Tests of a Dynamic Programming-Based Heuristic for Scheduling Forest Core Area Production over Large Landscapes

Potential simplifying assumptions are presented and tested for applying a heuristic model to schedule production of core area of older forest for the Chippewa National Forest in Minnesota, recognizing approximately 67,000 analysis units and 10 10-year planning periods. The model has strong ties to optimization modeling, utilizing dynamic programming to solve overlapping and linked subproblems. Emphasis of this research is on understanding tradeoffs between solution time and nearness to optimality when large landscapes are modeled. Results show that combinations of model-simplifying assumptions can help maintain an efficient balance between computation time and model optimality. Results are sensitive to values assumed for core area, suggesting that more spatial detail is recognized with higher core area values. Results are compared along an efficiency frontier, where improvements in either computation time or management schedule value cannot be achieved without a loss in the other. Understanding the tradeoff between optimality and computation efficiency is important for model use within a larger iterative system that integrates spatial values with aspatial, forestwide management constraints.