Latest Developments in the NWS Probabilistic Tropical Cyclone Storm Surge Model

The Meteorological Development Lab (MDL), in furtherance of the National Weather Service's mission to help protect life and property, developed a Probabilistic tropical cyclone storm Surge (P-Surge) model between 2003 and 2008 to provide tropical cyclone storm surge inundation guidance which accounts for uncertainty in the wind forecast. P-Surge uses a parametric wind model that incorporates forecast parameters, provided by the National Hurricane Center’s (NHC) official forecast, along with track, size, and intensity errors based on the official forecast's 5 year average errors. Thus P-Surge is based on the official forecast from NHC, which has historically outperformed other parametric wind guidance.

MDL overhauled P-Surge in 2014 by introducing gridded tidal predictions and updating the computational grids to improve the accuracy of the inundation guidance. Additional products were created to provide above ground level guidance. Finally, temporal products were produced to help users determine when they might be impacted by storm surge. MDL improved P-Surge between 2014 and 2018 by: creating hourly incremental 10, 20, 30, 40, 50 and 90% exceedance above datum products; creating cumulative 60, 70, and 80% exceedance above datum products; creating 6-hourly incremental and cumulative 90% exceedance above ground level products; extending the forecast to 102 hours (from 78) and the hind-cast to 24 hours (from 20); and providing NHC with coarser 2.5 km (vs 625 m) CONUS grids to assist with preparing NWS storm surge watches and warnings.

This talk will focus on P-Surge advancements since 2018. This includes developing ways to use the latest south Florida SLOSH basin, which has a significantly larger number of grid cells due to a higher resolution and broader extent. It also includes using different climatological error statistics based on whether the initial maximum wind speed of the storm is a tropical depression, tropical storm, or hurricane.