Combining environmental-socio-economic data with volunteer geographic information for mapping flood risk zones in Zhengzhou, Henan Province, China

Due to the occurrence of extreme weather phenomena, flooding caused by precipitation has become a frequent, widespread, and highly destructive natural disaster. For densely populated urban areas, efficient identification of people's emergency rescue needs and mapping of high flood risk areas are particularly important for decision makers to formulate disaster relief strategies. In this study, we employed a machine learning approach along with volunteer geographic information (VGI)1 data and environmental-socio-economic indicators to map the flood risk area in Zhengzhou. Firstly, the sentiment analysis tool based on artificial intelligence platform initially identified the public's emergency rescue needs. Secondly, the crowd-sourcing approach and geographical approach were used to evaluate the VGI data quality and the variable importance measures (VIMs)2 method to optimize the evaluation indicators. Lastly, the flood risk area was classified in four categories: flood environmental susceptibility, flood-induced risk, flood exposure risk, and flood mitigation capacity, subsequently resulting in a comprehensive risk map of Zhengzhou. The results showed that the accuracy of flood risk assessment increased from 85.7 % to 92.4 % after VGI quality evaluation. The most important indicators affecting flood risk assessment were population density, river system and nighttime light intensity. The comprehensive flood risk map better reflected the spatial distribution of high-risk areas in Zhengzhou. It allows emergency responders to quickly identify areas that require urgent attention.