Thanks for the answers. They bring clarity and more questions.
The various mentions of sea ice and sea ice area almost always refer to Sept extent.
Yeah, this is common practice in most climate papers I've read and maybe even in ASIF. I'm probably guilty of it on occasion. It is distressing to me because area is such an incomplete measure. Area and extent seems like pillars of ASI science when volume with shape information should be the pillar. I understand that volume is a much more difficult measure and area has larger, more agile set, but still. I believe it is wrong to use them as equivalent because it may lead to fundamental mistakes in the science.
The fingerprinting is how the variability is attributed. The same fingerprint is seen in the control runs as the runs under historical forcing and the historical observations.
This is what the paper says about the fingerprint:
Internal variability is determined from a long (1,800 years) control run of CESM1 with constant pre-industrial forcing (‘PI’ hereafter), as well as by the deviations of each ensemble member in 40-Forced from the ensemble mean.
After re-reading the paper (and failing to fully comprehend it) my questions increased.
This is my understanding of the experiment. They take a series of models, run them and calculate mean ASI area loss as a result of forcing. The ensamble (a sort of mean of the models) under predict the ASI area lost. Thus the difference between the ensemble and the observations must be "internal" variability.
I can agree with that if two assumptions hold. The author is beholden to the truth enough ( gotta love science) to say it more clearly than I could:
This approach makes the assumption that sea ice sensitivity can be observed without contamination by internal variability and that models appropriately capture the linkage between Arctic sea ice loss and global temperatures.
I'm convinced that neither of these assumptions hold, but that is based on my mental model of the Arctic. I will know over the next few years. If the Arctic recovers then it was internal variability. If it doesn't then we are in deep trouble.