Evaluating urban flood risk using hybrid method of TOPSIS and machine learning

In this study, a combination of machine learning and decision-making methods are used to evaluate flood risk in an urban setting. The destructive effects of floods are magnified in cities and as such, urban flooding has increasingly become an issue for regional and national governments. Accurate models of urban flood susceptibility are required to mitigate this hazard and build resilience in cities. To construct of the models, flood hazard maps were create by using three state-of-the-art machine learning methods. The TOPSIS decision-making tool was then employed for urban flood vulnerability analysis, which is based on socio-economic factors. Finally, an urban flood risk map is derived from the flood hazard and vulnerability maps. Jiroft City, Iran served as the subject of the study.

The results yielded by urban flood hazard modeling indicate urban drainage density and distance to urban drainages as the most important factors. In the case of Jiroft city, the lack of proper drainage systems and unplanned expansion of the metropolitan region along rivers are the most important reasons for flooding. The results also show that runoff potential and rainfall were not strong predictors of flood hazards. As might be expected, areas with a high or very high population density are most vulnerable to flooding. The integrated procedure outlined in this paper can be used as a first step in managing flood risk in regions without meteorological stations. It can be used by planners to guide urban development, plan appropriate urban drainage systems, provide floodwalls and other engineered protective structures, and protect buildings at risk.