Difference of Rainfall Distribution for Tropical Cyclones Over Land and Ocean and Rainfall Potential Derived from Satellite Observations and Its Implication on Hurricane Landfall Flooding Prediction

Current hurricane forecast models have very little skill to predict hurricane rainfall and inland flooding. It has been well known for years that the heavy rain and flooding of tropical cyclones over land bear a weak relationship to the maximum wind intensity. The Tropical Rainfall Measurement Mission (TRMM) based NASA Goddard Multi-satellite Precipitation Analysis (MPA) product is used to quantify the rainfall distribution in tropical cyclones that made landfall over the US Gulf coast during 1998-2004. A total of 37 tropical cyclones (TCs) were examined including 2680 3-hourly instantaneous MPA precipitation observations. Rainfall distributions for over-land and over-ocean observations are compared. It is found that the TC rainfall over ocean bears a strong relationship with the TC maximum wind intensity, while the relationship for over land conditions is much weaker. The rainfall potential is defined by using the satellite-derived rain rate, satellite-derived storm size, and the storm translation speed. In this study, the capability of the rainfall potential when the storm is over ocean to be used as a predictor for the storm's potential for inland flooding is examined. High correlations between the rain potential before landfall and the storm total landfall rain parameters are found by examining the 37 landfall TC dataset. Correlations are higher for the average rainfall potential during 1 day before landfall than averaging for other time period. A TC landfall rain index is introduced based on the rainfall potential study. This index can be used to predict the intensity of the TC landfall rain. Hurricane Katrina (2005) rainfall data is applied to verify the capability of using this index to forecast the TC landfall rain. The landfall rain forecast error for Katrina is within 10%.