Atmospheric particulate matter (PM) estimation from SeaWiFS imagery

Air quality monitoring by satellite imagery is a potential complementary tool to in-situ measurements. Deriving particulate matter (PM) from the satellite derived aerosol optical thickness (AOT) is not straightforward. The relation between the two quantities is not explicit or, if so, too many unknowns exist. The idea, proposed by Pelletier et al. [Pelletier, B., Santer, R., & Vidot, J. (in press). Retrieving of particulate matter from optical measurements: A semiparametric approach. Journal of Geophysical Research, 112. doi:10.1029/2005JD006737], is to involve in the associations PM-AOT auxiliary meteorological information using a statistical approach. We use this method first on optical ground measurements and second on SeaWiFS imagery in the framework of a regional European project. A SeaWiFS archive was processed both over land and over ocean to produce AOT. A database was developed to associate locally and daily SeaWiFS AOT and PM. A part of this database was used to train the statistical method; another part of it was used for validation. Results are promising. Overall, the method allows retrieving the PM10 with a coefficient of determination of 0.42 and the PM2.5 with a coefficient of determination of 0.48. The outputs of this work are PM10 SeaWiFS maps and a limited comparison with in-situ data is reported.