台北市致災性降雨之都市洪災預警模擬與分析—以2019年午後對流降雨為例

Abstract

在氣候變遷的影響下，台灣面臨極端天氣事件的風險增加，降雨強度與分布都將有所改變；在短延時強降雨發生頻率增加的趨勢下，考驗著台灣面對極端降雨的防洪能力。2019年在台北市區就有幾場因午後對流性降雨，在短時間內承載超過都市排水系統負載的雨量，因而導致多處積淹水事件發生。為提升都會區防洪能力，建立都市洪水預警（Urban Flood Early Warning, UFEW）是當前重要的研究課題。 目前都市淹水預警主要由氣象預報資料、水理數值模式之計算，以及相關專家的分析研判建構而成。本研究中選用中央氣象局發展之區域系集預報系統（WRF Ensemble Prediction System, WEPS）資料作為主要氣象預報資料來源，以防災預警應用為目標，利用模式產品統計（Model Output Statistics，MOS）的概念，將模式雨量預報數值以觀測雨量累積百分比(Observed Rainfall Accumulation Percentage Rank)進行調整後，作為SOBEK水里模式之雨量預報之輸入資料，提供給該模式進行一維下水道水位與二維地表漫地流淹水深度及範圍模擬之演算。 研究結果顯示，經由觀測累積百分比方法調整後的雨量預報數值，其在水文水理模式進行下水道水位與地表淹水範圍模擬結果，皆與實際觀測資料較為符合，意即此預報雨量之調整方法，可供未來研發預警系統時參考使用。未來應用於都市洪水預警時，將能有效提升防災應變、水情研判資料的準確性。

Under the impact of climate change, the higher risk of extreme weather accrues in Taiwan, increasing the intensity and changing the distribution of rainfall. With the rising frequency of short-term heavy rainfall, the flood prevention ability in Taiwan has been concerned. There were some afternoon thunderstorm rainfall-induced flood cases in 2019 in Taipei city central area, because the rainfall exceeded the load of the urban drainage system. In order to raise the ability of flood prevention in the urban areas, building a better urban flood early warning (UFEW) is an important issue nowadays. UFEW is based on higher accuracy weather forecast data, a more sophisticated hydrological numerical model, analysis by experts, and instant correction. WRF Ensemble Prediction System (WEPS) is selected as the main weather forecasting data resource in this study. To increase the application of disaster prevention, the Model Output Statistics (MOS) method based on observation rainfall quantity accumulation rank was chosen to modify the hourly rainfall quantity of forecasting data and put modified data into the SOBEK model to evaluate one-dimensional flow and two-dimensional overland situation. Then compared the original forecast result, modified one, and observed one could assess whether the method is applicable. The results show that the modified rainfall quantity forecast data increase the accuracy of rainfall quantity forecast, the water-level value in water gauge, and flood area prediction. Applying this MOS method to FEWS and other disaster prevention-related measures would be helpful.