Validation of Tropical Rainfall Potential (TRaP) Forecasts for Australian Tropical Cyclones

Tropical Rainfall Potential (TRaP) forecasts provide estimates of 24 h rainfall accumulation in landfalling tropical cyclones based on the advection of a field of satellite-estimated precipitation. Validation of TRaP forecasts for five Australian tropical cyclones during the 2003-04 season showed significant skill in predicting heavy rainfall. The predictions of maximum rain at landfall compared well with gauge observations in most cases. In terms of spatial rain coverage and amount, the TRaPs based on data from the Advanced Microwave Sounding Unit (AMSU) performed noticeably better than those based on the Special Sensor Microwave Imager (SSM/I), giving higher correlations with the observations, more accurate estimates of rain area and conditional rain rate, and lower root mean squared errors. The TRaPs performed neither better nor worse than mesoscale numerical weather prediction models. A decomposition of the TRaP error for regions of heavy rain suggests that only a small portion was related to errors in the track forecasts. Pattern errors, which relate to the shape, size, and fine scale structure of the forecast, accounted for about half of the total error, while rain volume error was about one third of the total error. These relate to errors in the satellite rain rate retrieval as well as the assumption of a steady state rain pattern. An ensemble of TRaP forecasts could account for some of these uncertainties, leading to more useful objective guidance.