fresh-fuchs

Stochastic, risk-aware forest landscape planning on the TSA29 mini instance (~100,000 ha subset of Williams Lake TSA29).

The research design is a nested decision problem:

• Outer (policy / administration). Landscape-scale policy on target species / development-type composition by area share and on harvest policy (AAC as f(development type, rotation age)), evaluated risk-sensitively on Monte-Carlo distributions of NPV using downside-risk measures (CVaR).
• Inner (enterprise / implementation). Harvest scheduling and replanting decisions (which species to plant) maximizing NPV subject to the outer policy constraints, formulated as a Model I linear program.

Full-Monte-Carlo outer problem: sample disturbance (fire) and — later — price realizations; solve the inner LP once per scenario (full foresight within a scenario); evaluate each policy on the resulting NPV distribution.

Reuses the UBC-FRESH ecosystem rather than re-implementing it: ws3 (wood-supply engine and LP machinery), femic (tsa29mini instance bundle and model bridge), fhops (harvest-cost estimation), nemora (DBH distribution fit and sampling), and freshforge (workflow + matrix orchestration and evidence). fresh-salvage provides economic calibration anchors (reference only).

Status

v0.1.0a1 (Phase 5 release). The end-to-end pipeline is implemented, validated, and tested: extended ws3 model build -> full-MC fire scenarios -> per-scenario inner LP (NPV max) -> NPV distribution -> policy grid search -> CVaR-based ranking, wrapped in freshforge workflows/matrices with evidence. See ROADMAP.md for the phase/issue tracker map and planning/v0.1.0a1-plan.md for the detailed master plan.

Quick Start

CI-safe synthetic end-to-end (no private data; the public-safe synthetic instance in fresh_fuchs.instance.synthetic):

pip install -e ".[dev,orchestration]"

from fresh_fuchs.orchestration import fuchs_workflow_spec, run_fuchs_workflow

spec = fuchs_workflow_spec(horizon=2, n_scenarios=3, master_seed=42)
result = run_fuchs_workflow(spec, workdir="outputs/synthetic")
assert result.ok

Real-bundle pipeline (requires the bundle extra and the annex bundle):

pip install -e ".[dev,bundle]"
fresh-fuchs --help
# build-model -> scenario-run -> policy-grid -> policy-rank

Documentation

Sphinx docs under docs/: installation, quickstart, model semantics, CLI reference, architecture, and development guides.

License

MIT, Copyright (c) 2026 UBC FRESH Lab.