Hi there.
@02:10Z, the crew is submerged by alarms & ECAM messages, possibly false alarms (those are not transmitted in the ACARS) like in the other past cases of Pitot failures (stall alarms most of the time, overspeed alarms occasionally), airspeeds are unreliable, no more A/P or A/T, alternate mode 2 (loss of many flight protections), no visual clues (night with bad visibility), possibly turbulent atmosphere... and lengthy/tricky procedures to implement... The anemometric failure is detected @02:10 but the failure of the 3 anemometric chains may have begun some tens of seconds before, beginning with the exclusion of a 1st ADIRU against the two others (which may have diverged in a similar/agreement fashion) ending with the disagreement between the two ADIRU remaining.
Some others have pointed out the plausibility of a crew rotation just before (@02:00Z), CPT going to rest.
At this point (@02:10-02:11), taking in account the large route variation of the AF447, wouldn't the worrying possibilities be: a bad reaction of automation (A/P) to the erroneous airspeeds before their unreliability can be detected (for example A/P & A/THR OFF with TOGA LK which would remain locked: overspeed risk) and/or drifts induced by A/P OFF (a growing bank angle, since the automation did not control it any longer), and/or the risk of overloaded attentions of the crew, stretched between procedures implementation and manual flight: as I see it, a first risk would be to badly react to a false stall alarm (frequent) meaning taking it for real and following the procedures accordingly (risk of overspeed), another risk would be not to pay attention to the growing bank angle, and let it drift in excess (say over 60°-70°) before trying to regain control. The high altitude overspeed/mach buffet has been somewhat debated on pprune, the other is similar to the Adams Air case (but less extreme: for AF 447 the roll would be controlled and the control of altitude close to be regained). Maybe there are other early (@02:10-02:11Z) risks in the chain of events ?
We now know that the crash area could be such that it imply a large route deviation (more than 135°), and a rapid descent (35000 ft over 40-50NM in less than 4mn45sec: -9° of average slope or ~7500 fpm in average). Beyond the high altitude stall/mach buffet, there are also the fully developped flat spin hypothesis, and the engine flame out. Apart the high altitude mach buffet or stall, could the terminal trajectory be explainable by a large initial roll/bank perturbation in cruise at high altitude like the Adam Air ? (possible causes: bank drift that remains unoticed due to task overload & spatial disorientation, only a wing stalls in overspeed, turbulences,...). The time needed to control the large initial bank perturbation would induce a large route deviation and loss of altitude. Maybe also that attempting to control the altitude before having controlled the excess of bank angle would be a bad reaction and would worsen the situation (Adams Air case I think). The terminal conditions described by the BEA reports (A/C en ligne de vol so 0° roll/bank, nose up/tail impact first, vertical compression) suggest that, in the vertical plane, there may have been a rapid loss of altitude initially (much larger than the average 7500 fpm, combined with the large roll pertubation, to control first) followed by a ressource (route stabilized toward E-34) in an attempt to control the altitude, close to its lowest point (vertical speed much lower than the average 7500 fpm), with an incidence (nose up) at the impact/pancacking with the surface. I think that the trajectory would pretty much look (horizontally) like the MM43's illustration (a blunt/short radius turn (with a rapid loss of altitude) and a rectilinear end (which would be the ressource attempt).
Jeff
Adams Air case:
Aircraft Accident Report AdamAir PK-KKW (see page 28, the bank angle exceeded 90°, the variations of altitude -35000 ft & route: +250° in ~1min)