Impact of 2D and 3D factors on urban flooding: Spatial characteristics and interpretable analysis of drivers

Abstract

Urban flooding poses a serious threat to both the ecological environment and human society. Previous studies identified natural and anthropogenic factors as contributors to urban flooding, but little attention has been paid to the influence of urban horizontal and vertical factors. To address this gap, we conducted a comparative analysis of the patterns in spatial distribution of urban flooding in two megacities in eastern China (Beijing and Guangzhou). We then used Pearson's correlation to investigate the associations between flooding events and multiple influencing factors. Finally, two scenarios were designed to quantify the relative contributions of each driver using the Light Gradient Boosting Machine (LightGBM) and Shapley (SHAP) interpretable models. The results show that: (1) urban flooding points in Guangzhou and Beijing are predominantly clustered in central areas, with mid-rise and high-density buildings presenting the highest flood risk. (2) in the base scenario, Annual precipitation (AP) is the primary influencing factor for urban flooding in both Beijing and Guangzhou. However, in the enhanced scenario, the addition of 2D and 3D (two-dimensional and three-dimensional) metrics shifts the main drivers to factors like Aggregation index (AI), Patch density (PD), and Building density (BD), significantly impacting urban flooding. This study highlights the critical impacts of horizontal and vertical urban structures and layouts, emphasizing the need for comprehensive urban planning and design strategies to effectively mitigate flood risk. It also provides new perspectives on urban flood risk management.