Eyeballing the QAR traces, there's a ≈2 second spike in the AoA topping at 20.92°. I suspect a vertical updraft pivoted the vanes entering the updraft. It would take ≈1 second before the entire 55m of aircraft would be completely in the updraft. The graph eventually shows AoA stabilised at -10°, which raises the question of how well the vane position in the graph is calibrated to wing AoA.