So, if you would want to tease out the contributions of AGW and natural variation to Arctic sea ice loss, what would your "designed research paradigm" be? How would you go about it?

I love it. Thank you Neven.

In general, it is very hard to tease out the contributions of AGW and natural variability to Arctic sea ice loss. As my teacher told me a long time ago : If it were easy, I would have done it myself.

However, Ding et al 2017 DOES provide a tool that we can use to 'tease out' particular variables out of a set of correlated variables.

They did it with this formula, where they 'teased out' the influence of Z200GL (the geopotential height over Greenland) :

B(x, y, t) = β(x, y) × Z200 GL (t)

Here, note that Z200 GL (t) is a scalar time-dependent variable.

Instead of Z200 GL (t) we may be able to 'tease out' the AGW influence using this very same method, but instead of Z200 GL we can insert another scalar time-dependent variable, like the global warming variable LOTI :

https://data.giss.nasa.gov/gistemp/graphs/We would just need to calculate :

B(x, y, t) = β(x, y) × LOTI (t)

and of course, need to re-calculate β(x, y) for each grid point and each variable against LOTI.

Fairly trivial to do if you have the data set up.

After that, we could re-run Exp.6. and see how sea ice is affected (by LOTI).

What we would have done in that case, would be to 'tease out' AGW out of the Arctic variables, and that would be a start (to see what AGW influence is on Arctic sea ice).

Of course, this method assumes that there is no "Arctic amplification" during the summer months.

But it would be a start...