Arctic Amplification

Various processes, including albedo change (the change in reflectivity when ice and snow are replaced by open land or ocean), amplify the warming which is observed in the Arctic. Yet the situation is complicated. Cloud cover can change, which also affects reflectivity and can reduce or increase Arctic warming. Atmospheric patterns can likewise change, as can the state of the atmosphere in general. Overall, although we know that the Arctic is warming faster than the planet as a whole, how great this amplification will be in the future remains uncertain.

The question also arose, how will Arctic warming patterns differ at different times of year? I decided to take a look at some data, to see what differences can be identified in what’s happened to Arctic temperature recently. Hence I determined the Arctic warming rate in two satellite data sets of temperature in the lower troposphere (from RSS and UAH), and one of surface temperature (from the NCAR/NCEP reanalysis). To make the results more comparable, I used surface data (from NCAR/NCEP) only since 1980. This is about the same period covered in the satellite lower-troposphere data (since 1979).

It’s important to bear in mind that these data sets measure different things. The lower troposphere is the lower layer of earth’s atmosphere, roughly the bottom 5 km or so. Its temperature will not be the same as that of the surface, and likewise its temperature changes will not be the same. Nonetheless, examining the rates of change will give us some insight into how Arctic temperature is changing.

For each data set, I isolated the data for each month of the year and estimated the warming rate by linear regression. For the RSS data, that gives this:

There are hints, but not really any significant evidence, of slower warming during February and faster during April. Overall, however, the warming rate at different times of year is surprisingly consistent.

The UAH data are also for the lower troposphere, giving this:

Again there’s a hint of faster warming during April, and for these data of slower warming during summer months. But again, the most surprising thing (to me) is that warming is so consistent throughout the year.

It’s also clear from the satellite data that the Arctic really has warmed faster than the globe as a whole — a lot faster.

The surface temperature estimates from NCAR/NCEP give this:

There are significant differences between the trends in different months. However, it’s not what I was expecting (although perhaps I should have). The fastest warming is during autumn, peaking in October. This is probably due to the loss of sea ice — during October (when there’s much less sea ice than in the recent past) the open ocean allows heat to be transferred from ocean to atmosphere, warming the surface air considerably. It also surprised me that the warming during October is so fast, with temperature increasing at a whopping 1.5 deg.C per decade.

Comparing the results from all three data sets gives this:

All results confirm the rapid warming of the Arctic. The lower-troposphere data show little difference in warming rates throughout the year. The surface data not only show considerably faster warming than the lower-troposphere data except during summer, they also show a marked, and very strong, annual pattern, with autumn especially heating up at an alarming rate.

16 responses to “Arctic Amplification

  1. john harkness

    Thanks for another great post.

    Does the difference between the surface and the lower troposphere rates of warming through the year suggest that there is a strong inversion in the Arctic in late summer/early fall that keeps the heat close to the surface?

  2. Inversions tend to keep the cold, not the warmth, close to the surface. If anything, the fact that surface warming exceeds lower atmospheric warming in the fall suggests that nighttime inversions are weakening, possibly due to more cloud or (less likely) as a direct result of more greenhouse gasses.

  3. In your blog post on The Annual Cycle of CO2, you point out that there has been a significant increase in the annual cycle, especially in high northern latitudes. Is this just an effect of the decline in snow and ice cover or could it be a significant feedback as well, contributing to the greater relative warming in autumn and winter?

    [Response: Honestly, I don’t know. My guess is that it’s related to biosphere activity, but that’s just a wild guess.]

  4. I don’t know the physics, but from living in a snowy country it seems obvious that snow or ice cools the surface temperature. As long as there is snow on the ground (or ice on the sea), it restricts temperature rise. I have heard Norwegian Meteorologists say that the snow “creates its own (local) climate”.
    There is also, I think, an effect of the greater variation in winter temperature. If you for instance look at the temperatures from Svalbard, the mean temperature rises above zero in June and drops below in October. This means that there is not so much room for temperature rise in the summer months. A winter month in Svalbard can be 12-13 degrees warmer than normal, but no kind of weather can cause such a big difference in the summer. Even when the ice disappears, you won’t see a July mean temperature of 18 degrees C in Longyearbyen.

  5. Tamino , it is indeed complicated by recent cloudier summers and especially by a lot of rain (no apparent warming). Adiabatic profiles from surface upwards infringe gradually earlier during the spring and later at fall, these may fool and give erroneous conclusions. An appropriate metric would be average upper air profile comparisons. Whereas the surface may be much warmer but cooling starts from that point upwards as opposed to not so long ago colder surfaces and a warmer (now disappearing) boundary layer (inversion) above.

  6. The annual cycle in the ice response, and the temperature manifestation of that response, is very interesting. There is little Arctic amplification (relative to the global or hemispheric mean) in the summer months because the extra energy goes into evaporation and melting, while at the same time the extra sensible heat content of the oceans will eventually work its way into the atmosphere and have implications for the timing of seasonal re-growth in ice.

    Some of the most interesting work that needs to be done related to this subject is at the interface of mid-latitude atmospheric dynamics. Most models predict a decrease in the surface pole-to-equator temperature gradient (particularly in the Northern Hemisphere), and increase in the upper troposphere/stratosphere pole-to-equator temperature gradient. Horizontal temperature gradients are equivalent to vertical gradients in wind velocity, implying implications for the Polar Jet, for example. Synoptic scale weather disturbances tend to form in the regions of maximum jet stream wind speed and propagate downstream. Through baroclinic instability, the potential energy associated with temperature gradients is converted into the energy in atmospheric eddies that dominate the heat and angular momentum transport poleward of the subsiding region of the Hadley cell.

  7. re: Earthfriend
    This effect is quite visible to skiers, wherein the slopes (which may have snowmaking or at least groomers that spread the snow around very night) can be noticeably cooler than areas at same altitude just across the road.

    For papers, see Impact of artificial snow and ski-slope grooming on snowpack properties and soil thermal regime in a sub-alpine ski area, which had:

    ‘On the ski slope, the snow depth was considerably larger than off-piste due to the artificial snow production (Fig. 1). At the off-piste site, the snow disappeared 4weeks earlier than on the ski slope.'</blockquote.and Enhanced ground cooling in periods with thin snow cover in the Swiss National Park, for example. Even a thin layer of snow can cool the ground, since it reflects sunlight, but doesn’t insolate the ground.

    I conjecture that this sort of snow-albedo effect was part of the drop into the LIA,m especially as seen in England and Western Europe, or anywhere else near the seasonal snow line. So, we had a millenially- unique steep drop in CO2, ~1525AD-1600AD. Bill Ruddiman hypothesized that drop as reforestration caused by a 50M person die-off in the Americas from disease and I think evidence has been accumulating for that. Following that, add volcanoes and Maunder Minimum, and CO2 that stays ~6ppm below where it usually was during 1100-1500AD, and (minimally) Milankovtich-lower insolation at 60degN and the result seems at least part of the LIA.

    One would not expect to see much effect in areas where (a) snow was on ground year-round or (b) where it never snowed, but where snow was seasonal. Snow on ground longer would keep it cooler, and hence sometimes keep snow from being rain. Again, on the slopes, I’ve been in conditions where snow was falling at bottom of slope, but across the road, it was just enough warmer to be snow.

    Why this matters: if Ruddiman is right, some of the CO2 drops of medium length (say 20-80 years, of which there are several in that graph), were human forcings where farms fell out of use from plague, got reforested, then, as population recovered or people migrated, got deforested again. I grew up on a PA farm in which trees were cleared for a pasture pre-1850. When Dad quit farming, circa 1970, in 20 years the trees had grown back.

    Obviously the 1600AD event is the strongest. This bears on the attribution problem of sorting out human signal from natural variation noise. Methane is related, but some different effects may be seen. Again, one would see teh amplified cooling effect primarily in the zeon of seasonal snow.

  8. Could the difference between the surface temperature data and data derived from satellite retrievals also partially due to the fact that the satellite data don’t cover the Arctic region North of 85 deg. latitude or so?

    [Response: Don’t know about UAH, but the RSS satellite arctic region extends from 60N to 82.5N. The missing area (from 82.5N to the pole) is only 6.4% of the entire area from 60N to the pole.]

    • Tamino,
      Comments made in the “snow+ice” post, using AVHRR data, suggest a moderate cooling in winter over the past decades North of 60N.
      And, until 2004, a cooling of -0.125°C per year for the Arctic north of 70 C, at least until 2004.

      In this post however, the NCEP/NCAR data you present suggests the ‘warming’ during winter (to the account of +0.08 C/yr or so)/

      So. could you please clarify what you considered “Arctic” in this post, specifically for the NCEP/NCAR, UAH and RSS data sets ?
      Or is there a mismatch in data sets (ayt least with AVHRR) on winter trends ?

      Mismatches in data sets are not uncommon, but they exist. they are always interesting to investigate.

      [Response: For RSS the Arctic is 60N to 82.5N. Don’t know about UAH but it’s probably the same. For NCAR/NCEP I used northe of the arctic circle (66.5N to the pole). In the article you refer to by Wang et al., they may make a statement about trend 70N and above, but their temperature analysis actually covers the region from 60N to the pole.

      I’m preparing a post comparing the Wang et al. results from AVHRR to NCAR/NCEP, over the same region (60N to the pole) and same time span (begin 1982 to end 2004). Yes there is a notable mismatch between the data sets.]

  9. Earthfriend, and others,
    Note that winter snow cover IS increasing moderately, even though summer snow cover is decreasing much faster.
    See Tamino’s “snow+ice” post on this.

    Also, if you want to get down to the physics and modeling aspect, here is a truely enlightening publication which shows how the Arctic can cool in winter, but warm in summer due to changes in snow cover :

    Click to access Cohenetal_ERL12.pdf

    This paper suggests cooling in the Boreal forest are in winter due to increased snow cover, as caused by reduced sea ice area in summer,

    Though we cannot conclude de?nitively that warming in the summer and autumn is forcing winter regional cooling, analysis of the most recent observational and modelling data supports links between strong regional cooling trends in the winter and warming trends in the prior seasons. A warmer, more moisture-laden Arctic atmosphere in the autumn contributes to an increase in Eurasian snow cover during that season. This change in snow cover dynamically forces negative AO conditions the following winter. We deduce that one main reason for models failing to capture the observed wintertime cooling is probably their poor representation of snow cover variability and the associated dynamical relationships with atmospheric circulation trends

    Note that the same poor model performance on snow cover variability may be at the root of IPCC GCMs underestimation of sea ice extent in summer.

    We are getting somewhere. Maybe Nature is not entirely arbitrary, but instead our models simply underestimate snow cover behavior in summer and winter alike….

  10. The extensive loss of sea ice contributes significantly to the ground level warming, while global warming simultaneously raises the extremity of atmospheric circulation both of which are hugely responsible for the increased temperatures within the Arctic regions and the consequential loss of sea ice. With the loss of sea ice, more heat is absorbed by the oceans due to the lessened albedo effect and the less reflectivity. As the ocean warms, as does the atmosphere above it. This ultimately leads to a change in the atmospheric circulation patterns. The circulation transports energy to the Arctic region, which increasing the temperatures further up in the atmosphere. In a usual situation, this would increase the capacity of the atmosphere to hold water vapour which would intensify the greenhouse effect. This however is not the case in the Arctic due to the fact that there is less moisture in the air thus meaning that there is a relatively smaller greenhouse effect than elsewhere. It is therefore the significant loss of sea ice and the consequential decrease in albedo that initiates the extentive warming of the Arctic region.

  11. Anxious about the Artic

    The Artic is mainly occupied or covered by an extent of Artic Sea Ice, it illustrates one of the most visible and large scale natural scales of our world, it is also relatively easy to monitoring and there is long term data. There is satellite data that shows shrinking since the late 1970s. The maximum extent of the ice is reached in the northern winter during March and the summer minimum is September. It is increasingly important to monitor the maximum and minimum extent of the ice so as to determine the rate of decline. Data shows that ice thickness and extent have been decreasing as well as more rapid melting. On 16 September this year (2012) there was a recorded all-time low of extent of Artic Sea Ice that was less than the previous low recorded in September 2007. 2007-2012 represent the lowest recorded extent of Artic sea ice that has been recorded. The recorded extent of sea Ice on the 16th September was 3.41 million km2, there has also been significant evidence of a decline in Ice volume. The main reason for changes in the cycle and more rapid melting can be attributed to anthropogenic forcings. “Why should this be of our concern you ask?” the extent of Artic ice has many implications. If the sea ice extent is less (i.e. more melting) then we will see aspects such as a decrease in sea salinity and its cascading impacts on marine ecosystems, sea level rise and its coastal region implications and a decrease in albedo which will further enhance positive feedback as a decrease in ice, will result in a decrease of albedo and more absorption of solar radiation, which will then result in an increase of temperatures, which will then feedback and cause more ice melt and so the feedback loop is initiated. There are also hidden impacts or indirect consequences that will occur if the Artic melts. Like a reduction is shipping of up to two thirds in places like South Africa as ships will be able to just sail across the Arctic Ocean. As briefly mentioned the Artic is a good indicator of what is happening with regard to changing climate as it is so observable and with it cascade the implications that could arise.

  12. The Arctic sea ice is decreasing at such a rapid rate, especially in the last few years, it is predicted that by 2030, maybe even 2020, the Arctic will be sea ice free.
    The earlier onset of spring and therefore earlier break up of sea ice will have a huge impact on the Arctic ecosystems and the species that live in the Arctic region. Polar bears, for example, will be greatly affected. The earlier break up of sea ice affects the polar bears’ dens by separating them from the spring hunting grounds; as young polar bears are not able to swim far, it could result in many cubs drowning. There is thus the possibility for an increase in endangered species in the Arctic regions and perhaps even extinctions in the future.
    The arctic sea ice volume is a sensitive indicator of climate change. While the decreasing of earth’s ice volume is a normal phenomenon of earth’s history, this decrease is happening at such a rapid rate (one that is even faster than the IPCC’s worst scenario predictions in 2000). Global warming caused by anthropogenic factors is therefore causing an unnatural change in the earth which is creating an uncertain future for all that inhabit it.

  13. From the represented graphs in this blog, it is clear that the increase of global warming is forcing the receding of the Arctic ice. The climate change is having a visible impact on the Arctic ice level and the processes that maintain the ice. The Arctic is warming at almost twice the rate of the rest of the earth, and this should get us worried because the arctic plays a huge role in the global atmospheric circulation, meaning that the warming of tr\his region will have an impact on the rest of the planet. The arctic ice reflects a large amount of solar energy back to the atmosphere, helping with the managing of the atmospheric circulation and cooling of the planet. The major concern s that the melting of arctic ice disturbs the global circulation process.

  14. Global warming and the thinning/melting of the ice caps has significant effects on surface temperatures in the Arctic as well as in the Antarctic. As global temperatures increase more surface ice is melted away during the Summer months of the year thus reducing the cooling effect produced by the albedo of snow and ice(incoming solar radiation is reflected off the surface of snow and ice back out into space). Moreover, global temperature increases have owed to longer periods of warming in the arctic ocean and thus sea ice does not have as much time to recover and regain it’s full winter extent as it had in previous years. Thus we see an even further increase in temperatures in these regions as the extent of snow and ice cover slowly deteriorates and thus the albedo effect produced by snow and ice is ultimately reduced. If this trend continues we may see a significant shift in global climates due to the huge role played by the arctic oceans in global circulation patterns and the distribution of heat throughout the atmosphere.

  15. The ECMWF is forecasting an Arctic Dipole for Oct 23-27. How much more MYI will be flushed out the Fram? Little by little Arctic Ice is being reduced to First and Second year ice that will melt out next summer to a new low levels in all measurements and perhaps totally disappear during late August or early September.