Can urban green space equitably regulate flood risk? a risk-adjusted spatial availability and multi-objective optimization framework for high-density cities

Rapid urbanization and intensifying rainfall have increased pluvial flood risk in high-density cities; however, urban green space (UGS) planning rarely integrates stormwater-regulation deficits, population-oriented potential spatial availability, distributional inequality, and land constraints. In central Chengdu, China, this study developed a framework linking Functional–Climatic Hydrological Units (FHUs)-based diagnosis, a risk-adjusted two-step floating catchment area model (Risk-2SFCA), population-weighted distributional assessment, and two-stage UGS optimization. FHUs were delineated from urban function, landscape morphology, and major road boundaries, and an FHU-level supply–demand balance was evaluated under six rainfall scenarios. Risk-2SFCA measured the alignment between effectiveness-adjusted UGS capacity and flood-susceptibility-adjusted population-equivalent demand within standardized network-based planning neighborhoods. Candidate sites were selected using a maximal covering location problem, followed by optimization of mean availability, inequality, and priority-weighted residual deficits using the differential-evolution-enhanced Non-dominated Sorting Genetic Algorithm II. Results showed that the mean FHU-level balance declined from 14.24 mm under 56.39 mm rainfall to −24.00 mm under 110.14 mm rainfall, and larger deficits were associated with greater observed inundation depth. Availability classifications were more sensitive to neighborhood thresholds than to impedance speed, while alternative population representations preserved broad citywide patterns but changed local and group-specific estimates. The entropy-weighted compromise allocated 4,202.26 ha of UGS among 141 candidate sites, with 74.2% of the allocated area concentrated in service-deficit zones. Within these zones, mean availability increased from 25.37 to 32.32, and the population-weighted Gini coefficient decreased from 0.86 to 0.81. The framework supports fine-grained, risk-sensitive, and distributionally informed green-infrastructure planning under land constraints.