Is the Western U.S. Snowpack Declining "Dramatically"?

Last week a paper (Mote et al., 2018)  was published in the journal Climate and Atmospheric Science claiming "dramatic" declines in western U.S snowpack (you can access it here).

The article had all kinds of scary details. The loss in water resources would be "comparable in volume to the West’s largest man-made reservoir, Lake Mead" and the losses would be so great that "new reservoirs cannot be built fast enough to offset the loss of snow storage."

The principal author's (Phil Mote) institution put out a press release that amplified the message, with the lead author noting that:

“It is a bigger decline than we had expected,”

The media headlined this "dramatic" loss of western snowpack, with hundreds of stories in major outlets around the world, allowing millions of people to learn about the bad news. Here are a few examples. I could show you a hundred more, but you get the point.

But fortunately, the "dramatic" headlines and all the hype are not correct.  

There has been no "dramatic" loss of western snowpack during the past century, but rather a relatively slow, steady decline.  And as I will demonstrate, the scary paper's own research supports a less apocalyptic interpretation.  As does other research in the peer-reviewed literature.

But let's get our terminology down straight.  What does dramatic mean?  Let's look at a typical definition (Oxford)

Since we are not talking about a thespian document, definition (2) is the one we want:  an event or circumstance that is sudden and striking.  Has the trend of snowpack over the western U.S. during the past decades been sudden and striking?

The answer to this question is really important.   Many politicians and activist organizations are claiming that we have experienced a rapid decline in western snowpack driven by global warming.  And an accurate knowledge of snowpack changes is clearly important for making decisions about water resources. And what about the future of western U.S. snowpack?

The Mote et al. paper uses two approaches to evaluate past snowpack changes over the western U.S.  The first is to examine snowpack changes based on direct measurements.  The problem with that approach is that there are only a limited number of stations and the number and distribution of such stations have changed considerably over time.

The second makes use of a snow/hydrology simulation model called VIC (Variable Infiltration Capacity) model, developed by Professor Dennis Lettenmaier of UCLA (and formerly the UW).  This model uses precipitation and temperature inputs (there is a LOT more of these than snowpack measurements) to simulate the changing snowpack.

Below is a figure from the Mote et al paper showing the snowpack (actually the snow-water-equivalent or SWE) over the western U.S. on April 1 each year from 1915 to 2014 using the VIC model approach.  They also fitted a line to the variation over time.  You will note that there is huge amount of variability year to year, with an apparent slow decline in snowpack over the past century.    Specifically, they found a 21% decline over the past century or 2.1% decline per decade.   Hardly seems dramatic.  I should note that April 1 snowpack is a frequently used measure, since in the west snowpack generally peaks around then, and thus April 1 snowpack is a good measure of the water availability for the upcoming warm season.

Now imagine their line was not there.  In fact, you don't have to imagine, I have done it for you!  There doesn't seem to be any decline during the past few decades...if anything, the snowpack seems to be increasing.

In fact, here is the same figure, with only the last 40 years shown.  No decline, dramatic or otherwise is apparent.  Where did that headline come from?

Now a completely independent analysis of long-term snowpack trends over the Northwest U.S. is found in a peer-reviewed paper in the Journal of Climate (A New Look at Snowpack Trends in the Cascade Mountains by Stoelinga et al...found here).   They used a statistical approach to secure the snowpack from temperature, precipitation, and streamflow instead of the physical model (VIC) mentioned above.  But the same general idea.  Their results for 1930 to 2007 are quite similar to those found in the Mote et al (2018) paper, with a 23% decline for the entire period, and increasing snowpack since 1975.  I repeat, increasing.

Furthermore, they went one step further and tried to remove natural variability (like the Pacific Decadal Oscillation) and got the April 1 snowpack trend shown below. Plenty of variability and a very slow decline (16% over the period).  About a 2% decline per decade...similar to  the Mote et al. VIC results.

Nothing large, nothing sudden, nothing dramatic.  2% loss per decade.  No acceleration of snowpack loss.    And as I will explain late, this make perfect sense considering that the Pacific Ocean is just west of us.

But what about snowpack observations over the West?

As noted by Mote et al., there is a major problem using such observations:  the number of measurement sites is small and their numbers and distributions have changes substantially over the past 50 years.  To illustrate, here is a figure from the supplementary material from the Mote et al paper, showing changes in the number of observations for various subregions.  Few observations before 1940 and a major increase in the 60s and 70s.  The numbers have been relatively stable since roughly 1975-1980.

With those issues noted, below is a plot from Mote et al of the observed April 1 snowpack for three western regions: the Cascades, the Rockies, and  California.  The circles are the average snowpack for each region (ignore the red xs and red line...that is for the VIC model which we already covered).  I removed a blue trend line from these figures--I want you to make your own appraisal of the trends.    Specifically, look at the period since 1980, when the observational network as relatively stable.   

It is clear that there is little April 1 snowpack trend in the observations for the last 35 years.  Yes, 2015 has a very poor snowpack...but that was an isolated outlier....for climate studies we must look at the trend...and there simply has not been much of trend.  Just a lot of variability.

As an independent check on the observed trend in April 1 snowpack, research meteorologist Mark Albright, past WA state climatologist, did his own analysis of the April 1 snowpack changes over Oregon, Washington, Idaho, and Montana using the USDA Snotel observing stations.  He considered the period 1984-2017, since the SNOTEL network expanded into the early 1980s.  A seen below, there is virtually no trend over that period (and I might note that 2018 looks like an above-normal year).

Now to beat a dead horse, here is one more observed record of mountain snowpack, one encompassing a very long period (1879-2017):  the snowpack at Donner Summit, high in the Sierra Nevada (this is from the Central Sierra Nevada site associated with the University of California, Berkeley).  The color bars are April 1 snowpack).   More snowpack in the late 1800s, but only a slightly downtrend during the past several decades.

Now you might ask, why has western snowpack been so stable if the earth is warming?  A very good question.

A major part of the answer is that the eastern Pacific has NOT warmed very much and our temperatures are substantially controlled by the eastern Pacific surface temperatures.  To illustrate the lack of warming, here are the surface air temperatures from the NASA/GISS website, showing the trends from 1977-2015.  The eastern Pacific actually cooled during that period.

This pattern is similar to that indicated in climate models driven by increasing CO2.  The Arctic warms up more than anywhere, land warms more quickly than oceans, and eastern oceans generally warm the slowest.

I know that some of you are unhappy with the above analysis, even though the evidence is pretty compelling that the snowpack has not been dramatically reduced the past few decades.  You have heard the constant drumbeat from the media, some activist groups, and a few scientists who should know better.

But before some start calling me names (e.g., climate contrarian or denier) or the Seattle Stranger does another hit piece, or someone complains to my Dean, let me explain that global warming will have major impacts on snowpack during future decades and especially after 2050.  Increasing CO2 will cause increasing warming during the upcoming century that will reduce snowpack substantially.   Some regional climate runs that my group and Professor Eric Salathe completed a few years ago, show major snowpack reductions (see graphic).

But the loss of snowpack during the snowpack has been modest and slow, and certainly not dramatic. And the fact that it has been going on for over a century suggests that part of it is probably natural and not driven by anthropogenically driven increases in greenhouse gases such as CO2.  The planet experienced a cool period (the Little Ice Age) from the 1600s to the late 1800s, that produced more snow over our region.  With the end of the cool period (probably due to natural causes), snowpack has slowly declined.

Scary headlines and claims of dramatic snowpack loss are counterproductive in many ways.   They are clearly not true and thus undermine the credibility of those claiming such losses (activist scientists, politicians, and advocacy organizations).   They can result in poor public policy and infrastructure planning.   They unnecessarily scare people and make them anxious, an increasing problem (two days ago the Seattle Times had a front page article about a UW Bothell class dealing with dealing with anxiety about climate change).

And then there is the moral/ethical dimension.  Scientists and the media must communicate our best understanding of the truth faithfully and not exaggerate/hype to get people to "do the right thing."   As I have learned personally, there is a real cost to telling "inconvenient truths", but society can only make wise decisions if it is provided with unvarnished information based on the best science, and including information about uncertainty.

Another issue regards the press releases of universities and other research institutions.  There is a tendency to go for dramatic headlines and hype to secure the "currency of the realm" for PR people--lots of clicks and attention.   But the contents of the research papers are often distorted in the process.  This was clearly an issue for the Oregon State University press release regarding the Mote et al paper, and it occurs all the time here at the University of Washington and at other instituions.

Finally, the spread of such hyped material says something about the current state of online and print media.  Apparently, few "reporters" bothered to read the paper they were headlining.  Few completed a reality check on the claims.   But they were attracted to the big "dramatic" headline and were happy put out the excessive claims as a way of getting attention and "clicks."  This is  more than an inconvenient truth, but is a challenge for our democracy, since a misinformed public will not make good decisions.