Are green spaces healthier in spring? Examining seasonal variations in the association between urban green spaces and emotional well-being in Wuhan, China.

Background: Urban green spaces (UGS) are widely recognized as environmental elements closely associated with emotional health. However, most existing studies rely on static or annual average indicators to examine associations between green spaces and mental health, while paying insufficient attention to seasonal dynamics, multidimensional UGS characteristics (including quantity, quality, and landscape patterns), and nonlinear and spatially heterogeneous associations with negative emotions. Clarifying these seasonally varying associations can provide a more practical basis for urban green space planning. Methods: Using geotagged Weibo data from Wuhan in 2022, we identified negative emotional expressions with a fine-tuned BERT model and constructed a population-standardized indicator of negative emotions. Based on remote sensing imagery, street view images, and POI data, we developed a multidimensional evaluation framework of UGS characteristics. To account for the particular social context of 2022, lockdown intensity was incorporated as a control variable in all models. We applied geographically weighted random forest (GWRF) and SHAP-based interpretation to examine seasonal associations between UGS characteristics and negative emotions, focusing on variable importance, nonlinear threshold patterns, and spatial heterogeneity across four seasons. Results: After population standardization of the negative emotion indicator and adjustment for lockdown intensity, negative emotions still showed clear seasonal fluctuations and significant spatial clustering. NDVI, green space accessibility, and green space openness were identified as key correlates of negative emotions. Among them, green space accessibility showed association with lower negative emotions in spring than in other seasons. Several major factors exhibited consistent nonlinear turning points across seasons, although the magnitudes and directions of their associations varied seasonally, with more pronounced changes in spring and milder changes in winter. In addition, associations between UGS characteristics and negative emotions displayed substantial spatial heterogeneity, while the broad spatial pattern remained similar across seasons. Conclusion: Associations between UGS characteristics and negative emotions are seasonally dynamic rather than constant throughout the year. In particular, spring may represent a priority period during which green space accessibility and usability are more strongly associated with lower negative emotions. These findings highlight the importance of incorporating seasonal dynamics into UGS research and provide evidence for more adaptive, context-sensitive green space planning.