Socioeconomic inequality in flood exposure and compound urban flood risk: A multistage assessment across 30,000 towns in China.

Increases in precipitation extremes and rapid urbanization are globally reshaping urban flood risks. However, few studies jointly assess within-city exposure inequalities and green-gray infrastructure capacities at national scale. A multistage framework integrating statistical modeling with machine learning is employed to analyze flood exposure of over 30,000 towns across 303 Chinese cities in 2000-2018. Flood hazard is derived from satellite-based inundation footprints, while population, development, greenness, and a gray-infrastructure proxy are compiled from remote-sensing and statistical sources. First, two within-city flood-exposure patterns are identified, the High-Development-skewed exposure pattern (HD) and the Low-Development-skewed exposure pattern (LD), with risk ratios around 1.4 for both skews. These patterns cluster spatially and show divergent town- and city-level drivers across socioeconomic and green-gray indicators. Secondly, a comparison of green-gray infrastructure capacity deficits reveals that green under-capacity is pattern-invariant, indicating a baseline requirement across cities. Gray infrastructure shortfalls are more prevalent under LD, by about 6 %. Thirdly, hazard, exposure, and vulnerability are integrated into a compound city-level risk index and modulated by an inequality factor, capturing both physical risk and distributional burden. Across China, risk hotspots align along four belts, the Sichuan-Chongqing-Yunnan corridor, the southeastern seaboard, the middle-lower Yangtze corridor, and parts of Northeast China. This result reflects hazard clustering in the humid south and southwest, concentrated exposure along coastal and inland corridors, and an eastward increasing vulnerability. This integrated framework provides an equity-aware, interpretable risk decomposition that supports transferable, resource-efficient flood mitigation prioritization in China and other regions worldwide.