Abstarct: Reliable information about river discharge plays a key role in water resource management, flood forecasting, irrigation planning, and ecological assessments. Therefore, estimating river discharge using remote sensing techniques is the objective of this research, particularly in small watersheds that have often gone unmeasured due to environmental or financial constraints. In this study, aimed at estimating discharge in low-width rivers, river discharge was calculated baxsed on parameters derived from remote sensing and hydraulic equations within the ranges of the Khoukhoreh, Adinan, and Saqez rivers. Model inputs included the average river width, estimated from Sentinel-2 images using normalized difference water indices (NDWI) and modified normalized difference water index (MNDWI), as well as the automatic water extraction index (AWEI). Depth and average velocity were baxsed on empirical equations with remote sensing inputs, and channel slope was estimated from the ALOS PALSAR digital elevation model. The channel roughness coefficient was obtained through visual interpretation of Sentinel-2 images, Google Earth images, and then using the Manning's roughness coefficient formula and Chow's coefficient tables. Initially, the widths obtained from water body detection indices were compared to select the most appropriate index. Subsequently, river discharge for the Saqez River at two sections, Adinan at two sections, and Khoukhoreh at one section was estimated baxsed on parameters derived from remote sensing images, average depth, and average flow velocity estimated from empirical methods using the Manning equation (Model 1) and the Berkeley equation (Model 2) The overall performance of Models 1 and 2 derived from the approach used in this study was satisfactory: the Normal Root Mean Squared Error (NRMSE) for discharge estimated from Model 1 before the Adinan-Pol, Qashlaq-Pol, Qabghlu, Kani Jashni (Cotton Valley), and Senthe gauging stations were 29.514%, 23.309%, 66.592%, 56.442%, and 56.916% respectively, compared to observed discharge from hydrometric stations. The estimated discharge from Model 2 and observed discharge at hydrometric stations was 25.495%, 20.056%, 37.563%, 56.129%, and 50.682%. Thus, it demonstrates the potential of quantifying river flow using remote sensing data. Overall, results showed that for medium and high flows, discharge estimates from Model 2 provided better estimates, while for low flows, Model 1 yielded better estimates. Furthermore, considering the assessment of average absolute percentage error in this approach, it can be concluded that this method shows greater error in estimated discharge during low water times compared to high water times.