Oscar Garcia
Forest Research Institute
New Zealand Forest Service
Simulation and Linear Programming (LP) are two commonly used tools in forest management planning. Because of their characteristics, they should be viewed as complementary rather than competitive approaches (Garcia 1981). LP models involve necessarily a considerable simplification of the management objectives and constraints, and many relevant factors that are difficult to quantify or not easily represented within the LP framework are excluded. Simulation, on the other hand, is usually limited to the evaluation of a very small proportion of all the possible alternatives, with the consequent danger of overlooking the most profitable management plans. With most known implementations it is not easy, however, to alternate between LP and simulation, because of a different conceptual model of the forest and different data requirements. This is particularly true with what Johnson and Scheurman (1977) call Model I, which forms the basis of most of the current LP-based forestry planning models.
Another factor that has delayed a voluntary acceptance of LP by forest managers is the difficulty in understanding these models and formulating the problem. Usage of LP-based forest planning systems has required an understanding of the mathematical and computational technicalities of LP systems. Managers have been dependent on a group of specialists for setting-up the problems and interpreting the "answers". They are reluctant to accept solutions printed-out by a computer, without full knowledge of the assumptions, etc. There is a need for systems that the forest managers could use by themselves, without the requirement of a detailed knowledge of the mathematical and computational technicalities involved. This is especially important with decentralised planning, as practised in the N.Z. Forest Service. More mathematical or computational problems, the challenge is one of human engineering.
FOLPI is an attempt at alleviating the deficiencies discussed above. It uses the same basic data as IFS, a forest simulator being used by the Forest Service (Garcia 1981). The planning problem is stated in forestry terms, and automatically translated into a Linear Programming formulation and solved with a standard LP package. The solution is then translated back in terms of the original problem, and reports identical to those used in IFS may be generated interactively.