I looked at this again last night and the only way the Inmarsat arcs work with the autopilot is if the plane was put on autopilot deadheading to New Amsterdam Island shortly after turning south for about 4.5 hours —
BUT then in order to hit that final arc it would have had to come off autopilot and that heading with about an hour of fuel left. It’s at that point that it would have to turn to the east toward Perth to hit that final Inmarsat arc and the new search area.
Whether he then put it back on autopilot after making those course corrections with an hour left, who knows, but that’s what they must be calling “being on autopilot”.
Very interesting. But are you doing anything to correct for the spherical to flat surface projection. Or, alternatively, have you determined that the mapping of straight lines works on the projection you are using.
As an example of this issue just look at the projection of any non-equatorial satellite on a flat map. No straight lines to be found.