The surprising tail end of the Arctic sea ice volume graph

The subject of previous post was the claim that there would not be any Arctic sea ice left after 2022, unless drastic actions would be taken. This claim was attributed to a speech at the University of Chicago by James Anderson in 2018. The Forbes article (that first-hand reported on this speech), referred to it as the speech from “Thursday” and the article was dated January 15, so that speech was likely held on Thursday January 11, 2018.

The Forbes article also mentioned that Anderson earlier received the Benton Medal for Distinguished
Public Service. With those two things in mind, I quickly found the pdf version of the slides of that speech (unfortunately not the text of the speech itself). The event was called the “Benton Lecture”, was indeed held on January 11, 2018 at the University of Chicago and presented by James Anderson. There were also things that I recognized from the Newsweek fact check and its sources, so that seems to be a match.

The main topic of the lecture was not the Arctic sea ice itself. Its title was “Climate, Chemistry, Technology, and Society: a University Responsibility” and the loss of Arctic permanent sea ice volume was only used as “Exhibit A” of what looks like evidence of irreversible changes to the climate structure.

This graph on slide 18 immediately caught my eye:

Benton Lecture 2018: graph on slide 18

This graph shows the evolution of minimum September Arctic sea ice volume. That makes sense, Anderson told in an AP fact check that Arctic sea ice volume was the basis for his prediction. That graph is an opportunity to see what data he was working with and how he envisioned this decline.

The data source of that graph is the PIOMAS dataset, luckily this is the same dataset that I used in some previous posts. When I first viewed that graph, I recognized most of it, but at that tail end it seemed a bit off. Looking closer, I saw a couple other things that didn’t add up.

The first discrepancy is found at the very end of the graph. That last data point of the curve is situated somewhat after 2015 and I expected it to be 2017. This because Anderson stated in the same AP article that his prediction of an ice-free Arctic was based on a projection of the latest observations of the Arctic sea ice volume. The last data that was available when Anderson gave that lecture in January 2018 was the minimum volume that occurred in September 2017. It would be logical that he used data until 2017 to illustrate his case.

However, that last point in this graph couldn’t possibly be the 2017 minimum volume. The 2017 volume in the PIOMAS dataset (4532 km3) is slightly higher than the 2016 volume (4400 km3). Just look at the actual graph of the minimum volume until 2017:

Chart 0022b: Minimum Arctic sea ice volume from 1979 until 2017

The Benton lecture graph doesn’t end with a slight uptick. On the contrary, it decreases at the end. Therefor, I initially assumed that this last data point must have been the 2016 value (the 2016 data point in the PIOMAS dataset is lower than the 2015 value).

The second discrepancy is the value of that last data point on that graph (which I assumed was 2016). It is so incredibly low, it is at roughly the same height as the lowest point on the graph (2012). In reality, the 2016 value is 4402 km3 while the last value in the Benton lecture graph is at roughly the same height as the 2012 value (3673 km3 according to PIOMAS). That is quite a difference.

But then, that last data point is situated so very close to 2015 on the x-axis. Is this really the 2016 data point? To figure this one out, I imported the image of that graph into graph digitizing software. After calibrating the axes, I found that the last data point has an x-value of 2015.3. Looking at the recognizable peaks and valleys, the other values coincide nicely with an actual year (for example, the x-value of the lowest point in the curve is indeed 2012). Then why does this last value (slightly) overshoot 2015?

That is odd, because the graph represents the annual minima and there is only one minimum per year. Is that last data point then 2015? That would make even less sense than 2016. In reality, the 2015 volume is even higher than the 2016 value. That would make the difference even greater.

The third discrepancy is the peak following the 2012 minimum. That specific peak in the Benton lecture graph is much lower than the 2008-2009 values, but I don’t see anything similar in the PIOMAS dataset. This peak in the PIOMAS dataset is the 2014 value and it is at roughly the same height as the 2008-2009 values, but in the Benton lecture graph it is well below it. This doesn’t make any sense.

I was puzzled by all these discrepancies and tried to explain them. I initially assumed that the 2015 value in 2017 might have been inaccurate and corrected later? I had to abandon this train of thought after looking at volume data on It became clear that the current 2015 value in the PIOMAS dataset was unchanged since September 2015.

I also tried to find Arctic minimum volume graphs with such a weird 2015 value, but my searches turned out to be fruitless. Other graphs that I found online didn’t seem to have a problem finding the correct 2015 value. So, although the Benton lecture graph claimed that it had the same source as other graphs online, it seems to be one of a kind.

The breakthrough came when I focused on the small graphs on slides 6, 7 and 19. Until then, I skipped over them because my initial assumption was that these were just thumbnails of the page-filling graph on slide 18.

However, looking closer, that isn’t entirely the case. The graph displayed on slide 19 is indeed a somewhat smaller and narrowed version of the bigger graph on slide 18, but the thumbnails on slides 6 and 7 are somewhat different.

This is the thumbnail from slide 6 (enlarged):

Benton Lecture 2018: enlarged thumbnail on slide 6 with orange trend line

Looking closer, it becomes clear that the last part of the graph is in orange instead of the yellow of the rest of the curve. The thumbnail on slide 7 also has on orange decreasing line at the end.

Then came the revelation. This is what shows up when I zoomed in even more on the last part of those curves:

Benton Lecture 2018: detail of tail end on thumbnails slide 6 vs 7

Hey! Those orange lines are clearly just DRAWN rather crudely behind an existing graph!

There is also a tiny difference between those two thumbnails. In the thumbnail on slide 6, the orange line meets the yellow line just below its highest point and it is comes very close to the x-axis. In the thumbnail on slide 7, the orange line meets the yellow line at its very top and it doesn’t come that close to the x-axis as the one on slide 6.

It is the title of those two thumbnails that shows the last piece of the puzzle (my emphasis):

Minimum September Arctic Sea ice Volume 1979 – 2013

From that moment on, it all started to make sense.

Let’s go back to the graph on slide 18 and check whether that last part of the curve is also drawn behind the 1979 – 2013 graph.

Zooming in on the last part of the graph, it is less clear because it is also yellow, not orange like the thumbnails on slide 6 and 7. Yes, this yellow line from 2013 on is also drawn behind an existing graph that went until 2013:

Benton Lecture 2018: detail of tail end of graph on slide 18

The line is drawn in the same color yellow, so the blending of the two lines is much better than in both thumbnails, but the trajectory until 2013 clearly has a black shadow on it, the part after 2013 doesn’t have any shadow whatsoever…

Now everything fell into place. We are not looking at data until 2017, not 2016, not even 2015. We are looking at data until 2013 and someone drew a yellow line to illustrate what comes after 2013…

It also explains why the graph ends at the year “2015.3”. Drawing a line is less accurate than plotting a graph with actual data, so the person who drew that line overshot 2015 slightly.

It also explains why the peak after the 2012 minimum is lower than the 2008-2009 values. This because that peak after 2012 on the Benton lecture graph is not 2014 as I expected, it is 2013 and the person who drew that line ignored all real world data between 2014 and 2017.

Let me get this straight. The guy who wanted to convey in a 2018 lecture that the Arctic minimum volume trend is evidence of irreversible changes and the Arctic would very likely go ice-free after 2022, illustrated this by an arbitrary line added to a 2013 graph, supposedly representing the data after 2013, while data until 2017 was already available back then?

via Trust, yet verify

April 28, 2023 at 03:43PM

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s