Estimating estuarine suspended sediment concentration through spectral indices and band ratios derived from Sentinel-2 data: a case of Umzimvubu Estuary, South Africa

The current study was aimed at evaluating the reliability and efficacy of selected remote sensing band ratios and indices in accurately estimating the spatial patterns of suspended sediment concentration level in Umzimvubu Estuary, Eastern Cape, South Africa. Sentinel-2 imagery was acquired on the 29th of March 2022. Band reflectance values were extracted from Sentinel -2 imagery, and laboratory measurements of suspended sediment concentration were obtained from samples collected from fifty (50) sampling points in the estuary on the 29th of March 2022. Sentinel-2 imagery was then validated with the field data in estimating and mapping the suspended sediment concentration. Several remote sensing band ratios Red/(Green plus Near-Infrared), Near-Infrared/Green, Red plus Near-Infrared/Green, Blue(Green plus Red)/Blue and Green plus Near-Infrared)/Blue and indices, that is the Normalised Difference Turbidity Index (NDTI), Normalized Difference Suspended Sediment Index (NDSSI) and Normalized Suspended Material Index (NSMI)) were then used to predict the suspended sediment concentration from Sentinel-2 imagery. The accuracy of band ratios and indices was evaluated by correlating the prediction against the observed suspended sediment concentration from Sentinel-2 imagery. A total of 50 points were randomly surveyed in the Umzimvubu estuary for analyzing suspended sediment concentration. Results indicate that the Blue (Green plus Red)/Blue, the Green plus Near-Infrared)/Blue and NMSI performed well based on their R-squared. The Blue (Green plus Red)/Blue and Green + Near-Infrared)/Blue band ratios had 0.86 and 0, 94, respectively. While NSMI yielded an R-squared of 0,76 and RMSE of 19,2 mg/L. The results in the current study indicate that Sentinel-2 imagery can reliably estimate the concentration of suspended sediment level in the Umzimvubu Estuary using band ratios and indices.Thesis (MSc) -- Faculty of Science and Agriculture, 202