I rather not give too much attention to fake skeptics, or climate risk deniers as I like to call them, but lately they are somehow finding it in themselves to come up with stuff that they think disproves Arctic sea ice loss. I've already posted about MASIE annual average nonsense (about which the last word hasn't been said), but in the past couple of days an even more spectacular and desperate attempt at downplaying Arctic sea ice loss has sprung up. I'm posting it as a reference, because I'm sure this will become a (short-lived) meme.
It all started with a blog post on fake skeptic blogger Paul Homewood's blog Not a Lot of People Know That (NALOPKT) about how the Danish Meteorological Institute (DMI) got rid of an old sea ice extent graph "simply because it gives the 'wrong' results". I decided to partake in the discussion because I knew what it was about and the reason the DMI removed that graph was because it had been replaced by another graph months ago, accompanied by the announcement that it at one point would be removed. The graph had clearly been neglected as it diverged more and more from other SIE graphs.
It was a nice exercise for me to drive my points home, but I had no interest in posting about it here, as it was a typical mountain-molehill type of misinformation, sprinkled with some suggestive conspiracy ideation, on the fringe of the fake skeptic echo chamber.
However, for some reason Anthony Watts elevated it to WUWT status:
The word 'inconvenient' you see there is spelled incorrectly and should read 'incorrect'. Let me explain in detail what this is about, and how pathetically these climate risk deniers are grasping at straws. It's all about these two DMI sea ice extent graphs, on the left the old SIE graph with a 30% threshold, and on the right the new one with a 15%threshold (as used by most organisations around the world):
Many months ago the old graph was replaced by the newer one. There was still a link to the old graph below the new graph, accompanied by a text saying it could continued to be be viewed "for a while" (see the image in this Great White Con blog post). It's quite clear from visual inspection alone that something was increasingly wrong with the old graph, and the logical explanation was that no one at DMI was correcting it because it had been replaced by the newer graph.
In fact, this was my first argument in the discussion on the NALOPKT blog, but this didn't convince the fake skeptics who rather believed that the DMI removed the plot because, and I quote (lest you think my imagination came up with it):
data does not correspond to the pet theory of funding institutions -> adjust the inconvenient data and ditch the rest (borrow Mike’s trick if necessary), do not try to come up with an scientific explanation.
How's that for conspiracy ideation, eh?
Watts voices the same 'concern':
Whether it is “skullduggery” or not as Homewood notes, climate science has this continuing habit of not showing adverse results, something Steve McIntyre has noted on more than one occasion through the years.
And so I decided to bundle a bunch of arguments to show that it's highly probable that it’s the old DMI SIE-30% graph that is incorrect:
1) The graph looks weird. There are two black trend lines, and a horizontal black line. There is no 2016 in the legend. The 2015 trend line contains strange dips that haven’t been corrected.
2) The trend line is much, much higher than all the other trend lines. Okay, if it was just a bit, but it’s way out there.
3) There is no SIE or SIA graph out there that looks remotely similar to the old SIE-30% graph.
4) Sea ice concentration maps, regional maps, satellite images, radar images, all show that sea ice cover is very low at the moment. There is no way it can be as high as the old SIE-30% graph suggests.
5) We know that DMI replaced this graph with the SIE-15% graph quite a while ago, and so there probably is no one to correct the old graph (like happened regularly in the past; I know, because alarmists would jump at the strange, steep dips). The graph is discontinued as announced many months ago.
That's the end of the story. It is highly, highly likely that the old DMI SIE-30% graph is incorrect, showing something that isn't even possible. Climate risk deniers are just too dumb/ignorant/dishonest to grasp this. First they jump on an incorrect graph that shows what they want to show, namely that Arctic sea ice is record high, and then they jump on the fact that the graph is removed by the organisation, as announced months ago, suggesting it is all a big conspiracy.
Mind you, Watts takes the time to highlight this nonsense, but he hides from his readers that new minimum records have been set last week for both Global sea ice area and extent, and the Arctic sea ice maximum record could very well be broken too in the coming weeks. None of that on WUWT. These things simply don't exist in the world of climate risk deniers.
If you find all this convincing enough, you can stop reading now. But below I will continue to explain the details by quoting extensively from my discussion with a couple of climate risk deniers on the NALOPKT blog. The difference between the 15% and 30% thresholds is quite an interesting thought experiment. Continue reading if you're interested, or spend your time more wisely. ;-)
Ron Clutz of the MASIE annual average recovery business then said something clever (he's not dumb):
Why is DMI’s 30% extent data the highest in 10 years, while their 15% extent is only average or lower? It may be that one or the other is in error, or maybe the drift ice is quite strong this year.
This seems to be the crux of the matter. Either one of the two graphs is in error, or they are both correct and for some unexplained reason show a massive disparity, because SIE-15% is currently lowest in the record, and SIE-30% is highest by a large margin. In this context, I'm quoting two more remarks by Anthony Watts:
Assuming both are right, we have a situation where there may be less ice in coastal regions and/or less 15% ice, but more of the 30% concentration. Given the fact that some of the mixed land/sea pixels can confuse the algorithm, there is good reason to think the 30% version is actually more reliable.
Since we are still learning about sea ice trends, factors, and effects (unless you are one of those who think the science is settled and nothing more to learn) it seems to me that this graph offers an important insight into change in the Arctic that can’t bee seen, that may be a precursor to change at the 15% concentration level.
Of course, Watts is just guessing from his gut, which is his speciality, and he clearly thinks that the old DMI SIE-30% graph is correct. He also clearly lacks the knowledge about Arctic sea ice to realize how nonsensical his statements are. For that you need to know what the difference between a 15% and 30% threshold is, and I must admit it requires a bit of thinking and logic.
So, what does that threshold thing mean? The NSIDC has a good explanation on its terminology page, but here is how I explained it over on the NALOPKT blog:
The Arctic is divided into grid cells. The size of these cells depends on the resolution (for instance, resolution for Cryosphere Today SIA is 25 x 25 km, for JAXA SIE it’s 6.25 x 6.25 km). A threshold of 15% means that when the total area of the grid cell that is covered with ice (in other words, the sea ice concentration), is 15% or higher, the total area (ie 100%) of the grid cell is counted. With a threshold of 30%, total ice-covered area being 30% or more, means that the entire grid cell is counted. A grid cell’s area isn’t counted if the sea ice concentration is less than the threshold.
What’s so counter-intuitive about the threshold business, is that a 30% threshold actually results in a lower number! Imagine you have a grid cell with a sea ice concentration of 25%, ie something between 15% and 30%. SIE with a 15% threshold will consider this grid cell to be 100% ice-covered. For SIE with a 30% threshold, however, the grid cell will not be counted.
See this nice diagram from the aforementioned NSIDC terminology page:
We know that the old DMI SIE-30% is highly likely incorrect, bordering on certainty, but let's assume that both graphs are correct, and for some reason the large disparity between the two is real. That's the line of thinking I then took to see where I would end up:
A large disparity would occur if there were many, many grid cells with a sea ice concentration between 15% and 30%, right? This would mean that those grid cells would be counted for SIE-15% and it would go up, but the SIE-30% trend line wouldn’t go up because those grid cells wouldn’t be counted.
But hold on, in this case it’s the SIE-15% trend line that goes low, and the SIE-30% trend line is high! It’s the other way round! Now, I’m confused.
So, let’s turn it around. What would happen if there were very, very few grid cells with a sea ice concentration between 15% and 30%? Well, they would be counted for both SIE-15% and SIE-30%. So, that doesn’t work either.
So, unless someone can come up with a hypothetical situation that could explain the large disparity, we have to conclude that one of the graphs is in error, as Ron suggested.
Luckily, one commenter (my guess he's a climate risk denier too, but not as outspoken as the others) Jaime took up the challenge and came up with such a hypothetical situation:
Using these graphs, we can only really compare one year to the next for the specific SIE being analysed. So what DMI are telling us is that the 30% more consolidated area has expanded in relation to previous years. They also tell us that the larger extent of sea-ice encompassing much of the penumbra of free-drifting ice at the fringes has not expanded much, if at all, maybe even shrunk a little bit, in comparison to other years where the 15% SIE has been measured.
So I’ll give you a hypothetical situation: 30% SIE has increased relative to recent years and 15% SIE has not. This is made possible by the fact that the smaller 30% SIE EXCLUDES a whole chunk of fractured sea ice area which is included in 15%. So the more consolidated core has grown whereas the larger extent including most of the penumbral, more fractured ice, has not. This can only mean that the 15-29% sea ice area has been ‘squeezed out’ in the process. Hence a significant increase in 30% SIE but no change – even a slight decline – in 15% SIE. As mentioned above, there may be sound geographical/meteorological reasons why this could be so.
Jaime rightly maintains that it's about a change between the relative disparity between the two calculations for the 15% and 30% thresholds from one year to the next, and I gladly go along with him on this route:
Jaime, thanks a lot for thinking along with me on this. I’m slowly getting tired of hypothesizing about this, and was hoping someone would make the effort of thinking about this.
So, you’ve come up with a hypothetical situation where SIE-15% stays the same and SIE-30% goes up. That’s great. Somehow I couldn’t get my head around it, but it makes sense now. Thanks for the explanation.
I’ve made this simplified illustration that probably won’t be posted here as the comments get narrower, so here’s the link. It’s showing two circles, equal in size, depicting two successive years. One has a core of 100% SIC and a band of 15-29% penumbra ice, as you call it, around it. The other – a year later – has a sea ice concentration of 30-100% everywhere:
From one year to the next, SIE-15% will remain the same as it doesn’t ‘see’ a difference between the core and the penumbra. But SIE-30% will go up because the second circle is much larger than the core of the previous year.
This is what you mean, right?
So far, so good. Basically what we’re talking about here, is what scientists call compactness. On the Arctic Sea Ice Blog I use a crude compactness measurement during the melting season which consists of dividing Cryosphere Today sea ice area by JAXA sea ice extent numbers. During the melting season it can give an idea of how many melt ponds there are on the ice pack and/or how compact the ice pack is, compared to other years (example).
Okay, so let’s have a look at other sources to see if we can see evidence of these geographical/meteorological reasons. Remember, we are looking for a band of penumbra ice, zones with 15-29% sea ice concentration. There needs to be less of that now than in previous years. In other words, other years need to show large bands of marginal ice with relatively low concentrations, and this year has to be super compact.
Here’s a comparison for of today’s Uni Bremen sea ice concentration map with those of other years in March 1st (sorry for the time difference, I don’t have time to manually retrieve all the SIC maps for Feb 21st). We’re looking for zones with light blue coloured ice, which stands for 25% according to the legend.
I don’t know about you, but I’m not seeing any great differences between this year and previous years that could explain the large disparity between SIE-30% going up and SIE-15% staying level. In fact, I’m seeing a couple of these zones in the current map (in the large version), in the Sea of Okhotsk, but also in the Barentsz Sea, and quite a lot (relatively speaking) off the coast of Newfoundland.
Also the compactness graph I mentioned above shows that this year isn’t extraordinarily more compact than usual. For that the 2016 trend line would have to be way above the others, near 100%, whereas it’s moving at the lower side of the pack.
The situation you describe might occur during summer (under extreme conditions for such a large disparity to occur), but it simply can’t happen during winter, I believe. Do you agree?
If yes, this would mean that we both agree that one of the graphs has to be incorrect.
But Jaime didn't agree and still believed his hypothetical situation could come about in real life:
It’s very difficult to just eyeball these maps and make a judgement as to whether the area of low concentration relative to more compact ice has decreased.
There’s plenty of open ocean on the Pacific side and likewise to the west of Greenland over to Scandinavia and Russia. Vast swathes of this open ocean were once regularly frozen over in winter before sea-ice started to decline. It doesn’t seem beyond the bounds of possibility that in these areas, in winter, previous years have seen a greater extent of scattered ice flows, which this year are much diminished but, at the same time, the area where ice is more consolidated (>30% concentration) is, compared to previous years, greater. This suggests an expanding inner core of sea-ice whose open water edges are, paradoxically, sharper and more devoid of an outer halo of less concentrated sea-ice. I don’t know whether there is a viable meteorological/oceanographical explanation for this, but as I say, I would not dismiss the possibility out of hand. It may be the case that ice has been dispersing much more quickly at the fringes this winter for some reason; maybe storm systems/ocean currents, I don’t know. Really, as Paul rightly says, we need DMI to tell us if there is a good technical reason to doubt that their >30% concentration data is faulty or whether they have just withdrawn the graph because it is ‘confusing’ people. We are just speculating on here, but thanks at least for agreeing that it is technically possible at least that both graphs could be correct.
So, I gave it another try and argued from another angle that his hypothetical situation is not what is causing the large disparity between the old DMI SIE-30% and new SIE-15% graph:
It’s not difficult to eyeball at all. Here’s the SIE-30% graph that was posted earlier by Pethefin:
The difference with the other black line (which I assume is for 2015) is approximately 1 million km2.
As we have noted already, there is no difference with last year on the DMI SIE-15% graph (and all other SIE and SIA graphs). In fact, this year is even lower than last year, but never mind. For the sake of simplicity we say it’s the same.
Now, remember the visualisation I made for your hypothetical situation? Two years, two identical circles, one with a core of 30-100% sea ice concentration and a penumbra zone of 15-29%, and the other with only 30-100% SIC everywhere.
From the perspective of a 15% threshold nothing changes from one year to the next (because the 15-29% penumbra SIC was counted anyway). Just like this year and last year which are practically the same.
From the perspective of a 30% threshold there is a change, namely that the entire 15-29% SIC zone has become 30-100% and thus is counted.
This means that compared to this year there had to be a total of approximately 1 million km2 of 15-29% SIC zones last year, right? They weren’t counted last year because 15-29% doesn’t surpass the 30% threshold. But this year they aren’t there, they have become 30-100%, and so the trend line on the old DMI SIE-30% shoots up.
Now, we can go to the Uni Bremen SIC archive and retrieve the SIC distribution maps for February 17th 2015 and 2016. These images are quite large (1517×2321 pixels), so it should be easy to see the 1 million km2 of 15-29% SIC zones in 2015 that aren’t there this year.
The blue/green of around 25% sea ice concentration according to the legend should really stand out on the 2015 map, but there is very little to be seen. In fact, the ice seems to be much more dispersed this year, with more blue/green in the Sea of Okhotsk and east of Svalbard.
Your hypothetical situation is interesting, but it can’t be applied to the current situation.
If this still doesn’t convince you, then think about Arctic sea ice area measurements for a minute. Instead of counting everything above a certain threshold as 100% ice cover, sea ice area is the total of percentages in all grid cells. Imagine, like you say, that last year there was a total of approximately 1 million km2 of zones with 15-29% SIC. This would mean that sea ice area would have been much, much lower than this year. But actually, this year SIA was 585K lower than 2015 on February 17th, according to Cryosphere Today!
I’m looking at Arctic sea ice every day. There isn’t a viable meteorological, geographical or oceanographical explanation for this, not this year, and not during winter, I don’t think. And I’m not sure if it could happen during summer either. Not this much. A small relative disparity, perhaps, who knows, but not this big. And for your theory to work the ice at the fringes would have to converge/compact a lot, not disperse.
No, both graphs can’t be correct at the same time. One of them is wrong, and it is highly, highly likely that it’s the old, discontinued, uncorrected DMI SIE-30% graph.
I'd like to leave it at that. :-)
A lot of words for just one paragraph of conspiracy ideation and the belief that there is a lot of ice in the Arctic right now that is invisible to satellite sensors. But alas, that's how it goes. A lie goes halfway around the world and so on...