Hybrid Statistical–Dynamical Probabilistic Prediction of Hurricane Landfall Winds

There is a growing need for-high resolution predictions of landfall impacts from hurricanes. Motivated by the needs of our clients in the energy and insurance sectors, Climate Forecast Applications Network (CFAN) has developed a hybrid statistical-dynamical technique to produce probabilistic forecasts of landfall winds at a resolution of 5 miles. The technique integrates the global model forecast ensembles from ECMWF and NCEP with a radial wind model, and NHC operational evaluation of maximum wind and R64. The global models provide forecasts of hurricane tracks and intensity, which are calibrated using model reforecasts (hindcasts) and historical track/intensity information. Coarse-resolution global model wind forecasts are downscaled using a radial wind model. The forecasted 2D wind fields are calibrated using forecasts of intensity (maximum wind) and R64. Real-time verification of intensity forecasts from the calibrated ECMWF and NCEP ensemble members plus the NHC forecasts provides the basis for selecting intensity forecasts to be used in 2D wind calibration. Forecasts are presented as instantaneous, hourly 2D wind fields and also swaths of 2D maximum wind fields. Uncertainty information from the ensembles is presented using interpercentile ranges. Forecasts are displayed using an interactive zoomable tool that enables the user to anticipate potential impacts. Examples and verification statistics are provided from the 2017, 2018 and 2019 Atlantic hurricane seasons.