耦合一維河川變量流模式及NTU-CAFIM模式之研發與應用

Abstract

近年來流域內的河川水位因暴漲而溢堤，導致堤內地區的嚴重淹水災害增加。若能有快速且精度良好的流域淹水模式，將可提升流域尺度的災害應變能力。現有的NTU-CAFIM模式(National Taiwan University Cellular Automata Flood Inundation Model)經過發展，已可提供快速且精度良好的都會區地表淹水模擬，如能將其與一維河川變量流模式進行耦合，便可提供即時的流域淹水模擬，對於防災應變有所助益。 本研究選取合適之一維河川變量流模式，將其耦合可模擬二維漫地流淹水之NTU-CAFIM模式，使一維河川變量流模式與NTU-CAFIM模式之間可以線源的溢堤方式進行動態水體交換。並針對兩模式之耦合制定一套銜接方式，包含漫地流河岸網格與河段於不同空間解析度下之銜接、河岸網格與河段間水體傳遞量之計算等。其中，一維河川變量流模式與NTU-CAFIM模式之水體傳遞量部分採用自由或潛沒堰流公式計算。 本研究選取兩個研究案例，以效率係數(NSE)、準確度(ACC)與敏感度(TPR)，比較耦合後的模式與HEC-RAS水理模式之模擬結果吻合程度，以此驗證本研究所制定的銜接方式。藉由小範圍理想案例以及具有真實地形的英國塞文河流域案例之模擬結果可知，耦合後的模式在河川斷面水位歷線、河川斷面流量歷線、二維漫地流觀測水位歷線、入流與溢流量歷線以及最大淹水範圍的模擬結果皆與HEC-RAS模式吻合。此外在理想案例以及英國塞文河流域案例上，耦合模式分別僅需HEC-RAS水理模式53%與59%之演算時間。以上結果表示耦合模式可合理模擬河川與漫地流之間的水體交換，且擁有較高的演算效率，在流域淹水模擬中具有相當的潛力。

In recent years, serious flooding in areas near the river due to river embankment overflowing is increasing. To promote disaster prevention and protection for the whole river basin, a fast and accurate inundation simulation model is required. The existed National Taiwan University cellular automata flood inundation model (NTU-CAFIM model) can fastly and accurately provide a prediction of flooding in urban areas, thus it is appropriate to be coupled with the 1D unsteady river flow model to provide real-time forecasting of flooding in the river basin. In this study, the 1D unsteady river flow model (RFM) is coupled with the NTU-CAFIM model (OFM), so that water can be dynamically exchanged through embankments of each river reach. Rules to describe exchanged water are developed for coupling the RFM and OFM, including the correspondence of different spatial resolutions between riverbank cells in the OFM and river reaches in the RFM, and the calculation of exchanged volume between riverbank cells in the OFM and river reaches in the RFM. In particular, the exchanged volume between the RFM and OFM through embankments along each river is calculated by either free or submerged weir flow formulas. To verify the water exchanging rules in this study, two case studies are selected, and the coupled model is compared with the HEC-RAS model by indicators such as Nash-Sutcliffe efficiency (NSE), accuracy (ACC), and sensitivity (TPR). Based on the simulation results of a small-scale ideal case and the case of the Severn River Basin in the United Kingdom with an area of 24km2 and actual complex terrains. The simulation results of the coupled model in the hydrographs of water level in river sections, the hydrographs of flow rate in river sections, the hydrographs of water levels in two-dimensional overland flow, the hydrographs of water overflowing, and the maximum flooding area are very similar with the HEC-RAS model. Furthermore, in the ideal case and the real case of the Severn River Basin, the coupled model is 47% and 41% faster than the HEC-RAS model. The above results indicate that the coupled model is as accurate as the HEC-RAS model with a significant reduction in its computational time. Hence, it is concluded that the coupled model has considerable potential to be applied to real-time simulation of flooding in the river basin.