Blue–Green Space Provision and Health-Related Well-Being Perception: Spatial Matching and Environmental Associations in Shanghai

In high-density cities, blue–green space provision and residents’ health-related well-being perceptions may exhibit spatial mismatches. Identifying these mismatches is important for improving land-resource allocation and prioritizing renewal interventions. Using Shanghai as a case study, this study analyzed 5840 geotagged Weibo posts from 2022–2024. An ERNIE–LSTM-CNN model was used to construct the Health-related Well-being Perception Index (HWPI), while remote sensing, GIS, and street-view data were integrated to construct the Blue–Green Infrastructure Index (BGI). BGI and HWPI were aggregated to 216 subdistricts/towns for spatial matching classification, with the resulting types represented on the 500 m grid. XGBoost and SHAP were used to identify key environmental factors and nonlinear contributions, with spatial interpretation informed by the GeoShapley framework. Matched types covered 64.73% of the study area, while mismatched types covered 35.27%; high-BGI–low-HWPI alone accounted for 29.87% and was concentrated around the central urban area and in parts of Pudong and southern Shanghai. PM2.5, distance to green space, relative humidity, and land surface temperature were the most important predictors. The findings support differentiated blue–green space management focused on accessibility, environmental quality, perceived benefits, and renewal priorities.