Development and Validation of a Robust Algorithm for Retrieving Aerosol Optical Depth Over Land From MODIS Data

Due to the limitation of surface “dark-target,” in some regions and over certain surface types, Moderate Resolution Imaging Spectroradiometer (MODIS) standard aerosol algorithm does not work very well. In this paper, we developed and validated a robust algorithm, which combines the revised minimum reflectance technique and the technique of the synergy of Terra and Aqua MODIS to eliminate the dependence on surface conditions. The rationale of our algorithm is to first identify the “clearest” day observations in certain temporal window, assume that the aerosol optical depth (AOD) does not change during the limited period (0-3 h) between Terra and Aqua overpass in this day, subsequently obtain the relationships between visible bands and 2.1 μm band surface reflectance, and finally retrieve the AOD of all observations in the temporal window. The algorithm was validated using year 2006 measurements from 13 Aerosol Robotic Network (AERONET) sites distributed in North China, Central Asia, eastern United States, and Western Europe. For AOD, 65.5% and 72.7% of retrievals within MOD04 expected error envelope for all levels of quality assurance confidence (QAC) and QAC = 3, respectively, R² of 0.87 and a less amount of gaps are found. In comparison to the MODIS aerosol products, the new algorithm on the whole shows similar accuracy over dark land surface, but gives a larger amount of retrievals over bright land surface. This indicates that the new algorithm can provide a more steady inversion for a variety of surface types. Especially over urban surface, such as Beijing in China, the inverted AOD is clearly better than that from MODIS aerosol products. Therefore, the new algorithm can provide an alternative way for AOD retrieving over regions where the standard dark target algorithm does not work well or higher spatial resolution is imperative.