| 摘要: | 由於臺灣地區每年受颱風伴隨的豪雨影響，造成山崩、土石流及洪水之患，嚴重危害民眾的生命財產安全，因此急需要有颱風洪水的即時預警系統之建置。本研究將結合氣象模式與水文模式兩個模式系統，在氣象中尺度數值預報模式方面將使用MM5/WRF 模式，於河川逕流水文模式將使用FLO-2D 模式。初期研究將使用氣象模式的降水輸出以單向方式(one-way)驅動河川逕流水文模式，如同 Li et al. (2005)的做法，針對過去重大颱風個案如賀伯(1996)、桃芝(2001)、納莉 (2001)、敏督利(2004)、及艾利(2004)颱風等，進行颱風暴雨引發洪水之模擬研究。待此氣象模式降水輸出單向方式驅動水文模式技術漸趨成熟後，我們將發展氣象模式與水文模式的雙向藕合系統(two-way coupled system)，直接考慮大氣降水輸出與河面蒸發、陸面蒸散、及植被葉蒸的雙向交互作用(two-way interactions)。於第一年計畫中，我們將對2001 年桃芝(Toraji)颱風、2004 年敏督利(Mindulle)颱風及2004 年艾利(Aere)颱風，進行測試。初步研究發現，在模擬時間30 小時內MM5 氣象模式皆有不錯的雨量掌握。氣象模式降雨輸出對於水文模式有顯著影響；颱風路徑的誤差及模式積分時間過長(超過24 小時)所導致的誤差皆為影響降雨分佈的因素。於第二年計畫中，我們將對這些誤差原因做進一步探討，並將結果延伸至其他颱風個案。於第三年計畫中，我們將發展氣象模式與水文模式的雙向藕合系統。本年度計畫為此三年度計畫中之第二年。 ; The heavy rainfalls associated with landfalling typhoons induce landslide, mudflow, and severe flooding, which cause tremendous property damage and human casualities over the Taiwan area. Thus, a strong demand is needed for the establishment of a real-time forecasting and warning system for typhoon-induced rainfall and flooding. This study will combine two modeling systems, a meteorological model (MM5/WRF) and a river runoff model (FLO-2D). Firstly, the coupled modeling system will use the one-way coupling approach, which uses the rainfall output from the MM5/WRF meteorological model as the input for the FLO-2D river runoff model as done in Li et al. (2005). Historical typhoon cases of Herb (1996), Toraji (2001), Nari (2001), Mindulle (2004), and Aere (2004) will be examined to calibrate the couple model system. Secondly when the one-way coupling technique of the meteorological model with the river runoff model is more established, we will develop a fully two-way coupled system, which considers two-way interactions between atmospheric precipitation, river surface evaporation, and land-surface transpiration. In the first year, we will investigate the applicability of this coupled modeling approach to Typhoons Toraji (2001), Mindulle (2004), and Aere (2004). The preliminary results show that MM5/WRF model has good predictive skill of typhoon-induced rainfalls in the first 24 hours. The rainfall forecast from the MM5/WRF meteorological model has significant impact on the performance of the FLO-2D model’s runoff forecast. The errors of typhoon tracks, imperfectness of precipitation parameterization, and long integration period (more than 24 hours) all contribute to deficiency of rainfall forecast. If the backwater mechanism after infiltration is not included in the hydrological model, the simulated stream flow would be further reduced. In the second year, we will examine the reasons for model deficiency in more depth and further expand our application to other typhoon cases. In the third year, we will develop the fully two-way coupled modeling system. This coming year will be the second year for the three-year project. ; 研究期間 9708 ~ 9807 |
