Climate change represents an intractable problem which urges a prompt intervention to be resolved. One of climate change's most destructive calamities is flash flooding. On that caveat, this study tries to identify the zones most susceptible to flash floods by utilizing machine learning technique. Initially, a digital elevation model (DEM) has been utilized to delineate a basin located in the city of New Cairo, Egypt, where the flash floods occur frequently. Subsequentially, 12 flood causative factors were calculated and mapped via ArcMap. Furthermore, the depth of runoff was calculated via HEC-RAS and employed as a flood causative factor. Additionally, both flooded points and flood causative factors have been employed by utilizing the “GARP”* approach to produce the Flood Hazard Map (FHM). The state-of-art in this study is to predict the hazard degree of flash floods by utilizing the simulated runoff depth which has been used as flood causative factor in the selected machine learning technique. The FHM anticipated approximately 20% of the total area to have a very high hazard of floods, whereas, more than two-thirds of the study area has expected to have low and very low flood hazard. In addition, the receiver operating characteristic “ROC” approach has been applied to examine the FHM, which estimated the area under curve as 96.88%. Finally, decision-makers can utilize the generated map to better comprehend repercussion of flooding and make adequate preparations to alleviate this risk. *GARP: “Genetic Algorithm for Rule-set Prediction”.
各種検索
サポート
ヘルプ