Long-term spatial forest management planning using the mean-value of disturbance

Uncertain events such as fire and insect disturbance affect the sustainable harvest from the forest. However, it is difficult to represent even the mean value of these disturbance processes within the Model I or Model II frameworks. In this thesis a method for representing uncertainty in large forest linear programs will be presented. A Model III formulation is used to generate prescriptions for stands or stand aggregates. These prescriptions can, in turn, be used in a Model I linear program, thus allowing the mean disturbance rates to be taken into account in a large-scale strategic forest management linear program. Once disturbance has been observed in terms of its mean-value, an attempt will be made to bound the expected value of stochastic solutions. Using the solutions to the mean-value problem and the two-stage stochastic linear programming representation of the Model I formulation with perfect information, a framework can be developed to generate estimates of the upper and lower bounds on the expected value of stochastic solutions.