When more green is not always better: Nonlinear and spatially dependent effects of green space morphology on subjective well-being in compact cities

In high-density urban environments, limited green space challenges residents' subjective well-being (SWB) and social interaction (SI), yet the nonlinear and spatially heterogeneous mechanisms linking spatial morphology to human experiences remain underexplored. This study investigates the nonlinear and spatially dependent effects of community green-space configurations on well-being outcomes. Using Suzhou, China, as a case study, we integrate high-resolution remote-sensing imagery, GIS-based park POIs, social media check-ins, and community-level statistics. Key landscape metrics include Green Distribution Equality (GDE), diversity (GD), shape index (SHAPE), patch density (PD), and mean patch area (MPS). Analyses were conducted using Ordinary Least Squares (OLS), Extreme Gradient Boosting (XGBoost), and spatial autocorrelation models (Moran's I and LISA). Results show that: (1) GDE and GD emerge as the strongest positive correlates of both SWB and SI, with pronounced effects on SI. (2) Nonlinear responses reveal diminishing marginal effects of GDE on SWB (plateauing after index ∼3), with MPS exerting maximal influence within an intermediate range. (3) Spatial patterns exhibit core-periphery differentiation; SWB displays a “southern high-northern low” gradient, with high–high clusters of SWB concentrated in Gusu District and SIP cores; SI shows “core agglomeration,” with high values aligning with commercial and transport nodes. The study develops a multi-source, data-driven framework that reveals the nonlinear threshold effects and spatially heterogeneous pathways through which urban green morphology influences SWB and SI, thereby extending existing research beyond linear and average-effect explanations and offering actionable insights for optimizing green-space design in compact cities.