A Statistical Atlantic Hurricane Model to Forecast 6-hour and 24-hour Intensity Changes

Currently, predicting which tropical cyclones that will develop into a major hurricane is extremely difficult. Contemporary models are often inaccurate or delayed in predicting major hurricane development. Average 24-hour intensity forecasts produce errors on the order of 5 m/s, however rapid intensification prediction needs improvement since 5% of the 24-hour intensity forecasts have errors greater than 20 m/s. This study utilized data from all 103 Atlantic hurricanes (i.e. tropical cyclones with wind speeds greater than 33 m/s) from 1988 – 2003. Dynamic and thermodynamic parameters were calculated during the Rapid Intensification Period (RIP) of each storm, which is defined as the 24-hour period with the maximum pressure reduction. These parameters were calculated at 6-hour time intervals beginning at 60 hours prior to the RIP and ending at 24 hours after the onset of the RIP.

The statistical model is highlighted by a two-step discriminant function analysis (DFA). First, by examining the characteristics of the hurricane, the DFA identifies whether the hurricane will become a major or minor hurricane. Secondly, the DFA examines the proximity of the hurricane to its RIP. Once the DFA classifies the hurricane accordingly, a collection of multiple regression equations were formulated based upon the characteristics of previous hurricanes at similar stages in their lifecycle in relationship to their RIP. Multiple regression equations were then applied at each time interval to forecast the 6-hour and 24-hour changes in the maximum sustained winds and the pressure reduction in the eye. Independent case studies were used to test the accuracy of the model.