I am not aware of any specific reference for this. My judgement would tell me that all four sides need checked simultaneously to ensure the structure can handle a drift on any side. There will be leeward and windward drift calculations and you could check a load case which involved a leeward drift on one side and a windward on the other. Myself, I would take the largest of either and put it on all four sides and go.

PE, SE
Eastern United States

"If a builder builds a house for someone, and does not construct it properly, and the house which he built falls in and kills its owner, then that builder shall be put to death!"
~Code of Hammurabi

I think the tendency is for "holes" in the roof where you have a lower roof surrounded on four sides by higher roofs, to fill up with snow.
I get real conservative with situations like this.

I did inspect a low roof once that had high roofs on two sides and nothing on the orthogonal ends (open) - it had high drifting on both sides pretty close to code calculated drifts.

Put a roof over the hole, then the problem goes away, and so do the roof leaks. :)

Mike McCann
MMC Engineering
http://mmcengineering.tripod.com

I saw pictures a few years ago of a scenario as you had described following a Buffalo, NY snowstorm that had dropped 7' overall on the city. The low area was probably on the order of 30' square and had basically filled to the brim with snow. I would agree with mssquared, put a higher roof on to eliminate the problem. If you really wanted daylighting around the perimeter of the inner space, raise it slightly above the other roof. You would still have snow drift issues with a raised roof, but it certainly would not be as problematic as the original layout.

It would be nice to put a higher roof on, however this lower portion is also the drain area for roof rain coming from other portions of the building. I have deciced to take the snow drift from the worst case leeward/windward direction and apply the maxiumum intensity over the entire roof area to be safe.