7.3
Using Probabilistic SLOSH Output to Improve Storm Surge Forecasting

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Thursday, 10 January 2013: 9:00 AM
Using Probabilistic SLOSH Output to Improve Storm Surge Forecasting
Room 18B (Austin Convention Center)
Erik R. Nielsen, Texas A&M University, College Station, TX; and J. Rhome

This research tests the viability of using probabilistic data derived from an ensemble of the Sea, Lake, and Overland Surges from Hurricanes (SLOSH) model, known as Probabilistic Storm Surge (P-Surge), to improve the storm surge forecasting at the National Hurricane Center (NHC). Specifically, this research aims to develop criteria for using P-Surge exceedance guidance in the creation of more accurate storm surge forecasts. The SLOSH P-Surge exceedance guidance has been evaluated using storm surge observations from Hurricanes Rita, Katrina, Wilma, Gustav, Ike, and Irene. Given the spatial irregularity of storm surge observations, ArcGIS was used as a means to accurately compile and relate the model output to observations obtained from the National Oceanic and Atmospheric Administration (NOAA), the Federal Emergency Management Agency (FEMA), and the United States Geologic Survey (USGS).

These observations were then used to evaluate the accuracy of P-Surge exceedance levels over all advisories in which P-Surge was available through landfall. RMSEs were used to compare the accuracy of higher probability exceedances as landfall approached in an effort to determine if P-Surge can be used as a viable, real time forecasting tool. The statistics from each storm were compiled to determine which exceedance is the most accurate at various lead times before landfall. Furthermore, if the most accurate exceedance varied relative to time before landfall, an average rate of migration was determined across all storms examined. Using current forecasting techniques as a baseline, the new forecasts were evaluated for any improvement or degradation of skill.

In this presentation the results and forecasting applications will be discussed. A methodology for using P-Surge exceedances to improve real-time storm surge forecasts is provided. The results presented here have the potential to produce more accurate storm surge products and forecasts at NHC. The research was enabled by the NOAA Office of Education Ernest F. Hollings Program.