ABSTRACT Predicting urban waterlogging risk influenced by land use dynamics has become increasingly important with growing urbanization. However, prior studies have scarcely considered the influence of land use patterns. The present study proposes a novel methodology for predicting waterlogging risk, while exploring the enhancement effect of incorporating landscape indices on prediction. The present study proposes an enhanced framework for predicting waterlogging risk, while exploring how incorporating landscape indices affects prediction. This framework integrates the MaxEnt and PLUS methods, which are specifically designed to incorporate landscape indices. It can enhance the predictive performance and interpretability of waterlogging risk. The integrated model was constructed using data from 2015, and its accuracy was validated with data from 2020. First, landscape indices were computed and the PLUS model was constructed based on the land use data from 2015. Subsequently, we trained two sets of MaxEnt models (with and without landscape indices) using the waterlogging data from 2015, and simulated the landscape indices and waterlogging risk in 2020 with the MaxEnt‐PLUS method being utilized for this purpose. The findings indicate that the incorporation of landscape indices resulted in an enhancement of the AUC value of the MaxEnt model from 0.919 to 0.928. Furthermore, the results of the waterlogging risk simulation are more consistent with actual waterlogging hotspots. Therefore, the calibrated PLUS model was deemed appropriate for the purpose of predicting future land use change. Then, the MaxEnt‐PLUS method was employed to project the waterlogging risk in 2030, and the changing patterns of the various risk areas during 2015–2030 were examined. In summary, the proposed method could enhance the reliability of waterlogging risk prediction, thereby providing decision support for sponge city planning and stormwater management.
Wu et al. (Wed,) studied this question.