Abstract:
The Elliott State Forest, located in the Coast Range of Oregon, is currently
revising their Habitat Conservation Plan (HCP). Many of the constraints in the HCP
are spatial, requiring identification of specific parcels in order to limit activity along
habitat reserves, limit harvest opening size, and to coordinate activities within harvest
units. To model the Elliott, the state forest planning team divided the 93,000 acres
into 17 management basins, 576 forest strata, 1900 logging settings, and 57,700
parcels. There are 63 prescriptions per forest strata with 20 alternative final harvest
ages. Due to the large number of integer variables necessary to represent the decision
variables in the harvest scheduling model, a heuristic modeling technique, simulated
annealing, was used to solve the 400,000 decision variable problem for a 30-period,
150-year planning horizon.
The modeling framework presented here allows for the evaluation of forest
management alternatives by spatially quantifying timber harvest, revenue stream, and
habitat outputs through time while tracking both stand- and landscape-level attributes.
It provides spatially explicit features and measurable trade-offs for a problem that
could not otherwise be accurately modeled using traditional forest planning
techniques. A feasible timber harvest schedule that meets all habitat constraints and
policy mandates was produced, estimating an NPV of $260 million for the Elliott State
Forest under the current HCP alternative.
Statistical inference involving extreme value theory can be used to obtain an
estimate of the global optimum to a heuristic solution. This method was evaluated for
use in this planning problem and found to be unreliable as a means for model
validation. The limitations of heuristic validation through extreme value theory are
highlighted, and suggestions for future research in this area are made.
