Linear programming and related approaches in forest planning.

Abstract

A number of approaches to forest management planning are surveyed and analysed. A unified treatment of the various models and optimisation techniques is attempted, aiming at a better understanding of their essential characteristics and interrelationships. Ways of improving computational efficiency are examined. Structures for the forestry submodel are classified into three basic types, and interpreted in terms of network flows. A utilisation / economics submodel, comprising an objective function and constraints representing supply requirements, yield regulation, terminal conditions, and other aspects of the management problem, completes the formulation. A new criterion for yield regulation, based on irreversible investment concepts, is discussed in detail. The optimisation methods proposed in the literature for solving forest planning problems are evaluated, together with other potentially useful techniques and strategies. Topics covered include the use of standard Linear Programming packages, the prospects of decomposition and partitioning methods, the potential for utilising oldest-first and other cutting priorities, the explicit use of time structure in Dynamic Linear Programming and Optimal Control Theory algorithms, and the Hoganson-Rose procedure and its relationship to Lagrangian relaxation.