Skip to main content
ARS Home » Midwest Area » Columbia, Missouri » Cropping Systems and Water Quality Research » Research » Publications at this Location » Publication #102036
Title: USE OF UNSUPERVISED CLUSTERING ALGORITHMS FOR DELINEATING WITHIN-FIELD MANAGEMENT ZONES

Author
item Fraisse, Clyde
item Sudduth, Kenneth - Ken
item Kitchen, Newell
item FRIDGEN, JON - U OF MO

Submitted to: American Society of Agricultural Engineers Meetings Papers
Publication Type: Other
Publication Acceptance Date: 7/22/1999
Publication Date: N/A
Citation: N/A

Interpretive Summary:

Technical Abstract: Site-specific management addresses the issue of managing large fields more efficiently by applying crop inputs in accordance with the specific requirements of a location. Development of agronomic strategies specific for areas of the field that are subject to a unique combination of potential yield-limiting factors would allow more accurate management of inputs. The objective of this research was to evaluate the use of unsupervised classification algorithms to delineate potential within-field management zones based on topographic attributes and soil electrical conductivity. Data collected in two fields located in central Missouri were used to test the proposed methodology. First, principal component analysis was used to determine which layers of data were most important in representing the within-field variability. Second, GIS-based unsupervised clustering algorithms were used to divide fields into potential management zones. Grain yield data were used to analyze the "goodness" of the zones defined. Elevation and soil EC are the most important attributes that should be taken into account when performing unsupervised classification in claypan soils. The ideal number of zones to use when dividing a field decreases if adequate moisture conditions are present throughout the cropping season. Additional layers of information that may be considered important in characterizing the yield variability, such as remote sensing images and soil fertility maps, can easily be added to the classification process. This classification procedure is fast, can be easily automated in commercially available GIS software, and has considerable advantages when compared to other methods for delineating within-field management zones.