Climate change is driving growth in extreme weather events through vast world regions. Therefore, one of the most destructive calamities of climate change is flash flooding. This research aims to predict the most prone areas to flash floods in urban regions using machine learning techniques. Moreover, the runoff depth has been calculated via HEC-RAS software to use it as a conditioning parameter. Subsequentially, seven environmental factors (elevation, slope, flow length, distance to streams, maximum runoff depth, land use, and curve number) were adopted in New Cairo city, Egypt, where the flash flood occurs persistently. A combination of spatial flooded points and environmental parameters has been utilized using a genetic algorithm for Rule-set Prediction (GARP) to generate a flood susceptibility map (FSM). Furthermore, the FSM accounted nearly fifth of the total area for very high prone to floods whilst 40.03% and 13.14% of the city have very low and low hazards, respectively. Additionally, the receiver operating characteristic curve (ROC) method was used to validate the FSM which gave 71.5% for the area under curve (AUC). Eventually, Decision-makers can use the produced flood danger map to better understand the consequences of flash floods and to prepare alternative plans to mitigate this hazard.
各種検索
サポート
ヘルプ