Synopsis:
A Reconstruction animation (not entirely accurate as it shows an attitude change (pitch-up) but no actual zoom):
YouTube - blenderpedia's Channel
.
Stick-shaker suddenly alerted pilots (at around 110kts) that they were approaching the stall (due autothrottle [A/T] not engaged/inadvertently disengaged). Surprised pilots added max power and the aircraft then pitched up violently. Cause of this scenario and outcome (as per the Colgan Q400 accident) is two-fold:
.
a. If autothrust isn't present on a captured glideslope, the speed will gradually and insidiously bleed off (depending upon the power set at the moment of A/T disconnect and any subsequent-to-disconnect gear/flap down selections). The autopilot's auto-trim function will attempt to maintain the "captured" ILS 3 degree glideslope by trading speed for a height-loss regulated to the standard 600fpm rate of descent. (justification for this is a recorded lowest speed on finals of a pre-impact low-point of 88kts recorded on GPS for THY Flt TK1951 ). The record shows 88kts GroundSpeed at 420ft and a last recorded position N52 22.8 E004 42.8 (right where the impact occurred). 88kts GS was the final recorded value in the software log from a guy with a Radarbox Mode-S receiver who is fairly near EHAM, and has coverage of aircraft on the ground there so he would be able to receive the Mode-S from where TC-JGE was just before impact. Naturally because the position, speed and height transmissions are not all sent at the same time you can't say the three things are coincident, but they will be within a second or so
.
b. Because of the cumulative effect of near-to-full pitch backtrim (stabilizer and elevator trim movement towards nose-up as speed decreased) AND the strong nose-up pitching couple of max power at a low IAS, adding full power at near to stall speed caused an abrupt nose-up pitch, a stick-shaker
....and then a stick-pusher input, a pilot throttle-back at the apex of the zoom and, soon thereafter, a panicky pilot's hard PULL (for post-stall ground avoidance). It's a nasty and confusing sequitur that's never ever trained for in a simulator. Justification for this scenario is the witness reports (inclusive of the "turbulence buffeting" felt pre-impact and the pre-impact pitch oscillations (on the way down from its zoom-climb) -as observed by the qualified witness on the adjacent freeway).
.
Causation: When you become used to the auto-throttle "taking care of" the aircraft's approach speed (per the FMS bug-set), there's no perceived (or real) need to monitor it closely for any required power adjustments, post-configuration increments etc. It (the IAS) and even the pitch attitude just "drops out of" the pilot's active scan..... and/or he/they become distracted anyway by a third member in the cockpit or another task (FMS keypadding etc). The type of scanning done nowadays is very passive and detached (i.e. no input required or feedback loop involved) rather than the old-school active pilot-involved-in-the-control-loop scan ("oops, I'm a bit low and slow - must add power/raise nose/trim and fine-tune that heading"). This automotive disassociation leads to a lack of due deference - and opens the door to any passing/insidious technical upcocks. We've not seen the end of these type accidents (and many have occurred to date - but never been acknowledged as such). So forget birdstrikes - in this instance. Inattention may be the real enemy