Recent Arctic Sea Ice Decline Is Not Accelerating

It is time to return to examining the relationship between atmospheric carbon dioxide and the average surface area of northern polar ice in September of the previous year. I therefore once again conducted a regression analysis of the relationship between these two variables—over the period from 1979 to 2025—using carbon dioxide concentrations measured at Mauna Loa and ice area data from the NSIDC.

Since 2018 (using data from 2017), the basic idea underlying the analysis I have carried out annually is derived from the following premises: (1) according to the climate change hypothesis, climate change is an accelerating process; (2) models predict that temperature increases in the Arctic will be faster than in the rest of the globe; and (3) changes in polar ice area serve as a good proxy for temperature changes in the northernmost parts of the Earth.

In the figure below, the years marked in blue indicate those starting points from which annual measurements have demonstrated the existence of such a relationship according to the criterion I use (P < 0.01 in two consecutive years). The height of the bars shows how many years after that starting point this statistical significance was achieved.

The years marked in red are those for which no such statistical significance has been found. In these cases, the height of the bars indicates the number of years available for the analysis.

As my esteemed reader notes, during the early years of the time series the minimum extent of polar ice decreased in such a way that—according to the regression analysis—it exhibited a clear statistical cause-and-effect relationship with atmospheric carbon dioxide concentrations. Moreover, this relationship strengthened almost year by year up to the time series beginning in 2001.

After that, however, a change occurred: although atmospheric carbon dioxide concentrations continued to rise steadily, the change in ice area did not reach statistical significance until last year. At that point, the observational series beginning in 2002 also became statistically significant, but only on the basis of an observational record twice as long as that required for the series beginning in the previous year.

This year, what was new was that the time series beginning in 2003 also reached statistical significance. It can therefore be concluded that northern sea ice has indeed melted in a statistically significant manner in recent years as well, but that the process—at least as measured by its annual minimum extent—has not been accelerating, as climate models have predicted.

* * *

In this respect, it is also interesting that although Arctic sea ice reached a larger minimum extent this year than in six other years (2007, 2012, 2016, 2020, 2003, and 2024), it melted rapidly in October, November, and December, and its surface area reached the smallest December average in the measurement record last month. This is shown in the figure below (the early part of which admittedly raises questions, but for which I am unable to provide an explanation).

Previous thoughts on the same topic:
Observations on Arctic Sea Ice Challenge the Notion of Particularly Rapid Melting
Exceptionally Warm July Falls Short of Records
Exceptionally Warm July Falls Short of Records

The original blogpost in Finnish:
Pohjoinen merijää sulaa sittenkin