Abstract: The changes in carbon sinks within urban agglomerations are crucial for national ecological security, and clarifying the driving mechanisms behind these changes is of significant importance for achieving the national 'dual carbon' goals (carbon peaking and carbon neutrality). This study combines the CASA carbon sink estimation model with the LightGBM-SHAP model, overcoming the limitations of traditional linear models and enabling multi-factor threshold identification and nonlinear impact analysis. Taking the Changsha-Zhuzhou-Xiangtan Urban Agglomeration, Hunan Province as a case study, the research first uses the CASA model to estimate carbon sinks from 2000 to 2022. Subsequently, it employs the LightGBM model in conjunction with the SHAP method to reveal the nonlinear impacts of various driving factors on carbon sink changes. The results indicate that the total carbon sink experienced a slight decline from 2000 to 2010, followed by a significant rebound from 2010 to 2022, with a cumulative increase of approximately 11.90%. The spatial pattern shifted from high-value clustering in the northeast to a more spatially balanced distribution. Land use configuration indicators, such as forest aggregation degree, cropland aggregation degree, and non-construction land connectivity, exerted significant influences on carbon sink changes, collectively accounting for 58.43% of the total importance value ratio. Moreover, these indicators demonstrated notable enhanced effects (synergistic effects; here, 'enhanced effects' is used to convey the meaning of significant improvement after exceeding thresholds) after surpassing their respective thresholds. Based on these findings, it is recommended to strengthen the construction of ecological corridors and high-standard farmland, optimize the spatial pattern of land use, and establish a multi-level collaborative governance system encompassing green urban integration, ecological restoration, and green heart protection. These measures aim to facilitate the coordinated advancement of green transformation goals within urban agglomerations and the national 'dual carbon' goals.