Dewhurst S FEMC2007

Recursive Graph Optimization for Achieving Spatial Structure Goals in Estate Models

Stephen Dewhurst, School of Forestry, Northern Arizona University, Box 85018 Flagstaff, AZ, 86001, USA, Stephen.dewhurst@nau.edu

Building upon the approach described in Campbell and Dewhurst (2007)*, graph optimization is used to develop spatially-explicit estate management simulations, which can include natural disturbance processes and management activities. This approach provides an efficient mechanism for incorporating targets for structural indicator metrics (e.g. patch size) into some types of estate models. Using this approach, both spatial and aspatial indicators, and management goals for them, may be specified. Solutions are then developed using standard heuristic optimization techniques. Implementation as a recursive algorithm using a dynamic change-based architecture is discussed. Analyses including up to 70,000 “stands” have been conducted in recent case studies. Mathematical examination of popular landscape indicator metrics reveals that some metrics lend themselves readily to this approach, while other metrics may be better addressed using other approaches. Practical application in wildlife habitat and natural disturbance modeling are discussed.

* K.A. Campbell and S.M. Dewhurst. A hierarchical simulation-through-optimization approach to forest disturbance modeling. Ecological Modelling, Volume 202, Issues 3-4, 10 April 2007, Pages 281-296