Exploring the impact of streetscapes on waterlogging risk from a 15-minute walkability perspective

Understanding the impact of streetscapes on waterlogging risk is critical for enhancing urban resilience. This study uses street view images and navigation data to investigate the spatial variability of waterlogging probability and dominant factors within a 15-min walkable radius of residential communities, applying interpretable machine learning. Focusing on Beijing's main urban area, the results reveal that within 150 m of waterlogging points, buildings have the greatest impact, followed by walls and roads, while sidewalks and plants play a minimal role. Streetscapes such as roads, walls, buildings, and sidewalks exhibit an inverted U-shaped trend in their impact on waterlogging risk, whereas sky and vegetation, particularly grass, significantly reduce the risk. Dominant factors vary spatially: in areas near airports and railways, dense buildings increase waterlogging risk, while walls and open sky mitigate it; in the eastern business district, walls and sky are the main contributors to waterlogging, but grass alleviates it; in decentralized areas, open skies reduce risk, while buildings exacerbate it; in the northern periphery, dispersed buildings and walls help reduce waterlogging; and along the Fifth Ring Road, walls, sky, and trees play a stronger role. Spatially, high-risk zones are concentrated along major roads, particularly between the 3rd and 4th Ring Roads in the south, forming a U-shaped pattern, while areas near the 2nd Ring Road show lower risk. These findings provide actionable insights for urban planning, streetscape design, and Waterlogging prevention strategies.