High Tech with a Human Touch
Photo by Hurrevac software
Wind Speed Probability Tool Analysis
The primary focus of the project is to examine the performance of the National Hurricane Center (NHC) Wind Speed Probability Forecast product. The 45th Weather Squadron makes extensive use of the NHC wind speed probability forecast product for advising operational customers in preparations associated with mitigating the risks from tropical cyclones at Cape Canaveral Air Force Station/Kennedy Space Center/Patrick Air Force Base (CCAFS/KSC/PAFB).
Specifically, the wind speed probability forecast work is intended to help customers make educated evacuation decisions, including timing for setting the various hurricane conditions for CCAFS/KSC/PAFB. An evaluation was conducted on the previous operational version of the NHC wind speed product but NHC introduced a new improved model for operational use in the 2010 hurricane season. The significant improvements in the 2010 wind speed probability forecast model mean that the old performance evaluation is no longer representative of the current operational product. This work is to document the performance of the current (new) NHC wind speed probability forecasts implemented during the 2010 hurricane season for tropical cyclones in the Atlantic Ocean, Gulf of Mexico and Caribbean Sea -- that threaten the United States and Antigua. Since performance analysis will likely be extended in future years as more hurricane seasons occur and larger sample sizes become available for improved performance evaluation, the Wind Speed Probability Forecast product will be designed for easy addition of additional tropical cyclones.
The project tasks to be accomplished include: Task 1: Verify the performance of the cumulative wind speed probability forecasts. This verification will include the following six (6) factors: 1) Verify the cumulative forecasts for all landfalling tropical cyclones threatening the Gulf Coast and the East Coast of the U.S., 2) Verify these forecasts for at least 2005-2011, 3) Include handling of the 'null' forecasts (0%), 4) Verify the forecasts for three time intervals: up to closest approach or landfall, after closest approach or landfall, and for the entire forecast period, 5) Determine if the performance for land threatening hurricanes for east central Florida centered on CCAFS/KSC/PAFB is significantly different from the entire Gulf and East U.S. coastline; if so, document the performance for East Central Florida; if not, document that the two areas had statistically similar performances, and, 6) Use the new National Hurricane Center forecast model implemented operationally in the 2010 hurricane season applied retroactively to previous seasons, if available. Landfalling tropical cyclones are those for which the NHC wind probability forecasts have values of 1% or greater for any of the forecast points located on the Gulf or East U.S. coast. Task 2: Document the performance of the new NHC interval wind speed probability forecasts with the same six factors as listed for the cumulative forecast documentation noted above in Task 1. Task 3: Determine and document if the performance of the new NHC cumulative and interval forecasts for landfalling tropical cyclones threatening Antigua, and nearby areas, is significantly different from the entire Gulf and East U.S. coastline, and if so, document the performance for the Antigua area, and if not, document that the two areas had statistically similar performances. Task 4: Develop, tune and verify a two-dimensional interpretation tool for the new interval and cumulative NHC wind speed probability forecasts. The two-dimensional tool will convert the probability forecast into five plain-language categories (very low, low, moderate, high and very high) based on the wind speed forecast probability at a point of interest and the probability relative to the rest of the forecast. The verification will be performed on data independent from the development and tuning work.
Our previous published work in this area can be found in Weather and Forecasting.