Snow depth and density: What’s in a foot of snow?

Margaret LeMone  |  5 November 2009  •  Every snowfall is different, including how much water is packed into the flakes and how that changes over the life of a storm.  This can make it very hard to figure out how much snow “really” falls in a given storm.  A rule of thumb is that 10 inches of snow equals about 1 inch of water, but this number can vary greatly—as you can investigate yourself the next time it snows.

Snowfall in south Boulder, 10.29.09The October 2009 snowstorm as seen from south Boulder early on the morning of October 29. (Photo by Matt Kelsch, UCAR/COMET.)

A snowstorm that hit Colorado on October 27–29 left Boulder with an official 23 inches (58 centimeters).  Snow measurements collected through the NWS Cooperative Observer Program are taken on an exposed board at roughly six-hour intervals; after each reading, the board is swept off and a new grand total is computed.  This total is typically larger than the actual depth on the ground, for several reasons:

  • Snow may be melting from below on warm ground.
  • Temperatures could be near or above freezing, so the snowpack may be melting even as snow falls.
  • As the snow piles up, its weight may compress the snow.

How can we unravel this confusion for the Colorado storm?  On 28 October at 10:00 p.m. MDT, I measured 17 inches of snow in my backyard.  On the 29th at 7 a.m., I measured 6 inches of new snow on the deck (wood).  But I only measured 16 inches total snow depth on the lawn.   Subsequent measurements saw the snow depth continuing to decrease (Figure 1).  What was going on?

The temperature for the entire period before late morning on 30 October (~hour 58 on Figure 1) was below freezing.  Nevertheless, the ground was warm before the storm, so there was some melting from below.  However, compression of the snow was probably the major factor in the short-term change between hours 22 and 31.

If the snow compresses, the snow at the bottom should be more closely packed than the snow at the top, since there would be more snow pushing down.  Also, there is more time for the earlier snow to settle. To see whether the snow at the bottom was denser, I did a qualitative measurement of the snow density on 28 October.

Figure 1: Snow depth as function of timeFigure 1. Snow depth as a function of time, based on averages of measurements in open areas around our house.

First, I scraped the snow so that I had a vertical slab with sides straight up and down.  Next, I took two glasses, and punched them through the snow horizontally to collect a sample at the two heights, starting at the top (qualitatively shown by the two circles in Figure 2).   Since the procedure and the glasses are identical, the volume of snow sampled should also be close to identical.   Then, I melted the snow, with a cover on the glasses to prevent evaporation.  Once the snow was melted, I put a drop of food color in the water and photographed the two glasses (Figure 3).


Figure 2: Slab of snow where glasses were insertedFigure 2.  Slab of snow showing where I inserted the glasses.

Figure 3: Melt water Figure 3.  The melt water from both samples.  Glass at left = top sample; glass at right = bottom sample.

From Figure 3, the bottom sample has more than twice the volume of water as the top sample:  The density (mass divided by volume) of the snow clearly is higher as you go down. 

So we have two pieces of evidence that the snow likely compressed:  the incredible shrinking snow (Figure 2), and the differences between the density between the top and bottom sample (Figure 3).

Does this conclusively mean that the snow was compressing toward the bottom?  This is what we expected, because there was more snow pushing down nine inches below the top of the snow than five inches below the top.  As noted previously, though, we could blame it all on compression only if we knew there was no melting and if the snow accumulated with a uniform density when it first fell. But as pointed out by UCAR’s Matt Kelsch, who was taking the cooperative measurements for Boulder:

The first night of snow had big wet flakes followed by graupel. The snow-to-liquid ratio was a  very dense 6:1. The snow transformed into much fluffier dendrites on Wednesday with a snow-to-liquid ratio of 20:1 and then a 30:1 ratio Wednesday night.

Though the precipitation succession may have been slightly different at our house, on the other side of town, the lesson is clear:  density stratification in the snow column is also related to the history of the snowfall.

To allow for this complication, a better approach to my experiment would have been to measure the snow density profile on the night of the 28th, and again at the same height above ground on the morning of the 29th.  This would help narrow down any real change in the layer of snow located at a given height.  It would also be important to measure the total water content, to estimate the amount of snow loss due to melting.  Ideally, we would carry out the experiment on a day when the air and ground were cold enough to keep melting minimal.  We would also want winds to be as light as possible, since high winds could blow snow away from one area and deposit it in another.

The author Gertrude Stein once said, “A rose is a rose is a rose.”  Clearly, you can’t say this about snow!  We eagerly await the next storm to take a more careful look at how snow depth and density change with time.  Why not try an experiment like this in your own yard?  If you do, let us know what you come up with (write to me at lemone@ucar.edu).

Thanks to Matt Kelsch and Bob Henson for their helpful comments.