Flood risk evaluation of the coastal city by the EWM-TOPSIS and machine learning hybrid method

The frequent occurrence of floods and waterlogging has significantly impacted coastal cities. Effective mapping of flood risk can enhance the precision of disaster risk reduction strategies. However, there is room for improvement in fields such as flood inundation data, evaluation objectivity, and classification refinement. This study, using Xiamen as a case study, advanced flood inundation data accuracy by generating a frequency -based flood inundation map from multi -year remote sensing imagery. This study integrated the EWM-TOPSIS model with a neural network model to evaluate the flood risk. The EWM-TOPSIS model was employed to assess the flood vulnerability and flood exposure, while the multi -class neural network model to simulate the flood hazard. This combined approach reduced subjectivity in the flood risk assessment of Xiamen and achieved a finer level of risk classification compared to traditional binary neural networks. The results indicated that the flood vulnerable areas of Xiamen are concentrated along waterways distant from roads, with high flood exposure in Tong'an and Xiang'an districts. The flood -prone areas in Xiamen are primarily located along the coastal areas characterized by extensive impermeable surfaces. High flood risk areas are mainly distributed in Tong'an and Xiang'an subdistricts, particularly in Xiangping and Xindian. The method developed offers support for accurate flood risk identification and decision -making in cities with limited data resources.