论文部分内容阅读
The Tropical Rainfall Potential (TRaP) technique presented by Kidder et al. In 2005 is a useful method to predict the accumulated typhoon rainfall. However, one of the traditional main shortcomings of TRaP is its lack of considering the topography influence. This is important as the interaction between high mountains and typhoon circulations is a key factor that affects the typhoon rainfall distribution. Thus, it is crucial to take account the terrain effect when applying the TRaP method for mountainous areas. This study is based on the TRaP technique but further considers the topographic effect with three steps: “rainfall re-distributing”, “total rainfall amount re-adjusting”, and “seasonal atmospheric circulation classifying”. SSM/I-derived rainfall data was used to construct the TRaP and to estimate the new rainfall patterns, which was subsequently validated with rainfall station observations. The results showed the improved TRaP technique could indeed obtain a much better performance than the original TRaP method. For the 22 dependent cases in Taiwan, the mean correlation coefficient increased from 0.07 to 0.66, and the RMSE decreased from 196.4mm to 130.8mm, respectively. The mean of the observed rainfall was 196.4mm. For the devastating 2009 Typhoon Morakot, the correlation rose from 0.37 to 0.91, and the RMSE dropped from 497.0mm to 297.2mm. The mean of the observed rainfall was 666.4mm. These results notably show that the TRaP technique can be significantly improved by additionally considering the terrain effect and seasonal circulation