Using MAIAC Aerosol Products to Estimate PM10 Concentrations in the Southeastern U.S

Acute and chronic exposure to particulate matter has been linked to various adverse health effects. High PM levels including inhalable particles (PM10) and fine particles (PM2.5) are commonly found in large urban centers in the developing world. Unlike PM2.5 whose routine ground monitoring is very sparse, PM10 is regularly measured in many large cities in developing countries. In this analysis, we evaluate the potential for satellite aerosol remote sensing product to estimate PM10 levels. We chose AOD values in 2003 retrieved by the Multiangle Implementation of Atmospheric Correction (MAIAC) algorithm based on MODIS measurements, which has a high spatial resolution of 1 km. Our study area is a 600 km x 600 km region centered in Atlanta, GA. Linear mixed effect (LME) models were developed with MAIAC AOD as the primary predictor variable, meteorology, PM10 emission locations and land use variables as secondary predictor variables. Daily PM10 concentrations measured at ~70 EPA air quality monitoring stations were used as the dependent variable. Model day of year was used as the grouping factor for the random effect of MAIAC AOD. We aggregated AOD and other covariates on 1 km, 3km, 5km and 10km resolution grids and similar LME models were developed for each spatial resolution to compare their abilities to capture the spatial patterns of PM10 mass concentrations at various scales. Our models show that MAIAC AOD, temperature, wind speed and PM 10 emissions source locations are statistically significant predictors of PM 10 at all the spatial scales. Model fitting R2 ranges from 0.35 in winter to 0.56 in the summer. Model performances show a slight decline as the grid resolution decreases. Although the performances of PM10 exposure models are not as good as those of PM2.5 models reported in the literature, these models can still provide spatially resolved PM10 levels at urban scale, which would enable preliminary PM10-related public health research in developing countries.