Spatiotemporal patterns and nonlinear drivers of MSW carbon emissions from the perspective of consumption and industrial structure upgrading: An explainable machine learning approach

Municipal solid waste (MSW) carbon emissions pose a major challenge to urban sustainable development, yet nonlinear drivers across different urban development contexts remain insufficiently understood. Based on consumption upgrading and industrial structure upgrading, this study classifies 284 prefecture-level and above cities in China into four development types, estimates MSW carbon emissions from treatment processes during 2010–2023, and constructs an XGBoost–SHAP model to identify key drivers, nonlinear thresholds, and interaction contributions. The results show that MSW carbon emissions followed an overall trajectory of “growth–decline–slight rebound,” with comprehensive upgrading cities, despite being the smallest group, maintaining the highest long-term emission levels. Urban construction scale and population size are identified as core drivers, whereas waste treatment structure upgrading shows a negative contribution. Comprehensive upgrading cities are characterized by joint drivers related to scale agglomeration, economic activity, and governance capacity; household consumption is a more prominent driver in consumption-led upgrading cities; environmental sanitation infrastructure emerges as a more prominent driver in industry-led upgrading cities; and scale-related factors remain the primary drivers in foundational cities. From the combined perspective of demand-side consumption changes and supply-side industrial structure adjustment, this study reveals heterogeneous drivers of MSW carbon emissions and provides insights for city-type-specific low-carbon MSW governance strategies.