Live measurement

Measure a pass recorded today

Everything else on this console was measured in August and frozen so it could be checked. This one runs now. Paste any public SatNOGS observation id, including one recorded in the last hour, and the same function the command line calls will download its waterfall, propagate the pass from the elements in its own record, read the frequency axis off the tick labels, decide whether the station corrected for Doppler, fit the offset and score it against permuted nulls.

Read-only, no credential, no GPU, and nothing is written to SatNOGS or anywhere else. Three outcomes are normal and distinct: an uncorrected capture gets an offset and a p value, a corrected one gets an offset with no null test, and an image that does not settle which gets UNRESOLVED, the answer that says skip this one.

THE SKY OVER THE STATIONNESW30°60°TCA 89.7°THE RECORD, AS THE RECEIVER WRITES ITRECEIVED FREQUENCY, OFFSET FROM 436.400 MHZ-28k-14k0+14k+28k0:001:102:213:314:41STATEPredictedTrackingFittedELEVATION0.0°89.7°89.7°DOPPLER SWING19.6 kHz19.6 kHzFITTED OFFSET+32.05 ppmobs 14740027 · OBJECT AF · M0EYT / 2E0NOG · window 4:41 · one loop is 14s

Paste an observation id

Any public id works, including one recorded minutes ago. Find one on the SatNOGS observation list. The endpoint downloads one image and fits one corridor, so expect tens of seconds, and it caches each id for a day so a second reader pays nothing.

Frozen shelf: 6 passes recorded after the snapshot

Measured with the same function at 2026-08-20 21:56:15.843009 UTC and baked into this build, so a button answers with no network call. Every one of them started after 2026-08-16 21:36:32.792753 UTC, which is when the corpus this project was built on stopped growing, so nothing here was available to any model in this repository. 3 of 6 settled the Doppler convention; the rest are UNRESOLVED, which is what most of a real queue looks like. Null distributions are 99 permutations.