Predicting future urban waterlogging-prone areas by coupling the maximum entropy and FLUS model

Urban waterlogging is a severe hazard that can directly damage environmental quality and human well-being. It would be desirable for hazard mitigation planning and sustainable urban design if potential waterlogging-prone areas under dynamic land use change could be appropriately predicted. However, previous related studies did not simultaneously consider the reliability of negative samples and the future influence of fine-scale land use change. To fill the knowledge gap, this research has developed a robust method for predicting future waterlogging-prone areas by coupling the maximum entropy (MAXENT) and the Future Land Use Simulation (FLUS) model. The former can ensure that no extra sampling bias will be introduced, while the latter can accurately forecast the spatio-temporal pattern of land use. This case study has confirmed the accuracy and feasibility of this method. It was found that the proportion of impervious surfaces, population density, and proportion of green areas are key spatial drivers behind urban waterlogging issues. In addition, the future hazard potential map provided by the MAXENT and FLUS implies that a large proportion of impervious surfaces will face huge waterlogging risks. Therefore, policymakers should focus more on places with a higher probability of urban waterlogging. In summary, this research is expected to offer a practical tool for future urban design and waterlogging risk prevention.

Automated summary

This research has developed a robust method for predicting future waterlogging-prone areas by coupling the maximum entropy (MAXENT) and the Future Land Use Simulation (FLUS) model. The study confirmed the accuracy and feasibility of this method and identified impervious surfaces, population density, and green areas as key drivers behind urban waterlogging issues.