Bathymetric data plays a major role in obtaining accurate results in hydrodynamic modeling of rivers, estuaries, and coasts. Bathymetries are commonly generated by spatial interpolation methods of data on a model grid. Sparse and limited data will impact the quality of the interpolated bathymetry. This study proposes an efficient spatial interpolation framework for producing a channel bathymetry from sparse, cross-sectional data. The proposed approach consists of three steps: (1) anisotropic bed topography data locations transformed to an orthogonal and smooth grid coordinate system that is aligned with its riverbanks and thalweg; (2) sample data are linearly interpolated to generate river bathymetry; and (3) the generated river bathymetry is converted into its original coordinates. The proposed approach was validated with a high spatial resolution topography of the Tieu estuarine branch. In addition, the proposed approach is compared with other spatial interpolation methods such as ordinary kriging, inverse distance weighting, and kriging with external drift. The proposed approach gives a nearly unbiased topography and a strongly reduced RMSE compared with the other methods. In addition, it accurately reproduces the thalweg. The proposed approach appears to be efficiently applicable for regions with sparse cross-sections. Moreover, river topography generated by the proposed approach is smooth including important morphologic features, making it suitable for two- and three-dimensional hydrodynamic modeling.