Door Seal, Outflow Valve?
bblank
The range and bearing accuracy of a military radar depends upon the type (PRF,PW,Scan Rate, band etc) - but the greatest factor would be the distance of the radar head from the scene of the incident and whether it was on all-round scan (at the time) and how long until/whether it was switched to sector across the area of interest (thereby increasing the paint-rate). With no skin paints from the civil ATC radar, it was suddenly a matter for the investigators of trying to marry up a myriad of radar plots (secondary giving way to a number of primary returns deduced from radar tapes from an unknown number of military sites). The inaccuracies that creep in are related to time-locking the radar sweeps (remember the Valujet debate over that). In addition of course, you have the loss of echo persistence when the primary target dissolves into a number of tumbling pieces. Some primary radars use Doppler or MTI circuits to guess ahead and can easily pick up another (different) piece of nearby wreckage that happens to be presenting a good cross-section at that instant. The resulting deduced track-plot of "debris" might be totally skewed by invalid data. And so the end result of any multi-radar correlated plot might be so much wishful thinking (or at best, the best guess, and possibly useless for any meaningful deductions).
The speed variation might be the best indication of what eventually happened. One explanation of the airspeed discrepancies 750-850kph may be that as they climbed, a door or hatch may have part-opened / lost its seal and started a vibration. Varying the speed within a reasonable band can give a crew the clue as to whether it's aerodynamic or not (or maybe engine-related). Let's say that they concluded that a door or hatch was part-open (seal blown or something lodged in the seal) and that seal finally gave way and caused alarming noise/vibration. If the engineer then went aft and maybe did something a bit adventurous or foolhardy?.... Or maybe it was just the noise-investigation varying-of-speed prelude to a door/cargo-door letting go (of its own accord) as they climbed higher. It may have been (post-maint) some thermal/acoustic blanket stuck in an outflow valve (that's happened in the not too far distant past). Probably a good idea NOT to climb higher in such a situation (i.e. keep the pressure differential down, descend at low IAS).
Inflight trouble-shooting rarely works out happily.....best done on air-tests without pax.