Comparison and Evaluation of Different MODIS Aerosol Optical Depth Products Over the Beijing-Tianjin-Hebei Region in China

• Published in: IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
• Main Authors: Jing Wei, Lin Sun
• Format: Article in Journal/Newspaper
• Language: English
• Published: IEEE 2017
• Subjects: Aerosol optical depth (AOD); Beijing-Tianjin-Hebei; dark target (DT); deep blue (DB); moderate resolution imaging spectroradiometer (MODIS); high-resolution aerosol retrieval algorithm with a prior LSR database support (HARLS); Ocean engineering; TC1501-1800; Geophysics. Cosmic physics; QC801-809; Aerosol Robotic Network
• Online Access: https://doi.org/10.1109/JSTARS.2016.2595624; https://doaj.org/article/31d0fcf67df94df8b53b35e662c9ab4f

Many aerosol retrieval algorithms based on the remote sensing technology have been developed and applied to produce aerosol optical depth (AOD) products for different satellite sensors. The dark target (DT) and deep blue (DB) algorithms are two main MODIS aerosol retrieval algorithms that are suitable for dark or bright areas. The estimation of land surface reflectance (LSR) is necessary to improve the accuracy of AOD retrievals. Therefore, in this paper, a new procedure to improve LSR estimation using MODIS surface reflectance products is developed. A new high-resolution <;1000 m> aerosol retrieval algorithm with a priori LSR database support (HARLS) is proposed. The purpose of this paper is to evaluate the spatial adaptability of different MODIS AOD products produced by the above three algorithms. The Beijing-Tianjin-Hebei (Jing-Jin-Ji) region, which features complex surface structures and serious air pollution, was chosen as the study area, and the different AOD products are validated using aerosol robotic network (AERONET) AOD ground measurements from four stations located in dark and bright areas. Compared with the DT retrievals (R &#x2248; 0.88 - 0.95), the C6 DB AOD retrievals yield a stronger correlation (R &#x2248; 0.94 - 0.97) with AERONET AOD and lower RMSE, MRE and MAE values, resulting in approximately 20%-30% less average overestimation. The C6 DT&DBAODresultsshowaretrievalquality (R &#x2248; 0.93 - 0.97) similar to that of DB, with approximately 50%-70% of the collections falling within the expected error (EE). Moreover, DT&DB is much better than DT, with more than approximately 10%-20% of the collections falling within the EE. However, HARLS achieves a high correlation (R &#x2248; 0.93 - 0.96) with the AERONET AODs, with low RMSE (&#x2248; 0.118 - 0.128) and MAE (&#x2248; 0.09 - 0.12) and small offsets (intercept 0.00 - 0.04). HARLS retrievals exhibited 7%-8% less uncertainty than the C6 DB retrievals, 37%-38% less uncertainty than the C6 DT&DB ...