| 摘要: | 近十多年來颱風路徑預報之進步是大家有目共睹的，尤其數值預報模式在颱風路徑預報上更是逐年有所進步。根據中央氣象局公佈2006年18個颱風之24、48及72小時颱風位置預測的平均誤差分別為104、194及304公里，比過去3年之平均誤差130、230及342 公里，相較之下其準確率有明顯提升，而根據大陸中央氣象台之消息，今年其24小時、 48小時和72小時颱風路徑預報誤差分別為112公里、194公里和318公里，中央氣象局明顯較優，且今年(2008年)截至9月底為止，氣象局的24及48小時颱風路徑預報準確度領先其他國家。但是今年(2008年)七月卡玫基侵台時，降雨超乎預期，重創中台灣，造成農漁業損失嚴重，引發民眾質疑氣象局預報不準確。由中央氣象局颱風降雨及強度預報來看，並沒有跟上路徑預報進步的腳步，其進步的程度相對的緩慢，因此準確的颱風降水預報是民眾的期望，也一直是科學界努力之目標。由於颱風主要的生命期多半是處在廣大的洋面上，而這些地方都是最缺乏完整觀測資料的區域，因此中央氣象局預計需再花費上千萬之預算購置海上觀測儀器，然而氣象衛星觀測儀器多，觀測區域完整，且資料接收不需花費昂貴之經費，在颱風研究方面的應用便更加顯得重要。對於侵台前之颱風降水預報，氣象雷達掃描之區域有限，而模式亦有初始化及參數化之問題，無法很準確地預測強烈的降水，故發展被動式微波輻射計估算降水預報是刻不容緩之議題。最近Kidder等(2005)使用SSM/I(Special Sensor Microwave /Imager)、 TMI(TRMM Microwave Imager)及AMSU(Advanced Microwave Sounding Unit)等被動微波儀估算，參考美國官方之颱風最佳路徑，預測豪大雨之24小時潛勢預報，此技術 (Tropical Rainfall Potential，TRaP)提供一個快速又實用的方法，應用此方法預測降雨量比中尺度數值預測模式更接近於實際觀測(Ferraro等，2005)。本研究的前期研究曾利用 SSM/I 估算梅雨期及颱風降雨量，而陳等(2002)亦使用TMI 及AMSU-A 估算降水，結果與地面之觀測比較都有不錯之結果，故皆可用於颱風降水之估算。就颱風降水預測而言，本研究先使用持續法，利用中央氣象局之路徑預測資料，可得到六、十二、二十四小時後之潛勢降水預報，並與地面觀測之累積雨量比較。颱風之生成與發展和海洋所蓄含的能量有相當密切的關係。另本研究亦將探討颱風螺旋雨帶移動與地形效應對降水之影響，將颱風預報降水技術提升，並能自動化作業，提供未來對颱風降水預報與防災之參考。 The progress in typhoon position forecast is evident, especially for the aid of weather numeral models. According to Central Weather Bureau, the average position errors of 24, 48 and 72 hours forecast times are 104, 194 and 304 km for 18 typhoons in 2006, respectively. Comparing to the previous 3 year average position errors, which are 130, 230 and 342 km for 24, 48 and 72 hours forecast times respectively, it has shown impressive improvements. For this year 2008, the CWB’s position forecast errors of 24 and 48 hours times has evidently shown a leading role compared with other countries. But when typhoon Kalmaegi hit Taiwan, it brings unpredictable large amount of rainfall and caused serious floods and damages on the facilities of fishery and agriculture. CWB was blamed for not issuing an early warning to people. Pursue of better accuracy of quantitative precipitation forecast is always a scientific study goal, especially for disastrous rainfall from landing typhoons. Typhoons get birth and develop on ocean where traditional observations are scarce. Thus satellite observation data are very crucial to tropical cyclone’s monitoring. While the typhoons are approaching offshore, fewer traditional observations around typhoons are available. Radar observations of rainfall structures are although valuable, but only could be used for landing typhoon observation. Satellite-borne microwave radiometers can measure instantaneous rain rates through the entire cloud area of typhoons which more suitable for usage on rainfall forecast. This study uses SSM/I data to estimate typhoon precipitation. Chen, et al. (2002) had used TMI and AMSU-A data to estimate MEI-YU rainfall. It showed a good relationship between ground truths and estimations. Another microwave radiometer AMSR-E, one of AQUA instruments which were launched in Jun 2002, will be added to estimate the rainfall of Northwest Pacific’s typhoons. According to original Tropical Rainfall Potential (TRaP, Ferraro et al., 2005), it is assumed that estimated rain rates are correct and no change happened on the intensity and rain area size during typhoon movements. First we use JTWC’s typhoon best track data to predict rainfall potential of 3, 6, 12 and 24-hour forecast. The results are validated through rain gauges data. The second, air-sea parameters will be added to estimate the intensity and rainfall change of typhoons, and further be applied to predict rainfall potential. Finally terrain effect will be considered to estimate the typhoon precipitation. 研究期間：9908 ~ 10007 |
