Retrieval of total suspended solids from remote sensing reflectance in highly eutrophic lakes in Hanoi (Vietnam)

Abstract

ABSTRACT High concentrations of suspended solids can cause water quality deterioration, leading to ecological degradation of the aquatic environment. In highly eutrophic waters, they are important carriers of pollutants and nutrients. Therefore, quantifying total suspended solids concentration () is critical for any water quality monitoring program. This study proposes two band ratio algorithms for estimating in highly eutrophic lakes based on a data set of 180 in situ samples of remote-sensing reflectance, , and measured simultaneously across 10 dates in four lakes and reservoirs in Hanoi (Vietnam). The in situ ranges from 2.7 to 73.3 mg l−1, averaging at 24.9 mg l−1, and is highly correlated to water transparency (the correlation coefficient, = 0.73), chl-a concentration ( = 0.78), and the trophic state index ( = 0.89). Our analyses demonstrate the appropriateness of the near-infrared/red spectral ratio, , model (N = 124) for estimating in the study lakes (the determination coefficient, = 0.84). The model was then verified with a small error (the root mean square error, RMSE = 4.1 mg l−1) using independent multi-date-obtained data sets (N = 56). The model is further optimized to establish a regional estimation model for Landsat 8/OLI and Sentinel 2/MSI bands using the 865 and 655 (or 665) nm bands. The band ratios showed reasonable estimations when applying to the respective spectral band ratios, i.e. the OLI’s band 5/band 4 ( = 0.58; RMSE = 7.7) and the MSI band 8/band 4 ( = 0.73; RMSE = 6.1).