WildGap

Find where nature data is missing, then go look.

WildGap is a biodiversity survey copilot for school eco-clubs and citizen scientists. It compares recent and historical observation coverage across H3 cells, adds weather and climate context, and turns a candidate monitoring gap into a safe, shareable 60-minute field mission.

Public repository: github.com/Fink692/wildgap

Live app: wildgap-habitat-2026.fink692.chatgpt.site

WildGap ranks observation coverage gaps. It does not estimate habitat health, species abundance, population decline, or ecological causation.

What works

• Global place search through Open-Meteo Geocoding
• Focused seven-cell H3 analysis for 2, 5, and 10 km areas
• Live GBIF occurrence counts for the latest 12 months versus the preceding 36 months
• Transparent survey-priority score and confidence level
• Previous-month climate context and seven-day survey windows from Open-Meteo
• Interactive MapLibre/OpenFreeMap map plus an equivalent keyboard-accessible ranked list
• Portable, printable mission links that work without an account
• Field cards with score provenance, weather, readiness checks, a timed protocol, and optional evidence completion
• Device-local mission history with complete, account-free share links
• Explicitly labeled Winnipeg demo snapshot for outage-safe judging

Run locally

Requirements: Node.js 22+ and pnpm.

pnpm install
cp .env.example .env.local
pnpm dev

Open http://localhost:3000. No environment variables are required for the portable demo.

Account-free mission storage

WildGap has no sign-in, user account, or mission database. Missions stay in the browser's local storage, and the complete validated mission payload is embedded in its share URL. Anyone holding a complete link can read that mission, so users should share it intentionally and avoid putting sensitive information in evidence URLs. Opening a portable link saves that mission on the receiving device for offline-friendly return visits.

Scoring method

For each cell:

• D: inverse percentile of recent observation density among nearby cells.
• C: decrease from the preceding three-year annualized observation rate.
• T: evidence that plants, fungi, birds, or insects were historically/nearby observed but are recently under-observed.
• gapScore = round(100 × (0.55D + 0.30C + 0.15T)).

Confidence is High at 100+ baseline records, Medium at 20–99, and Low below 20. WildGap suppresses ranking when the area has fewer than 50 total baseline records or fewer than three comparable cells.

Architecture

flowchart LR
  U[Citizen scientist] --> E[Next.js explorer]
  E --> A[Analysis route]
  A --> H[H3 cell geometry]
  A --> G[GBIF occurrence API]
  A --> W[Open-Meteo APIs]
  A --> S[Transparent scoring]
  S --> M[Map + ranked list]
  M --> F[Field mission]
  F --> P[Portable URL/local storage]

Loading

Data sources and attribution

• GBIF occurrence API. Individual records retain their publisher licenses and citations.
• Open-Meteo geocoding, historical weather, and forecast APIs; weather data is CC BY 4.0 with attribution.
• OpenFreeMap map style, OpenMapTiles, and OpenStreetMap contributors.
• H3 for spatial indexing.

Quality checks

pnpm test
pnpm typecheck
pnpm build

Submission materials

• docs/DEVPOST.md — prepared submission write-up
• docs/VIDEO_SCRIPT.md — final 66-second product-cut guide
• docs/PILOT_PROTOCOL.md — optional future outing and tester protocol
• docs/TEST_RESULTS.md — technical verification record and intentionally blank future field sections
• docs/SUBMISSION_CHECKLIST.md — final operations checklist
• docs/OUTREACH.md — ethical public-voting copy
• docs/assets/architecture.svg and docs/assets/screenshots — upload-ready visuals
• /opengraph-image — generated 1200×630 social card

License

MIT for original code. Third-party data and libraries retain their own licenses.