Submitted to: International Conference on Precision Agriculture Abstracts & Proceedings
Publication Type: Abstract Only
Publication Acceptance Date: 7/19/2000
Publication Date: N/A
Technical Abstract: In order to attain maximum efficiency of crop inputs, site-specific management zones should represent a somewhat homogenous combination of potential yield limiting factors. The objective of this study was to investigate the variability of soil and landscape attributes within sub- field management zones created using unsupervised classification. Potential management zones were delineated on two claypan soil (Mollic Endoaqualf) fields with the fuzzy k-means unsupervised classification algorithm using the parameters of apparent soil electrical conductivity (soil ECa), elevation, and slope. From the clustering results, reduction in grain yield variance and measures of cluster performance indicated that between four and six management zones was the appropriate number of zones for the two fields. Parameter loadings within a principal components analysis demonstrated a relationship between the clustering parameters and soil fertility factors (e.g., pH, soil-test K and P) for one of the two fields. An examination of the reduction in variance for these same soil fertility factors with zone delineation verified the principle components analysis results. For the other field, patterns of soil fertility and lime followed patterns of historical management, rather than soil properties associated with natural soil properties (soil ECa, slope, and elevation). Dependence between the parameters used for creating zones and the other factors that will be a basis for management plans is requisite for producers to gain value from delineating unique management zones.