Accurate estimation of river discharge is essential for sustainable water resources management, particularly in arid and semi-arid regions where hydrometric monitoring networks are often sparse and water systems are highly vulnerable to climatic variability and human pressures. This study investigates the integration of remote sensing data, hydrometric observations, and machine learning algorithms for estimating river discharge in the Karun River, one of the most important river systems in southwestern Iran. A seven-year dataset, including field measurements of discharge and turbidity from three hydrometric stations, was combined with Sentinel-2 multispectral imagery. Remote sensing reflectance values were extracted using the C2RCC atmospheric correction processor, and their relationships with turbidity and discharge were analyzed across Sentinel-2 MSI spectral bands.

The results showed a strong positive relationship between river discharge and turbidity, with a coefficient of determination of approximately 0.85, indicating that flow variability plays a major role in controlling suspended sediment dynamics in the river. Spectral analysis demonstrated that the visible to near-infrared bands, particularly B4 to B8A, were more sensitive to turbidity variations and therefore provided useful predictors for discharge estimation. Several machine learning models, including K-nearest neighbors, random forest, gradient boosting, multilayer perceptron, XGBoost, and support vector machine, were developed and evaluated using standard performance indicators.

Among the tested models, XGBoost achieved the highest accuracy during the training phase, while the random forest model provided the best generalization performance during testing, with an R² of 0.90 and an RMSE of 94.6 m³/s. The findings demonstrate that combining Sentinel-2 remote sensing data with hydrometric observations and machine learning techniques can provide an effective and cost-efficient approach for river discharge estimation, especially in data-limited arid environments. This framework can support river monitoring, water resources planning, and climate-resilient management strategies.