LB2 is a study sim and as such details the systems to a higher fidelity. For avionics and mission equipment package, LB2 has no equal by default of design. (I'm a systems guy really - comes with my background of software engineering).
The FM still stands up, although it doesn't have all the parameters as AH/CH it does let you do auto-rotations. Engine output is handled much better. CH's handling of forces around different parts of the rotor almost make up for this (that and other effects LB2 doesn't do).
Weather, wind in particualr is modeled fairly well too. Coming in gusts and is localised (different regions have different weather). LB2 just has a simple constant delta-v for wind. Apaches are not as sluggish as presented in LB2, but then the Jane's game doesn't have many cultural objects to avoid. Then again CH has FM issues too, the sidways flight speed and tail authority is underweight. (But potentially correctable).
EECH doesn't have anywhere near the avionic levels of the real choppers (but then LB2 is also a lightweight representation of the 64 delta). But it does a fairly good job of acknowledging the need for SEEING targets you've killed (they remain on your target list until LOS is re-established and swept by sensors to confirm their destruction and removal from said list). In LB2 you hid behind a hill, draw a PFZ around a rapier site, pop 5 missiles and bang-bang-bang all the targets die and vanish from your TSD. That wouldn't happen in CH, there's no way of knowing if you had destroyed the site without looking to confirm it. A small detail yes.
Weapons modelling, again it's a bit of a mixed bag for both sims. LB2 has the edge here I think because gravity plays a bigger role in their balistic path. Also the way EECH is programmed for networking means you can't arbitrarily shoot Kentucky windgage style.
But EECH does try to model armour effectivness to a small degree, it's not always one shot one kill. And proximity damage is modelled which neither are present in LB2.
As far as terrain goes, LB2 is quite dated now. I don't like like muddy textures in sims, I think they look awful; blury and unconvincing in what they are trying to do. Personal tastes aside, I know the actual terrain in Razorworks games comes from real places. Although looking "simmy", those rivers, powerlines, general features and elevations are accurate. LB2 is a height map from a larger region and repainted with the textured features painted on, it's made up (but higher mesh res). See my feature on Terrain In Simulation at SimHQ.
Both AH and LB2 were designed for 640x480. LB2 features a user friendly squashed everything in cockpit, AH is what they actually look like but difficult to use on a monitor.
So, they both do things the other doesn't, which leads into my parting shot on this topic.
It's a bit of mismatch to compare them, a study sim verses a mid-level sim, both of which had very different goals in the minds of the designers. Both of which are very good.
Ultimatley at the end of the day if you're having fun, you are the winner.
Nobody is making any money out of either sim today. Austin Skunkworks is no more, Razorworks was lucky to get CH out given the poor sales of AH and sims in general. Piracy does make a difference when figures are so marginal it seems.
My assertion is:
AH != LB2
LB2 != AH
But I'd also assert:
LB2 == AH == hours of enjoyment
I could go into more details, but that will have to wait.