Substantial Differences in Source Contributions to Carbon Emissions and Health Damage Revealed by Adjoint Modeling

China's dual strategy to mitigate climate change and air pollution is constrained by insufficient data on the distinct sources of carbon emissions and associated health damages. This study utilizes adjoint emission sensitivity modeling with the CMAQ Adjoint model, alongside an exposure-response model and a multiregional input-output model, to perform high-resolution source attribution across 53 production sectors and fuel/process combinations, as well as 42 consumption economic sectors. Our analysis uncovers significant discrepancies between sources of CO₂ emissions and PM_2.5-related premature mortality, with monetized health damages surpassing climate impacts in over half of the subsectors examined. Additionally, more than one third of the CO₂ emissions and health damages are outsourced from well-developed coastal provinces to less-developed inland provinces, though the regions absorbing these burdens differ geographically. CO₂ emissions are primarily shifted to the northwestern region, which relies heavily on coal as an energy source, while PM_2.5-related deaths are concentrated in the central region, the heavy industrial hub of China with high population densities. These findings demonstrate that high population densities and lower adoption of control technologies exacerbate health damages, particularly in downstream provinces that bear the majority of emission leakages. The CMAQ Adjoint model’s capability to evaluate the marginal benefits of emission reductions at a granular level enabled precise source attribution by incorporating emission profiles, population density, and atmospheric conditions. This research underscores the critical advantage of adjoint modeling in integrating health and climate impacts, advocating for tailored mitigation strategies that address emissions from both production and consumption sides to achieve balanced decarbonization and effective health risk mitigation in China.