The quality of outcomes from tabu search when enhanced to address contemporary forest harvest scheduling problems

Abstract

Tabu search is a metaheuristic search process that is useful for locating high-quality solutions to complex combinatorial forest harvest scheduling problems. In this work, we illustrate the quality of outcomes produced by a tabu search process when it is applied to four forest management problems, and we statistically compare the effectiveness of tabu search when it uses a range of parameters and enhancements. This illustration of the variability in results is novel to the use of tabu search to address complex agriculture and forest management problems. The addition of enhancements, which allow $n$-opt moves through the solution space, reversion back to high-quality positions in the solution space, and randomly designated tabu states (tenures), all seem to improve the ability of tabu search to locate near-optimal solutions with regularity. In addition, we assess whether the second-best selection from the tabu neighborhood has any value in improving the outcomes from the search process. We found that the use of 2-opt exchange moves seems essential for developing forest harvest schedules of higher quality, and the use of search reversion can often help improve the quality of solutions. The selection of the second-best solution from the tabu search neighborhood, while of little value when used exclusively, may be valuable when used sparingly and in conjunction with other search process enhancements.