Quantifying the impact of rainfall spatial heterogeneity and patterns on urban flooding by integrating machine learning algorithm and hydrodynamic-hydrological model

Urban flood simulation serves as a critical tool for risk assessment and management. However, the potential impact of rainfall spatial heterogeneity is often ignored in urban flood simulation. In this study, we developed a rainfall spatial distribution model that generates spatially non-uniform rainfall based on the multilayer perceptron (MLP) neural network and adaptive moment estimation (Adam) optimizer. Subsequently, rainfall scenarios were designed by the proposed model and Chicago approach. The impact of rainfall spatial heterogeneity and patterns on urban flooding was quantitatively analyzed through the hydrodynamic-hydrological model. Using the central urban area of Zhengzhou, China as a case study, the results showed that the rainfall spatial distribution model exhibited excellent performance, achieving a coefficient of determination (R²) of 0.968. Considering rainfall spatial heterogeneity and pattern, ninety rainfall scenarios were designed and input into the urban flood model. We observed that urban flood inundation escalated with increasing rainfall peak coefficients. Neglecting the rainfall spatial heterogeneity led to a systematic underestimation of flooding in the study area, resulting in an average reduction of 3.97% in inundation volume and 2.77% in inundation area. Furthermore, the underestimation of inundation volume caused by neglecting rainfall spatial heterogeneity intensified with increasing peak coefficients. Specifically, when the rainfall peak coefficient increased from 0.2 to 0.8, the magnitude of underestimation rose by 6.85%. This study provides a new approach for considering the impacts of spatial variations in rainfall on flooding and offers an important reference for future urban flood warning, disaster prevention and mitigation efforts.