Project
FRESH-Fuchs risk-aware forest planning
A stochastic, risk-aware forest landscape planning model on the TSA29 mini instance, coupling a policy-level Monte Carlo outer problem with an enterprise-level harvest-scheduling LP.
FRESH-Fuchs is a stochastic, risk-aware forest landscape planning model built on the TSA29 mini instance, a roughly 100,000-hectare subset of the Williams Lake Timber Supply Area.
The research design is a nested decision problem. The outer policy level sets species and development-type composition and harvest policy, evaluated risk-sensitively on Monte Carlo distributions of net present value using downside-risk measures such as CVaR. The inner enterprise level solves a Model I linear program for harvest scheduling and replanting under those policy constraints.
The full Monte Carlo outer problem samples fire disturbance and, later, price realizations, solving the inner LP once per scenario with full foresight within each scenario.
The implementation reuses the FRESH ecosystem: WS3 for wood supply and LP machinery, FEMIC for the TSA29 mini instance, FHOPS for harvest-cost estimation, Nemora for diameter-distribution fitting and sampling, FRESHForge for workflow and matrix orchestration, and FRESH-Salvage for economic calibration anchors.
Status
- Ongoing research and software project
People And Roles
- FRESH Lab: project team
Outputs
- GitHub repository
- Documentation
- Public-safe synthetic instance
- End-to-end Monte Carlo and inner-LP workflow
- Policy grid search and CVaR-based ranking