Frequency comparisons are withheld. A week's traffic measured over a couple of observed days is not a small sample, it is a different quantity: divide it by a seven-day baseline and every carrier looks cut in half by our own downtime. The observed counts are still shown, the bars are left empty rather than drawn against a ratio the window cannot support, and every carrier reads UNKNOWN until enough days accumulate.
No reference period yet. Until backfill-baseline
finishes, this page can say whether a carrier is flying, but not whether
it is flying less than it used to. Every carrier therefore reads
NO BASELINE, and the frequency bars are left empty rather than drawn
against a zero that would look like a collapse.
Carriers that have gone silent for longer than the threshold for their scope, dated from their last observed flight. Silent days during a coverage outage are not counted.
Services observed operating again after a stop, in the last 30 days.
Weekly departures against a frozen pre-escalation baseline. Traffic is scaled to the days we actually observed before it is compared, so the left figure is a full week only when enough of the week was seen — otherwise it is the raw count, labelled with the days behind it, and no bar is drawn. A carrier only reads SUSPENDED after three consecutive observed silent days; blind days neither confirm a suspension nor clear one.
Raw ADS-B counts against the winter 2025–26 baseline. None of the checks behind the stop register has been applied (return leg, airport boards, carrier visibility), so seasonal routes and receiver gaps show up here. This is not a list of suspensions; those are on the front page and in v1/suspensions.json.