Modeling Spatial Distribution of Suspended Sediments in Hoa Binh Reservoir Using Sentinel-2 Image and Co-Kriging

Abstract

Knowledge about the spatial and temporal distributions of total suspended sediments (TSS) play a crucial role in effectively managing pollution and optimizing reservoir operations. This study utilized an in-situ dataset of TSS collected from 79 sampling points within the Hoa Binh Reservoir, coupled with Sentinel-2 (S2) satellite imagery acquired on three different dates: 12/07/2023, 19/12/2022, and 15/10/2022 to map the spatial distribution of TSS across the reservoir during these survey dates. By leveraging the correlation between TSS and the spectral ratio of band 3 to band 4 of the S2 images (with a correlation coefficient of 0.90), this band ratio was incorporated as an auxiliary variable to enhance the accuracy of Co-kriging predictions, with a notable improvement in the determination coefficients ( ) and the standard errors (RMSE), increasing from 0.08-0.46 to 0.90-0.95 and from 0.17-0.61 mg/L to 0.09-0.18 mg/L, respectively. The resulting TSS maps revealed two distinct trends: a) elevated TSS levels were observed near discharge rivers, while lower concentrations were found in the vicinity of the dam; b) TSS during the wet season (averaging at 1.85 mg/L) were higher compared to those in the dry season (averaging at 0.75 mg/L). These findings highlight the efficacy of utilizing S2 satellite imagery as a valuable tool for accurately mapping TSS spatial distribution across the Hoa Binh Reservoir.