Delineating vineyard zones by fuzzy K-means algorithm based on grape sampling variables

This study describes a method for delineating management zones using interpolated maps of grape characteristics recorded in 2013 and 2014 in a Godello vineyard located in the Bierzo Denomination of Origin (León, Northwest Spain). Ten variables were analyzed and recorded for the sampled vines (50 vines/ha). Interpolated maps reflecting each variable and year were created by spatial interpolation (kriging) from the sampled points. Principal component analysis was used to detect relationships between variables and to select the variables to be used to create the cluster classification. Using the fuzzy k-means classification algorithm implemented in the Management Zone Analyst (MZA v.1.0.0) software, several zones were delineated by combining the studied variables. The results delineated 2 different management areas composed of 3 zones each based on winery objectives: (1) to increase grape production (combining the yield for 2013 and 2014); and (2) to improve grape composition (combining the pH for 2013 and 2014).