Skip to content

DMI Explain Why Graph Was Withdrawn

February 25, 2016

By Paul Homewood   




It’s only taken them a week, but finally DMI have published an explanation as to why they withdrew their old style graph of 30% ice coverage:



DMI has removed the old sea ice extent graph to focus the attention on the new graph that is based on data from an improved algorithm.
However, the removal of the old sea ice extent graph was done at an unfortunate time, namely, during a period where it seemed that the new and old ice extent plots disagreed (see figure 1). Naturally this has led to discussion among our dedicated followers, about the "true" ice extent. The apparent elevated sea ice extent in the data from the old extent algorithm was an artifact, caused by a new and higher resolution coast mask.

Most of our sea ice extent followers know that the old plot includes a coastal mask, inside which sea ice was is accounted for. In summer 2015 this mask was refined and the masked region was subsequently smaller, thus leaving mo re area for classified sea ice and open water. The difference in masked area, before and after summer 2015, is approximately 1.4 million km2. This corresponds to difference of the blue coast lines in figure 2, showing the old and new coastal masks in the left and right panels, respectively. The difference may be difficult to detect on the figure, but the area is quite significant. The increasing sea ice extent that is caused by the new coast mask is not great during summer, because sea ice has a relative short line of contact with land during summer. But the new and finer coast mask will result in increasingly more sea ice, compared to previous years during winter, as the coast line with sea ice contact is increasing. This is the reason for an increasing sea ice extent during current freeze-up period, relative to previous winters. A comparison of the 2015/2016 sea ice extent with previous years does therefore not make sense (see figure 1-left).


    Plots of sea ice types February 22 2015 (left) and February 22 2016 (right) using 2 different coast masks.     The mask used before summer 2015(left) is wider than the new mask (right), corresponding to approximately 1.4 million km2 less area under the new mask.


Because of the deprecated status of the old plot in the past year, DMI has not been monitoring these irregularities. The old plot should, of cause, have been removed when the mask was replaced. DMI apologizes for the confusion and inconvenience this has caused.



As I stated at the time, my real complaint was that they withdrew the graph with no explanation. It is only the power of the internet that has now forced this clarification.

Hopefully, other such organisations will learn that they too owe the public a much more open and transparent approach in future.

  1. Ian George permalink
    February 26, 2016 1:26 am

    I note the new graph only goes back to 2012. Can they (or will they) adjust the new graph to show details back to 2005, as was the original ice extent graph.
    First we had changes to the GISS maps adjusting the global temps, then the sea temp adjustments, then the sea-level adjustments, then the use of satellites to ‘estimate’ cyclones/hurricanes strength, and now this. And it’s all worse then we thought.

  2. Sceptical Sam permalink
    February 26, 2016 2:12 am

    I can’t seem to find where DMI clarifies that the “Mean Value 1979-2000” has also been adjusted to line up with the change or whether it’s using old data as a comparator for the new? If so, then that’s truly wacky – but no more than I’d expect with this unscientific “science”.

  3. Pethefin permalink
    February 26, 2016 6:37 am

    They finally did what they should have started with. And it was not just because they discontinued the graph, the reinterpretation of the fall 2015 should have been explained right away. In other words, an apology was in order, although it was a surprise that they gave one, climate science is not exactly known for admitting mistakes.

    • AndyG55 permalink
      February 26, 2016 8:10 am

      So the 30% DMI graph was NOT incorrect, like some AGW tart claimed. 😉

      • Pethefin permalink
        February 26, 2016 8:40 am

        Well, I would say yes and no. The graph was not “broken” in that the algorithm was not “broken”, as the alarmists claimed. The graph was “wrong” in that due to the changed coastal masking, the un-updated algorithm was monitoring a larger area than before and therefore showing too high amounts of sea-ice compared with the previous years (measured with the old masking). The graph was in other words no longer measuring the same thing as it had in the past.

        The interesting thing with the explanation is the fact they updated the coastal mask, which as far the DMI tells us is not being used for the 15 % coverage. So what is it used for? Are they keeping the rest of their 30 % products? If yes, will they explain their methods in reconciling the old measurements with the new?

      • AndyG55 permalink
        February 26, 2016 9:14 am

        I do have to wonder WHY they would play with one of the longer standing CONSISTENT methodologies by changing the algorithm and masking

        No need to unless you don’t want that consistency of methodology.

        They had already started a “new” methodology, why alter the old one ????

      • Pethefin permalink
        February 26, 2016 9:27 am

        I think the scientific AGW-community is beginning to realize that something needs to be done in order save credibility. Mann et. friends suddenly admitted the “pause”
        Now Stern admitted that the climate models need to be replaced:
        Unfortunately, this will most likely mean that the climate science methodology will become quite inconsistent, even more so than before. What they should do is to embrace skepticisms as all good scientists do, and start questioning.

      • Pethefin permalink
        February 27, 2016 9:38 am

        I should have added a smiley after the Stern reference since unsurprisingly a devote alarmist that he is, Stern naturally demands more alarming models 🙂

  4. Streetcred permalink
    February 26, 2016 9:03 am

    Have they recast the historical data as well?

  5. David Richardson permalink
    February 26, 2016 9:22 am

    The explanation re the masking is fine, if a little belated. BUT it now leaves no 30% output and surely no real historical comparison? Is that to come?

    As Ian George says above – we are being presented with various re-writings of history and even re-wiring of the present.

    • AndyG55 permalink
      February 26, 2016 9:25 am

      I reiterate from above….

      I do have to wonder WHY they would play with one of the longer standing CONSISTENT methodologies by changing the algorithm and masking

      No need to unless you don’t want that consistency of methodology.

      They had already started a “new” methodology, why alter the old one ????

  6. February 26, 2016 12:02 pm

    Now, which shell is the pear under?

    • February 26, 2016 12:03 pm

      That would be: which shell is the PEA under.

    • Sceptical Sam permalink
      February 26, 2016 1:10 pm

      Pea, pear? Who cares.

      It’s always the green one, on the left.

      Every time.

  7. A C Osborn permalink
    February 26, 2016 1:45 pm

    As I commented on WUWT the MASIE data is more consisitent with the 30% position than the 15% position and MASIE is supposed to be the latest Dog’s whatsits.

  8. Pethefin permalink
    February 27, 2016 9:39 am

    Does anyone know whether the DMI uses a coastal mask of some kind in their 15 % coverage product?

Comments are closed.

%d bloggers like this: