Answer: you melt it. Just take what falls in the rain gauge, melt it, and see how much water there is. Simple procedure, but it does require human intervention. Unfortunately, as we move closer to a fully automated observing network, we lose the ability to determine the amount of water associated with snowfall.
The automated techniques don't work too well for measuring snow in its raw form, either. The standard technique is to use a level surface, stick a ruler in the snow and see how deep it is, and then brush off the surface to allow the next batch of snow to accumulate. Try that with a cheap, automated device, and then have it try to distinguish between snow that fell and snow that drifted.
The total depth of snow added to the ground is rarely as large as the amount of snow that falls. Several things happen. First, as snow first hits the ground, it might melt. Eventually, if it snows hard enough, the snow starts "sticking". Second, as snow gets added to the snowpack it compresses the snow beneath it, so six inches plus six inches might only equal ten inches. Finally, if some rain is mixed in it might melt some of the snow, which then might re-freeze into a hard crust.
All of these processes suggest that as snow lies on the ground, it becomes denser and denser. But how much water is in fresh snow? A good average number is that about a foot of snow is equivalent to one inch of water. But this number is highly variable, and in particular is strongly dependent on temperature. If the temperature is above freezing, an inch of liquid might correspond to six inches of snow. This is called wet snow. If the temperature is around 30 degrees, ten inches of snow might equal an inch of rain. And if the temperature is only 10 degrees Fahrenheit, it might take 18 to 24 inches of snow to equal one inch of liquid water. This is called dry snow, or powder snow.
Snow forecasts are made by first forecasting the amount of equivalent liquid water that will fall. The numerical weather prediction models give you precipitation in liquid form, in hundredths of inches. Next, the forecaster takes the forecasted precipitation amount, decides how much of it will melt before sticking, converts the rest to an equivalent snow depth based on the temperature at which the snow will be accumulating, and comes out with the snow forecast.
Try it yourself. Use the precipitation forecasts and temperature forecasts contained in this FOUS bulletin from Bangor, Maine on 12Z Nov. 14, 1997 to predict the total accumulated snowfall due to the upcoming storm:
OUTPUT FROM ETA 12Z NOV 14 97 TTPTTR1R2R3 VVVLI PSDDFF HHT1T3T5 BGR//624767 -2324 170313 39949697 06000828064 00822 120413 40979695 12011898851 04423 080419 41999696 18018918848 03020 010320 38989695 24008928840 00319 000120 36989792 30004918634 01018 023616 33009790 36001898022 00216 053614 30009790 42000897727 00813 030112 26999589 48003917724 00511 050111 22999488
How much snow actually fell? We don't know; it's an automated station.
Technical: E-mail John Fulton < email@example.com >
Scientific: E-mail John Nielsen-Gammon < firstname.lastname@example.org >
Copyright © 1996-2003 Texas A&M University, Texas A&M Atmospheric Sciences Department and Dr. John Nielsen-Gammon. All rights reserved.