Soil spatial variability : the process of mapping soil as a continuum
Graham, Jennifer Lynn
MetadataShow full item record
The United States Department of Agriculture and Natural Resources Conservation Service soil surveys are the main sources that help determine the type of soils that are in a field, their productivity, and their price value for taxing purposes. Current soil maps are choropleth maps that contain soil polygons placed over an aerial photograph. These maps have been carefully drawn out to best represent the types of soil present in the area. Once the soil is classified, it is then represented by a polygon that correlates with its position in the field relative to other known locations. Users often assume that these polygons are homogeneous, when they really are not. Soil is a continuous phenomenon that gradually changes over space, not abruptly as portrayed on the present soil maps. In order to portray such an outcome, the idea of a fuzzy soil map has been suggested. Crisp classification observations fall into only one class, "yes" or "no". Fuzzy classification allows a soil sample to have membership in more than one class at a time, with membership declining gradually over space. The end result is a more complex map, but one that describes soil units as non-homogenous units by their highest membership value. This research reports on the application of fuzzy classification using a fuzzy - k - means algorithm to generate the membership values. There were 892 observations sampled from a local grain production farm in DeKalb County, IL. At each location the elevation, aspect, gradient, and plan and profile curvature were measured. Each observation's geographic coordinates were also recorded at that time. The maps generated produced an extremely high amount of variability. It is suggested that this high variability is due to the large scale that was used to produce the maps and that a smaller scale should be used in future research to gain a useful soil management map.