1. medical occurrence on non-autopilot downwind resulting in slow descent and inability to send mayday, such as stroke or heart attack
2. mechanical failure but this does not explain lack of Mayday call
3. intended non-autopilot low circuit due to weather in apparent limited but possibly intermittent visibility but with grey indescript (no contrast) clouds ahead and sudden (re)entry in IMC resulting in disorientation and CFIT
4. A DIY approach in solid IMC, with an error made on the altitude.
I think it would be logical to not be on autopilot if in a bad weather circuit
It would depend on the "bad weather". A visual circuit would be flown by hand, but in any high workload situation, and that includes any approach flown for real (in IMC), one would use the autopilot. Of course some pilots will fly by hand in IMC for the practice, too, but I think this highly qualified pilot (especially coming from the USA/FAA stable where they don't go for the European "life should be hard" way of doing everything by hand with the GPS turned off to make life extra hard) would have known that pilot workload minimisation is the key to safety and would have used every piece of kit he had.
My guess is #4, and a straight flight into terrain in which the pilot never suspected anything is wrong.
I don't believe it was a
lateral navigation error because he would have had better nav kit than a 1980s 747.
So, the big Q is why was he too low?
QNH was reported as 1012 which rules out the easy mistake of forgetting to set the altimeter to the local QNH, after coming off the enroute section which would have been flown with a setting of 1013 - the difference is just 30ft.
Did he descend on QNH instead of QFE? LOAV is 765ft up and that is plenty in this case to account for the error. I have no idea if QFE is used out there. I killed myself once (on a simulator) at EGKB by going missed and forgetting to reset the altimeter from QFE back to QNH. So, today I never use QFE at all and if this is passed by the airport I ignore it. I think QFE is a British-only thing (the RAF uses it) and this pilot did some some British connections. It would suprise me if this was the reason though because no real instrument pilot I know ever uses QFE.
Did he set the autopilot to descend to 1500ft instead of 2500ft? It's easily done.
Did the autopilot fail to capture the assumed preset altitude? The KFC225 autopilot (which I have, and which was fitted to Mirages of a certain generation) is not only prone to servo failures (I consider this unlikely in this case because a pitch failure is annunciated within ~ 10 seconds because the pitch trim usually runs away) but it also has some subtle "finger trouble" issues whereby you can dial in a preset altitude and normally this should activate the ALT ARM mode but this doesn't always happen, so you descend at some defined rate but there is no stop... done that a few times and one has to watch it.
The relative lack of damage to the hull suggests that the impact was in level or nearly level flight, not in any kind of steep descent. Unless the pilot pulled up at the last moment.
Anyway, Vienna ATC radar will have the answers to most of these questions, immediately after the crash, but we won't see this data for a very long time. One N-reg crash in France was nearly 2 years ago and still no publication - very very irresponsible.