Global validation and comparison of MODIS DT and POLDER-3 GRASP aerosol products over ocean

Abstract

The Moderate Resolution Imaging Spectroradiometer (MODIS) and the third generation of Polarization and Directionality of Earth's Reflectances (POLDER-3) are two types of sensors, which have their own advantages in sensor design. MODIS is a conventional multi-spectral imager with single-angle observations and a wide spectral range of surface reflection bands from blue to shortwave infrared, while POLDER-3 is a multi-angle polarization instrument with narrower spectral range from blue to shortwave near-infrared. It is highly anticipated to understand the quantitative capability of these two sensors in aerosol retrievals. For this purpose, this study conducts a long-time series comparison and evaluation of over-ocean aerosol products from these two types of sensors during the period 2005–2013, corresponding to the entire lifetime of the POLDER-3 products. The MODIS aerosol products based on Dark Target (DT) algorithm and two POLDER-3 products based on General Retrieval of Atmosphere and Surface Properties (GRASP) “High Precision” (GRASP/HP) and “Models” (GRASP/Models) algorithm are adopted in this study. Two types of ground-based measurements from automatic Aerosol Robotic Network (AERONET) data and shipborne Maritime Aerosol Network (MAN) data are employed to validate four parameters common to both satellite products, including total Aerosol Optical Depth (AOD), Ångström Exponent (AE), fine-mode AOD (AODF) and coarse-mode AOD (AODC). The overall validation results show GRASP/Models has similar performance with MODIS DT in total AOD products. This indicates that the multi-angle polarization observations of POLDER-3 can have the same advantages as the multi-spectral observations of MODIS in total AOD retrievals. Furthermore, the statistical parameters of the validation results based on AERONET data show that GRASP/Models is slightly better than MODIS DT. However, for the validation results using MAN data, it is interesting to find that MODIS DT is slightly better than GRASP/Models in all aspects. This may be due that the manual measurements of MAN can consciously avoid more cloud contamination. Meanwhile, it suggests that MODIS DT algorithm over ocean needs further improvements in cloud mask and quality control. In addition, from the validation results of both types of ground-based data, it shows that GRASP/HP significantly overestimates AOD, but has better validation accuracy than GRASP/Models and MODIS DT in AE products, which implies that multi-angle polarization sensor has great potential in aerosol properties retrieval. For AODF and AODC, the statistical parameters of validation results for all three products show little difference. GRASP/HP is slightly better than GRASP/Models and MODIS DT in AODF products, while the AODC products of GRASP/Models and MODIS DT show slightly superior to that of GRASP/HP. This may suggest that although the higher AE validation accuracy of GRASP/HP, the higher total AOD accuracy of GRASP/Models and MODIS DT still dominates the validation results of AODF and AODC. The results could contribute to the application of over-ocean aerosol products of these two satellite in climate research, as well as to the improvement of aerosol retrieval algorithms over ocean for instruments such as single viewing spectrometer and multi-angle polarimeter.