Skip to content

The Arctic sea ice might be heading for a record low season. With six weeks to go until the end of the melt season the decline is comparable with 2007 – the lowest season recorded in the satellite era (post-1979). Since the late 1970s the Arctic sea ice extent has been declining at about 12 per cent per decade.

Raw data. Arctic sea ice graphs. Cryosphere Today. Arctic sea ice extent today.

However, it should be noted that the start of satellite measurements obviously caught a long-term decline already in progress and its relationship with anthropogenic factors is not straightforward.

A new study reported in Environmental Research Letters reports the problem of using computer models to study changes in environmental parameters like Arctic sea ice. An ensemble of climate models captures the trend (1979 -2010) but no individual model does, which is a worrying commentary on using model ensembles.

The authors of the paper comment on the reasons for this: “The potential sources of this discrepancy include: observational uncertainty, physical model limitations and vigorous natural climate variability.” This is actually another way of saying, “Everything we don’t know.”

The author’s solution to this dilemma? More models! This time looking at the influence of the Arctic oscillation (AO), Atlantic multi-decadal oscillation (AMO) and the Atlantic meridional overturning circulation (AMOC.)

They conclude: There is little evidence of a relationship between the AO and Sea Ice Extent (SIE) in the models. However, we find that both the AMO and AMOC indices are significantly correlated with SIE in all the models considered. Using sensitivity statistics derived from the models, assuming a linear relationship, we attribute 0.5–3.1%/decade of the 10.1%/decade decline in September SIE (1979–2010) to AMO driven variability.

The problem is of course that since satellite sea ice measurements started there has been only three decades so calibrating models with real-world observations is impossible as observations have not captured the extent of natural variability in such a short time. In their conclusions the authors admit to this saying that due to chance alone there might be a significant correlation between the AO and SIE for periods of about or less than 58 years. Then again, it is possible that a correlation in the real world might be due to chance, or it might not.

The graphs produced by the researchers are interesting, and I reproduce a few of them. Click on the image to enlarge.


They are scatter plots of the AMO index in March and September. In most of them I think given the scatter of the data points it’s very unwise to rely on the least squares fit to derive a relationship. It reminds me of deriving a cloud sensitivity relationship that could not be justified from the data.

The problem is, of course, that the 5 – 30% of the Arctic sea ice decline being due to natural variability will enter the media and become a firmer figure than the very tentative suggestion it is. Other research that attributes 40% of SIE variability to natural causes or even its entirety will be sidelined.