Impact of street-block built environment on the synergistic effect of carbon emissions and pollution in Hefei based on machine learning

Human sustainable development faces dual challenges from climate change and air pollution, making the coordinated promotion of pollution reduction and carbon emission reduction a global consensus. This study exemplifies the main urban area of Hefei City to investigate the influencing factors of the carbon-pollution synergy effect from the perspective of the built environment. Initially, high-resolution urban carbon emission distributions and PM2.5 concentration distributions were derived through the inversion of remote sensing data and energy statistics, utilizing nighttime light data corrected with land-use functional coefficients to enhance spatial accuracy. Multi-source spatial data for 2018 were harmonized to a unified 500 m grid to ensure analytical consistency. Subsequently, the coupling coordination degree model was employed to examine the synergy between carbon emissions and air pollution at the urban block scale, revealing spatially aggregated patterns of synergistic intensity. Finally, the XGBoost model, chosen for its superior capability in handling nonlinear relationships and high-dimensional data, was utilized to quantify the correlations between the five dimensions of built environment factors—geographical location, land use function, development intensity, traffic accessibility, and environmental quality—and the carbon-pollution synergy effect. The SHAP method was further applied to interpret the model and assess the contribution of each built environment indicator. The findings reveal two key aspects: 1) A significant synergy exists between carbon emission intensity and PM2.5 concentration in the main urban area of Hefei City, spatially forming a belt-shaped region encompassing “Xinzhan District-Old Urban Area and Baohe District-Economic Development Zone”, with a gradual decrease towards the northwest and southeast, as confirmed by spatial autocorrelation analysis. 2) The relative contributions of the five types of built environment indicator variables to the synergy effect, ranked in descending order, are: land use function>development intensity>environmental quality>traffic accessibility> geographical location. Notably, the floor area ratio, industrial land proportion, vegetation coverage, public service facility land proportion, and water coverage emerge as the five most influential built environment indicators. This study addresses the gap in block-level research. It holds substantial significance for elucidating the underlying influence mechanisms of the built environment on the carbon-pollution synergy effect. Furthermore, the findings provide a scientific basis for formulating targeted urban planning strategies and policies, ultimately supporting the goal of achieving coordinated pollution reduction and carbon emission reduction.

人类可持续发展面临气候变化与空气污染的双重挑战，推动减污降碳协同增效已成为全球共识。本研究以合肥市主城区为研究区域，从建成环境视角探究碳污染协同效应的影响因素。首先，本研究通过遥感数据与能源统计反演，获取高精度城市碳排放分布与PM2.5浓度分布；为提升空间精度，采用土地利用功能系数对夜间灯光数据进行校正。将2018年多源空间数据统一聚合至500米网格尺度，以保障分析的一致性。随后，采用耦合协调度模型在城市街区尺度下分析碳排放与空气污染的协同关系，揭示协同强度的空间集聚特征。最后，鉴于XGBoost（Extreme Gradient Boosting）模型在处理非线性关系与高维数据方面的优异性能，本研究采用该模型量化建成环境五大维度——地理位置、土地利用功能、开发强度、交通可达性与环境质量——与碳污染协同效应的关联关系，并进一步借助SHAP（SHapley Additive exPlanations）方法对模型进行解释，评估各项建成环境指标的贡献度。研究结果主要包含两点：其一，合肥市主城区碳排放强度与PM2.5浓度存在显著协同效应，经空间自相关分析验证，其空间分布形成新站区—老城区与包河区—经济开发区的带状区域，且向西北、东南方向逐步递减。其二，五类建成环境指标变量对协同效应的相对贡献度从高到低依次为：土地利用功能＞开发强度＞环境质量＞交通可达性＞地理位置。其中，容积率、工业用地占比、植被覆盖率、公共服务设施用地占比与水域覆盖率为影响程度最高的五项建成环境指标。本研究填补了街区尺度相关研究的空白，对于阐明建成环境对碳污染协同效应的内在作用机制具有重要意义。同时，本研究结果可为制定精准化城市规划策略与政策提供科学依据，助力实现减污降碳协同增效的目标。