Nonlinear correlations and scale differences of factors influencing spatial variation in green space carbon sequestration in mainland China

Human activities have significantly contributed to greenhouse gas emissions, thereby exacerbating climate-related challenges. As the principal type of vegetation-covered terrain, green spaces are crucial in absorbing atmospheric carbon dioxide. This study examined the spatiotemporal patterns and determinants of carbon sequestration in green spaces across mainland China from 2003 to 2023, utilizing multi-source data and applying the XGBoost model coupled with SHAP for analytical interpretation. The findings reveal: (1) A consistent annual increase in carbon sequestration, peaking at 8.45 billion tons in 2023, with marked spatial heterogeneity aligned with the Hu Huanyong Line theory. (2) The carbon sequestration in green spaces is influenced by natural environmental conditions, climatic factors, and landscape patterns, exhibiting pronounced nonlinear characteristics. (3) The factors driving carbon sequestration are scale-dependent: at broader spatial scales, natural environment and climate are predominant, whereas at medium and smaller scales, landscape patterns have a more pronounced impact. • The amount of carbon sequestered in China reached 8.45 billion tons and showing significant spatial variations. • Green space carbon sequestration is primarily influenced by natural environment, climate and landscape pattern including NDVI, PLAND, precipitation, COHESION, TE, elevation and temperature. • Landscape patterns affect green space carbon sequestration, with area-based indicators having a greater impact than structure-based ones. • Different indicators exhibit varying nonlinear effects on green space carbon sequestration, and influencing factors differ across scales.