Abstract
A model was constructed to address the harvest scheduling problem under risk of occurrence of catastrophic events. The model was made up of an optimization and a simulation model. The optimization model provided an optimal harvest pattern under deterministic conditions. The simulation model modified the optimal harvest pattern according to a function of random risk assessing the performance of the system through net present value (NPV). The model was run a large number of times estimating mean, standard deviation and a frequency distribution for NPV. The model was run for a case study considering a forest estate of 8,412 ha located in Canterbury in New Zealand. The random component of the system was windthrow occurring at different times and with different intensities. Storms were generated randomly over a planning horizon of 50 years considering an average return period of 28 years between two successive storms. The intensity of damage was assumed to be proportional to the historical damage. As a result, NPV after taxes regarding the management of the forest estate was reduced 11 percent on average. Windthrow brought about economic losses due to reduced harvest following windthrow, reduced recovery (80%), increased total establishment costs, and the fact that trees were harvested before optimal rotation.
