I am still not convinced by most theories posted here on why there was a nose-up input as soon as the autopilot dropped out due to unreliable speed data. The most rational so far:
- it was an involuntary action, an unwanted side effect of putting in a left roll demand to counteract a right wing low tendency, or
- It was the response to a false indicated speed increase,
I find (1) unconvincing because it was one hell of an input, sufficient to cause a 7000 fpm climb and, apparently, two stall warnings, sounds sort of deliberate to me.
similarly (2) sounds wrong because, while we don't know for sure what the indications were on the PF's panel, due to the lack of recorded data, we do know that the other two systems showed abrupt speed decreases, and the previous high altitude pitot icing incidents, including the
Air Caraibes precursor, also only registered speed decreases. So it would be extraordinary for the unforunate AF co-pilot to be the only one to have been presented with the opposite.
That same Air Caraibes report also said that at the time of the beginning of the anomalous air data, the indicated altitude rapidly dropped 300 ft (presumably due the lack of a vaild Mach number to correct the static pressure (? ) the BEA has already told us that with no vaild M, the stall warning system uses a value near zero...) Maybe the PF was reacting to that, admittedly 300 ft isn't much, but I suspect it looks a lot on the tape, and it would have been rapid. Perhaps this is the trigger we've been looking for.