PERSEUS Forest Intelligence

CONUS multi-model forest-carbon explorer covering all 48 conterminous states with deep focal coverage for Maine, Indiana, and Georgia (Minnesota, Oregon, Washington, and Idaho now also carry multi-engine series). Pick a state, drill into multi-model growth curves with uncertainty ribbons and the FIA above-ground reference; toggle the LANDIS biomass spatial layer for Maine (total + balsam fir, red spruce, red maple, pine) plus a growing library of CONUS rasters (LCMS disturbance, TreeMap forest type, relative density, BGI/CSI/CSPI, asymptotic AGB, LANDFIRE canopy height, Hansen forest loss).

Built on the perseus_db carbon-pool-harmonized database (current export schema perseus_api_v1). Static React + Vite + MapLibre, deployed via GitHub Pages.

Live app: https://holoros.github.io/perseus-forest-intelligence/ Latest deployed version: v1.3 (gh-pages, May 2026) Latest release tag: v0.56 (release tagging is currently behind the deployed build; see Known limitations).

Develop / build

npm install
npm run dev        # local dev server
npm run build      # static bundle -> dist/

Refresh data

public/api/ is the canonical source of truth (as of v1.6). The deployed data was assembled from the June hybrid / treemap-spatial / 48-state pipeline and committed here directly. The upstream perseus_db export (48_export_api.py) is currently partial (11 states, 2 scenario buckets, no yc_treemap_spatial_v1) and does not reproduce v1.6. Do NOT wholesale cp the DB export over public/api/ — it would regress the site to 8-state / 2-bucket data. Edit public/api/ directly, or merge specific files only after verifying they match the deployed schema. Re-syncing perseus_db to reproduce v1.6 is tracked in issue #24.

The public/raster/ overlays are produced by scripts/50_raster_image_overlays.sh. To add or update overlays, regenerate the affected PNGs + bounds JSON, drop them into public/raster/, commit, and push to main.

Deploy

Continuous deploy is live (v1.6, 2 June 2026). Pages Source is "GitHub Actions" and .github/workflows/deploy-pages.yml runs on every push to main: it builds with Vite and publishes dist/ via the Pages artifact path. A clean build reproduces the deployed site byte-for-byte, so pushes to main are non-regressive. The legacy gh-pages branch is no longer the serving source. To deploy: commit to main and push; the workflow does the rest.

What's in here

• CONUS choropleth (engines per state); 48-state coverage with focal ME/IN/GA highlight and multi-engine coverage for ME (32), WA (8), GA (7), MN (7), IN (6), ID (6), OR (5).
• Multi-model growth curves with metric and management controls (reserve, managed harvest, managed intensive, managed conservation), FIA reference, toggleable uncertainty ribbons (CEM lo/hi, FVS q10/q90), engine hover, URL deep-links.
• YC engine: FIA empirical Chapman-Richards yield curves stratified by forest type x EPA Level III ecoregion x ownership, with owner-specific harvest regimes and a climate sensitivity ribbon driven by a state-specific CSI envelope (eastern domain) plus modeled extrapolation for western and plains states.
• Tier B spatial layer (Maine): LANDIS biomass total + 4 species, timestep + opacity, layer-aware legend.
• CONUS raster overlays: LCMS disturbance cause (2022), TreeMap forest type group, relative density, BGI/CSI/CSPI productivity, asymptotic AGB, LANDFIRE canopy height, Hansen forest loss, harvest intensity, expected removal.
• Per-state HTML reports for GA, ID, IN, ME, MN, OR, WA.
• Stumpage view: $/MBF and $/cord by state, 120K observations 1977 to 2026, deflated to real $24 with the BLS CPI-U.
• Methods page: CONUS B1.1 vs B1.3 inventory stratification analysis at n=48.

Engines

The current export catalogues 34 engines across 48 states and 70+ metrics covering carbon stocks, fluxes, volume, biodiversity, mortality, disturbance, and ownership. See public/api/meta.json for the canonical catalog.

Methods and data provenance

• CONUS yield-curve methodology: docs/yc_engine_provenance.md and the in-app methods page at https://holoros.github.io/perseus-forest-intelligence/methods/.
• Inventory stratification analysis (CONUS B1.1 vs B1.3 at n=48): public/methods/inventory-stratification/.
• Per-state HTML reports: public/reports/{GA,ID,IN,ME,MN,OR,WA}_report.html.
• All scientific outputs cite perseus_db schema perseus_api_v1 (see public/api/meta.json for the current snapshot).

Known limitations

• perseus_db export is partial: the upstream DB reproduces only 11 states / 2 scenario buckets and lacks yc_treemap_spatial_v1; public/api/ is canonical until the DB is re-synced (issue #24).
• Western climate band: CSI rasters cover only the eastern domain (lon -90 to -52); 18 western and plains states use a modeled transfer from the national climate-embedding PCs plus latitude (R^2 about 0.3 to 0.4) and are flagged domain=modeled.
• YC managed scenarios are owner-class default rotations, not explicit per-stand silvicultural prescriptions.

Citation

If you use PERSEUS Forest Intelligence in research, please cite the underlying perseus_db data product and the CRSF yield-curve pipeline. Contact: Aaron Weiskittel ([email protected]).

Contributing

See CONTRIBUTING.md. Important: do not enable push: on .github/workflows/deploy-pages.yml until the v1.3 features in gh-pages are reconciled into main source. The v0.73-source and v1.3-deployed tags anchor both states.

Changelog

See CHANGELOG.md for the source (main) and deploy (gh-pages) timelines.