Investigating cloud contamination in MODIS and VIIRS AOD retrievals and the impacts on air quality applications

Satellite remote sensing can provide the air quality monitoring and near-operational forecast by relating satellite retrieved AOD to PM2.5. We examine the probability and magnitude of cloud contamination in MODIS Aerosol Optical Depth (AOD) retrieval then relate the AOD bias to uncertainties in the remote sensed air quality monitoring. The study uses a one-year global dataset of the collocated MODIS and CALIOP sensors. Our assessment shows that filtering with MODIS AACF (Aerosol algorithm Cloud Fraction) equal to zero, 64.22% of land and 17.42% of ocean MODIS AOD retrieval cases are still cloud contaminated identified by collocated CALIOP cloud detection. The aerosol retrieval near cloud pixels affects on derived AOD with positive biases, which will directly lead to an overestimation of the PM2.5 levels. We demonstrate that cloud contamination in the MODIS AOD retrieval over land results in an over estimation of the global mean AOD 0.13 to 0.23, equivalent to over estimating PM2.5 by 6μg/m3.