Terrestrial pedologists often use on terrain models to visualize the spatial variability of landscapes. Topographic maps and digital elevation models (DEMs) are available for most of the terrestrial United States, often at scales suitable for pedology (e.g. 1:24,000). Subaqueous terrain models, are usually availible at coarser scales and thus are not well suited for pedology. To create a subaqueous DEM at a scale of 1:24,000 we collected bathymetry using an an acoustical sounder combined with a global positioning system. Soundings were collected every 10 m at a speed of 10 m/s along a grid of North-South and East-West transects. Using a popular Geographical Information System software package, we employed geostatistical techniques, polynomial trend surfaces, and Triangular Irregular Networks as tools for modeling the bathymetric data. We modeled the data using each modeling technique to determine the effects of technique, data density, and transect spacing on the DEMs. A low prediction error is desirable when a DEM is to be used to predict elevations whereas a smooth landscape is desirable when the DEM is to be used to derive slopes. Comparisons of the DEMs were based on a traditional metric, root-mean-square (RMS) error, and a novel metric we created to quantify landscape smoothness. Kriging constantly provided the lowest RMS while polynomial surfaces consistently provided the smoothest landscapes. Bathymetric point density had only a small effect on the smoothness of the DEM whereas the distance between transects had a tremendous effect on the smoothness of the DEM. Using kriging and a transect spacing ten times greater than the DEM cell size, we were able to achieve an acceptable balance between low RMS and DEM smoothness. These guidelines paired with these newly available technologies should assist pedologists in creating subaqueous terrain models at finer scales scales.