Scoring methodology

Events are ingested from eight official and scientific sources -- USGS Earthquake Hazards, NOAA National Weather Service, NOAA National Hurricane Center, NOAA Storm Prediction Center, NASA EONET, GDACS (EU/UN), OpenFEMA, and UN OCHA ReliefWeb. Each source is fetched on a rolling basis (feeds refresh roughly every 2 minutes when the site is active), then normalized into a common event model, de-duplicated across sources (the same hurricane reported by NHC and GDACS becomes one event with merged attribution), and validated for coordinates and timestamps before display. Every event page lists its underlying sources with links and their last-updated times. See Data Sources.

Answers: how dangerous is this event, right now, to people near it? It blends four weighted inputs:

Official alert level (40%). The highest active government alert attached to the event (advisory → watch → warning → emergency), taken directly from the issuing agency.
Physical intensity (30%). Type-specific magnitude: earthquake magnitude and depth, hurricane category and wind speed, flood stage, fire growth rate, etc., scaled to the historical range for that hazard type.
Population proximity (20%). Whether the event footprint touches populated places, using the event geometry (warning polygon, forecast cone, or radius).
Trajectory (10%). Whether the event is intensifying or weakening between updates.

Bands: 0-39 advisory (blue), 40-54 watch (amber), 55-71 warning (orange), 72-84 emergency (red), 85-100 critical (deep red). These colors are used consistently on the map, feed, and event pages.

Answers: how unusual is this event for its region and season? It compares the event's intensity against the typical range for that hazard type in that part of the world at that time of year (e.g., a magnitude-5 earthquake is routine for Japan but highly anomalous for the U.S. East Coast). Inputs: deviation from regional baseline frequency, deviation from seasonal norms, and rarity of the event type at that location. High anomaly scores flag events worth attention precisely because they are off-pattern -- they are not a measure of danger by themselves.

Answers: how much could this event disrupt commodities, sectors, or supply chains?It maps the event's location and type against known economic exposure: commodity production regions (energy, agriculture, metals), transport chokepoints (ports, rail corridors), manufacturing concentrations, and insured-value density. Severity and projected duration scale the result. Where live market data is available, observed price moves in related instruments are shown alongside -- clearly labeled as observations, not predictions.

This score is context, not advice. It is never a recommendation to buy or sell anything.

Sources can be delayed, revised, or temporarily offline; coverage is strongest for the U.S. and events tracked by international agencies. Scores update only as fast as the underlying feeds. The platform does not guarantee that every disaster, warning, or hazard is detected or displayed, and it is never a substitute for official emergency alerts. Population and economic exposure use approximations appropriate for triage, not for engineering or actuarial use.