Combined Retrievals of Turbidity from Sentinel-2A/B and Landsat-8/9 in the Taihu Lake through Machine Learning

Lakes play a crucial role in the earth’s ecosystems and human activities. While turbidity is not direct biochemical indicator of lake water quality, it relatively easy to measure indicates trophic status health. Although ocean color satellites have been widely used monitor parameters, their coarse spatial resolution makes hard capture fine variability lakes. The combination Sentinel-2 Landsat provides an opportunity with high temporal resolution. This study aims generate consistent products Taihu Lake from 2018 2022 using Multispectral Instrument (MSI) on board Sentinel-2A/B Operational Land Imager (OLI) Landsat-8/9. We first tested performance three atmospheric correction methods retrieve reflectance MSI OLI images. found that Rayleigh subtraction SWIR band Rayleigh-corrected can most (the coefficient determination (R2) > 0.84, mean absolution percentage error (MAPE) < 7%, median (ME) 0.0035, slope 0.92). Machine learning models outperformed existing semi-analytical retrieval algorithm retrieving (MSI: R2 = 0.92, MAPE 18.78%, OLI: 0.93, 16.20%). consistency same-day images was also satisfactory (N 3110 26.48%). distribution exhibited obvious seasonal 2022. results show potential when combined inland quality.