Climate change and industrialization in recent years threaten water resources, degrading water quantity and quality. Various research activities are in progress, to cope with such crises in water resources. Also, various numerical models in hydraulics and hydrology are applied to simulate water quantity and quality in rivers and lakes. Reliable numerical simulation depends primarily on accurate bathymetry data of rivers and lakes. Spatial interpolation is thus used to make up the entire topography of the computational domain with a limited number of bathymetry data. However, conventional methods of spatial interpolation cannot properly make up the river topography, influenced by flow direction. Therefore, in this study, we developed and tested an anisotropic spatial interpolation method consisting of statistics-based interpolation (i.e. kriging) and coordinate transformation, and compared the new method with conventional interpolation methods (i.e. inverse distance weighting and spline). A free-surface finite element computational fluid dynamic model (TELEMAC, EDF, France) was also applied to test water flow over the interpolated topography. The anisotropic ordinary kriging method, combined with coordinate transformation from the cartesian (XY) to the curvilinear orthogonal (SN) system, was found to be the most reliable and accurate interpolation method. The anisotropic spatial interpolation for a river topography will enhance accuracy and reliability of hydraulic and hydrological simulation for effective water resource management.