With the oppressive heat striking across the world, I thought I might provide a little relief to readers plagued by high temperatures by discussing something they might welcome right about now: snow.
Snow is crucially important in many regions because it’s a source of water, and let’s face it, water is important for life. When snow accumulates during winter, it builds up a reservoir of water which is then released to the environment during snowmelt. This is critical, especially in areas like the western U.S. where snowpack forms a huge part of their water supply. Snowmelt feeds rivers and lakes, and ultimately humans, and does so at the right time to sustain a healthy ecosystem and healthy agriculture.
When it comes to snowmelt, the timing is right because life has evolved to make the best of existing conditions. Changes to the amount of water stored in the snowpack, and to the time of year at which it’s released, can have profound effects — effects which are not generally good. They’re bad.
What has been happening to the snowpack in the western U.S.? I retrieved data from SNOTEL, which monitors the amount of water stored in the snowpack on a daily basis at locations throughout the region. Since snowpack is so important to supplying the Colorado River, and the Colorado River is a crucial water supply for so much of the U.S. west, I decided to start with SNOTEL stations in Colorado. There are 115 of them.
Here’s typical data, for the first Colorado station (in alphabetical order), Apishapa:
The x-axis is time (in years), the y-axis is snow water equivalent (SWE) in inches. That’s the depth of water you’d have if the entire snowpack melted instantly. It’s a true measure of how much water is stored in the snowpack, much better than snow depth which varies a lot depending on whether the snow is densely packed or light and fluffy.
Clearly the SWE changes a lot, and just as clearly there’s a buildup every year during the snow season and a melt-out every year as summer approaches. There are hints from the above graph that the total amount has been declining over time, but without some genuine analysis we can’t have confidence in that conclusion. We also can’t tell, just by looking, whether or not the timing of snowmelt has been changing.
To gauge the timing of snowmelt, I identified the day on which the main snowpack (the big continuous bulge each year) ended. I call this the “snow-out day.” As an example, here’s a close-up on the data since 2010 with a red dot marking when the main outburst declines to zero:
To gauge the size of the snowpack, I recorded the maximum value of SWE for each snow season. It’s not a perfect measure, but is surely indicative of changes in the amount of water stored in a given year’s snowpack.
Then I looked for trends in those variables, estimating the linear trend in two ways. I used least-squares regression, and I also used a non-parametric method, the Theil-Sen slope estimate. Both provide an estimate of the rate of change, as well as a “p-value,” as estimate of the statistical significance of the hypothesis that the rate of change isn’t zero. Values less than 0.05 indicate less than 5% change the result is due to random variation when the true rate is actually zero.
Here’s the result for snow-out day:
The red line is from least-squares regression, the blue line from Theil-Sen estimation. At the bottom is a summary giving the rates and p-values, first for least-squares regression, then for Theil-Sen.
Both tests return a statistically significant result; the snowpack has been melting out earlier at Apishapa, Colorado, probably more than 0.8 days per year (0.80 according to least squares, 0.87 according to Theil-Sen). The net change over the period of record indicates that snow-out at this location is happening just about a month earlier than it was in the early 1980s.
Here’s the result for maximum SWE:
Again both tests suggest decline, but neither reaches “statistical significance” (at 95% confidence, a p-value of 0.05 or less). Least squares comes close, with p-value 0.07 for 93% confidence, but I like to use the “de facto” standard of 95% confidence because I know how easy it is to draw mistaken conclusions when our statistical tests are too lax.
Of course that’s just one station in Colorado. I did similar calcuations for each station with at least 20 years data, of which there are 80. Here are the estimated rates for each, for the day of snow-out (on the left) and the maximum SWE (on the right):
Dots mark the estimated rates, and the “whiskers” extend to the 95% confidence intervals. Those which reach statistical significance (at 95% confidence) are shown in red.
If the variations were just random, we would expect four stations to show statistical significance at 95% confidence for each variable, about two being low and two high, just by random accident. But 15 stations show statistically significant rates of snow-out day, all of them trending earlier, and 15 reached significance for maximum SWE, all of them trending lower. This is not an accident. It’s the result of climate change. Man-made climate change.
This has consequences. It reduces the available water supply in those states which depend on the Colorado River for part of their water supply (and there are many such states). It also has other environmental consequences, of which I’m not aware, but I suspect that the increase in wildfire in the west is one of those. These effects are bad for human society. For us.
There are other states with SNOTEL sites, but at the moment their website doesn’t seem to be functioning to supply the data (I expect the problem to be fixed soon). Other states represent different regions, and as we’ve seen global warming hasn’t affected them all in the same way. And hotter temperatures don’t necessarily mean less snow; hotter air can hold more water vapor, enabling heavier snowfalls. The temperature-snowfall relationship is complicated, although the temperature-snowmelt relationship is rather direct.
In any case, it’s yet another example of the fact that the “ideal” climate for Earth is: stable. When climate changes, when rainfall and drought and temperature and storminess and even snowpack changes, living things have to adapt to new situations. We (and other life forms) have adapted to conditions as they were. We’re already having trouble from the way things are. The worst is yet to come as we try to survive the way things will be.
This blog is made possible by readers like you; join others by donating at My Wee Dragon.