In the early 1980s, there was a growing recognition that a probabilistic forecast should accompany the deterministic tropical cyclone track forecast issued by the National Hurricane Center (NHC). In 1983, the strike probabilities started being issued as an operational advisory product for selected coastal and island locations from 12 to 72 hours. Somewhat later, the probabilities for the 24, 48 and 72 forecast track positions were added. Finally, in 2001, an array of position probabilities at one degree resolution were being generated for the entire Atlantic basin. Except for periodic updating the track error statistics on which the probabilities was based, very little has changed in how the probabilities are generated.

Much has changed in the twenty years since these position probabilities were first issued. Because of the continuing increase in coastal development, decision makers need more than the probabilities that a tropical cyclone might pass within 65 nmi of their area. A more useful product is the probability of experiencing winds exceeding a particular threshold (for example, hurricane force), that takes into account the uncertainties in the storm track, intensity and size forecasts.

In an effort to meet this need, a new tropical cyclone probability program is being developed which takes in to account the position, intensity and size forecast errors to estimate the chance that a specified location will experience 34, 50, 64 and 100 kt winds within a specified time interval. This new program uses a Monte Carlo method that directly samples from the forecast error distributions of position, intensity and radii of these intensities. In this paper, we will discuss the program's methodology, the results for selected cases from the 2003 Atlantic and West Pacific hurricane seasons, conclusions and future research.