Forest Road Decommissioning Policies Determined using Deterministic and Stochastic Dynamic Programming

Extensive timber harvesting and the accompanying road construction in the Pacific Northwest region have decreased the quality of fish-bearing streams. The decommissioning of abandoned forest roads increases stream quality by decreasing erosion and downstream sedimentation. Road removal treatments have been performed in many locations. However, the management of these treatments has been generally site-specific, with little investigation of how the treatments will affect the entire watershed. Land managers have a need to design a watershed wide management policy to reduce sedimentation, while maintaining overall costs within a reasonable limit. Identifying the trade-offs of the costs of different treatment policies associated with net reduction of sediment can be quantified. This work further develops optimization approaches to manage road decommissioning projects. Previous work in deterministic dynamic programming and genetic algorithmes did not incorporate the uncertainty of the effectiveness of the road treatments. Stochastic dynamic programming is used to determine the road treatment policy that maximizes the expected sediment saved. This approach is used to determine a policy for the Lost Man Creek Watershed in Northern California containing 691 road segments and road crossings. The model determines the optimal treatment level for each road segment and road crossing while considering a budgetary constraint.