind profilers, soundings, and surface observations) will improve hail representation within a NWP model, and (2) application of advanced data-mining techniques to NWP ensemble output will improve predictions of hail size and coverage. The specific goals of SHARP will be to: (1) produce and verify 0-2 hour ensemble forecasts of severe hail for a variety of events; (2) assess the ability of these forecasts to correctly represent hail as observed by dual-polarization Doppler radars and other instruments; and (3) accurately predict the geographic extent and size of hail reaching the surface, verifying the forecasts against high-quality observational data sets such as those produced by the NOAA Severe Hazards Analysis and Verification Experiment (SHAVE). Forecast ensemble output will be compared against radar-based extrapolation methods. The intellectual merit of this effort lies in progress toward application of advanced data assimilation (using multiple-moment microphysics and dual-polarization radar data) and data mining for short-term hail prediction. Though data mining has been applied to precipitation forecasts at the continental scale, far more novel is its use alongside advanced Ensemble Kalman Filter (EnKF) DA for convective-scale NWP. This effort builds upon previous work by the PIs in the fields of DA, storm-scale ensemble prediction, and data mining. This project is expected to advance the science of severe weather prediction, defining a paradigm linking data assimilation and data mining that could be applied to predict other convective-scale hazards such as downbursts and tornadoes. With sufficient computing resources, the techniques and algorithms developed in this project could be applied to support real-time severe weather warning operations, as envisioned in the NWS "Warn-on-Forecast" paradigm.
Broader Impacts of will include enhanced cross-disciplinary connections disciplinary and potential economic benefits to sectors including aviation that are especially vulnerable to hail. Results will aid in improving lead-time and user confidence in hail warnings, giving vulnerable industries increased opportunity to mitigate damage and individuals additional time to seek shelter. Collaboration with the National Severe Storms Laboratory will enable comparison and testing of results against operational, radar-based extrapolation methods and enable transfer of knowledge and tools to NWS forecasters responsible for severe weather warnings. The PIs will provide interdisciplinary training to graduate students, as well as undergraduate students through the OU/CAPS Research Experiences for Undergraduates (REU) program, with which the PIs have a strong history of involvement.
