Insights on public health promotion based on urban green space runoff mitigation: Type differences, spatial patterns and preferable planning

With the backdrop of extreme rainfall and rapid urbanization, quantifying the runoff mitigation capacity of different types of urban green spaces (UGS) and spatially representing them holds significant importance for UGS design and planning aimed at sustainability and at managing stormwater-related public-health risks in healthy urban environments. This paper, taking Kunming City, China as a case study, develops a physics-guided, XGBoost-augmented SCS-CN framework that integrates normalized difference vegetation index (NDVI) and terrain metrics while enforcing basic hydrologic feasibility constraints. Using multi-source spatial data that includes 30 m Land Use and Land Cover Change (LUCC), resampled Hydrologic Soil Groups (HSG), and UGS types classified under Standard for Classification of Urban Green Space (CJJ/T 85–2017) and an event rainfall set spanning 1–166 mm, we evaluate runoff reduction across UGS types and compare the augmented model against the baseline Soil Conservation Service Curve Number (SCS-CN) approach. Results show that the augmented model outperforms the baseline, with the most stable gains under moderate storms (approximately 20–120 mm), while maintaining physical consistency. Type-based comparisons reveal that EG-category UGS sustain higher and more stable runoff-reduction coefficients across most rainfall intervals, whereas several G-category and affiliated UGS show earlier efficiency decline as rainfall intensifies. At the city scale, runoff-reduction coefficients exhibit an inflection around 20–30 mm, which indicates a transition from a plateau phase to a sustained decline with larger storms. Feature attribution indicates that NDVI dominates runoff reduction under light to moderate rainfall, while rainfall intensity and slope become the primary drivers during heavy rainfall, with key interaction effects shifting around the mid-range storm depth. These findings support health-oriented planning strategies that prioritize canopy and soil-based retention for frequent small to moderate events and coordinate green–gray infrastructure to buffer overflow and associated public health exposures during heavy storms.