研究期間：10208~10307;The aim of this 3-year research project is to improve the tropical rainfall potential technique (TRaP) for obtaining a better accuracy in typhoon rainfall potential estimation, especially for the usage in Taiwan area. The TRaP was proposed by Kidder et al.(2005) to estimate the 24-h rainfall potential by summing the rainfall rates retrieved from satellite microwave observations. The technique is easily used since its quite simple assumptions. On the other hand, it simple assumptions also could possibly cause huge errors. During the first year’s tasks, the considerations of the rainband rotation of typhoons and the terrain effect to typhoons were added into our improved TRaP (I-TRaP), and then significantly reduce the TRaP’s errors. During the second year’s tasks, the previous tasks will be continuously refined. Especially the sensitivity of the typhoon paths will be focused, and also an ensemble forecasting model will be established. Predicted typhoon paths from international operational weather agencies will be gathered and used to I-TRaP to estimate the rainfall potential, and then compared to the ground truths to understand the sensitivity in influencing the rainfall potential results. Meanwhile, the paths will be compared to the best tracks to evaluate their accuracy and give individual weightings to construct the ensemble forecasting models. The sensitivity analysis could be used for the future reliability evaluation for I-TRaP estimated results. Meanwhile, the ensemble rainfall potential could help us to obtain more stable rainfall forecasting.