Spatial prediction and mapping of water quality of Owabi reservoir from satellite imageries and machine learning models

• DOI: 10.1016/j.ejrs.2021.06.006
• Publication Date: 2021-06-29T12:25:09Z
• Authors: Yvonne Yeboah Adusei; Jonathan Quaye-Ba...; Albert Amatey Adj...; Alex Appiah Mensah
• Keywords: Turbidity; Alkalinity; Total dissolved solids; Total suspended solids

Estimation and mapping of surface water quality are vital for the planning sustainable management inland reservoirs. The study aimed at retrieving parameters (WQPs) Owabi Dam reservoir from Sentinel-2 (S2) Landsat 8 (L8) satellite data, using random forests (RF), support vector machines (SVM) multiple linear regression (MLR) models. Water samples 45 systematic plots were analysed pH, turbidity, alkalinity, total dissolved solids oxygen. performances all three models compared in terms adjusted coefficient determination (R2.adj), root mean square error (RMSE) repeated k-fold cross-validation procedure. To determine status quality, pixel-level predictions used to compute model-assisted estimates WQPs with reference values World Health Organization. Generally, produced more accurate results S2 L8. On average, inter-sensor relative efficiency showed that outperformed L8 by 67% reservoir. gave highest accuracy RF (R2.adj = 95–99%, RMSE 0.02–3.03) least MLR 55–91%, 0.03–3.14). Compared RF, SVM similar but slightly higher RMSEs (0.03–3.99). estimated pH (7.06), (39.19 mg/L) alkalinity (179.60 within acceptable limits, except turbidity (33.49 which exceeded thresholds. recommended monitoring