(8h) A Hybrid Approach for Developing Time-Space Resolved Estimates of Ambient Air Pollution Exposure

Integrating satellite remote sensing, data assimilation techniques and spatiotemporal Kriging, we demonstrate how to develop time-space resolved estimates of exposure to ambient particulate of different sizes (PM) for the health effect studies. The proposed methodology is implemented in three sequential steps. First, using the empirical relationship between aerosol optical depth (AOD) from satellites and in situ PM measurements at EPA sites, predict PM for all valid AOD data points. Second, employ data assimilation techniques to validate and improve the predicted PM. Third, employ spatiotemporal Kriging to develop exposure estimates (using the predicted PM) at the required spatiotemporal scales, such as daily PM estimate at any point location.

To demonstrate the application of the Hybrid approach, daily estimates of PM (PM_2.5, PM₁₀ and PM_10-2.5) will be developed from 2000 to 2009 at 2.5km spatial resolution for Cleveland Metropolitan Statistical Area (MSA). The 2km AOD (~2.3 million data points at 2km spatial resolution from 2000 to 2009) were extracted using the data from Moderate Resolution Imaging Spectroradiometer (MODIS), aboard Terra and Aqua satellites. In situ measurements of PM were acquired from EPA and meteorological data from the National Climatic Data center. Using these data a systematic grid of daily PM was developed at 2.5km spatial resolution for the Cleveland MSA from 2000 to 2005; 90% of the predicted values were robust. The proposed methodology is likely advance the exposure modeling to the next level by providing robust and reliable PM estimate at the (desired) spatiotemporal scales of health data.