How residential density constrains nature-based cooling benefits and their distributive justice: Implications for climate-adaptive cities

Global warming has intensified extreme heat events, posing escalating risks to urban health and livability. Rapid urbanization and residential densification have reduced available greenspace, and declining cooling efficiency (CE) has further weakened an already constrained capacity for urban heat mitigation. Existing research provides limited understanding of how CE diverges across varying residential densities. To address this gap, this study develops a grid-based CE model to quantify greenspace cooling efficiency across more than 300,000 single-family residential parcels in Southeastern Wisconsin. A multi-level Geo-XAI framework integrating CatBoost, GeoShapley, and GNNWR is further applied to characterize the nonlinear and spatially varying relationships between CE and greenspace quantity, quality, and spatial configuration. Scenario simulations then evaluate changes in Cooling Benefit (CB) and population-weighted cooling-benefit inequality across residential-density categories. The results show that: (1) low-density areas exhibit stronger cooling effects than high-density areas (7.24 vs. 3.92), with the gap widening over time; (2) greenspace area is the dominant predictor of CE, while high-density areas are more sensitive to greenspace configuration and vegetation quality; and (3) optimization of CE produces the largest proportional increase in CB in high-density areas and generally reduces cooling-benefit inequality, although the low-density group shows a slight increase in cooling-benefit inequality. Overall, the study reveals density-dependent differences in greenspace cooling performance, providing critical evidence for targeted greenspace strategies that support more climate-adaptive and equitable urban environments.