In the AF447 disaster, one of the elements that contributed to the crash was an "excessive" nose-up input shortly after the disconnection of the auto-pilot.

This caused the aeroplane to climb and lose airspeed, precipitating the ultimately fatal stall.

  • Why did the pilot flying pull back (excessively) on his stick? Did the investigation find probable causes?

He was at the same time correcting for turbulence-induced roll. Would a nose-up input in such circumstances be appropriate?

  • If so, how easy is it for a pilot to make an excessive input of that kind without realizing it, and what would normally prevent that from happening?

  • If not, is there any explanation for why he might have made it?

Most of the crew's actions in the incident are comprehensible, even if they were incorrect responses. They make sense in the context of the situation, and it's possible to have a fairly clear picture why they made the judgements and decisions they did.

Is there a good explanation for that initial nose-up input, that makes sense of it too in the context?

  • 3
    $\begingroup$ I'm voting to close this question as off-topic because accident speculation is explicitly called out as off-topic in Help/On-topic $\endgroup$ Commented Sep 20, 2017 at 21:24
  • 2
    $\begingroup$ We have a few questions about AF447 already; this one might have some relevant info. And the official report is available, that has all the objective data. $\endgroup$
    – Pondlife
    Commented Sep 20, 2017 at 21:45
  • 1
    $\begingroup$ He didn't recognize the stall, the captain did but by the time he realized that the control inputs were wrong it was too late. This accident resulted in more stall training and hand flying for pilots when automation fails. $\endgroup$
    – Ron Beyer
    Commented Sep 20, 2017 at 21:45
  • 2
    $\begingroup$ Daniele, have you read the report issued by the French accident investigators? It's pretty thorough (BEA). If that body after an exhaustive investigation, could not come up with the why, who do you think can? The "why" being nailed down requires something like mind reading. If you have not read the report, I suggest you do that. Once you've read it, consider editing your question to reflect what you learn. $\endgroup$ Commented Sep 20, 2017 at 22:11
  • 1
    $\begingroup$ How are questions like Would a nose-up input in such circumstances be appropriate?, how easy is it for a pilot to make an excessive input of that kind without realising it, and what would normally prevent that from happening? or is there any explanation for xxx? "opinion-based"? I am not interested in guesses as to the contents of his mind; I am asking about piloting responses to conditions. Anyone who imagines those are a matter of opinion doesn't understand aviation. Or perhaps, questions. If the answer is that his initial action remains unexplained, that's also a possible answer. $\endgroup$ Commented Sep 21, 2017 at 7:52

2 Answers 2


It's not accident speculation when it has been investigated.

The full accident report is available online with a very adequate summary on Wikipedia.

There are many reasons, but from the human factors study:

The final BEA report points to the Human Computer Interface (HCI) of the Airbus as a possible factor contributing to the crash. It provides an explanation for most of the pitch-up inputs by the pilot flying (PF), namely that the Flight Director (FD) display was misleading.

The pitch-up input at the beginning of the fatal sequence of events appears to be the consequence of an altimeter error.

The crew present in the cockpit at the time of the pitot blockage lacked experience/training, and were bombarded with conflicting warnings and indications.

As I've been reminded in a comment, they were also in darkness over an ocean with no external visual references.

A note on how human factors studies can relate cause and effect. There is a certain reaction time that we are unable to exceed, roughly it's 200 milliseconds for a trained person anticipating an event. It's longer otherwise and there are many studies on that. When a control input is put up against the various indications, and taking into account the reaction times to the different aural and visual warnings, then scientifically the probable cause can be known within reason.

From page 42 in the BEA report:

The drop in Mach also impacts the SAT and thus the true air speed and the wind speed.

In the following table, the case an A330-200 flying at FL 350 at Mach 0.8 in standard atmosphere with a 30 kt head wind is given as an example to illustrate the consequences of pitot icing that would result in a drop in Mach from 0.8 to 0.3.

enter image description here

During Pitot probe de-icing, the same variations occur in the opposite direction.

From the table, you can see how the BEA may have assessed that the likely reason for the initial nose up reaction by the pilot flying was to correct a ~300' altitude error.

They also say (p. 157):

The report specified that the stall was caused by inappropriate crew reactions to erroneous speed and Mach displays that resulted from blockage of the Pitot probes through atmospheric icing. The report stated that contrary to standard operating procedures, the crew had not switched on the Pitot probe heating.


First Officer Bonin's actions are almost a textbook example of the startle effect.

Simply put, he was shocked by the sudden disconnect of the autopilot that he mentally froze and couldn't think straight. He appears to already have been in a state of anxiety due to the St Elmos fire immediately beforehand, and the thunderstorm in the vicinity. It is a very interesting but sad human factors discussion.

A good explanation can be found here.

  • $\begingroup$ @KorvinStarmast Edited to make it clear that I was referring to Bonin. Physically he did not freeze, but mentally he was stuck on his initial reflex, and could not take a step back to figure out what was happening. $\endgroup$
    – Ben
    Commented Sep 21, 2017 at 0:29
  • $\begingroup$ Yeah, I see the point you are making. Also fits better with your link. $\endgroup$ Commented Sep 21, 2017 at 2:14
  • $\begingroup$ @Ben I don't believe this is entirely correct. Although the startle effect is certainly implicated in the case, far from freezing, Bonin appeared to make a number of immediate corrections. Some were appropriate in kind if not degree (e.g. he overcorrected for roll) but the initial nose-up input of those first few seconds is harder to fathom. $\endgroup$ Commented Sep 22, 2017 at 16:29

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .