The accuracy of numerical models analyzing hydrologic and hydraulic processes depends largely on how well the input terrain data represents the earth's surface. Modelers obtain terrain data for a study area by performing surveys or by gathering historical survey data. If a modeler desires to create a predictive model to simulate the addition of man-made features such as channels, embankments and pits, he must modify the terrain data to include these features. Doing this by hand is tedious and time consuming. In 2001 Christensen implemented a tool in the Surface-water Modeling System (SMS) software package for integrating man-made geometric features into surveyed terrain data. He called this process feature stamping. While Christensen's feature stamping algorithms decrease the time and effort required to integrate geometric features into existing terrain data, they only function on centerline-based features having a constant trapezoidal cross-section. In addition to placing geometric limitations on the features they stamp, Christensen's feature stamping algorithms also possess several instabilities. These instabilities arise when stamping features that leave the bounds of the terrain data, and when modifying and re-stamping features that have already been stamped. This thesis presents changes and enhancements made to Christensen's feature stamping algorithms. These changes and enhancements completely eliminate the instabilities found in Christensen's feature stamping algorithms and make it possible for numerical modelers to stamp more complex geometric features having compound slopes, asymmetric cross-sections and varying cross-sections along their length. Finally, additional feature stamping algorithms make it possible to stamp radial features such as mounds and pits.