The increasing availability of LIDAR and other high-resolution digital elevation models (DEMs) offers the potential for digital soil mapping with unprecedented spatial resolution. To date the promise of improved surveys is only that---a potential. Given the much higher acquisition and data processing costs associated with high-resolution DEMs, there is an immediate need to know if increased DEM resolution will in fact yield better surveys. This study is the first to use the SoLIM expert system predictive mapping model with a very high-resolution DEM, and so far as we know, it represents the first application of this genre of model based on very high-resolution terrain data.

A SoLIM model was developed and applied to 400 acres of the UW-Platteville Pioneer Farm located in Lafayette, Wisconsin, U.S.A., based on a one-foot DEM. The study area consisted of very subdued topography with only modest slopes and almost nonexistent profile curvature. Three different soils were recognized and inferenced using the SoLIM methodology. These consisted of moderately well to well drained soils. The parent materials were mainly loess over clayey pedisedimet over loamy-skeletal residuum from the underlying Galena dolostone formation. A narrow drainage way with silty alluvium soils was not inferenced but was delineated using GPS in the field. Two soils were recognized in this drainage way, a somewhat poorly drained and a poorly drained soil. A Giddings probe was used to examine the soils in 22 locations for verification purposes. Soils data was recorded along with the surface topography including the slope gradient, aspect, profile curvature, and planform curvature. Also, 11 backhoe pits were dug in several different landform positions and detailed profile soil descriptions were prepared. Primary characterization data for pedons S04WI-065-001 and SO4WI-065-002 have been completed from the Soil Survey Laboratory in Lincoln, Nebraska. Results show that the SoLIM map captured very subtle relationships between the soils and the landform positions that they occupy. For example, all of the more convex positions (noses) on the shoulders have been delineated consistently regardless of size and location. Saddle areas far too small to be seen in standard DEM products (e.g. 10m) were well-captured by the high resolution DEM and were expressed in the resulting maps. Use of a wetness index computed over a suitably chosen neighborhood size proved effective in separating two soils that would otherwise have been difficult if not impossible to map correctly. Our results show that despite the very subdued relief of this area, the 1-foot DEM was able to represent topography very well, and that resulting terrain derivatives could be used in SoLIM to produce accurate maps which revealed abundant fine-scale structure in soil patterns. This supports the potential of high-resolution DEMs for precision agriculture and similar applications demanding detailed spatial soils information.