MIKE MOSS SAYS:      Michael,      Forecasting snowfall amounts is nearly always difficult to do accurately, but the basic process is to first assess the amount of liquid equivalent precipitation a system is likely to produce (quite a challenge in itself in many cases, this is done by estimating the rate of precipitation that may occur along with its expected duration). Once you have a reasonable projection of how much precipitation will occur, the next step is trying to figure out how that will be partitioned by precipitation type. In our part of North Carolina, we are often in a "borderline" region where the precip type changes during the event, often involving more than two types (snow, sleet, freezing rain, and/or rain) and sometimes back and forth between different types multiple times. A mixture of multiple precipitation types will result in lower snowfall accumulations than a purely snow event, since freezing rain accumulates at about the same rate as rain, sleet accumulates at a rate that is greater than the rain equivalent (around 2-3 inches of sleet per inch of water when melted) but usually notably less than snow, and snow can cover a very wide range of ratios relative to water depending on the vertical profile of temperature and humidity in which it is produced.

Even for a rare snow-only event here, the combination of temperatures and moisture content of the air as you move upward through the atmosphere can make a big difference, and it is possible for snow to fall that is quite wet and would only amount to about 4-5 inches of snow per inch of liquid water, while with unusually cold profiles for our part of the country the shape of the snow crystals and the way that they interface on the ground can trap a lot of airspaces and result in snow depth that is 20-25 times that of the liquid water equivalent. So, part of the overall process is to examine forecast profiles of temperature and humidity as they are expected to evolve through the event, and assess what the snow to liquid ratios will be. Applying those ratios to the amount of liquid precipitation, and attempting to account for any expected periods of rain or sleet that will be mixed with the snow, and throwing in some consideration of what ground temperatures are likely to be, we can come up with a range of potential snowfall amounts.

Of course, that's only part of the story. The other tough part is to get the locations reasonably correct. A relatively small (in the overall scheme of things) deviation in the track of a low or the position of a front or the speed or intensity of an upper level jet streak, for example, can shift the heaviest precipitation, which often occurs in fairly narrow streaks or bands, by tens of miles and/or several hours, with major implications for the accuracy of a forecast for a specific location.