That event was my senior thesis. Some of the wrf simulation actually did an excellent job in mainly showing snow.
Wrf simulations maximized precip over BUF at the observed lake temperature. Lake temperature decreased modeled precip overall. Lake temperature increased resulted in the band concentrating more precip over the lake but less at BUF
The extremely deep mixed layer (lake induced equilibrium level in excess of 500mb) allowed the sensible heat of the lake to only heat up the surface temperature by 1 to 1.5 degree Celsius as the sensible heat gets mixed upward through the mixed layer. A more shallow cold air mass would have been more modified by the sensible lake heat.
The land breeze circulation aided in drawing low level cooler and drier air from northern pa/ne ohio/sw ny underneath the band. This air wasn't as modified by the lakes and wet bulb and ice bulb temperature were fairly close to freezing.