Evaluation of LaSRC aerosol optical depth from Landsat-8 and Sentinel-2 in Guangdong-Hong Kong-Macao greater bay area, China. (1st July 2022)

• Record Type: Journal Article
• Authors: Jin, Yangyang; Hao, Zengzhou; Huang, Haiqing; Wang, Tianyu; Mao, Zhihua; Pan, Delu
• Is Part Of: Atmospheric environment. Volume 280(2022)
• Publication Date: 2022-07-01
• Subjects: LaSRC -- AOD -- Landsat-8 -- Sentinel-2 -- GBA; Air -- Pollution -- Periodicals; Air -- Pollution -- Meteorological aspects -- Periodicals; 551.51
• DOI: 10.1016/j.atmosenv.2022.119128
• Languages: English
• ISSNs: 1352-2310

Abstract

Abstract: The aerosol optical depth (AOD) is a key factor to assess atmospheric environment, and accurate monitoring of it is conducive to the research and environmental protection. The Land Surface Reflectance Code (LaSRC) algorithm has been used to retrieve AOD from fine resolution satellite instruments, including the Landsat-8 and Sentinel-2. This study evaluated the AODs against the in situ data (i.e., Aerosol Robotic Network and Sun-Sky Radiometer Observation NETwork measurements). The temporal window and wind speed were jointly applied to determine the spatial window size (i.e., 6 km) centered on the in situ site. And a three-step strategy for calculating the average of satellite retrieved AODs within the spatial window was proposed in this study. As a result, both STEPs 2 and 3 had improvements compared to STEP 1 and the three-step strategy achieved the best performance in terms of the highest coefficient of determination (R 2 : 0.754) and within expected error (EE: 60.81%), the lowest root mean square error (RMSE: 0.150) and mean absolute error (MAE: 0.113). Then the contemporaneous AOD matchups were used for validation and evaluation of the LaSRC retrieved AODs. The Sentinel-2 AOD retrievals outperformed the Landsat-8 AOD retrievals with lower RMSE (0.138 and 0.173, respectively), lower MAE (0.102 and 0.136, respectively), higher R 2 (0.818 and 0.707, respectively) and higher within EE (64.36% and 53.19%, respectively). However, there are 31.91% of Landsat-8 matchups falling below EE and 32.67% of Sentinel-2 collocations falling above EE, which were mainly detected at HKP and HKS sites. And the site-scale validation results revealed that the LaSRC algorithm performed well over inland urban areas, but should be developed further over coastal and water areas in the next generation. Highlights: A three-step strategy, which considered weighted averages, was fit to validate the mid/high resolution satellite retrievals. Temporal window and wind speed are jointly to determine the spatial window size. LaSRC AODs from Sentinel-2 outperformed that from Landsat-8. LaSRC algorithm needs further improvement over coastal urban areas.