Abstract
The uneven-aged management problem is to determine the management regime for an existing stand over an infinite time horizon, without ever clearcutting it. A new solution method to solve this problem is presented and demonstrated on simulated Norway spruce stands with different initial stand structures. Diameter distributions were used as a convenient way to describe these structures. The method, which puts much effort into determining the timing of harvest activities, is based on Tabu search and greedy heuristics. The growth dynamics were described with a single tree simulator. Different problem approaches for this problem were adopted to maximize the net present value (NPV) of harvested trees, with or without steady-state constraints. With no such constraints, a finite time horizon approximates the general management problem, while the steady-state constraints involved fixed and equilibrium endpoints. The fixed endpoint was a reverse J-shaped diameter distribution, and its purpose was to mimic the structure of a virgin stand. Analysis of economic efficiency, productivity. and managerial implications were made. With the method used and for the endpoint problems, the conversion strategy, conversion length. and steady-state diameter distribution were determined simultaneously. The Kolmogorov-Smirnov measure was used to describe the similarity between diameter distributions and was put directly into the constraints. The infinite approximation is suggested as the best problem formulation as it, in contrast to a steady-state approach, does not assume that a resource system reaches a climax, eliminates the need to define a diameter class width, and involves no fixed cutting cycles. The even-aged management problem, which is to determine the best period for clearcutting and the best thinning regime that precedes the clearcut, while recognizing the soil expectation value, was analyzed for comparison. The method is flexible, independent of the kind of growth simulator used, and can, for example, be directly applied to a diameter class model. FOR. SCI. 46(3): 452–463.