Takata's analysis suggesting a turn back toward Fernando de Noronha is interesting and well thought out, but I think there is one factor that has not been considered. By 0210Z the aircraft was evidently just NW of a very large CB cell and the hypothesized turn would have taken the aircraft right through this cell complex, with tops exceeding 50,000 feet. See the BEA flight path points plotted on the probable radar image reconstructed by Tim Vasquez at:
Jet Crash forum • View topic - Crash location
I suggest a different final location here. The now-discredited 0214Z "ACARS" location of N3.5777, W30.3744 that was attributed to FAB has always been intriguing because it is quite close to the location where bodies were recovered by FAB on 6 June (see the FAB cumulative PowerPoint presentation of 6-6-09). Now that it is clear that the aircraft transmitted position reports at ten minute intervals throughout the flight, we must ask ourselves what else was contained in those position reports.
Although I have no idea how the A330 system is configured, the aircraft GPS tracking devices with which I am familiar provide (in addition to latitude and longitude) ground speed, heading, and altitude. See for example the devices offered by:
EMS Sky Connect
Thus, in addition to latitude and longitude, the heading of the aircraft at 0210Z was probably known to BEA by the morning of 1 June. These headings are calculated by the GPS receiver based on the azimuth between present and previous positions, and these positions are typically calculated at about 1 second intervals. My hypothesis: BEA actually calculated the N3.5777, W30.3744 position from the heading at 0210Z, and projected that heading forward to 0214Z when the last ACARS maintenance message was sent. But that position requires impossibly fast speeds to cover the 38.2 nm distance between 0210Z and the N3.5777, W30.3744 position (573 kt, ~Mach 0.98). Could it be that BEA simply used an erroneous assumed speed for the calculation?
If you use the last calculated ground speed of AF447 between 0200Z and 0210Z, 463 kt, and project it forward along the path between the 0210Z and the "0214Z" FAB location, the aircraft would have covered about 30.9 nm and the 0214Z location would be 3° 27.8'N, 30° 25.0'W, about 7.3 nm south of the FAB 0214Z position. Assuming the aircraft made only one turn to the left, projecting the path from 0210Z back to intersect the planned flight path would place the hypothesized turn at about 2° 34.5'N, 30° 44.8'W. This would place the crash location south of the recovery location of the first bodies on 6 June, as appears to be required by the drift direction of the debris field.
Alternatively, perhaps the 0214Z ACARS maintenance time stamp is truncated, and the actual time might have been as late as 02:14:59Z. If BEA made this assumption to make a best estimate of where the wreckage might have been to aid the SAR teams on the morning of 1 June, from the 0210Z position at 463 kt the aircraft would cover 38.6 nm in five minutes, placing it only 800 m north of the "0214Z ACARS-FAB N3.5777, W30.3744" position.
I am unable to post images in this forum for some reason, but you can see a map of this scenaio on page 2 of:
Jet Crash forum • View topic - Crash location
-rer47