台灣由於山勢陡峻以及颱風降雨雨勢急促，時常引發水災、山崩、土石流等嚴重災害，造成無數生命及財產的損失。而工程上的防洪方式則受到許多限制，因此最有效的方式莫過於有效的災害預警，及時的掌握豪雨及颱風動態，以期能達到及早防汛，降低災害的效果。 本研究首先研究定量降水預報(Quantitative Precipitation Forecasting；QPF)之適用性，以鳳山溪流域為研究範圍，選取於2019~2020年間發生之7場降雨類型為春雨、梅雨及颱風之降雨事件，分析於有效降雨期間與實際降雨量之相關性。結果表示，QPF降雨資料對於洪水發生時序掌握度頗佳，使用於洪水預報為可行。並驗證模式之率定曲線與水利署公告之率定曲線使用於RT即時校正(Real-Time Correction,RT)模式之適用性，結果顯示，於有效降雨延時，使用模式率定曲線模擬出之水位與測站水位之最低相關度為0.860，水利署公告率定曲線最低相關度為0.867，差異不大。此外評估RT系統使用於洪水預警系統之成效，發現經使用RT系統即時校正後，預估水位達到預警水位及水位退去能力為佳。 ;Due to the steep mountains and Typhoon packs cloudburst in Taiwan, serious disasters such as floods, landslides, and mudslides occur frequently, causing lives and property losses. However, engineering flood control are limited to many restrictions. Therefore, a more effective approaches measures is to develope effective disaster warning systems and timely grasp the torrential rain and typhoon dynamics, so as to achieve the early flood prevention and the reduction of disasters. In order to improve the the applicability of Quantitative Precipitation Forecasting (QPF), the Fengshan river of as taken as basin as the research scope, this Seven rainfall events that occurred between 2019 and 2020 as spring rain, plum rain and typhoon. The correlation were selected between the effective rainfall period and the Station rainfall was analyzed. The results show that although the QPF rainfall data has a good grasp of the timing of flood occurrence. and it is feasible to use it in flood forecasting. It is also verified that the Rating-Curve of the model and the Rating-Curve announced by the Water Resources Agent (WRA) are both adequate for RT (Real-Time correction,RT) mode. The results show that the minimum correlation between the water level simulated by the model Rating-Curve and the station water level is 0.860 for the effective rainfall period, while the water level simulated by the WRA Rating-Curve and the station water level is 0.867. In addition, the effectiveness of RT system used in flood warning system was evaluated, and it was found that the ability of warning water level to start to rise and recede was better with the RT system.
