Planning under uncertainty at the forest level: A systems approach

Abstract

Large-scale long-range forest management problems have been successfully analysed for decades using linear programming models. Existing larger systems, such as MELA, FOLPI, FORPLAN, GAYA-LP and Spectrum, are based on a model formulation that is known as model I or the closely related model II. This paper shows how the model I formulation can be extended to incorporate stochastic phenomena. The uncertainty problem is given as a programme with recourse, i.e. the formulation takes account of the fact that the decision maker is able to observe the state of the system over time and subsequently make adaptations. Basic for the formulation is the expression of the stochastic process as a collection of scenarios. After the basic model has been formulated the implications for systems design are indicated. The approach is applied to a small sample forest where the consequences of different objectives and constraints are illustrated. The limitations of the method, among which model size is prominent, are discussed. It is also noted that not all stochastic processes are amenable to analysis with the suggested approach. The procedure requires global, not standwise, processes and there should be no feedback between actions and scenario probabilities.

Keywords:

• Linear programming
• long-range forest planning
• mixed integer programming
• scenario
• stochastic

Acknowledgments

I am indebted to PhD student Sofia Backéus at the Department of Forest Resource Management and Geomatics, SLU, for discussions on the nature of climate change and its transformation into models. The study was performed in co-operation with the Heureka research programme.