Exploring spatio-temporal heterogeneity of rural settlement patterns on carbon emission across more than 2800 Chinese counties using multiple supervised machine learning models.

China, the world's largest carbon emitter, plays a pivotal role in achieving carbon neutrality. This study systematically analyzes the impact of landscape indices on carbon emissions from rural settlements across more than 2800 counties using eight supervised machine learning models. To assess variable influences under diverse conditions, we also employed the SHapley Additive exPlanations (SHAP) and Accumulated Local Effects (ALE) methods. From 2000 to 2020, carbon emissions in China increased significantly, with the highest regional growth in the Northeast, surging by 259.52% to 10.199 million tons per year. After identifying the Gradient Boosted Regression Trees (GBRT) model as most effective, our findings reveal that the Mean Patch Area (MPA) index had a greater influence on emissions compared to Patch Density (PD), Edge Density (ED), and Aggregation Index (AI). Each index demonstrated unique impact characteristics and varied trends across different regions. These findings are crucial for crafting targeted environmental policies and advancing sustainable development goals.