Adding to what Mappy and PSU already said, if you check out the 3k NAM at the onset of precip you can get some clues there. The peaks along the Blue Ridge for example are only a couple degrees colder than the valley (Staunton/Harrisonburg area), but that's enough to have a large difference. Ice will barely accrete right at 32 degrees, you get closer to the upper 20s and it's a very different story. Later on when the warm nose does finally push through at ~925mb, the ridges and valley all switch over to rain around the same time.
TL;DR: In this specific setup the high elevations accrete more ice because of conditions at the start, not because they hold onto the cold air longer.
Thats also a large part of why these marginal events are so hard to forecast. Small temp swings can make for large changes in precip type/duration.