Accounting for uncertainty in forest management models

Abstract

Forests and their ecosystem services are subjected to uncertain factors, causing drastic changes in forest production and/or market conditions, the impacts of which cannot be precisely estimated beforehand. We introduce a theoretical framework, based on control theory, for robust optimization of forest management under uncertainty. Forest owners herein regard their decision support system only as an approximation to an unknown, true, model. Concerns about model misspecification incite them to seek the single harvesting rule that works well over a set of models statistically similar to their approximation. Accounting for mistrust of decision support systems in modelling harvesting behavior is particularly relevant in view of uncertainty induced by climate change. We use a stylized forest model to explore the effects of uncertainty on harvesting decisions, also considering the role of information aimed at reducing, or making forest owners aware of, such uncertainty. The simulation results demonstrate that model uncertainty affects harvesting intensity and thus forest development. Harvest levels are lower for forest managers concerned with model uncertainty, favoring stand volume over harvest revenues. Further, information release affects the level of perceived uncertainty and thus harvesting behavior and forest development, reaffirming the importance of information as a forest policy instrument.