Buell M FEMC2007

Linking LP Models with Spatially-Explicit Models for Forest Planning

Michael Buell, Cortex Consultants Inc., Suite 2a - 1218 Langley Street, Victoria, BC, V8W 1W2, Canada, mbuell@cortex.ca

Aspatial (or quasi-spatial) linear programming (LP) models of forest resource development and timber production are driven by a management objective, produce an “optimum” schedule, quick and inexpensive to solve, and are extensible to financial and log allocation analyses. However, these aspatial models cannot adequately represent the spatial relationships between critical features on the ground or patterns of disturbance.

Spatial models (raster-based and polygon-based) can explicitly represent these critical spatial relationships and are extensible to a wide range of habitat models and landscape metrics, but are typically driven by heuristics (e.g., extend a road network) and are unlikely to be optimal. Furthermore, spatially-explicit approaches are much less efficient than the aspatial models.

We have developed a methodology that links an aspatial LP-based model with a spatially-explicit sub-model, and preserves the advantages of both approaches. Reduced-cost coefficients from the LP model guide the spatial location of the disturbance schedule, ensuring that the spatial solution remains optimal.