how "stable" is "stable flight" in "turbulent air"
Several posters made reference to "stable/level flight" before AF 447 "a/p disconnect"....
If my understanding is correct, for the same trajectory, the controls/adjustments performed by the "a/p" during a flight through "turbulent air" are quite different in certain aspects, from those made during a flight through "non-turbulent air".
Assuming a straight line "overall" trajectory:
a) in "non-turbulent air", the flight is "a straight line", with almost an "inertial", no significant controls/actions from the "a/p".
b) while in "turbulent air", the trajectory under the effect of the turbulences is a concatenation of a sequence of "up" "down", "left" "right" "level" segments, with a distribution as a consequence of the "a/p controls" of "up/down", "down/up", "right/left", "left/right", and "level" in between, which is at best ONLY an "approximation" of "a straight line", and thus of course NOT "a straight line".
The segments are shorter, or longer, depending on the degree of turbulence, and "a/p actions".
In terms of actions, the "a/p" "actions/controls" continuously try to counter the "up", "down", "left", "right" a/c motion driven by the turbulence, in an attempt to keep the flight level/straight.
It seems clear that before the "a/p disconnect" the "a/c" was in "turbulent air".
The trajectory the "a/c" followed immediately after the "a/p disconnect" depends on which type of segment and which position were the control surfaces of the "a/c" at the very moment of the "a/p disconnect".
My understanding is that the "a/c" would continue "level" under inertia and position of control surfaces as left after the "a/p disconnect", ONLY and ONLY if the "a/c" were on a "level" segment.
But the probability of being on a "level" segment is 1/5, which is 20%, which is quite low (maximum probability is 100%).
The probability of being on a "NON-level" segment - "up" or "down" or "left" or "right" - is 4/5, that is 80%. That is quite high!!!