Evaluation of Various Atmospheric Correction Methods in the Processing of Landsat8/OLI Data in Jiaozhou Bay
Liu Xiao-yan,Shen Chen,Cui Wen-Xi,Yang Qian,Yu Ding-feng,Gao Hao,Yang Lei,Zhou Yan,Zhao Xin-xing
DOI: https://doi.org/10.3964/j.issn.1000-0593(2022)08-2513-09
2022-01-01
Spectroscopy and spectral analysis
Abstract:In ocean color remote sensing research, it is the key to obtaining the accurate remote sensing reflectance spectrum (R-rs (lambda)) data to retrieve marine biogeophysical parameters from ocean optical satellite data. In practice, R-rs is calculated according to the radiance received by the remote sensing instrument after the correction of atmospheric absorption and scattering and the correction of solar distance and solar elevation angle. Therefore, the atmospheric correction of satellite data is one of the key factors for obtaining accurate water remote sensing reflectance spectral data, which is also an important problem in the research of ocean color remote sensing. Jiaozhou Bay is a semi-closed bay in the west of the Yellow Sea of China and an important representative of the northern temperate zone bay ecosystem. A large range of Marine ranching areas are planned in this sea area, and the water's bio-optical properties are complex. Landsat is the Landsatellite program of NASA in the United States. It was initially developed to observe the land. However, its advantage of high spatial resolution (30 m) is outstanding in Marine remote sensing monitoring, which makes it become one of the data sources that can not be ignored for satellite remote sensing to monitor rivers, lakes, inland bays and other water bodies. Based on the Quality Assurance system-QA Score, we evaluate the results of five atmospheric correction algorithms in processingLandsat8/OLI data in Jiaozhou Bay. Those five atmospheric correction algorithms are NASA's ( National Aeronautics and Space Administration) standard near-infrared atmospheric correction algorithm (Seadas adopted it as the Default atmospheric correction algorithm, recorded as Seadas Default in this paper). Acolite default atmospheric correction algorithm-Dark Spectrum Fitting (recorded as Acolite DSF in this paper) and the Exponential extrapolation method of Acolite, which is recorded as Acolite SWIR, Acolite Red/NIR, Acolite NIR/SWIR respectively according to the different bands used in the Exponential extrapolation algorithm. The analysis results show that the probability (83. 95%) of QA score of R-rs (lambda) data obtained by Seadas Default atmospheric correction algorithm in Jiaozhou Bay is much higher than that of Acolite DSF (49. 66%) Acolite SWIR (4. 13%) Acolite Red/NIR (7. 25%), and Acolite NIR/SWIR (1. 38%D. The atmospheric correction algorithm of Acolite DSF is superior to that of Acolite SWIR, Acolite Red/ NIR and Acolite NIR/SWIR. Finally, MODIS/Aqua satellite data were used to compare and analyze the R-rs (lambda) data at 443, 483, 561 and 655 nm obtained by Seadas Default and Acolite DSF atmospheric correction algorithm respectively. The results show that the atmospheric corrected R-rs (lambda) results obtained by the Seadas Default algorithm are better than that obtained by the Acolite DSF algorithm at all the bands. Based on the results of this study, we suggested that the NASA standard near-infrared atmospheric correction algorithm would be the first choice when applying Landsat8/OLI data to do remote sensing application research in Jiaozhou Bay and its adjacent waters areas.