Abstract
The purpose of this study is to examine the current state of water quality in the Nura-Sarysu river basin and predict its future trends. To achieve this purpose, a comprehensive monitoring approach has been used, including satellite imagery and machine learning. For instance, Sentinel-2 (10 m resolution) and Landsat 8/9 (30 m resolution) satellite imagery have been used for estimating the Normalized Difference Water Index (NDWI). This index has helped in identifying water bodies in the study area. In addition, the study has used critical water quality parameters, including heavy metals (Cd 0,15 mg/L, Pb 0,10 mg/L), biochemical oxygen demand (BOD5 10-15 mg/L), and salinity (0,5 g/L). It has been found that there are significant changes in the water quality parameters in the Nura-Sarysu river basin compared with the natural state. A significant correlation (R2 ≈ 0,82-0,87) has been found between satellite imagery and sensor-based estimations. In addition, a machine learning model has also been used for estimating temporal changes in water quality parameters. This model has shown high accuracy (R2 ≈ 0,88, RMSE ≈ 0,09-0,12). This study has shown the potential of the proposed approach for improving the assessment of water quality parameters in the Nura-Sarysu river basin in Kazakhstan.