Thursday, March 13, 2014

Snow Sponge

What happens when rain, and particularly heavy rain, falls upon a mountain snow pack?   Does it rapidly melt the snow?  Contribute to flooding?  Or what?

 It turns out that the answer to this question is quite important, with implications for what will happen to our area rivers under global warming.

Consider what happened during the past week.  The Cascade snow pack was hugely enhanced during late February and then we had an intense, warm rain event with a freezing level way above the Cascade crest. National Weather Service forecast Doug McDonnal created a wonderful graphic showing what happened for the Snoqualmie River drainage last week (see below). 

The precipitation at 9 SNOTEL observing  sites in the drainage (Alpine Meadows, Skookum Creek, Stevens Pass, Olallie Meadows, Mount Gardner, Tinkham Creek, Cougar Mountain, Lynn Lake, and Sawmill Ridge)  is shown by the green line.  There was 12 inches of precipitation (liquid water equivalent) that fell over the drainage with the biggest deluge on March 8.  As shown by the surface temperatures (black squares) nearly all of the precipitation fell as rain).  The depth of the snow pack increased by five inches early on, while the temperatures were cool enough for snow, but stagnated and then started to decrease as the temperatures warmed above freezing. By the end of the period (March 10th), the snow depth was down by 10 inches.

But now the interesting part.  The amount of water content of the snow pack (SWE--snow water equivalent), or to be more specific for this pot, the change in SWE from the initial time, is shown by the purple line. And here you see something that might be surprising:  the Cascade snow pack, hit HARD by torrential warm rain on March 8th, had  an increase in water content.  It did not melt off rapidly.  That snow pack was amazingly robust.

Now something just as interesting:  take a look at the flow in the Snoqualmie River at Snoqualmie Falls (red line, vertical scale at right). Roughly 8 inches of precipitation (mainly rain) drove the river up to roughly 17,000 cubic feet per second (cfs) by March 6th.  The rain pretty much stopped for two days, resulting in the river flow plummeting, followed by a period of intense rainfall  (and very warm temperatures) on the 8th (atmospheric river or pineapple express), which drove the river level to 29,000 cfs and associated flooding. The snow pack hardly changed with the onslaught.   How could this be?  Wouldn't warm rain greatly lessen the snow pack.  The answer:   not necessarily.

It turns out that mountain snow packs have much in common with a sponge: both have some capacity to store liquid water.  According to my colleague Professor Dennis Lettenmaier, a well-known hydrology expert, liquid water can be as much as 6% of the volume of a snow pack.  Thus, with the snow depths at the Snotel sites for the Snoqualmie Valley drainage varying from 60 to 130 inches before the rain, the snow pack was probably capable of holding 3-8 inches of liquid water.  More rain than that and precipitation would be running out the snow pack on the bottom and then into streams and rivers.  Some melt would be going on, but considering that this was a cold rain at such altitudes, the melting was minimal.

So let's think about what the plot above is showing us. For the first day there was snow at higher elevations and the snow water equivalent (SWE) at the above stations increased, while the river flow remained constant.   Then temperatures warmed and rain started falling on the snow at both high and low elevations.  Initially, the snow soaked up the rain and the snow pack (SWE) increased (March 2-4).  With subsequent warming temperatures and heavier rain, the snow pack had become saturated and the water flowed through the snow pack into the stream/river system.  Rain also fell on the lower portions of the river basin without snow.  Both contributions led to increased flow in the river.   With warm temperatures and little rain, melting snow led to a slight drop in SWE over March 6 and 7th.  Then the atmospheric river hit on March 8, with heavier rain and warming.  The saturated snow pack soaked up a bit (SWE went up slightly), but rapidly could not take any more water and large amounts of water ran through the snow layer into the river system. The snow depth lessened but the water content of the snow pack only dropped a small amount.

If cold weather returns after a rain event, some or all of the water can freeze in the snow pack, contributing to our favorite "cascade concrete."

The moral of the story is that snow can help buffer moderate to heavy rain for a while--until it becomes saturated. Thus, snow can lessen the pulse of precipitation hitting local rivers, at least temporarily.

Regional climate simulations for later in the century suggest there will be less snow pack and heavier precipitation in November.  With less snow buffering the rain and more intense rain, the potential for more serious flooding exists.


  1. Is there a difference in the melting rate of snow packs with different SWE? Does a drier snow pack melt more slowly because of the entrained air insulating the snow, or does the wetter pack melt more slowly because of the greater mass and specific heat capacity? Or does it all even out?

  2. I'm not sure what the generally accepted definition of cascade concrete is, but as a skiier, I tend to think of it as snow my skis penetrate into, but then won't allow me to turn easily... similar to concrete when you're mixing it (NOT when its set). This type of snow doesn't present itself when a very wet snow pack freezes - instead "cascade concrete" in my mind happens when there is a heavy dumping of wet snow, or when temperatures spike up after low density snow is on the ground - this gloppy soupy mixture is cascade concrete. When we get lots of rain, and the snow pack soaks it all up, then it freezes, it's actually quite good for skiing, since once the top portion melts during the day, it's as if everything has been groomed - even in the back country! Spring skiing! Heavenly stuff.

  3. There are competing models for this sort of liquid/solid portion of the phase diagram, analogous (in part) with the percolation models of water/soil. One model (similar perhaps to the "sponge" model here) says that it would take days for water to percolate through soil, and yet a flood water pulse shows up in streams within hours of heavy rainfall. Provision should be made for saturation in both soils, and solid water, for concerted passage. That is, the situation where one drop of water in, forces another drop out, hence essentially zero delay. There remains lots of room for folks to study and improve current, somewhat over-simple, models of surface water flow.

  4. Based on the difference between the purple and green lines, by my calcs there were 13 inches of precip of which approximately 1 1/2 inches stayed in the snowpack (on a net basis). That means 11 1/2 inches of precip going into the Snoqualmie. Wow!

  5. Some old-time SNOTEL guys have a rule of thumb. If there is more than a meter of snow on the ground, it acts as a sponge to rain, SWE increases, and flooding is lessened. If there is less than a meter of snow on the ground, snow will melt, SWE will decrease, and flooding will be more intense.

  6. Great post! I think it would be interesting to add a snowpack energetics component to this explanation. I believe that to produce melt the snowpack must also be isothermal at 0 degrees C. It seems that a large mountain snowpack could buffer the downstream effects of a atmospheric river (AR), but the AR could also significantly reduce the cold content of the snowpack, making it easier to melt later on in the season.

  7. Interesting research about snow's capacity to act like a sponge and for rain's contribution to increasing SWE's. Less snow on the ground accompanied by heavy rain, more flooding vs. a high snowpack with the same heavy rain.

    One thing...Future computer modeling for later in this century should show a significantly higher margin of error - there may be little change or even a decrease in temperature - its hard to tell. The warming related to the sun will have the most effect on our terrestrial temperatures as it always has. Modeling correlated with CO2 conc. will not produce a reliable result for temperatures - as the relationship between CO2 and terrestrial temperatures is not statistically significant. Think back to why the warmer global temperatures during the period 1200-1400 when CO2 concentrations were considerably lower than now.

  8. Fascinating entry and discussion. I watch the river flow data carefully and signed up for a USGS email alert at higher levels. This led me to drive up to the Middle Fork valley to take pictures at near peak flow on March 8. This exact question was in my head as I wondered what was happening to the snow pack in the mountains.