Every point below is a row of satcom_observations.csv (WS-004 canon). Nothing is simulated. Time is UTC, 7–8 March 2014.
Dashed lines mark the 10 usable BTO events from the validated sequence (table below). Raw BTO is shown, not corrected for bias. Within the same minute the log carries more than one BTO level (for example about 9,9xx and 14,9xx µs near 16:41–16:42). I show them as recorded and do not assign a cause.
BTO status VALIDATED absolute-calibrated, default bias −495,679 µs. Calibration max benchmark residual 59 µs against a 60 µs tolerance. 10 usable BTO events, 12 Table 6 rows, 2 BFO-only rows, 2 excluded log-on rows.
Those manifests describe the June 2026 restart, where BFO residuals were blocked. Your later frozen BTO/BFO scorer output (engine v06) is not in the folders I could read, so no scorer residuals or scorer-derived position appear here.
From mh370_satcom_canonical_events_v1_0.csv. 12 canonical observations, 6 restart transients excluded, 1 terminal BFO-only row. Corrections of −4,600 µs are applied to the two log-on requests.
11 validated states, 16:30 to 00:20 UTC. Latitude and longitude of the sub-satellite point are computed here from the ECEF position. Perth ground station (ECEF km): −2368.8, 4881.1, −3342.0.
The 00:20 sub-satellite point (0.53°N, 64.46°E) matches the centre of the circle I fitted to your reference arc (0.53°N, 64.33°E), which supports treating the arc as a circle about the satellite.
From satcom_events.csv (WS-002 chat export). Residual is observed minus predicted BTO for the diagnostic path family used in June, not a located aircraft. Blank means the file has no value.