Integrating short-term spatially feasible harvest plans with long- term harvest schedules using Monte-Carlo integer programming and linear programming

Graduation date: 1989Procedures are developed for integrating short-term,\ud area-based plans with long-term, strata-based harvest\ud schedules. These procedures were tested on a 4000 hectare\ud tract of forest land in British Columbia. It is shown that\ud a combination of these two approaches to forest planning\ud provides a spatially feasible short-term solution that is\ud also feasible in terms of strategic harvest goals over the\ud long-term.\ud Three basic steps are used to combine the area and\ud strata-based plans. First, a 15-decade strata-based plan\ud is solved with linear programming (LP) to establish\ud strategic harvest goals. This LP determines the\ud nondeclining yield of volume and the minimum net revenue\ud that can be produced each decade, and it also determines\ud how much of the forest must be partitioned into harvest\ud units to supply a 3-decade area-based plan.\ud Second, using the strategic harvest goals as\ud guidelines, 3-decade area-based plans are generated using a\ud random search technique called Monte-Carlo integer\ud programming (MCIP). The area-based plans are selected from\ud 45 harvest units which are accessed by 51 major road\ud projects. Both the harvest units and the road projects are\ud specified as strict binary variables. Adjacency\ud constraints prohibit any two adjoining units from being cut\ud in the same decade.\ud Third, the area-based plans become the first 3-\ud decade solution for 15-decade integrated plans that are\ud solved using LP. This is accomplished by using\ud coordinated allocation choices within the FORPLAN model.\ud It was found that MCIP is an effective technique for\ud generating feasible solutions to large integer problems\ud encountered in area-based planning. No guarantee of\ud optimality can be assured, however, the technique is a\ud major improvement over manual methods, and it is a\ud practical alternative to mixed integer programming where\ud the cost of finding the true optimum is prohibitive.\ud It was also found that the discount rate is not a\ud factor in determining the optimal basis for an LP that\ud maximizes present net worth and includes a nondeclining net\ud revenue constraint