Shheletar names aircraft, so the method has to be defensible. It logs two rules at Seletar — the night curfew and the daytime noise-abatement floors. Every curfew movement goes through the two checks below before it reaches the record; the altitude rule is measured differently.
Did more than one independent source see it? Provisional → Confirmed
Does the leg look like a permitted medevac or emergency? Flagged → or counted
A movement publishes the same night it happens, marked Provisional. The next morning we check it against other sources and either upgrade it to Confirmed or leave it Provisional. A movement's record number never changes.
Tail number, operator, route and the exact minute, live on the record the same night. So far, only we have seen it.
We saw it with our own antenna. We confirm by checking whether someone else's antenna saw it too. Two independent witnesses means Confirmed. Only us means it stays Provisional.
An independent witness was found. The movement carries a green confirmed badge on the public record.
No independent witness yet. It stays flagged and on the record, never dropped, and can still be upgraded later.
Medical evacuations, diversions and emergencies are permitted inside the curfew. We test each flight leg, because the same jet can fly a medical leg one night and a plain charter the next. For every movement we check what each source reported, then run two tests.
“Squawk” just means the identifier a plane broadcasts over the radio so ground systems know who it is. The callsign is set per flight, so it's the strongest signal that this leg is medical.
Used only when there's no medical callsign. It's a weaker signal, since operators fly mixed fleets.
When an aircraft keeps turning up on medical-looking legs, we phone the operator and ask directly. Where they confirm a specific tail is a dedicated air ambulance, we whitelist that airframe — every leg it flies is treated as exempt, not just the ones that squawk a medical callsign.
Any signal, from any source, is a match.
If any source, on any sighting, shows a medical signal, we treat the whole leg as medical. Once a leg looks medical it stays that way: a later data update can't un-flag it.
Something looked medical. Marked with a note on the record.
No medical signal on the leg. It goes into the tally.
The curfew is about time. The second rule is about height. Seletar's published charts set a minimum altitude over three residential Noise Abatement Areas; flying below that floor while inside the area is a breach. It is a daytime, geometry-and-altitude rule, measured from the same public ADS-B tracks.
| Noise Abatement Area | Over | Minimum altitude |
|---|---|---|
| Area 1 | Yishun | 1,500 ft |
| Area 2 | Yio Chu Kang | 1,500 ft |
| Area 3 | Ang Mo Kio | 700 ft — closest to the runway, so its floor is lowest |
We test the flown path point-by-point against the exact coordinate corners of each area, taken from the published chart. Only points inside an area can bust its floor.
ADS-B broadcasts pressure altitude against a fixed standard datum. We re-reference every reading to the day's actual sea-level pressure — the QNH — so the figure matches what the pilot read on the altimeter. Only a track below the floor by more than the measurement margin is recorded; a reading within the margin is held back, not asserted.
No landing or medevac exemption.
Unlike the curfew, the floors are a hard performance gate: an aircraft that cannot hold the minimum altitude is not meant to operate here at all. There is no “necessary for take-off or landing” carve-out for these areas, and medical-evacuation status — a curfew-only exemption — does not apply.
Below the floor, beyond measurement doubt.
Kept out of the public altitude record, on purpose:
What we can't see. Public data does not show an ATC descent clearance, confirm which areas were active that day, or settle runway dependence. We publish the measurement and the caveat together — and an operator can dispute any record with their own data.
An exempt-looking flight is never hidden. It stays on the public record, just outside the violation tally, with a note explaining why.
A charter jet that occasionally flies medical is exempted only on the leg it squawks a medical callsign. It's never blanket-cleared as “a medevac”.
Public ADS-B data can't confirm or rule out a real emergency. Where an operator declares an exemption we say so; otherwise we mark it possible and let you judge.
We use only public flight data — and the day's public weather, to correct altitude — never noise readings or anonymous reports. Here's what each source contributes, and whether it counts as independent corroboration:
| Source | Role | What it gives the record |
|---|---|---|
| Our receiver | Primary | Detects the aircraft over Seletar in real time. It's the first sighting of every curfew movement. |
| adsb.lol | Coverage | Additional overnight coverage; its next-day archive gives the authoritative route, origin / destination and path. |
| airplanes.live | Coverage | Additional overnight coverage and bearing on the same movement. |
| adsbdb | Lookup | Route and aircraft-owner attribution: who the operator is. |
| FR24 override | Lookup | Operator-name override where attribution needs correcting. |
| OpenSky | Independent | Someone else's antenna; its sighting is the independent witness that confirms a movement. |
| Flightradar24 API | Independent | A separate commercial network we don't feed. We cross-check every historical movement against it: a match is a second independent witness (the record then reads "recorded on 2 networks"), and where it disagrees with adsb.lol on the origin or destination, we flag the conflict rather than pick a side. |
| IEM ASOS (weather) | Altimetry | The day's sea-level pressure (QNH) at Seletar. Used only for the altitude rule, to correct ADS-B pressure altitude to true height; each altitude record notes the QNH used. Not a noise measurement. |
We log what flew and when, nothing more. Shheletar reads open ADS-B records; it doesn't measure noise or take complaints. The curfew window is 2200–0700 SGT, a purely time-based quiet period. As a benefit of the doubt, movements within 5 min of the 22:00 start or 15 min of the 07:00 end are not flagged.
Where the data is uncertain, the record says so rather than guessing. That's why we publish the method: you don't have to take our word for any single movement. You can check it yourself.
Every record is run through these checks before it publishes. Leave your email for the morning-after alert, or browse the full log now.