Spatial forest plan development using heuristic processes that are initiated with a relaxed linear programming solution
MetadataShow full item record
Linear programming is often used in forest planning models; however, these models usually lack spatial (e.g., harvest adjacency) constraints. To accommodate adjacency constraints, a linear problem often becomes a mixed-integer linear programming model because binary variables are needed to represent harvest decisions. Hence, heuristic methods have been suggested to use for forest planning problems that involve complex spatial relationships. In this study, heuristic methods (threshold accepting and tabu search) were used with a high-quality initial solution acquired from a relaxed linear programming solution rather than randomly initiated traditional heuristics. A western and a southern U.S. forest were used as study areas. A mixed-integer programming solution and randomly initiated heuristic solutions were used to compare the results. The findings of the study suggested seeding heuristics with a high-quality relaxed initial starting point provided better solutions than randomly initiated heuristics.