DisasterIntel fuses weather, seismic, social, and global alert signals into verified, prioritized incidents — giving responders a single pane of truth.
Raw feeds become actionable intelligence through verification, anomaly detection, and incident fusion.
Heuristic weighting, time-window correlation, and location overlap across all sources for confidence scoring.
Spike detection across time-series data to catch emerging disaster signals before they trend.
Merges repeated events by disaster type, location key, and time window into single responder-facing incidents.
Incident-level scoring with priority bucketing, impact radius estimation, and recommended responder actions.
Time-weighted confidence decay so stale events lose priority and fresh signals surface first.
Firecrawl-powered search to corroborate or flag incidents with external evidence and stance inference.
Entry point: ingestion/pipeline.py
Async ingestion from Open-Meteo, USGS, GDACS, IMD, Reddit, and Firecrawl/news simultaneously.
Source trust weighting, time-window correlation, location overlap, and social credibility heuristics.
Statistical spike detection across event time-series to surface emerging threats.
Merges correlated events into unified incidents based on type, location, and time window.
Adds priority bucket, AI relevance index, impact radius, verification status, and responder actions.
Writes to latest_intelligence.json and timestamped snapshots. Backend API serves the dashboard.
/api/intelligence/latest/api/incidents/priority/api/map/layers/api/timeline/api/incidents/{id}/verify/api/ingestion/runSix source adapters feed the intelligence pipeline.
Weather
ActiveSeismic
ActiveGlobal Alerts
ActiveWeather (India)
LimitedSocial
ActiveNews / Verify
ActiveStart the backend, trigger ingestion, and open the GIS dashboard to see live disaster intelligence.