Ice Over

We’re witnessing a remarkable decline in Arctic sea ice. The annual minimum is taking a nosedive:

Bear in mind that we haven’t yet reached the minimum this year, so the 2012 value will end up being even lower than the value plotted.

But that’s not the only change taking place. We can investigate in greater detail by studying how the cycles have changed over time. One way to do this is with windowed Fourier analysis. I took each 2-year time span (nearby spans overlapping by 1 year) and fit a 2nd-order Fourier series in order to quantify the size and shape of the annual cycle.

Here’s how the amplitude (actually semi-amplitude, which is just half the full amplitude) of the fundamental Fourier component changed:

The dramatic low extent in 2007 ushered in a new era of greater amplitude. But it’s not just the main Fourier component whose amplitude increased, so did the 2nd harmonic:

In fact the 2nd harmonic amplitude increased more — proportionally — than the 1st harmonic (the fundamental), so the relative amplitude (the ratio of their amplitudes) also increased:

That’s not all. We can use windowed Fourier analysis to measure the phase of each component of the Fourier series. We can then compute the relative phase for higher harmonics, which is their phase when the fundamental is at phase zero. Here’s the relative phase (in cycles) of the 2nd harmonic:

Note that it’s always near 0.5. This means that when the fundamental is at maximum (during late winter) the 2nd harmonic is at minimum so they partly cancel, but when the fundamental is at minimum (during late summer) the 2nd harmonic is too so they reinforce each other. That makes the minimum more “pointy” while the maximum is more flat.

Note also that the relative phase has gotten closer to 0.5 lately. That’s because the annual cycle has become more symmetrical, with the rise to maximum and the fall to minimum occurring at more nearly the same rate.

The summer minimum has consistently occurred around mid-September, but the change in relative phase is because the winter maximum has migrated to later in the year. It’s now just about 6 months after the summer minimum, whereas before it occured a little bit earlier.

We can see this plainly if we compute the average annual cycle for the data before 2007, and compare that to the average annual cycle after 2007. Here they are, with pre-2007 data in blue and post-2007 in red:

The most obvious feature is that the minimum has declined much more than the maximum. But it’s also evident that the winter max is now around mid-March, when it used to happen in early March or even late February.

Part of the reason for the greater decline at minimum is simple geometry, due to the placement of land masses in the northern hemisphere. But there may be other contributing factors as well. As for the phase change, with maximum occurring slightly later in the year, I don’t know why.

25 responses to “Ice Over

  1. As for the phase change, with maximum occurring slightly later in the year, I don’t know why.

    Could it be due to the ice no longer extending as far south where the effects of the seasonal progression would be felt earlier? It’s clear there’s been some decrease in winter maxima. I would assume most of that would be along the southern fringes of the ice pack.

  2. I agree with ohioclimate.

    Also, an increasing role of water temperature should delay everything. In previous years, with thicker multi-year ice, air temperatures would have had a greater relative role compared to now?

  3. Regarding sea ice maximums occurring later in the season– there is some hint at a relationship to an increasing frequency of the Dipole Anomaly and the shattering of the traditional closed Arctic vortex in the winter. What this means is that colder air is pushed out of the central arctic to lower latitudes for more extended periods later in the season and expanding the sea ice. We saw this for example this past winter when a persistent northeast wind blew across the Bering Sea during March, and a great majority of the late season sea ice growth came from the Bering.

    Of course, late season sea ice is thinner and thus is more prone to rapid melting once the summer melt really kicks in. This late season first year sea ice growth will become the norm in the future as the Arctic fluctuates between an ice-free summer and late winter ice growth that looks impressive but melts very fast during the early summer melt.

  4. Alex the Seal

    Could it be that as thicker “old” ice decreases it gets easier to melt? Thus extent is more variable?

  5. Does this years collapse in arctic sea ice extent support the PIOMASS sea ice volume data+model. I know that William Connelly has criticized PIOMASS in the past as not being realistic. So though I’ve not discounted it I’ve only ever watched it being a bit more circumspect than usual. But it seams to me that this years collapse in arctic sea ice extent is at least partly made possible by the large fall in volume that preceded it.

  6. Another scary plot to try: five day mean of the rate of change, plotted vs time of year. This year has the greatest rate of change on record for this time of year, or at least it did a few days ago. There is a long way to go yet..

  7. Glenn Tamblyn


    PIOMas was cross validated against data from ICESAT some years ago. In in the last year or so, CRYOSAT2 has produced data validating PIOMas’s latest results. So that is pretty good confirmation of the methods used by PIOMas.

    Their next round of results, for August, are due out next week. Will be very interesting. Some food for thought (or a post by Tamino next week). Not only have extent figures crashed this year, so have area figures. Area is 2 out of the 3 dimensions that make up ice volume. If area is crashing this hard, what is happening to thickness, the missing dimension. If we use ArcticROOS Area and assume that thickness declines by a similar ratio, then volume has declined by 50% over the last month. That would have us pushing the 3000 km^3 mark by end of August. Thickness may not have declined by as much as the areal measurements, but since the ice floes are fairly flat, most melt happens on the top & bottom surfaces, impacting more on thickness that area.

    If August volume figures push near the 3000 mark, mid September (or maybe late September this year) will likely go down to 2000. Half the remaining ice at minimum lost in one year. Next year could be really really interesting.

    We will have some idea next week.

  8. Wow, this is amazing.

  9. I think regional data might be more illuminating for tracking maxima and minima variability.

    Looking at sea ice area on Cyrosphere Today, the post-2007 amplitude increase appears to be almost entirely a function of an abrupt change in melt behaviour in the Arctic Basin.

    Almost all regions within the enclosed Arctic tend to return to the same ice covering at maximum whereas marginal regions show variability and generally tend towards lower maxima in recent years (e.g. Sea of Okhotsk), although some show little trend (e.g. Bering Sea, which had a record high maximum this year… all gone now).

    These marginal regions mainly control the maximum, whereas they always reach near zero at minimum so don’t have much effect on those trends. The enclosed regions mainly control the minimum, but don’t have much effect on maximum because they always tend to return to the same value.

  10. So to summarize the denialist position:

    1) The Arctic ice isn’t really shrinking
    2) …but even if it is, it’s happened before
    3) …but even if it hasn’t happened before, it’s no big deal
    4) …but even if it is a big deal, it’s just part of a natural cycle
    5) …but even if it is from CO2, there’s nothing we can do about it now
    6) …but even if we can stop it now, we shouldn’t, because an ice-free Arctic will free the area for exploitation
    7) …but that’s a moot point, really, because the Arctic ice isn’t really shrinking

    Yes, you really can smell the desperation.

    • The scary thing is that position #5 may be true. If the volume graphs are correct, then there’s in fact little we can do to prevent a seasonally ice-free Arctic. And if the slight to moderate global warming we have seen until now is enough to melt the sea ice in late summer, how many decades might it take to an anually ice-free Arctic?

      When the positive impacts of a radical reduction of our GHG emissions eventually become effective, the ice might already be gone.

    • 8) but if it is shrinking, it will be good for us anyway!

  11. Climate Weenie

    “the reason for the greater decline at minimum”

    Is age of ice. First year ice melts more rapidly than the lost multi-year ice. But during winter, ice still forms over nearly the same area.

    “As for the phase change, with maximum occurring slightly later in the year, I don’t know why.”

    Is geometry. The sun rises (albeit briefly) on the Arctic circle on the morning after Winter Solstice. The sun doesn’t rise on the North Pole until the morning of the Spring Equinox. So if the area of ice is more constrained toward the pole, the time of ice maximum will shift toward the equinox ( March 21-22 ).

  12. Statistics predicted that the sea ice system was unstable and would go out of control, but once the system is out of control, “statistics” does not predict how the system will behave (unless there is data from similar systems that have gone out of control for similar reasons).
    One reason the Arctic system has gone out of control is that it has warmed so much that (some) latent transported in from the south no longer condenses out before it gets to the Arctic Sea ice. Sea ice melt is no longer constrained by local insolation. Now, sea ice melt can be driven by latent heat from the south. For the past few summers, we have seen substantial rain in what was one of the driest places in the world. In 40 years, the Arctic has gone from desert to ocean.(Well, OK! an ocean, peat bogs, and some rock.) If you think that is weird, wait a couple of years until the Arctic starts exporting substantial amounts of latent heat during the summer.

  13. Juday Juday Juday has yet to comment. But why wait? It’s all natural, folks!

  14. From the talk page of wikipedia

    “Since I’m new here, I just want to bring up the fact arctic sea ice has come roaring back since the minimum level set in summer of 2007. The article does not really discuss how the decades-long trend of decreasing ice has been completely reversed over the last 2 to 2.5 years. Surely the article has to come to terms with the facts, correct? [3] The NSIDC has a nice graphic I think we can use. [4] I’m wondering if the article does not deserve a rewrite. What do you think? RonCram (talk) 23:54, 4 April 2010 (UTC)”

    [Response: It looks like RonCram will have to wait another few years to make idiotic statements like “roaring back” and “trend … over the last 2 to 2.5 years.”]

  15. I’ve tried my hand at projecting the September 2012 average extent (which will be a little higher than the actual minimum). If the September average does end up around 3.6 million sq km, the actual minimum (as you are using) should end up around 3.5 million sq km. And, yes, the later it arrives the lower it will be.

    2012 Arctic sea ice minimum, part 2: September 2012 projected at 3.6 million sq km, 700K below previous low in 2007

    In my previous discussion of the extraordinary 2012 melt, I noted the eclipse of the old daily record on August 24, three weeks ahead of the 2007 pace. But I also gave a series of short-term projections for the September extent average, which is the metric typically used to track the decline in Arctic sea ice. The 2012 September projection now stands at 3.56 (+/- .0.13) million sq km, slightly down from my previous projection of 3.67 million sq km. That’s more than 700,000 sq km less than the previous 2007 record of 4.30 million sq km.

  16. David B. Benson

    The maximum is also partly determined by how much sea ice exits through Davis & Fram straits.

    The Titanic encountered some on its last night.

    • No, the Titanic encountered an iceberg calved from a glacier or ice shelf. Very very different to sea ice.

  17. Sceptical Wombat

    Tamino said
    “The most obvious feature is that the minimum has declined much more than the maximum. But it’s also evident that the winter max is now around mid-March, when it used to happen in early March or even late February.”

    This would explain why Watts and Co got their chance to salivate at the possibility that sea ice extent might get to equal or even exceed climatology for a day or two earlier in April. You could bet that if that had happened they would not have doubted NSIDC’s accuracy.

  18. This new record, coming after a low in the suns output confirms one thing – there is climate gate like scandal, the scandal being the people are taken in by charlatans the likes of Anthony Watts.

  19. one more thing while I’m venting, we should be angry at a sceptics, for years these charlatans like monckton, watts and morano, conned people and underplayed the existence of climate change ! Now it turns out – ops it was real after all, but double ops it’s too late to do anything about it. History will not be kind to these people!

  20. > double ops it’s too late
    Wrong. Don’t be fooled, The selfish say it’s too late to justify doing nothing. There’s plenty of time for us to take pains now to reduce suffering later on by others.

    • Hank’s point deserves repeating, and it’s what keeps me going whenever I grow pessimistic through thinking about how we’ve already made a mess of the planet and its future.

      Things can always get worse if we don’t act, whether it’s today or tomorrow or the day after, and the longer we don’t act to fix the problem the worse it will be. And the longer that we don’t act, the more we’ll bring the problems of our grandchildren forward to plague our own children, and even ourselves if we’re not already superannuated.

      But if by “too late” we mean the point at which we can get away with neglible impact on our biosphere, then yes, we’ve already passed that milestone…