River discharge serves as a critical variable in basin hydrological processes, playing a vital role in flood control, water resource planning, and management. Due to harsh natural environments and climatic conditions, discharge observation stations are scarce in the Yangtze River source region located on the eastern edge of the Qinghai-Tibet Plateau, making it a typical data-deficient area. Therefore, conducting discharge estimation in this region is of significant importance for ensuring water security in the Yangtze River basin and protecting regional ecosystems. This study establishes virtual stations upstream and downstream of the Zhimenda hydrological station in the Yangtze headwaters (ZMD_1 and ZMD_2) as reference points. Utilizing Sentinel-2 and Jason-3, and Sentinel-3A satellite remote sensing data. A novel quantile matching approach was employed to fuse remotely sensed water level and river width data. The Manning"s equation was modified based on generalized cross-section profiles to conduct quantitative discharge estimation for the Yangtze headwaters. Results indicate that the new method achieves high esti-mation accuracy at both virtual stations, with Nash"s efficiency coefficients (NSE) exceeding 0.74. The root mean square error (RMSE) of estimated discharges was 302.13 m3/s and 316.46 m3/s, respectively, with relative root mean square errors (RRMSE) of 30.0% and 32.8%. Overall, discharge estimation accuracy at the ZMD_1 virtual station outperformed that at ZMD_2. The ZMD_2 virtual station results exhibited significant fluctuations, primarily due to the presence of mid-channel bars within the ZMD_2 buffer zone, which compromised the accuracy of flow inversion at the virtual station. This study, based on multi-source satellite remote sensing for estimating river flows in the Yangtze River source region, provides theoretical methods and technical references for flow estimation in data-scarce areas.