Geostatistical inference of Sentinel-2 spectral reflectance patterns to water quality indicators in the Setumo dam, South Africa

Abstract

Monitoring water quality is of utmost importance for protection from pollution and ecological functioning of inland water body. In South Africa, water quality has been identified as a major environmental issue experienced in many water bodies. Accurate estimation of reservoir's water quality indicators through conventional techniques, such as those employed by several authors in the reservoirs under investigation, has proven to be time-consuming, expensive, labour-intensive, and impractical, especially when the sample area is large or inaccessible. Sentinel-2 imagery, in conjunction with field-based water quality indicator data were used to map spatial patterns in water quality indicators. The water quality indicators investigated in this study were electrical conductivity (EC), total dissolved solids (TDS), pH and turbidity. Linear regression analysis was employed to establish the relationship between each water quality indicator and the spectral reflectance patterns of Sentinel-2 bands. The empirical models were generated from the linear regression results that showed strong relationship between each water quality indicator and spectral reflectance patterns of Senstinel-2 bands. The linear regression analysis results between the in-situ data and Sentinel-2 data revealed the response of EC, TDS, pH and turbidity to red, shortwave infrared, near-infrared and green spectral channels respectively. Therefore, demonstration of the band correlations with in-situ data suggests that Sentinel-2 sensor provides invaluable information for water quality monitoring.