A tri-system urban waterlogging risk assessment framework based on GIS- game theory combination weight: a case of Zhengzhou City

The urban waterlogging risk is sensitive to complex interactions among spatially distributed H-E-V (Hazards, Exposure, and Vulnerability) factors and is affected by indicator weighting and risk calculation models. To investigate and relieve these sensitivities, a tri-system risk assessment framework is devised, consisting of indicator, griding, and risk assessment systems. The entropy, CRITIC, and game theory (GT) are integrated to weigh the fifteen indicators for the Zhengzhou City area. The risk assessment method based on GT and additive model achieved the best performance: (1) 90.39% actual waterlogged sites are in the High and Very High-risk areas; (2) the area under the ROC curve (AUC) value is 0.791; (3) it is competitive in Accuracy, Zero–One Loss, Recall, and F1-score. In the analyzing results, the total High and Very-High risk areas account for 20.34% of the city. The first three districts with High and Very-High risks are Xinzheng, Jinshui, and Zhongmou, with average areas of 231.94, 171.06, and 167.53 square kilometers, respectively. The risks incur a positive spatial autocorrelation, and inhibit a High-High-Value clustering pattern with a confidence level of 99%. The devised tri-system framework can contribute to urban waterlogging prevention, post-disaster resettlement, and rescue efforts.