Tree-selection algorithms for optimizing thinning using a distance-dependent growth model

Abstract

When a distance-dependent yield model is used in growth simulation and optimization, the stand is described by a plot on which the trees are described by species, DBH, height, age, coordinates, and other characteristics. Optimizing a distance-dependent model requires that the trees to be removed in a thinning treatment be specified individually and that a special algorithm be developed for this selection. This algorithm affects the formulation of the optimization problem. In this study, we compared four different problem formulations for optimizing a distance-dependent yield model: (1) Harvest percentages in different diameter classes were utilized together with a tree-selection algorithm proposed earlier; this algorithm removed trees on the basis of competitive status. The algorithm was fixed, i.e., it was not amenable to numerical optimization. (2) Minimum distances between remaining trees were used as decision variables. (3) Minimum distances were used with harvest percentages and the existing tree-selection algorithm. (4) The tree-selection algorithm was made dependent on two parameters, which were optimized together with harvest percentages in different diameter classes. Based on case optimizations in pure and mixed conifer stands, the fourth formulation was considered to be the best one from the standpoint of simplicity of optimization and in terms of the objective function value.