Improved O3 attribution and health impact assessment using RSM-based inverse emission inventories in the greater bay area.

Ground-level ozone (O3) remains a persistent air-quality challenge in large urbanized regions, driven by nonlinear photochemical interactions between nitrogen oxides (NOX) and volatile organic compounds (VOCs), as well as regional transport across administrative boundaries. In the Guangdong-Hong Kong-Macao Greater Bay Area (GBA), differences in emission structures and inventory methodologies among jurisdictions, together with COVID-19-related activity disruptions in 2020, introduce substantial uncertainties into bottom-up precursor estimates and limit the reliability of O3 attribution and exposure assessment. In this study, a Polynomial Function based Response Surface Model (pf-RSM) was embedded in an emissions inversion framework to iteratively refine NOX and VOCs emissions in the GBA for 2017 and 2020. The framework produced inversion-based emission inventories that corrected systematic biases in bottom-up estimates, including a 2.9 × 104 t overestimation of NOX and a 6.1 × 104 t underestimation of VOCs in 2017. These corrections substantially improved O3 simulations, reducing normalized mean error by 19-24% and enhancing spatial agreement with observations. With the refined emission fields, O3 concentrations were quantitatively decomposed to local photochemical formation and regional transport, revealing a structural transition in the O3 formation regime: local production declined by 27-37% in inland manufacturing centers, whereas regional inflow accounted for 35-45% of total O3 in coastal and downwind cities. Emission and meteorological changes together prevented approximately 5600 premature deaths between 2017 and 2020, but substantial inter-city heterogeneity was observed, with changes in population size and structure offsetting up to 60% of the achievable health gains in several regions. Cities heavily influenced by transported pollution from upwind industrial areas experienced limited improvements despite local emission controls. Taken together, these results demonstrate that the pf-RSM approach provides a scalable, integrated framework for refining emissions, diagnosing O3 formation mechanisms, and quantifying city-specific exposure and health outcomes, thereby supporting equitable, population-focused O3 mitigation in large coastal urban regions.