Seminar: Tracking Air Pollution from Space Using Artificial Intelligence

Abstract:

Exposure to ambient air pollution, including fine particulate matter (PM_2.5) and trace gases like ozone (O₃) and nitrogen dioxide (NO₂) at the ground level, poses serious threats to environmental quality and public health, significantly increasing the risk of death. Satellite remote sensing allows for generating spatially continuous PM_2.5 data, but many current datasets have overall low accuracies with coarse spatial resolutions and large gaps limited by data sources and models. Air pollution levels in developing countries like China have experienced dramatic changes over the past couple of decades. To reveal the spatiotemporal variations, artificial intelligence, including machine and deep learning, is extended by considering the spatiotemporal heterogeneity of air pollution to generate long-term and high spatiotemporal-resolution datasets of outdoor air pollutants from big data that integrate ground-based measurements, satellite remote sensing products, atmospheric reanalysis, and model simulations. The application and fidelity of the dataset are demonstrated by analyzing their spatial distributions and temporal variations of surface air pollution, exposure risk to the public, and the COVID-19 pandemic. These novel products have been widely employed to address a variety of atmospheric, environmental, ecological, and public health issues.