Thread by @mikarantane

🚨New paper alert 🚨

I'm excited to announce that our open-access paper "The Arctic has warmed nearly four times faster than the globe since 1979" is now out in @CommsEarth: nature.com/articles/s4324…

#ArcticAmplification #Arctic

Thread of the main results:

1/13

We report that the Arctic has warmed four times faster than the globe in the last 43 years.

The four-fold Arctic amplification (AA) ratio is higher than what is generally reported in literature and media.

2/13

One reason for the higher ratio is that we defined the Arctic properly (with the Arctic Circle, 66.5°N), and calculated the trends since 1979 when satellite data became available.

With these parameters, the observed AA ratio is 3.7-4.1, depending slightly on the dataset.

3/13

More importantly, the unabated warming in the Arctic over the last two decades has raised the moving 43-year Arctic amplification ratio from two to four. Earlier estimates of 2/3x warming may be thus outdated.

4/13

Regionally, the warming has been even stronger. Areas in the Barents Sea near Novaya Zemlya have warmed up to seven times the global average, as recently reported in @SciReports by @Ketil_Isaksen et al: nature.com/articles/s4159…

5/13

Or course, the observed AA ratio depends on the area definition of the Arctic and the time window used for the linear trends.

The higher the Arctic latitude threshold, the stronger AA, as a larger proportion of the area is ocean (where AA is the strongest).

6/13

Furthermore, we found evidence that climate models, when considered as a group, cannot reproduce the observed four-fold AA ratio over 1979-2021.

We used ensembles from CMIP5, CMIP6, MPI-GE and CanESM5 models.

Similar result was recently reported in GRL: agupubs.onlinelibrary.wiley.com/doi/10.1029/20…

Those models that show plausible Arctic warming trend typically have too much global warming. In contrast, those models that simulate global warming close to that observed, generally have too weak Arctic warming.

8/13

Even when allowing a longer time window for the models than for observations (because internal variability is not in phase in models and observations), the difference between simulated and observed 43-year AA ratios was found to be statistically significant (except for CanESM5).

Arctic amplification has seasonal cycle, with lower AA in summer and higher AA in late autumn/winter.

CMIP6 models systematically underestimate the observed AA, especially during the melting season.

A particularly interesting finding was the anomalously high AA in April.

10/13

It's important to stress that the model-observations discrepancy is smaller when using a longer time window for the warming trends (i.e. 1950-2021).

The plots below show the AA ratio as a function of the starting year of the linear trends.

11/13

This indicates that internal climate variability may have contributed considerably to the observed AA ratio over 1979-2021. It is also possible that climate models have biases which makes them systematically underestimate the Arctic amplification. More research is needed.

12/13

End. Huge thanks to my co-authors Alexey Karpechko, @anttilip, @kalle_nordling, @OttoHyvarinenIL, Kimmo Ruosteenoja, Timo Vihma and @AriJLaaksonen.

This was a group effort with diverse contribution from various units of @ilmatiede. So glad that the paper finally came out!