Evaluation of atmospheric correction algorithms for salt lake water assessment: Accuracy, band-specific effects, and sensor consistency

Atmospheric correction plays an important role in satellite monitoring of lake water quality. However, different atmospheric correction algorithms yield significantly different accuracy for inland lake waters beset by shallowness and turbidity. Finding a suitable algorithm for a specific lake is critical for quantitative satellite water-environmental monitoring. This study used Landsat 8 and Sentinel 2 L1 level data of Ebinur Lake in arid northwest China on May 19, 2021. Atmospheric corrections were performed using FLAASH, QUAC, 6S, Acolite-DSF and Acolite-EXP algorithms. The Sentinel 2 reflectance product verified the consistency of the algorithms. Quasi-simultaneously measured hyperspectral data determined the algorithm applicable to Ebinur Lake waters. The results indicate that the Acolite-DSF algorithm has good consistency and high accuracy in the atmospheric correction of Landsat 8 and Sentinel 2 images. Extracting the atmospheric correction of Landsat 8 images found relative error at 0.3 in the Blue, Green, and Red bands and 0.5 in the NIR band. For comparison, the relative errors of Sentinel 2 in all bands are 0.3. Therefore, these four bands of Landsat 8 and Sentinel 2 data are recommended for temporal monitoring of water-environmental parameters in Ebinur Lake. Besides identifying the suitable atmospheric correction algorithm for Ebinur Lake, this study analyzed the atmospheric correction errors of common wavebands for remote sensing monitoring of water bodies, especially applicable for inland salt lakes of arid regions.