SLFinAZ,
We don't actually know yet that they were on a type of approach that actually had a glideslope. They may have been on an NDB or twin NDB approach - for which the minima is typically around 3-500ft AGL, depending on terrain etc - or the Russian equivalent of a PAR or GCA approach - the radar-based "talkdown." (A PAR does have a glidelsope; the pilot follows it by reference to controller instructions. It is not displayed in the flight deck.)
If they were on a PAR approach, for example, and it was calibrated and working correctly, the controller would have been telling them for quite some time prior to the collision to climb.
If they were on a twin NDB approach..well, I would hope there would be no pilot in the world who would deliberately go so far below the published minima on purpose "for a look", as indicated above by captplaystation. Maybe sneak a hundred feet lower, but not 4 or 5 hundred.
Another possibility was that they were in clear air above the fog bank and attempted to visually navigate to the runway, underneath it. That sort of approach is generally unsafe, considered bad airmanship, and illegal (usually) even in a light, slow, nimble aircraft like, say, a 172. It would be highly unlikely anyone would attempt it, regardless of outside or internal pressure, in anything at all, and especially not a jet airliner. There's no future in attempting to navigate visually, at low level, when the visibility is 500M. Or even 1000M.
There are indications that there were communication difficulties between the crew and the controller/s. Especially in (readbacks?) of numbers. There was a suggestion in a television broadcast from a senior ranking Russian officer that the controller had been telling the pilot to "level out" at some stage during the approach.
I would be very interested to know what altimeter setting the crew used, whether it was QNH or QFE based, and whether the crew knew which it was; the approach in use, the charts used for that approach - and especially what reference was used for altitudes or heights on those charts; any cross-checking of height against the radio altimeter, and any discussions on the landing minima and missed approach procedure.
I believe that it's likely the crux of the reason for descent below the minimum will be associated with these factors. It seems likely the crew did not realize there was anything wrong until (possibly) they actually hit the trees or maybe a second or two prior. I would not be at all surprised if a QNH reference was selected on the altimeter sub-scale, but the crew inadvertently flew the approach using QFE reference. Or (very simply) the wrong QFE reference was set, due to language differences. It would have to be about 15- 20 Hpa out. Maybe more. (The difference between QFE and QNH for this aerodrome is approx 29Hpa.)
I would imagine the reason the approach was commenced in the first place under the circumstances would be to satisfy themselves and (especially) their passengers that at least they'd had a look, expecting it to be unsuccessful. Some would consider that a reasonable course of action provided the approach and missed approach was briefed for and the correct altitudes displayed and used.
All this talk about cutting down trees on the approach, or a NDB/beacon at a possibly non-standard distance from the aerodrome, or what blind-landing aids were or were not fitted, is red herrings, I'm afraid.