BREAKING NEWS: airliner missing within Egyptian FIR

[Etud_lAvia]

Interpreting FR24 Data

I'm not an aviation professional (though I have performed engineering work on an IFR trainer and ground support systems for military avionics).

But I do know something about time series of discretely sampled data, and forensic analysis of sampled data records.

ADS was never intended to substitute for the FDR (we all know that, but ...).

ADS data can give a quite accurate portrayal of aircraft dynamics under conditions of near steady-state, or smooth gradual variation.

But the nature of these samples, as collected by FR24, makes them highly error-prone during sudden and extreme variations, such as the accident flight evidently experienced.

In recent years, we have seen cases in which the FR24 data indicated dramatic jumps in parameter values, which proved to be highly erroneous when compared with more reliable information.

Some commenters have already pointed out that certain sensors on which the ADS reports are based can give skewed readings under extreme conditions.

A big spike in the numbers is a reliable indication of ... a big spike in the numbers. Does it reveal what the aircraft was physically doing? Maybe ... maybe not.
_____________________________________________________

I have just confirmed that ADS-B (Downlink Format 17) does NOT include a time stamp, as others here have reported.

Accordingly, it seems that the FR24 time stamps are assigned by the FR24 receiving point and may be in error by one second or more.

This has some very important consequences. For one, FR24 samples sorted by time stamp might not correspond to the order in the aircraft sent the reports!

Also, it obviously means that any dynamic rate-of-change computation based on comparison of (supposedly) successive samples can be in error by hundreds (or even thousands) or percent.

Using such data to infer the dynamics of an airframe in a gross upset or other extreme maneuver is like ... walking blindfold while suffering vertigo.