Identifying landscape patterns at different scales as driving factors for urban flooding

• Fragmented and dispersed landscape patterns are more effective in reducing flood occurrences. • Landscape configuration plays a more crucial role in flood events than landscape composition. • Flood mitigation requires prioritizing certain planning factors at varying spatial scales. • We introduce a novel multiscale and multifactor analysis framework to enhance urban flood mitigation planning. Climate change and rapid urbanization have led to increasingly frequent urban flooding, causing substantial losses. While previous studies have examined the impact of land use types on flooding, few studies have explored how the spatial distribution and configuration of land use (landscape patterns) influence urban flooding across different scales. This study addresses this gap by investigating the effects of landscape patterns on urban flood events in Chengdu, China. We constructed a comprehensive dataset comprising 28 flood influencing factors, including landscape pattern, topographic, and hydrological characteristics. Using Principal Component Analysis (PCA), we classified these variables and applied stepwise Poisson regression to evaluate how landscape patterns affect urban flooding. Our findings show that key influencing factors vary by scales: at the 1 km scale, topographic factors were most important; at the 2 km scale, impervious areas had the largest impact; and at the 3 km scale, landscape configuration factors were dominant. In particular, the mean patch area and cohesion were consistently significant across all scales, indicating that more fragmented and dispersed landscapes tend to reduce flooding occurrence. We conclude that scale is an important determinant for properly understanding the contribution of landscape patterns to urban flood mitigation.