Our research group uses earth observation data from MODIS instruments to construct spatiotemporal exposure models for PM2.5 air pollution and temperature to be used in human epidemiologic studies for public health. We will present how these satellite-derived estimates of aerosol optical depth (AOD) and land surface temperature (LST) are key predictors and how we reconstruct surface conditions in the absence of satellite data. We will also present results from applying machine-learning algorithms (XGBoost, Gradient Boosted Models, and Random Forests) to characterize AOD estimates from MAIAC versus AERONET stations over the US Northeast and how the resulting correction for retrieval-biases can improve prediction of ground conditions. We demonstrate how machine learning with quality control and spatial features substantially improves satellite-derived AOD products for air pollution modeling.
Visit Host: Alexei.I.Lyapustin@nasa.gov