2005 Volume 4 Pages 49-69
A growth pattern of an individual tree differs due to such effects as a local geography condition and competition among trees. In such a case, clustering growth patterns of individual trees becomes effective to describe a whole forest stand growth. We proposed a clustering method for tree growth in an even-aged forest stand. Our method consists of I) estimating parameters of a growth curve, 2) clustering a set of estimated parameters by the k-means method, 3) multivariate analysis of variance (MANOVA) on a set of the parameters, and 4) searching for an optimal number of clusters by the smallest information criterion.
