Using dynamic programming and overlapping subproblems to address adjacency in large harvest scheduling problems

The adjacency problem for forest management scheduling is formulated as a dynamic programming problem. To overcome typical model-size problems, a solution strategy is proposed which links dynamic programming formulations for overlapping subproblems. Subproblems are solved sequentially in a moving-windows fashion. Solutions are near-optimal for three large but simplified test cases. Model size is sensitive to the size of the moving window. Using relatively small windows produced solutions that are closer to optimal than those found using two other modeling approaches. Solutions are also insensitive to the side of the forest used to start the moving-windows process, and microcomputer solution times are not a major consideration. The strategy offers potential to treat adjacency measures as valued management outputs. It will likely fit well into broader forestwide models which decompose the forestwide problem into parts to help address other forestwide constraints.