A hierarchical simulation-through-optimization approach to forest disturbance modelling

We describe a stochastic–heuristic forest management model which has been adapted from a timber harvest scheduling model, to model natural disturbances based on the concept of Markov chain probabilities. Using a simulation-through-optimization approach allowed us the advantage of modelling complex disturbance problems while at the same time overcoming the extensive processing times that can be associated with conventional simulation techniques. We implemented the model for a 1.2 million ha study area in the central interior of British Columbia, parameterizing the model with historical wildfire probabilities calculated from historical spatial data. A sensitivity analysis of model parameters was carried out to evaluate the performance of disturbance model. In addition, the effects of scale on disturbance simulation were explored by modelling disturbance processes at the forest, habitat, and stand scale. The disturbance model was found to be relatively robust to changes in disturbance probabilities, but was affected by the initial landscape condition. We were able to validate the disturbance model against the results of other published papers, which indicated our model produces similar results for disturbances in the sub-boreal as other models of disturbances in the boreal forest. Using the simulation-through-optimization approach, the efficiencies inherent in heuristic optimization techniques can be used to simulate the effects of natural disturbance processes and management policies.