Gale-Forced Storm Prediction Over The Oceans

Forecasts for severe weather and individual storm warnings have steadily been improving over land however, the same cannot be said for these systems over the ocean.  Due to the lack of radar input, it is more difficult to provide accurate warning over the ocean for navigational hazards. Geostationary data and numerical weather prediction (NWP) models are the primary tools utilized by forecasters when outside of radar range. 

This Geostationary Operational Environmental Satellite-R (GOES-R) project will help improve the ability of forecasters to provide warnings to mariners.  The proposed algorithm is similar to the ProbSevere model, developed by the Cooperative Institute for Meteorological Satellite Studies (CIMSS), which calculates the likelihood that a particular radar-identified precipitation cell will produce severe weather (wind speed > 50 knots or hail > 1 inch) in the next 60 minutes.  ProbSevere collocates precipitating cells with the glaciation rate and growth rate from GOES IR imagery, the maximum expected size of hail (MESH) from radar, and the effective bulk shear and convective available potential energy (CAPE) from NWP. In the future, lightning intensity will also be a predictor.  Our goal is to create a similar model but without the help of radar.  Instead, storms will be identified based on geostationary data, and including lightning data from the GOES-R Lightning Mapper (GLM).  The ProbSevere product for oceans will calculate the likelihood that the storm will produce gale-forced winds (34 kts), instead of meeting the severe criteria for storms over land. A database of high-wind events observed from buoys has been created to train the Bayesian ProbSevere model over oceans. Real-time conditions are compared to the historical events in the database to determine the likelihood of the storm having high surface wind speeds.  This product will help forecasters to improve predictability of potentially hazardous storms over oceans.