Constraint Technology Applied to Forest Treatment Scheduling

Forest treatment planning and scheduling is an important part of forest resource management. It is a complex task requiring expertise and integration of multi-disciplinary fields. In this paper, we outline a real life problem from the ECOPLAN project called the Long Term Forest Treatment Scheduling Problem (LTFTSP). A review of optimization techniques applicable to forest treatment problems in general is presented, and contrasted with our case. The review suggests that long term scheduling is difficult because of the prohibitive size and complexity inherent to the problem. Based on experience from the successful resolution of a simplified problem, we advocate the use of iterative improvement techniques as a solution strategy. Iterative improvement techniques will in general benefit from high quality initial solutions. We show how a Constraint Satisfaction Problem formulation of the LTFTSP can be used to generate initial solutions. A key element to success is the use of a forest simulator for knowledge based definition of variable domains. The initial solution generator will be used as a module in an integrated forest treatment scheduling system which is under development in the ECOPLAN project. ∗Intelligent Systems Lab, Simon Fraser University, Burnaby, B.C., Canada V5A 1S6, [email protected], http://www.cs.sfu.ca/cs/research/groups/ISL/ †SINTEF Applied Mathematics, Departmernt of Optimisation, Oslo, Norway, [email protected], http://www.oslo.sintef.no/am/ ‡SINTEF Applied Mathematics, Department of Spatial Modelling, Oslo, Norway, [email protected], http://www.oslo.sintef.no/am/