Quantitative Evaluation of Dust and Black Carbon Column Concentration in the MERRA-2 Reanalysis Dataset Using Satellite-Based Component Retrievals

Abstract

The aerosol optical property products of Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2) reanalysis dataset have been extensively investigated on a global or regional scale. However, the understanding of MERRA-2 aerosol component products on an extensive temporal and spatial scale is inadequate. Recently, the aerosol component products have been derived from the observations of Polarization and Directionality of the Earth’s Reflectances/Polarization and Anisotropy of Reflectance for Atmospheric Science coupled with observations from a Lidar (POLDER/PARASOL). This study presents a quantitative evaluation of the MERRA-2 reanalysis dust and black carbon (BC) column concentration using independent satellite-based aerosol component concentration retrievals. Both GRASP/Component and MERRA-2 reanalysis products can capture well the temporal variation in dust column concentration over the dust emission resource and downwind dust-dominated regions with the correlation coefficient (R) varying from 0.80 to 0.98. MERRA-2 reanalysis dust products present higher column concentration than GRASP/Component dust retrievals with relative differences of about 20~70%, except in the Taklamakan Desert and Bay of Bengal, where the relative differences can be negative. The differences in dust column concentration over the African dust regions are larger than that over the Asian dust regions. Similar temporal variations in BC column concentration are characterized by both GRASP/Component BC retrievals and MERRA-2 BC products with R of about 0.70~0.90, except in the North China Plain region. We should pay more caution with the regional applicability of MERRA-2 component products when large differences and high correlation coefficients are obtained simultaneously. The results are favorable for identifying the behavior of MERRA-2 reanalysis component estimation in a new view and demonstrate a practical application of the satellite-based component retrievals, which could make more contributions to the improvement of model estimation in the near future.