Riverbank erosion by the Brahmaputra-Jamuna River has displaced thousands of people from their homes and pushed them into abject poverty in Bangladesh. Consequently, a better system for predicting riverbank erosion might be one tactic to mitigate the consequences of such a disaster. Numerical simulation of riverbed evolution can be a solution of that which necessitate high-resolution bathymetry data. The river bathymetry measurements taken by the Bangladesh Water Development Board (BWDB) at 4–6 km intervals are too coarse for numerical simulation. An appealing alternative to the costly and time-consuming classical bathymetric mapping of braided rivers is the data fusion of measured data with spectral reflectance obtained from satellite images. The data fusion method was used to construct 10 m resolution river bathymetry using sentinel 2 images and limited measured data. The estimated bathymetry has been evaluated for the adequacy of river erosion estimation using a 2D hydro-morphological model for the years 2017–2019. On average, between 2017 and 2019, the estimated bathymetry showed a depth of 10, 9, or 10 m, with a maximum depth of 17.5, 16, or 19.5 m. The simulated model’s R2 value was close to 0.96 in relation to a discharge at Bahadurabad. The NSE, PBIAS, and RRMSE for the Mathura, Bahadurabad, Serajgani, and Chilmmari stations, respectively, show the level of satisfaction. Kappa statistics was used to measure the accuracy of erosion predictions, and the results for the three years 2017–2019 were around 67%, 78%, and 72%, respectively.