Modeling timber and non-timber trade-offs in spatially-explicit forest planning

by 1971- To?th, Sa?ndor F

Forest management is inherently a multiple-objective decision making process. Timber production, recreation, wildlife habitat and watershed protection are some of the many uses of forests. Providing the public with an optimal bundle of timber and nontimber benefits is challenging because many of these benefits conflict. Modeling the conflicting objectives, quantifying the trade-offs between them, and identifying efficient management alternatives can facilitate consensus between the stakeholders that represent the various forest uses. Spatially-explicit harvest scheduling provides an excellent modeling environment for such analyses. This work evaluates several existing and one proposed multi-criteria mathematical programming techniques as applied to a two-, then to a three-objective spatially-explicit forest planning problem. The objectives of these models were (1) to maximize the net revenues of the forest, (2) to maximize the minimum amount of mature forest habitat in large patches that evolve across the landscape and over time given various harvest schedules, and (3), in the three-objective model, to minimize the perimeter of these patches. The comparison criteria were the number of efficient alternatives found, the time to find them, the ability of the user to filter the alternatives, and the potential of the techniques to handle n-objective problems. In both the two- and the three-objective case, the proposed Alpha-Delta and the traditional Weighted Method showed promising computational performance. It is recommended that these two methods should be employed in concert to make full use of their respective advantages. iii Formulating the perimeter-minimizing criterion itself was the second objective of this study. Two 0-1 programming formulations are introduced that allow the forest planner to increase the amount of interior habitat relative to edge habitat by minimizing the boundary of the mature forest patches. Both formulations improved the shape of the patches and resulted in fewer and larger patches with more temporal overlap between them. Finally, a procedure is introduced that strengthens the area-based adjacency constraints that restrict the size of harvest openings in spatial forest planning models. The results from test runs suggest, however, that the proposed, theoretically “better” mathematical programming formulation does not necessarily lead to shorter solution times. iv