Shoveling Snow Can Make You Nutty
Shoveling Snow Can Make You Nutty
I have lived for extended amounts of time in five places: North Carolina, Florida, metropolitan Washington D.C., Michigan, and Colorado. Of these places, I have shoveled, by far, the most snow when in D.C. In fact, I have never shoveled snow in Michigan – the cold state. Many times I have stood in more than 18 inches of snow in D.C., but I have never seen more than 8 inches of snow in Ann Arbor. Growing up in Cary, N.C., I saw deeper snow than I have ever seen in Ann Arbor.
One of the things that I learned, while anxiously waiting for snow-day cancellations in North Carolina, is that real snow came from the south, not the north. Almost seems nutty, doesn’t it? But big snowstorms require a lot of water, and if it is to snow a lot in North Carolina or Washington D.C., then the water for the storms has to come from the Gulf of Mexico and the Atlantic Ocean. Such storms always move up the coast from the south and west towards the north and east. The temperatures during these snowstorms are generally in the upper 20s (Fahrenheit). If it is much colder, more like Michigan, the air can’t hold as much water. Washington D.C.-like storms don’t happen very often in Michigan.
My experience is that “cold” and “lots of snow” do not go together with simplicity. “Lots of snow” happens when it’s pretty close to freezing, and the air can hold the largest amount of water that can still fall as frozen puffy water. For snow in the eastern part of the United States, there is a sweet spot in temperature - above freezing it rains, far below freezing it does not snow a lot, and just a little bit below freezing, it’s just right. If you look across most East Coast storms, you see you are on the edge: the warm side of the storm is rain, the cold side is snow, and in between, ice and the dreaded wintry mix.
At the National Climatic Data Center there is a nice mapping tool that shows different measures of snow extremes. I made a map of one-day extremes, that is, the most snow on a single day. My figure is not as clear as if you go play with your own maps, but I think that it shows the point I want to make. Compare southern Michigan and the region around Washington, D.C. The darker purples around D.C. show a region with higher snow extremes. It’s also a warmer region. Looking to the west of D.C. and down the Appalachian mountains to the south, the contrast becomes even larger, with the appearance of the pinks and reds of the highest snow. (Sorry, you have to go to the site to see the color scale.) Of course, there is a little bit of elevation with those mountains that makes the comparison more complicated.
Figure 1: One-day snowfall extremes from National Climate Data Center U.S. Snow Climatology. You have to go to their interactive mapper to see the color scale, but basically oranges and yellows are low. Purples and pinks are high, and blue is in between.
This tried-and-true relation between temperature and moisture and snow helps us understand what the climate models are telling us. This relationship has been true since the Creation so it really shouldn’t seem nutty that we could see more blizzards in a warmer world. But, let’s look more closely.
There is a new paper entitled Controls of Global Snow Under a Changed Climate. The paper is by Sarah Kapnick and Tom Delworth. (Here’s a popular summary of the article. And here’s a good AP report from Seth Borenstein) This paper is a nicely designed set of model simulations investigating snow with 1990 conditions and with doubled carbon dioxide.
The paper takes a global perspective. To be explicit, globally, the total amount of snow is simulated as decreasing. If you look at the continental United States (Figure 2), there is a stark reduction of snow. Looking globally, however, at high mountain peaks, where it remains cold enough to snow, there is an increase of snowfall. This is true even if there is a decrease at lower altitudes in the same mountains. At the lower altitudes the precipitation is falling as rain. Not only is there a reduction of snow, there is also a large reduction of snow-cover area. There is far less snow remaining on the ground as we go into spring. In Greenland, eastern Siberia, and coastal Antarctica, where it is still cold enough to snow, there are increases in snowfall.
Figure 2 shows the geographic distribution of snowfall change in response to carbon dioxide doubling as a percentage snow in the 1990 simulation. The only regions in the United States with snowfall gains are found in Alaska.
As for eastern U.S. blizzards, Kapnick and Delworth leave this for a future study. However, we are already observing the trend emerge of more extreme snowstorms. If I were to be presumptuous, in Figure 2, I would draw a line from Chattanooga, Tennessee to Montpelier, Vermont and predict that along this transition line, the frequency of large snow events will increase. This will be the region of the new sweet spot where it is just cold enough to snow in a warmer climate.
A couple of blogs ago, I took a more regional look at this same problem – that is, where it might snow more in a warmer environment. I showed the example of the area around the Great Lakes. We already observe a clear signal in northern Michigan where it remains cold enough to snow but too warm for the Great Lakes to freeze (Figure 3). Here, and in the anecdotal experience I reported above, we see that coldest and snowiest don’t go together. We also see as the climate warms, changes in snow that are consistent with this experience. That is, where it is warmer there is less snow and less snow cover, and where we are close to freezing and it is moist, there are big snow events. Then the snow melts, because it is warmer. So climate models, like those described in Kapnick and Delworth, are not projecting any fundamental change in the way the atmosphere works. They project the current, well-observed relation of temperature and snow into the future and show a likely scenario of how snow will change. The weather forecast models of the here and now are more and more often predicting record and near-record snowfall. They, too, work from the same relation between temperature and snow. They are a consistent piece of information. It is the consistency between the known behavior, the emerging observed trends, present-day weather forecasts, and projections of conditions in the future that provides us with confidence that we have a substantive vision of the future. Doesn’t seem so nutty after it’s all said and done.
Figure 3: Snow in a warming world: Mean seasonal snowfall (inches) across the Midwest for a) 1961-1990 (left) and b) 1981-2010 (right) periods. Figures courtesy of Midwest Regional Climate Center. We look at the last 30 years compared to the previous 30 years, we see this signal emerging. Decreases in snow and snow amount are marching up from the south (red arrow), and lake effect snow is increasing around the Great Lakes (blue circle).