During the recent discussion of ice albedo and how strongly the warming influence of northern hemisphere sea ice loss outweighs the cooling influence of southern hemisphere sea ice gain, it was mentioned by several readers that snow loss in the northern hemisphere is also a major warming influence, a potent amplifying feedback of global warming.

The Rutgers Univ. Global Snow Lab has data on snow cover during the satellite era. Like most climate-related variables, snow cover shows a strong seasonal cycle, with more snow in winter and less in summer. When looking for trends, it’s useful to compute anomaly, the difference between a given month’s value and the average for the same month throughout the data record. Here’s the anomaly data for northern hemisphere snow cover:

There is a clear, and statistically significant, overall decline in snow cover. In spite of what we often hear from the fake skeptics — who love to bellow about any big snowfall (even when it hasn’t happened yet) as though it were disproof of global warming — the actual trend in snow cover is one of decline.

More important, the declining trend is strongest when it really counts — when incoming sunlight is strongest during summer. Snow is very highly reflective, and when present tends to reflect much of the incoming solar energy back to space, which has a cooling influence on the climate. As snow cover declines, so does this cooling influence.

In fact the downward trend in snow cover is strongest during the month of June, when solar input is also strongest:

The trend accounts for a net loss of over 5 million km^2 June snow cover since 1979. That’s considerably larger than the loss of Arctic sea ice over the same time span.

In fact the snow loss has been greatest during all the months when it counts most, i.e., when solar input is strongest:

The graph shows the net snow cover change due to trend since 1979. For those (like me) who prefer a time series plot, here it is:

A few months (most notably December) have shown snow cover increase, albeit not statistically significant. But the increases have all been in winter when solar input is weakest so in terms of climate feedback, it counts least. There’s no doubt that the albedo feedback due to snow cover change has been sizeable, and has tended to warm the planet.

It has tended to warm the Arctic more than any other region of the planet. Which is probably one of the reasons that the Arctic is warming so much faster than the globe as a whole.

The cryosphere — the frozen parts of earth — is the “canary in a coal mine” of earth’s climate. Every part of it, snow cover, the sea ice, the great ice sheets in Greenland and Antartica, the mountain glaciers, ice cover on lakes and rivers, even the permafrost, is sending a message loud and clear. Our entire planet is changing. Heed its warning.

54 responses to “Snow

  1. Pete Van de Water

    Is not the increase in the winter months consistent with an overall warming of the atmosphere which then can hold more water vapor?

    • David B. Benson

      I think so.

    • If it were a measure of total snow, then maybe that would be right, though it’s complicated.

      However, this is a measure of snow cover – ie the area covered with snow – and says nothing about depth – ie the total amount of snow. As a first guess I would have expected snow cover in the winter months to decline, as the snow line moved north due to generally milder winters.

      So it is surprising to me, I have to acknowledge that. The questions I would ask are:

      1. Where has the increase happened? Is it pushing the snow line to the south?
      2. Is this anything to do with the hypothesis that reduced Arctic sea-ice could lead to colder European winters?

      Also, one has to bear in mind that the increase is not statistically significant. It may be that a decrease will emerge from the climate variability noise over the next decade or so.

      • “However, this is a measure of snow cover – ie the area covered with snow – and says nothing about depth – ie the total amount of snow.”

        As a (former) inhabitant of snow country, I think that’s a little too simple. Thickness does translate to snow cover over time, since (once melting starts) a thick pack takes much longer to melt than a thin one.

        (A tendency exacerbated, I think, by an albedo-mediated non-linearity: once melt uncovers a dark object, it tends to create a tiny, relatively warm ‘nano-climate’ which in turn widens the melt area. For example, a half-inch of fresh snow over a grassy lawn, followed by a sunny (but not too warm) morning can produce a blanket of snow perforated at close intervals by the tallest grass stems–each of which is encased in a tiny ‘melt cylinder’ surrounding it like a jacket, but not touching it.)

        So thicker snow will produce prolonged melt times where it exists, and these locations will show up as increased extents during the melt period. Having picked that nit, though, it’s hard to see how it could account for an increase in December extents; that’s the *start* of the snow season in many areas, and not part of the melting season in any.

        But perhaps we can elaborate Pete’s concept a bit: what if the increased precipitation early in the season (combined, perhaps, with higher temperatures in really northerly locations, which would augment the ‘water vapor feedback’ we are talking about) helped the snow cover reach depths that got it past that non-linearity I mentioned? Four-inch cover will definitely survive a warm spell much better than a one-inch cover.

        Thinking out loud, as it were…

    • There is that, and another effect – essentially, the warming of the Arctic has the effect of reducing the temperature gradient between Arctic and mid latitudes; as I understand it, this allows for greater air movement and exchange between North and mid latitudes. Hence the possibility of more cold snaps. We’ve certainly noticed that in the UK in recent winters.

      (Google ‘Rossby Waves’ for a more coherent explanation).

  2. faustusnotes

    Is it possible we’ll see a negative feedback here, with the increased arctic ice melt leading to greater snow falls in the Northern Hemisphere, and consequent (partial) recovery of snow cover?

  3. Pete, yup.

  4. Looking at the snow cover trend by month plot, I expect to see a big WUWT post on increases in snow cover in the month of December.

    • Groundhog Day!!!
      Me –
      Brian Dodge says:
      February 10, 2010 at 4:47 pm

      I downloaded the North American Rutgers data of monthly snow cover from 1967 to 2009, and normalized the data to the monthly averages and then calculated the deviations. (Normalizing the data supresses noise caused by small relative changes in larger absolute snowfall cover numbers – e.g, if I lose 8 kg, its only 10 percent of my body weight, but if a sick 32 kg child loses 8 kg, that’s a serious change). A plot is at [imagenerd is defunct – plot is now at

      Reply –
      Steve Goddard says:
      February 10, 2010 at 5:50 pm

      Brian Dodge,

      I’m guessing that you aren’t located in Washington D.C.

      When I tried to point out that one swallow doesn’t make a summer, and that one heavy snowfall isn’t (another) final nail in the coffin of global warming, I got a “You are not authorized to access this server” error message from the Site That Doesn’t Ban Commentary.

      • I didn’t have much luck trying to comment, ca. September 25, on the August 11 (I think) post about how the Great Arctic Cyclone was going to put an early end to the melt season. But *surely* it’s just a matter of it being an old thread; Brutus is, after all, an honorable man.

  5. When I examined the snow cover data a few weeks back, there were two points that tumbled out that go unmentioned here.
    The first is rather minor. Prior to the satellite era, the Rutgers Uni data gives some rather strange-looking numbers for Greenland snow cover.See this SekeRob graph. If Greenland is excluded from an analysis, statements about ‘record snow anomalies’ can be put a little more robustly.

    The second point is that the rate of loss of summer snow is surely accelerating. To make the point more visual, I (very roughly) framed the retreat North of June’s snows in terms of miles. Since the 1970s, June’s snowcover has retreated 500 miles, an average of 18 miles per year. But the average since the mid-2000s is double than (38 miles per year) with this year’s June melt a whopping 71 miles further North than the previous ‘record’ set in 2011. The anomaly graph I came out with is here (usually 2 clicks to ‘download your attachment’)

  6. Pete Van de Water, I don’t see why it should be.

    If it’s below freezing and you get precipitation, you get snow. Assuming all places get precip at least once a month, then you’d still expect snow area to shrink in winter if you just had a uniform warming. Especially since we’ve seen more warming in winter than summer.

    You would, however, expect thicker snow. Perhaps it’s a case of thicker snow not melting out during daytime which would explain why the area hasn’t shrunk in December. I suspect that the weather has something to do with it tbh. If circulation patterns mean that you get more precip on land versus ocean, then you could theoretically have the same amount of snow but greater coverage because if it snows on the ocean you don’t get snow cover…

    This is just speculation though.

  7. The bump in late 80s – early 90s… why? development of unleaded petrol and low-sulfur diesel?

    • jyyh,
      The “bump” was directly due to a change in the trend of snow cover in the late winter months (Jan into March). Up to that point, that part of the year had been on average losing snow cover but since the late 1980s has gone into reverse & been gaining snow cover. It was only after 2000 that the annual reductions in late spring/early summer snowcover grew big enough to exceed the combined increases in winter & autumn snowcover.

      • thanks, not a chemical effect on snow crystal formation in the clouds then since it’s ‘only’ a seasonal effect…

    • Lewis Cleverdon

      There is a further underlying component to the snow-cover ‘bump’ that radically changes the prognosis both for snow cover and ice cap density. It is most readily seen in Alguo Dai’s series of NCAR graphics of the global PDSI progression 1950 to 2099, but I’ve yet to track the link. Here’s one for a 2000 to 2099 series:-
      The images need enlarging for clarity, and to be seen in the context that the US dust-bowl droughts were mostly at 3 to 4, occasionally spiking to 6, on the PDSI index [Palmer Drought Severity Index].

      They depict what is actually a generally northward migration of gradually increasing global precipitation which, with ongoing warming, is increasingly delivering unprecedented rains on high latitudes. As I understand it the notable bump in snow cover was the point where that increase was a/. still falling as snow, and b/. was not yet being offset by greater snowmelt due to warming in summer.

      The consequences of that migration are becoming discernable in many fields: albedo loss, permafrost melt, forest combustion, tropical & temperate soil desiccation and, by putting rainwater onto the GIC and into the drainage canals now feeding its moulins, in the rate of its mass-loss and resulting SLR. Given that the migration directly powers the acceleration of at least four and arguably five of the seven interactive mega-feedbacks, it surely warrants a title acknowledging its significance. I’d suggest “Mother of Feedbacks” would be about right.



  8. Excellent timely post Tamino!
    Current surface temperatures are out of whack warm in the Arctic:
    Making the subarctic regions quite warmer as well.
    sst’s are likewise very warm especially NE Atlantic:

    NE Pacific as well. Precluding ENSO, SST’s are for the largest part much warmer where the (sub) Arctic lands without snow cover Eastward atmospheres hits the sea.

    All this without a large sunny Arctic summer, Mostly cloudy, still ongoing mostly cloudy. Now the clouds keep winter at Bay and sea water more open. This article graphs show a trend which has been slowly but gradually happening. Reaching this point is not a surprise, nor a “cycle”.

    I use to remember people criticizing me because they thought Arctic ice didn’t play any much of a role about Southern weather. Well, now for those living in temperate zones enjoy perhaps the warmest fall in history.

  9. With more open water in a seasonally ice-free Arctic, would one not perhaps expect increased moisture in the atmosphere, increased storm activity, and perhaps increased rain and snow? In a complex system which has positive and negative feedbacks, the non-negligible possibility exists, that the increased snow fall at the right time may increase the speed at which ice may grow back. This is uncertain, though, as it the opposite may happens also: if all the new moisture is dumped into a warm ocean where it melts and stormier waters deepen the surface mixed layer which then mixes the melt vertically to reduce the vertical stratification, which then allows entrainment of more warmer water mixing it towards the surface, we may speed up the disappearance the ice rather than growing it back.

    So, all I am saying, there is lots of unsettled science, physics, and feedbacks that are settled. Not all feedback have to point in the same direction.

    • I was suprised by this finding of a feedback pointing in a direction which I was not expecting: “The study also points to an unexpected relationship between snow cover and sea ice loss. The more snow on top of the ice at the beginning of the sea ice melt season, the greater potential there is for melt ponds to form, and thereby speed up sea ice loss, the study found.”

      • Oh, absolutely. Same physics when you place dust or a tiny piece of dark rock onto a glacier. It makes a hole as the dark rock or dust absorbs more heat in summer the same way that melt-ponds with their higher albedo. Once they drain in summer, the sudden release of fresh-water stratifies the very surface water which then traps more heat at the surface (takes more work by the wind to mix it down against the ocean density gradient) which then leads to fast melting of what’s left of the drained pond.

        Now, if thin ice in winter is loaded with lots of snow, the snow can “sink” the ice which then becomes flooded, forms a slushy mixture between ice and snow, which can freeze making the ice thicker. This is what happens in some areas around Antarctica and may happen on occasion in the “new Arctic” also.

        The caveat in all of this is that it matters when and where the snow falls as well as when and where it melts. Very tricky to get all these things right in models.

    • Andreas, as Tamino’s graphs shows, there is less snowing in the Arctic over the years. In great part due to the warming of fall and spring causing more rain and especially less ice extent. Snow on ground or ice drifts all over the place and lasts all winter, if winter is shorter there is less of it to go around. So the models should show this if they are correct.

  10. In 2007 NASA scientist Jay Zwally said about Arctic sea ice and Greenland melt: “the canary has died”. There’s less and less doubt about all these poor birds – and about out own fate.

    • You really gotta wonder who it is that has (and is) the bird brain in all this.

      At least canaries are smart enough to stay out of the mine unless forcably put there by humans.

      We’re not only not smart enough to stay out of the collapsing mine, but we keep digging it deeper (thinking maybe we’ll escape through China?)

    • Sceptical Wombat

      The problem with the canary in the coalmine metaphor is that it tends to understate the seriousness of the problem. I prefer to think of the Arctic Ocean as the genie in the bottle. The sea ice was not just an indicator of global warming it provided a barrier between the atmosphere and the ocean. It dramatically reduced the ability of winds to whip up waves, thus helping to preserve the stratification of the ocean, it slowed down evaporation thus reducing humidity levels in the air column and consequently keeping the greenhouse effect relatively low, it stopped the ocean from warming the atmosphere – keeping the air column cold and dense which in turned help power the jet stream and of course it stopped the sun from warming the ocean.

      Taking it away has among other things slowed the jet stream, increased the amplitude of Rossby waves and slowed their rate of movement, resulting in heat waves and cold snaps of longer duration.

      The summer storm in the Arctic is often mentioned as a possible contributor to increased melting. The interesting question is whether its duration was extended by its ability to pull heat out of the ocean – it will be interesting to see whether this phenomenon is repeated in the next few years.

      The Arctic Ocean is highly stratified with a roughly 50 metre thick layer of cold relatively fresh water above a thermocline where the water temperature and salinity both increase with depth. If that stratification breaks down then both the heat and the salt from the deeper water will become available to melt more ice.

  11. Changes in the snow pack change availability of water for drinking, transportation, and agriculture (irrigation). This affects our infrastructure -now.

    Soil moisture from melting snow is vital to agriculture. The timing of this moisture is important to large scale agriculture – now..

    • To expand on this a bit: Early loss of snowcover/albedo change is only part of the story. In the Rocky Mountains early loss of snow will be and is a serious water supply issue as Aaron notes, but is also a significant contributor to our longer and more intense fire seasons. This only compounds the problem as burned areas: release CO2 during fires, are removed from the carbon storage equation for a time after the fires, seriously impact both water quality and water storage ability, and then, last but not least nothing absorbs heat better than blackened forest. After two months of smoke in much of Western Montana and central Idaho, we’re living the dream (nightmare?) here.

  12. You could compute a Lower Bound for the combined NH sea-ice-plus-snow albedo effect by simply replacing the arctic ‘sea-ice’ monthly area with the ‘sea-ice plus NH snow’ monthly area, and re-running your sea-ice analysis.

    It would be interesting to see how this lower bound on the NH snow albedo effect compares to the sea-ice effect in each hemisphere.

    • That would be an interesting computation. Sounds like something of the sort was done in the paper linked above by MMM, above–though just the abstract is available free, apparently.

  13. Is it possible to attach a value in terms of climate sensitivity to this trend in snow? As I understand it, climate models are using a figure of around 0.25 W/m2/*C as total earth albedo change. Since there seems to be an accelerating rate of snow loss, should the figure be revised?

    My interest is in trying to substantiate the total inadequacy of the climate contrarians’ desired figure of 0-0.5*C as a figure for climate sensitivity. Developed on this page

    Any help gratefully appreciated.

    • Richard,

      The ice/snow albedo feedback is pretty small globally, but very important regionally. There’s a few points to consider here…one is that the area of the globe covered in persistent ice or snow is relatively small, and generally weighted toward regions with little solar radiation to begin with. The other factor is that there is a seasonal structure in how the ice and albedo feedback is manifested as amplified temperature (say, relative to the hemispheric or global mean). Third, the surface albedo is of second-order importance to the total planetary albedo…you need to know something about the overlying cloud structure, which is a tough problem.

      You should be able to get a good first-order estimate of climate sensitivity if you only know about the water vapor+lapse rate response, and the uncertainty is generally dominated by the shortwave component of the cloud response. I would not expect new figures of snow loss to substantially change climate sensitivity estimates.

      • Chris

        Many thanks. Even though snow/ice albedo is a small part of total planetary albedo, is it not possible to put a figure on the 5 million Km2 loss since 1979? Or about 8 million since 1970? I suppose not. Too small.

        I understand that CERES is able to calculate earth total albedo. There appears to be no long term trend. Yet. Happily.

    • Climate sensitivity in the absence of feedbacks is

      λ = T / (4 F)

      where T is mean global annual surface temperature and F is absorbed climate flux density. For Earth, these figures are roughly T = 288 K and F = 240 watts per square meter. Thus lambda is about 288/960 = 0.3 K/W/m^2. But with feedbacks it is more likely to be closer to 0.75 K/W/m^2. The denialist case, usually made by Upper-Class Twit of the Year Christopher Monckton, is that there are no feedbacks, or the feedbacks don’t matter, or lambda is “fixed by the laws of physics” at the figure they want.

  14. Lewis Cleverdon

    Tamino – thanks for this info – the graphs in particular are very helpful for showing to those without relevant scientific training (i.e. 99% ?). I note an intriguing similarity in the latter half of the ’68 to 2012 June snow cover to the Piomass sea-ice loss plot that seats an exponential curve so neatly.

    Further to Brad’s comment above, I wonder if you’d possibly spare the time to assemble the track of albedo loss from snow cover decline with that from from sea-ice loss, and convert it to a CO2e metric ?

    A track record even just back to 1980 of the effect of planetary albedo loss due to cryosphere decline would be gold-dust in terms of promoting the urgency of commensurate action.



  15. A report in today’s (October 6) Weekend Australian from notes that according to the NSIDC Antarctic sea ice extent set a new record on September 26 and that the monthly average for September ice was also a record. This, the report states, continues the increase in Antarctic sea ice of about 1 per cent in Antarctic sea ice per decade. The report also notes that this is at odds with predictions from climate change models that continue to forecast long term decline. Dr Rob Masson from the Australian Antarctic Climate and Ecosystems Division said that forecasts from the “most authoritative climate change models did not indicate the present expansion (in Antarctic sea ice)”. Dr Masson is aboard a ship in the Antarctic to “pick the gap between what we are observing and what the models are telling us”. Is this, apparently unexpected, increase in Antarctic sea ice due to changes in Antarctic snow cover?

    • Ian, Dr Mason should study in person Ice shelves, the let loose floating ones, their influence over further ice formation, how they disintegrate and seed further ice extent, unlike Greenland which sheds ice mostly going towards the warmer South, Antarctica ice is influenced by a strong Eastward current keeping the ice more next to it. I also like the idea of more snowing, because of drifting snow over frozen ice, and Antarctica coast has loads of strong winds.

    • David B. Benson

      As I understand it, the increase in the circumpolar wind speed, together with the Coriolis effect, tends to push the ice further north. This increases the extent without necessarily increasing the volume.

    • Where is this moldy climate model forecast for long-term decline in that faraway time period otherwise known as 2000-2012?

  16. My wife and I are day-tripping today, so moderation will be delayed until our return.

  17. Day-Tripper

    Tamino’s 3rd chart [the anomalies bars], inspired to run them from 1966 through 2012 to see if weather and climate were getting very obvious.

    Weather for sure and some odd years [volcano impact maybe too and super El Nino’s], but the last 5 years it’s unavoidable to eyeball the signal farmers are aint happy at all about. For sure they’re in the Oh-Shit! department.

    • On Greenland comment by Al Roger up in thread, I’ve removed the Island from the charts and added percent anomaly labels and and actual snow cover line plot. Hit Esc to stop animation, F5 to restart. (Only viewable in Firefox and Chrome)

  18. Where can I find global sulfate emissions estimates through 2010, or 2011 if possible? The lastest figures I can find are 2005. I’d like to have that extra 5 or 6 years…

  19. I just found out that NCEP/NCAR actually does have downward solar radiation flux data (daily), available here :

    That means that in pinciple, we should be able to calculate quite accurately not just how much extra heat was absorbed by ice loss (and snow cover loss), but also when and where that heat was absorbed.

    NOAA rocks !

  20. Speaking of ice shelves letting loose, Thwaites Glacier, Antarctica had some changes over the Austral Winter

  21. Extensive snow showers covering wide areas may be mistaken as sea ice when the sea temperatures are cold enough to prevent snow flakes from melting. I have a near live explanation with the real thing…