Jump to Main Content
Analyzing explanatory factors of urban pluvial floods in Shanghai using geographically weighted regression
- Wang, Congxiao, Du, Shiqiang, Wen, Jiahong, Zhang, Ming, Gu, Honghuan, Shi, Yong, Xu, Hui
- Stochastic environmental research and risk assessment 2017 v.31 no.7 pp. 1777-1790
- autocorrelation, cities, climate change, disasters, environmental factors, floods, issues and policy, models, risk, rivers, social welfare, urbanization, China
- In the context of climate change and rapid urbanization, urban pluvial floods pose an increasing threat to human wellbeing and security in the cities of China. A valuable aid to managing this problem lies in understanding the roles of environmental factors in influencing the occurrence of pluvial floods. This study presents a spatial analysis of records of inundated streets in the inner city of Shanghai during 1997–2013. A geographically weighted regression (GWR) is employed to examine the spatially explicit relationships between inundation frequency and spatial explanatory factors, and an ordinary least squares regression (OLS) is used to validate the GWR results. Results from the GWR model show that the inundation frequency is negatively related to elevation, pipeline density, and river density, and is positively related to road/square ratio and shantytown ratio. The green ratio is another significant explanatory factor for inundation frequency, and its coefficients range from −1.11 to 0.81. In comparison with the OLS model, the GWR model has better performance as it has higher R², and lower corrected Akaike information criterion and mean square error values, as well as insignificant spatial autocorrelation of the model residuals. Additionally, the GWR model reveals detailed site-specific roles of the related factors in influencing street inundation. These findings demonstrate that the GWR model is a useful tool for investigating spatially explicit causes of disasters. The results also provide guidance for policy makers aiming to mitigate urban pluvial flood risks.