Measured earliness

“Faster than the press, really?” Fair question — the only honest answer is a measurement. For every corroborated fire of the last 72 hours, this page compares two public, verifiable timestamps:

• First kanari signal: the time of the first satellite pass that saw the fire (VIIRS, GOES or Meteosat — UTC time supplied by NASA and EUMETSAT, not by us);
• First press article: the publication time of the first article detected by our monitoring (GDELT) mentioning this fire.

The gap between the two is the measured lead. It probably underestimates the real lead over public awareness (a published article is not yet read), and it says NOTHING about a lead over emergency services: firefighters have their own channels (emergency calls, watchtowers, cameras) that are often faster than the press.

Measured cases (rolling 72 h)

No measurable case right now: no fire of the last 72 hours has both a satellite detection and a dated press article yet. Come back during an active episode.

Case study: the Diois fire (Drôme, France), August 3, 2026

On August 3, 2026 at 16:00 UTC, the Meteosat MTG geostationary satellite picked up a hotspot in the south-east of the Drôme department (Diois area, anchored to Serres, the nearest town in our gazetteer). Within 2.5 hours, kanari accumulated 174 satellite detections on this fire, with a peak measured intensity of 1,193 MW — the most powerful French fire recorded in our archive since it opened. Every detection is timestamped by EUMETSAT (one image every 10 minutes), not by us.

On August 3, 2026, kanari measured a 1,193 MW fire in the Drôme (France): 174 satellite detections in 2.5 hours, the first at 16:00 UTC. Source: kanari.io, Meteosat MTG data (EUMETSAT), CC BY 4.0.

This case shows the measurement density of geostationary monitoring, not a lead over official alerts: we have no reliable press timestamp for this fire. The full chronology is public: permanent fire page.

Limits

• Press monitoring (GDELT) does not see everything: a fire may be covered by a local outlet without appearing here.
• The “first signal” is the first satellite pass: actual ignition happens earlier.
• The sample is short (72 h) and varies with the news cycle: this is a measuring instrument, not a sales pitch.