Optimal upgrading of forest road networks: Scenario analysis vs. stochastic modelling

A two-stage stochastic model with binary choice in the first stage has been developed to optimise the upgrading of a real-world forest road network, geographically located in the middle of Sweden. We have compared the model with solutions to the same problem from deterministic scenario analysis. Upgrade policies can be achieved swiftly using both approaches, since the road network is of moderate size. However, a deterministic approach is considerably faster when larger problems are involved. The study here, furthermore, indicates that deterministic scenario analysis provides us with quick, near-optimal solutions to the stochastic model, which are of reasonable good quality. We conclude that the model used here is rather insensitive to uncertainty in critical period length, such as the length of spring thaw, at least when applied to the medium-sized problem presented here. Nevertheless, we strongly recommend the use of the stochastic model whenever possible, since the stochastic and deterministic solutions differ, due to the hedging effect in the stochastic solution.