How fake skeptics fool themselves, part infinity: Sea Ice version

Suppose you were suffering from a serious illness which attacked many of your internal organs. It’s likely to be fatal, so your wife is very concerned about the results of your latest medical tests. They indicate that your heart, lungs, and liver are degrading rapidly, but that your spleen has actually improved. In fact your spleen didn’t show any signs of degradation, even over the long term. So you go home and tell your wife, “Great news, honey! My spleen is in better shape than it was before!”

That wouldn’t really be an honest report of your overall health, would it?

But that’s exactly what Anthony Watts has graced us — or should I say, disgraced himself — with, in his latest post about sea ice at WUWT. He crows about how much sea ice there is in the Arctic, and how it has reached “near normal” levels (remind you of anything?). But if Arctic sea ice is disappearing in a “death spiral,” how could one possibly do so? Well, if your heart, lungs, and liver are degrading rapidly, how could you possibly report that your health has improved? Talk about your spleen!

It’s easy. First, look only at a brief moment of time, quite ignoring what really matters, the trend. Second, look only at a small region of the Arctic, quite ignoring, well, most of it. Be sure to pick the one region that happens to show signs of improvement at this particular moment. It’s a classic strategy called “cherry-picking.” As for an overall survey of the state of Arctic sea ice and its statistically significant trends — we can’t have that! It would show just how dire the situation is.

Which goes to show that Anthony Watts is no skeptic. When he finds some excuse — any excuse — to blind himself to the obvious (that Arctic sea ice has been and continues to be one of the most compelling signs of global warming) Watts has no skepticism to apply. Instead he’ll bring his usual level of infinite gullibility. Honestly, it’s pathetic.

But it’s the same strategy that almost every fake skeptic has used (or should I say, abused) and many of them make their modus operandi. And well they should, because if you want to deceive the naive (including yourself), it works.

What has really been happening to sea ice lately that can tell us about the recent course of global warming?

A good place to start is take a close look at sea ice extent data from the National Snow and Ice Data Center. The data consist of monthly averages for each hemisphere. This enables us to estimate trends in sea ice extent, not only overall, but separately for each month of the year. We can also investigate whether trends have changed recently. I’ll do so with a simple strategy: compare the trend using data up to 2000.0, to that using data since 2000.0. I’ll also take the trend estimated using just the earlier data and extrapolate that to the present, to give a visual impression of how sea ice has departed from its previous trend since the year 2000 began.

Here’s the overall result for the northern hemisphere:

The thin black line is Arctic sea ice anomaly (anomalies are computed relative to the entire time span). The blue line gives the trend estimated using only the data prior to 2000, which has been extended to the present to show what would have happened if the trend had continued unchanged. The red line is a smoothed version of the actual data, to show what did happen.

There’s been a clear and sizeable decline in Arctic sea ice, which was already clear (and sizeable!) by the year 2000. Since then, Arctic sea ice declined even faster than it had previously — the change in its rate is statistically significant. The loss of Arctic sea ice is one of the most dramatic consequences of global warming, and is happening quite a bit faster than computer models have predicted.

Things are different for the southern hemisphere:

Antarctic sea ice has shown an increase which is statistically significant but wasn’t so prior to 2000. Although it has risen a little more since 2000 than the pre-2000 trend line would indicate, the departure from its pre-2000 trend is not significant (not even close), so there’s no sign of acceleration in Antarctic sea ice gain.

Arctic sea ice loss greatly outweighs Antarctic sea ice gain, as is evident from looking at global sea ice:

It’s also interesting to study how the trends vary throughout the year by estimating them for individual months. Here are the monthly trends (Arctic in red, Antarctic in blue, global in black):

Clearly the Arctic loss is greater than the Antarctic gain for every month of the year, so globally sea ice is on the decline. We can also compare monthly trends using data prior to 2000, to monthly trends using data after 2000. For the Arctic we get this:

This shows that the acceleration of Arctic sea ice loss has mainly been a summer-autumn phenomenon. Again things are different in the Antarctic:

There really hasn’t been much change in the growth rate of Antarctic sea ice since 2000, although there are signs of a possible acceleration of December increase.

A clear picture emerges: the main difference in sea ice changes since 2000 is that the Arctic is losing sea ice much faster than it was before — the situation has not only worsened, it has worsened a lot faster than expected. Not to see that it’s in a death spiral, you have to be delusional.

I have some advice for Anthony Watts. Just stop talking about sea ice, because every time you do, not only do you make yourself look like an idiot, you make yourself look like a dishonest and/or delusional idiot.

What’s happened lately to earth’s ice (the cryosphere) in general, and Arctic sea ice in particular, is immensely powerful evidence of rapid and unprecedented global warming. Anyone who tells you different, is selling something.


Another version of the same data:

42 responses to “How fake skeptics fool themselves, part infinity: Sea Ice version

  1. So far Arctic sea ice coverage on WUWT looks a bit like Arctic Gish Gallops to me. Now that we mention it, did he offer his many readers an explanation/correction for his crude combining of sea ice area anomaly graphs from pre- and post-1979 satellite data? Any news on that USS Skate picture of which the date had been ‘revisioned’? You know, the one that was used for years by pseudoskeptics to triumphantly exclaim that the North Pole was ice free in 1957. No updates? How surprising! That’s really not like Anthony Watts at all!

    It’ll be interesting to see how Anthony handles this melting season, especially if it emulates the last five melting seasons. I still think he should ask Steven Goddard to cover it this season. Or Joe Bastardi. Or both.

    But it’s difficult for him. If he doesn’t say anything, people will notice. Maybe even his loyal herd would. So the least he can or must do, is copypaste his Arctic Gish Gallops. You know, hide the dying trees with a big cardboard cutout depicting a forest. ‘NSIDC’s monthly analysis was posted…blah blah blah…the WUWT poll entry for SEARCH…blah blah blah…North Pole cam still showing ice…blah blah blah…look the DMI temps, still flat in summer!…blah blah blah…Al Gore said that in… blah blah blah…Why isn’t that graph updated?! What’s being hidden from us?!…blah blah blah…Thank God it’s October…blah blah blah…it’s snowing in Tadjikistan!…blah blah blah’. Give the guy a break. He has to keep this up for five-six months, doing the same thing every year, while the Arctic sea ice melts its way to a crescendo.

    But weather is still important, so maybe this year the Arctic spares our friends with the strong craving for psychological denial. Things are looking good on the Pacific side of the Arctic, not so good on the Atlantic side. Who knows?

    Either way, looking at the long-term, some mysterious negative feedback or cycle (AMO?) needs to start kicking in real soon now. Not only for the sake of pseudo-skeptic credibility (or whatever’s left of it), but for everyone’s sake.

    Because if the Arctic continues its walk along the edge of the cliff…

    • One nit: For the record, the Skate date was 1959, not 1957.
      Also, you can change “was used” to “is used”; Bastardi was pimping the Skate on Twitter within the last couple of weeks.

      • I know it was 1959, but wasn’t it 1957 according to the myth? And then 1958?

        See for instance here, where the title says: USS Skate reaches the North Pole in 1958 and finds open water! (but check the link in your browser address bar)

        Watts isn’t interested in all that. Sure, he helped spread the myth, but why correct it? No, he wants to know: “Is revisionism going on with the date of the famous USS Skate photo in the Arctic?”

        But actually ‘revisionism’ is the term we Dutch people use for ‘Holocaust denial’, so I really feel insulted by Anthony Watts.

      • capitalclimate: Not only did Anthony and his feeders get the location of the Skate picture wrong, they got the date wrong, repeatedly.

      • “See for instance here…”

        Gee, thanks Neven! That link had a nasty bit of javascript, opened a new window and spawned new tabs in the old whenever I tried to close the new window or close tabs in the old. I think I should probably fumigate my computer now.

      • Re the USS Skate – For those unfamiliar with Neven’s “commenter’s icon,” it is a actually a detail from that much-waved image of the Skate in the Arctic, the one that some allege was taken at the (geographical) North Pole when Skate arrived there back in the late 50s during the Arctic winter.

      • Gee, thanks Neven! That link had a nasty bit of javascript, opened a new window and spawned new tabs in the old whenever I tried to close the new window or close tabs in the old. I think I should probably fumigate my computer now.

        Sorry, Tim! I didn’t have any problems there.

      • @Timothy Chase: Firefox + NoScript +AdblockPlus = very, very few of the experiences you’re describing (none to date, in my own experience) and virtually no ads, too.

  2. Here’s another take on the disappearing Arctic sea ice. I’ve prepared a figure showing daily PIOMAS data (Arctic ice volume from 1979 to present) showing the annual minima and maxima (filled circles), with fits to the trends of these extremes fitted by an exponential curve:

    I tried a few other fits but the 3-parameter exponential shown above gives the best fit overall. I also did the fit for the maximum annual ice volume, which is also decreasing at a comparable rate.

    The value of a in these fits is the equilibrium ice volume. For the annual minimum pre-industrial volume implied by this fit, a= 15,812 km^3 of ice; for the annual maximum pre-industrial volume, a= 32,659 km^3.

    The annual minimum curve extrapolates to zero ice in 2015 (this date is the parameter “b” of the fit). The exponential fit also has the maximum ice volume going to zero in 2028, although this is quite an extrapolation and should be taken with a large grain of salt.

    The minimum figure looks pretty robust, though. For example, a quadratic polynomial fit (to the ice volume, not the log) somewhat underestimates the recent decline, and still gives zero ice in 2018.


  3. Nice work, Phil. Here’s an overview of several graphs with different fits, based on PIOMAS data.

  4. Well I see another Cyclist has joined the Jerk Offensive, Woy Spencer is now an Arctic sea ice newbie.

    In Woy’s WTFUWT? graphic I do wonder why Woy didn’t include a cubic polynomial fit (you know, just for the fun of it, because, well just because Woy likes his fellow fake skeptics to just play with it in a rhythmic cyclic sort of way), like Woy always does with his very flawed UAB temperature dateset?

    Oh, I see why Woy didn’t do that this time, the cubic would curve down at the end, meaning we all could all assume the AO will continue heading south, meaning further future Arctic sea ice losses.

    Woy is also of the impression that Arctic sea ice has only been declining for the past 20 years, when it has been clearly been declining since the start of the dataset in 1979. Woy can’t even count.

    But that’s what you get for being a newbie Woy.

  5. Having spent a fair amount of time debating with skeptics about sea ice dynamics on WUWT, I think it’s accurate to say that the only people posting there who have a clue about what’s going in the Arctic are Dr. Walt Meier and Dr. Julienne Stroeve, who occasionally attempt to set the conversation back to some balance of actual scientific basis. Joe Bastardi is a favorite among skeptical faithful on WUWT, and it’s fair to say he may know weather, but should simply not talk sbout sea ice at all. He, along with other skeptics, got all excited when 2008 and 2009 did not continue the steep decline we saw in 2007, even so much as claiming a “recovery” was underway, and that 2007 was obviously just a fluke event. Of course, along came 2010, and then 2011, and we saw that 2007 was no fluke. What will this summer bring? Still too early of course, but the low extent in the Barants and Kara are hints we may see 2007’s record once more challenged. The thin ice in the Bering sea will get eaten up pretty rapidly this spring and early summer. But no matter what happens this summer, the long term trend is quite clear and we will likely see an ice free summer arctic by 2030 at the latest.

  6. Paula Thomas

    shhh. Anthony Watts will say you got it wrong cos there’s a square metre of ice floating somewhere in the Arctic in 2030!!
    Seriously though that is impressive work which I will try to repeat (to confirm it).

  7. rabiddoomsayer

    Nil volume by 2030 er wow. With winter ice there is the possibility of huge extent of very thin ice and I suspect the possibility (probabiility) of some huge swings between years depending upon conditions.

    We see some big freeze ups when conditions change. There is a huge heat dump, eventually to space, in winter as the ice freezes.

  8. You certainly don’t need appreciable statistical chops to be able to predict Watts & Co…

  9. Cycle plots provide a different way to visualize area and extent trends in each month of the year:

    Our local New England cryosphere follows larger trends. 126 years of ice-out dates on Lake Winnepesaukee, in global context:

  10. Well Tony the Paper Tiger does it again see;

    Crack in the Antarctic!

    Wher Tony states;

    “And, there’s only a 40 year historical context for these observations.”

    Hmm, so Tony can’t count either. By my math we have 33 complete years (1979-2011) of Antarctica sea ice extent/area data.

    Tony can’t even be bothered by that West Antarctica ice shelf loss, because, well because Antarctica’s sea ice area/extent is growing at a rather slow rate, but who cares how slow that rate is relative to the loss rates elsewhere (meaning the Arctic), because it’s growing rather slowly, don’t you know?

    But here’s the real kicker, the West Antarctica ice shelves are shrinking according to the UT@A paper.

    So, for example, if the ice shelf is only 200 ft above water (from every aerial photograph I’ve ever seen, those above water ice shelf’s do seem to be at least a few hunderd feet above water), and it’s density is 0.92, that means that the ice shelf is 2,500 ft thick overall.

    So replacing 2,500 ft of ice shelf with even 25 ft thick sea ice is a net loss of 99% in sea ice folume at those retreating locations.

    So could this be the real reason for Antarctica’s rather slow rate of increase in overall sea ice ara/extent? You decide.

  11. I’ve taken some criticism elswhere for my blog piece “Climate officials and climate provisionals”: . The piece criticises the official view of climate change as lagging behind the real world (as Tamino shows above). It also reports the evidence on Arctic sea-ice and possible methane releases that was given to the House of Commons Environmental Audit Committee recently.

    Worryingly here is a quote from a letter my MP has received from the Met Office on my behalf:

    “Carbon dioxide and methane release from permafrost is an area of active research at the Met Office Hadley Centre. A simple framework has been developed for estimating the amount of carbon dioxide and methane release from permafrost. and to estimate the impact of this release on the global mean temperature. We expect this work to be published within the next 2 months. This is a step towards full representation or the permafrost climate feedback within the more complex Hadley Centre climate models – the outputs of which are used by the IPCC – which we plan to achieve within the next 2 years.”

    “Currently, no work has been undertaken to incorporate methane release from ocean hydrates lnto Hadley Centre climate models.”

    “I hope this helps.”

    Is this as worrying as I think it is?

    Perhaps my post was ill expressed but is “official” science really in safe hands?

    P.S. About this time last year Chris Huhne, who was then Secretary of State for Energy and Climate Change, told me this work had already been done. If he didn’t know ….

  12. Why are the error bars in the last graph for the months July to November smaller than the other months? I would have expected them all to be the same size.

  13. I think you might be doing Anthony Watts an injustice.
    It’s true that the emphasis in his post is on the increased ice in the Bering Sea.
    But right at the end he writes:

    “The Bering Sea stands in stark contrast to the rest of the Arctic ice cap, where sea ice extent was below average in both January and February. Ice cover was down drastically on the Atlantic Ocean side of the Arctic, including the Kara, Barents, and Laptev Seas, where air temperatures were 4 to 8 degrees Celsius (7 to 14 degrees Fahrenheit) above the norm.”

    [Response: He didn’t write that. He just copy-pasted a post by NSIDC which included that text.]

  14. Carbon dioxide is well mixed in the global atmosphere, but the ice melt is entirely in the northern hemisphere where most of the land and almost all the manufacturing is. The IPCC has, thankfully, recently highlighted the role of black carbon (soot) from manufacturing in Arctic ice melt and also the need to adress the carbon black air polution which is also a health issue. Black soot falls out with precipitation, and little of it makes it across the equatorial region to the Anarctic. I’m looking forward to futher studies on the role of Arctic ice melt from carbon black, One indication that it’s significant, I think, is the fact that Acrtic sea ice anomaly is mainly a summer phenomena when there is sunlight. The melting, of course, changes the albedo, resulting in further melting and plays a role in the northern hemisphere warming. Signifcant reduction in CO2 emissions this decade seems unlikely for political and economic reasons. Reduction of carbon black would be much less costly and have health benefits as well. Let’s all get behind it.

    [Response: Please spare us absurd statements like “ice melt is entirely in the northern hemisphere.” Both the Greenland and Antarctic ice sheets are losing mass at an alarming rate, glaciers worldwide are disappearing, snowpack is melting earlier in the season, as is ice cover on lakes and rivers. Also absurd is your claim that “Acrtic sea ice anomaly is mainly a summer phenomena.” It’s a year-round phenomenon, although it happens to be worse during summer and autumn.

    The cryosphere — all of it — is screaming about global warming. Changing the focus from CO2 to black carbon is, for some, a foolish diversion; for others, it’s a disingenuous attempt to avoid responsibility.]

    • Doug,

      Keep in mind that there is a lot of interesting physics in the seasonality of the ice-albedo response (in terms of extra energy storage), as well as in the temperature manifestation of that feedback.

      For instance, many regions of the Arctic tend to get pegged to near-freezing in the summer as most of the energy is dumped into melt or evaporation, so you don’t expect too much Arctic amplification in the summertime, even if the albedo response is well at work. Earlier sea ice retreat is associated with ice albedo feedback and delayed ice advance is consistent with an enhanced ocean heat response. Ian Eisenman also showed that there is asymmetry in the seasonal response between the Arctic and Antarctica that is simply due to the non-climatic fact that Arctic ice is usually constrained by a lot of landmass, whereas Antarctic sea ice can simply stretch out into the Southern Ocean (simulations actually have winter Antarctic sea ice going down faster in the future)

      There is also interesting spatial structure in how multi-year and first-year ice is being lost. In the Canadian Archipegalo for example, 1998, not 2007, was the lowest year on record. This is because of a circulation regime that helped prevent sea ice import from the Arctic Ocean (in addition to thermodynamic forcing), which was not operating in the 2007 season.

      It shows that people who only look at individual regions of the Arctic may miss out on a lot of what is going on, and vice versa, those who only look at the “Arctic average” will miss out on a lot of the interesting dynamics that is at work aside from the increase in temperature.

  15. Last year I tried to elicit, from the Scandinavian Troll Collective over at Deltoid, a wager regarding the trajectory of Arctic sea ice volume over the next several decades. They all put their tails between their legs and ran away wimpering.

    I wonder whether Watts might be prepared to put his apparent optimism about the sea ice on the table, and accept one of the options? I’m still prepared to put a thousand* US dollars down, with the same relative wager proportions. If Watts is prepared to engage me, I’d be happy to clarify the conditions here – they’d be essentially the same as those on that rat’s tooth thread at Deltoid.

    All he needs to do is to post a “yes!” on this thread and I’ll reiterate my conditions here.

    [*I would have used the same original cash amount as my original wager, had I not done a lot of mulling in the interim about the likely security (or rather, the lack thereof…) of tertiary economy instruments. Frankly, I doubt that most electronic accounts based on UD dollars are sound beyond even the initial 2025 milestone, and I have better things to do in the near term with that much of my own money, even though I am sure that I would win if only escrow accounts (or any other sort) could be guaranteed…]

  16. I wonder when that idiot anthony watts will ever start admitting he’s was wrong and see the damage that he’s done by contributing to climate change skepticism. There has been so much evidence of continued and accelerating change over the past couple of years – I wonder at what point he is going to change his tune?

    When the 1998 record is broken on he’s precious UAH record?
    When antarctic ice area turns and starts decreasing?
    When there is an ice free arctic summer?

    We’ve had records broken time and again all in a period with a weak sun and strong La-Nina’s over the past couple of years.

    What’s it going to take I wonder?

    • Horatio Algeranon

      I wonder at what point he is going to change his tune?

      Prolly not before Hell freezes over.

      But Horatio seems to remember reading somewhere (on WUWT?) that even Hell is in recovery, so maybe there’s hope.

      • Susan Anderson

        Hellzapoppin’ Getting hotter doesn’t sound gud. But all those guys think their wealth shows they are in God’s palm. They don’t read the New Testament it seems.

      • Hard to read Scripture–or honest stats, either–through the eye of a needle.

    • Greg, Think of it in the context of the experience on evolution.
      Every time a gap in the fossil record is filled by a new transitional fossil, creationists don’t say “oh, crap, there’s a missing link”, they say “wonderful, now there are TWO gaps!”

  17. Good questions, Greg.

    Maybe when energy companies have diversified enough that they are starting to make more money from non-fossil sources?

    Or is that just too cynical crazy naive all of the above (in varying senses)?

    • Andrew Dodds

      I get the impression that the more die-hard skeptics have now reached a position where even the economics of the situation don’t count; you might as well ask at what point the creationists will give up on the ‘young earth’ model.

  18. Here’s a different view of southern sea ice, a cycle plot of Uni Bremen extent by month for 1973-2011, with robust regression lines (more outlier-resistant and efficient than OLS with these data).

    Significant sea ice extent increases occur only in three austral winter months: July, August and October. In January (austral summer) ice extent significantly decreased. Other months show no significant linear trends.

    [Response: Here is a similar graph showing both hemispheres, using data from NSIDC. The Bremen data seems to have problems in the very early data, which would explain the apparent outliers very early on and argue for greater reliability of the NSIDC data in that time period. I disagree that robust regression is necessary in this case, and it’s certainly not more efficient.]

    [Response 2: Ah … I note the Bremen data extend back to 1973. In that case those very high early values may be accurate.]

  19. Setting aside the choices of time series, why are you sure robust regression is not more efficient? The robust regression standard errors are less than OLS for 7 out of 12 months, and very close for the others.

    Outliers occur both early & late in the UB extent data which makes robust regression look reasonable to me *for these data.* Even with Gaussian errors, the robust estimates should be about 95% as efficient as OLS, and better than OLS given heavy-tailed errors.

    [Response: I was referring to the NSIDC data. For the Bremen data, the extreme early outliers may indeed make robust regression a better choice. Then again — they may *not* — for reasons I’ll give in a post on the topic I’m working on at the moment. In any case, robust regression is certainly reasonable! Stay tuned.]

  20. The thing I find most striking about the figures Tamino has posted:
    ice trend in Arctic is most negative in late boreal summer early fall;
    ice trend in Antarctic is least positive in late austral summer early fall.
    Er, rather the trend I’m referring to is the difference between the pre-2000 and post-2000 years. And as Tamino points out, December shows the biggest increase in Antarctica — when the sun is highest in the sky there.

    I realize there are a lot of dynamics (inherited effects from previous years, feedbacks from open water reducing albedo, and others), but perhaps a simple explanation is adequate? CO2 prevents heat from escaping — thus the biggest global warming signals are observed after the peak of summer in each hemisphere. Solar changes are not responsible for recent acceleration of ice loss (especially wrt the current wimpy solar cycle), as evidenced by the warming signals being weaker at and before the peak of summer. Too simple?

  21. Halldor Bjornsson

    Thanks Tamino.
    The last graph is brilliant!

    [Response: I wish I could take the credit, but I was just extending a commenter’s graph to include the northern hemisphere.]

  22. I dont really know how long NSIDC will get away with missleading graphs .
    I agree with Tamino’s first graph , correct ! . Yet missleading !
    How about this statement . Every scientist’s and researcher’s dream is to have base period as long as possible for fair graphs .
    So if NSIDC wont give it to us , I ask Tamino to do it for us .
    Take base period 1980 – 2011 and the decadal loss of ice will shring by 30% .
    Basicly every year from now on when we get minimum around 5 mill , the running average will decrease by about 30000 sqkm , and after 10 years it will become irrelevant .
    So please Tamino , it wont take you long , overlay your first graph by a red line , which will represent base period 1980 – 2011 , ops I mean green line .

    • You seem very confused. NSIDC publishes absolute numbers for sea ice extent, not anomalies with respect to some base period so your complaint about the base period not being as long as possible doesn’t apply to them. As for Tamino, as he wrote, in producing the first graph “anomalies are computed relative to the entire time span”, i.e. the base period he used to convert the absolute numbers to anomalies already includes 1980-2011. That said, changing the base period wouldn’t change the decadal loss of ice at all – the same amount of ice would have been lost in, for example, 2001-2011 regardless of whether your baseline for calculating anomalies was 1980-2000 or 1990-2010 or July 22, 1997 any other time period.

  23. How about this statement . Every scientist’s and researcher’s dream is that people understand a base period is a convenience for comparison to some part of the past, and which base period line happens to be drawn makes no difference to anyone capable of understanding graphs.

  24. Our host did a real good paper on the weaknesses of straight linear regression analysis in the face of non-linear and climatically driven data. It’s well worth the read:

    Click to access FosterReport.pdf