[A-CG41] Satellite Earth Environment Observation

[ACG41-05] Reconstruction of PM_2.5 Concentrations in East Asia and Assessment of Potential Exposure Levels from 1981 to 2020

*Shuai Yin¹ (1.Aerospace Information Research Institute, Chinese Academy of Sciences)

Air pollution is a major global environmental issue that negatively impacts both human health and the earth-atmosphere system. According to the World Health Organization (WHO), nearly 99% of the global population breathes air that exceeds WHO standards, leading to approximately seven million premature deaths annually. PM2.5, referring to particulate matter with a diameter of less than 2.5 um, is widely regarded as the most representative air pollutant. Since the early 2000s, global and regional PM2.5 estimates have primarily relied on satellite data, resulting in a scarcity of spatial distribution datasets for ambient PM2.5 prior to 2000. The lack of continuous and consistent PM2.5 datasets has become a major obstacle to long-term epidemiological research and policy assessments spanning several decades. In this study, we combined ground-based measurements with satellite- and model-derived PM2.5 estimates to develop a PM2.5 distribution dataset for East Asia (EA) spanning 1981 to 2020. The proposed framework employs three machine learning approaches, with a novel wide-deep model designed to integrate and enhance prediction results. Using the constructed dataset, we analyzed PM2.5 exposure levels in EA over the past four decades. With rapid economic growth and rising emissions from fossil fuel consumption, EA has experienced severe and persistent air pollution since the 1980s. The findings of this study offer crucial insights and theoretical support for quantifying and characterizing the spatiotemporal variations in PM2.5 across EA. Moreover, these results have important implications for developing more effective mitigation strategies to reduce the adverse health effects of PM2.5 exposure in the future.
The results demonstrate that the framework effectively capitalized on the advantages of satellite observations (higher accuracy) and model-based estimations (longer temporal coverage) for assessing surface air pollutants. The reconstructed PM2.5 concentrations showed strong agreement with ground-based measurements, achieving a coefficient of determination (R2) of 0.99 and a root-mean-square error (RMSE) of 1.38 ug m-3, outperforming satellite-based PM2.5 estimates. Furthermore, as more ground measurements were incorporated into the model for training, the average RMSE decreased to 0.83 ug m-3 in Japan and 1.50 ug m-3 on the Korean Peninsula. Simultaneously, on the basis of the reconstructed datasets, we further analyzed PM2.5 exposure levels in EA from 1981 to 2020. Since 2000, rising anthropogenic emissions have significantly deteriorated air quality in the region, with nearly 50% of the population living in areas where the annual average PM2.5 concentration exceeded 50 ug m-3 during 2009-2010. Although local authorities have implemented various mitigation strategies to reduce ambient PM2.5 levels, achieving compliance with WHO air quality guidelines remains challenging across EA. Moreover, factors such as population aging and climate change could further elevate PM2.5 exposure risks in the future. To mitigate the health impacts associated with PM2.5, policymakers in EA must account for these factors and develop more effective air quality management strategies.