ALBERT

Description: The Multi-Angle Imaging Spectro-Radiometer (MISR) instrument on NASA's Terra satellite can provide more reliable Aerosol Optical Depth (AOD, τ) and more particle information, such as constraints on particle size (Angström exponent or ANG, α), particle shape, and single-scattering albedo (SSA, ω), than many other satellite instruments. However, MISR's ability to retrieve aerosol properties is weakened at low AOD levels. When aerosol-type information content is low, many candidate aerosol mixtures can match the observed radiances. We propose an algorithm to improve MISR aerosol retrievals by constraining MISR mixtures' ANG and absorbing AOD (AAOD) with Goddard Chemistry Aerosol Radiation and Transport (GOCART) model-simulated aerosol properties. To demonstrate this approach, we calculated MISR aerosol optical properties over the contiguous US from 2006 to 2009. Sensitivities associated with the thresholds of MISR-GOCART differences were analyzed according to the agreement between our results (AOD, ANG, and AAOD) and AErosol RObotic NETwork (AERONET) observations. Overall, our AOD has a good agreement with AERONET because the MISR AOD retrieval is not sensitive to different mixtures under many retrieval conditions. The correlation coefficient (r) between our ANG and AERONET improves to 0.45 from 0.29 for the MISR Version 22 standard product and 0.43 for GOCART when all data points are included. However, when only cases having AOD 〉 0.2, the MISR product itself has r ~ 0.40, and when only AOD 〉 0.2 and the best-fitting mixture are considered, r ~ 0.49. So as expected, the ANG improvement occurs primarily when the model constraint is applied in cases where the particle type information content of the MISR radiances is low. Regression analysis for AAOD shows that MISR Version 22 and GOCART misestimate AERONET by a ratio (mean retrieved AAOD to mean AERONET AAOD) of 0.5; our method improves this ratio to 0.74. Large discrepancies are found through an inter-comparison of the spatial-temporal patterns of MISR, GOCART, and our adjusted aerosol optical properties. We attribute these differences to (1) GOCART underestimations of AOD and ANG in polluted regions due to the emissions inventories used, and not considering the fine particles such as nitrate, (2) a lack of certain aerosol mixtures in the Version 22 algorithm climatology, (3) a lack of sensitivity in the MISR radiances to particle type under some conditions, and (4) parameters and thresholds used in our method.