Improving PM<sub>2.5</sub> retrievals in the San Joaquin Valley using A-Train Multi-Satellite Observations
Abstract. This paper demonstrates the use of a combination of multi-platform satellite observations and statistical data analysis to dramatically improve the correlation between satellite observed aerosol optical depth (AOD) and ground-level retrieved PM2.5. The target area is California's San Joaquin Valley which has a history of poor particulate air quality and where such correlations have not yielded good results. We have used MODIS AOD, OMI AOD, AAOD (absorption aerosol optical depth) and NO2 concentration, and a seasonal parameter in a generalized additive model (GAM) to improve retrieved/observed PM2.5 correlations (r2 at six individual sites and for a data set combining all sites. For the combined data set using the GAM, r2 improved to 0.69 compared with an r2 of 0.27 for a simple linear regression of MODIS AOD to surface PM. Parameter sensitivities and the effect of multi-platform data on the sample size are discussed. Particularly noteworthy is the fact that the PM retrieved using the GAM captures many of the PM exceedences that were not seen in the simple linear regression model.