A tool to interpret the National Hurricane Center wind probability forecasts

The new wind probability forecast product from the National Hurricane Center has several advantages over the previous methods to communicate uncertainty of the tropical cyclone forecast. These advantages include providing the probabilistic threat from the entire wind field surrounding the tropical cyclone, providing the full range of probabilities, and providing the probabilities over the entire area that is forecast to be will be influenced by the tropical cyclone. The National Hurricane Center began issuing this product in 2005.

However, the new wind probability forecast product has a drawback. The operationally important probabilities change for the varying forecast intervals. For example, a 6% probability in the 12-hour forecast for 64 Kt sustained wind or greater would represent a low risk of occurrence. However, that same 6% probability for 64 Kt winds or greater in the 120-hour forecast represents a high risk of occurrence. While the wind probability product performs well and is technically accurate, this sliding scale of operationally important probabilities can hinder interpretation of the product by non-meteorological decision makers and the public.

A tool to help interpret the sliding scale of operationally important probabilities from the National Hurricane Center wind probability forecasts was developed. This tool uses the observed distribution of wind probability forecasts issued by the National Hurricane Center to quasi-objectively convert the numerical probabilities into five Probability Interpretation Categories (PIC): 1) very-low, 2) low, 3) moderate, 4) high, and 5) very-high. Tables to convert the probability forecast into the corresponding PIC were developed for each of the three wind speed thresholds (≥ 34, ≥ 50, and ≥ 64 kt), each of the seven forecast intervals (12, 24, 36, 48, 72, 96, and 120 hours), and for both the incremental and cumulative wind probability forecast products. A verification of this method was conducted on an independent set of forecasts and indicated that the method performs fairly well, however, refined tuning of the probability thresholds is needed. In addition, an entirely different method to account for the sliding scale of operationally important forecasts has been conceived and is planned for development by the Spring of 2011.