The work being conducted for this project is a continuation of an effort started under a North Carolina State University Cooperative Project. In the earlier project, a model was developed to predict precipitation amounts from East Coast hurricanes making landfall. The model predicts mean precipitation within bins which are part of a variable grid system and is based on 15 hurricanes, each of which were tracked for 24 hours after making landfall (or until reaching tropical depression or extratropical stage). Hurricane track forecasts for the model were provided by the National Hurricane Center. The results from model tests using data from Hurricane Hugo indicated that the model rainfall predictions were most accurate along the storm track, while the model tended to overpredict rainfall amounts to the right of the track.
The objectives of this later project were to develop a model for Gulf Coast hurricanes, to improve the East Coast model, if possible, and to develop a model that combines East Coast and Gulf Coast hurricanes. The Gulf Coast model was based on 25 hurricanes which were added to data from 16 East Coast hurricanes to develop the combined model. Both the Gulf Coast and combined model were tested using rainfall totals from Hurricane Andrew. Overall, the Gulf Coast model provided the lowest median of the absolute differences between the predicted and observed rainfall totals, had a higher R-squared value, and an isohyetal distribution that was consistent with the actual distribution. By varying individual hurricane characteristics one at a time, it was found (not unexpectedly) that the forward speed was the most important variable in the determination of the storm total precipitation. Also of importance was the diameter of the largest closed isobar and the heading of the hurricane.