seakobotumile04-eng/geoguard-ai

GeoGuard AI is a Botswana-first environmental decision-intelligence platform for biodiversity monitoring, risk prediction, intervention simulation, incident response, and human-led conservation action. # GeoGuard AI 3.0 — Environmental Decision Intelligence **See the signals. Simulate the future. Protect what connects us.** GeoGuard AI is a Botswana-born environmental decision-intelligence platform. It consolidates ecological signals, preserves provenance, generates transparent baseline predictions, compares non-harmful intervention scenarios, ranks possible responses and requires authorised human decisions before action. This is a complete full-stack FastAPI prototype, not a static landing page. ## Intelligence cycle **Collect → Validate → Consolidate → Analyse → Predict → Simulate → Recommend → Verify → Respond → Learn** ## Functional workspaces 1. Cinematic landing experience 2. Ecosystem Digital Situation Room 3. Ecological Signal Constellation 4. Living intelligence map 5. Prediction Centre 6. Intervention Simulation Laboratory 7. Recommendation comparison and human decision 8. Evidence Ledger 9. Biodiversity evidence explorer 10. Incident-response centre 11. Field-network monitoring 12. GeoGuard evidence assistant 13. Reports and low-bandwidth exports 14. Responsible AI centre 15. Data-source and model-transparency centre 16. Accessibility, permission and low-bandwidth settings ## Current data adapters - Open-Meteo — current weather and seven-day forecasts - iNaturalist — research-grade biodiversity observations and images - GBIF — species occurrence records - NASA EONET — significant natural-event metadata - NASA FIRMS — active-fire and thermal-anomaly adapter; requires `NASA_FIRMS_MAP_KEY` - Copernicus Data Space — Sentinel-2 catalogue metadata adapter - GeoGuard prototype field network — explicitly labelled generated telemetry for workflow demonstration - Human incident reports — local SQLite workflow with audit history GeoGuard does not fabricate missing external environmental values. When a provider is unavailable, the data state is shown as unavailable or incomplete. ## Prediction engine The included models are transparent rules-based basel …