Use of Satellite Observations for Improved Air Pollution Exposure Estimates

Accurate exposure estimate is an essential part of long-term cohort studies that examine health effects of air pollution. Many of these estimates rely on air quality data from sparse monitoring stations. More recent investigations, however, have been utilizing Air Quality Models (AQM) to provide the pollution concentration. Therefore, to have a more accurate exposure estimate, it is imperative to improve the accuracy of AQM predictions. One of the challenges for having accurate air quality predictions has been the uncertainties in estimating the biogenic hydrocarbon emissions. Biogenic volatile organic compounds, BVOCs, play a critical role in atmospheric chemistry, particularly in ozone and particulate matter (PM) formation. BVOCs are the dominant summertime source of reactive hydrocarbon in eastern/southeastern United States. Despite significant efforts in the past to improve BVOC estimates, the errors in emission inventories remain a concern. This is partly due to the diversity of the land use/land cover (LU/LC) and complex weather pattern over east/southeast United States, and partly due to model errors in simulating clouds. VOC emissions are highly sensitive to light and errors in model simulated clouds impact the amount of Photosynthetically Active Radiation (PAR) reaching the canopy and thereby significantly impact the emission estimates.

To improve BVOC emission estimates, model derived PAR is replaced with the satellite-based PAR. Model of Emissions of Gases and Aerosols from Nature (MEGAN) is used to generate BVOC estimates. The Geostationary Operational Environmental Satellite (GOES) visible channel observations is utilized to generate PAR. The satellite-based PAR is evaluated against surface observations for summers of 2006 and 2013 and then used to generate BVOC emission estimates for the summer of 2013. This period coincides with the NASA Discover-AQ air quality field campaign over Texas. The emission estimates are evaluated against available observations and the impact on air quality predictions is examined. Preliminary results from this study will be presented.