I know that I've burned up much of my credibility during the recent exchange with Qinghua Ding and Eric Steig.
So I don't expect anyone to take the following notes seriously.
First of all, the subject is important.
If Arctic sea ice declines to record levels, this paper suggests that 30-50% is due to natural variability, and thus that it could revert if the "weather" changes.
However, I can't deny that I believe Ding et al 2017 still may have some serious flaws which were summarized best by Michael Mann :
https://twitter.com/MichaelEMann/status/841362467603255298"study doesn't support that conclusion. Atmospheric circulation changes may have anthropogenic component."
Let's take this one step at a time.
Ding et al 2017 has a claim that :
Internal variability dominates the Arctic summer circulation trend and may be responsible for about 30–50% of the overall decline in September sea ice since 1979.
This claim consists of two parts :
1) Cause :
Here, we present evidence that trends in summertime atmospheric circulation may have contributed as much as 60% to the September sea-ice extent decline since 1979.
Now, this claim is based on the Exp.5. and Exp.6. differences.
Here, Ding et al 2017 runs a model between two 'climate' regimes : One with the ERA climate data, and one with ERA data adjusted to exclude the Z200 geopotential height trend since 1979, and also adjusted 'temperature' and 'LW downwelling radiation' and 'humidity' variables, which are adjusted to the extent that Z200 changed. Their regression method causes these variables to knock out 65% of the long term trend. This all suggests that Z200 long term changes are the "cause" of the trend in temperature, humidity and LW downwelling radiation. But it may very well be that the "cause" is the long term trend in 'temperature' which may very well be (entirely or partially) anthropogenic (see Mann's note).
Which brings us to 'attribution' :
2) Attribution :
Ding et al 2017 suggests that 70% of the 60% ice loss from (1) is caused by natural forcing, leading to their final conclusion of some 40% natural cause for ice loss in the Arctic.
This (70%) is based on their experiment 7 and 8.
There, they took out the long term trend in 'high latitude wind', assuming that (30%) was anthropogenic.
However, the 'high latitude winds' may very well be caused by increase in Z200GL geopotential height increase. After all, increase in geopotential height over the Arctic leads to reduced cyclonic winds. If you assume that Z200GL increase is largely (or half, according to Ding et al 2014) caused by natural variability, then these long term changes in 'high latitude winds' that cause 30% ice loss may be caused by 'natural variability' and the remaining 70% may be caused by anthropogenic forcings.
Which turns the tables around since now 'natural variability' only causes 30% of the 60% (from point (1)) and thus only 18 % could be caused by 'natural variability', and only 9 % if half of Z200GL is caused by natural forcing.
Needless to say that there is much to argue about this paper's findings.