Performance of DSCOVR EPIC Atmospheric Correction Algorithm

The Earth Polychromatic Imaging Camera (EPIC) on board the Deep Space Climate Observatory provides offers a unique view of the Earth and thus is of high interest to climate, atmosphere, hydrology, and ecology studies. The unique orbit and choice of spectral channels also pose challenges for aerosol retrievals and atmospheric correction: (1) the ubiquitous presence of partial cloudiness due to EPIC's large pixel size; and (2) the potential ambiguity between aerosols and thin clouds without an infrared channel. We adopted the MultiAngle Implementation of Atmospheric Correction (MAIAC) algorithm originally developed for MODIS to EPIC data. The key improvement of MAIAC is the synergy of spectral, spatial, and temporal constraints to better determine cloud mask and contribution of surface reflectance. The aerosol optical depth (AOD) retrievals based on 2016-2017 EPIC observations are compared against the AERONET data. It is found that EPIC AODs averaged over the 7x7 windows around AERONET sites are correlated with AERONET values with a correlation of 0.69. The mean root mean square error (rmse) is 0.17 and the mean bias is 0.03. The atmospherically corrected surface reflectance and the calculated Normalized Difference Vegetation Index also show plausible diurnal variation and seasonal cycle.