**Whereas summer sea ice is too heterogeneous for large-scale rapid shifts in sea ice area to occur over a few years, **

... Is the above highlighted quote the explanation (or part of it) ?

My guess is that the authors are thinking about the cumulative effect of (a) the gradual complete loss of MYI compounded by (b) the even more gradual regression in the date at which re-freeze begins.

Even if we lost all the MYI in the next couple of years, the date at which re-freeze commences isn't going to change much in the short term. The effect of this is that the new ice in some areas will still clock up significant Freezing Degree Days compared to those areas in which the refreeze didn't get going until (say) February.

I think that accounts for the heterogeneity which is being discussed in that quote.

As for this year's varying behaviour viz-a-viz volume & extent, I really wouldn't like to hazard a guess. There's always going to be a certain level of disconnect between these metrics.

I don't know if you noticed, but on the ASIB a couple of months ago, one of the contributors hypothesised that, since PIOMAS and NSIDC gave differing "projection down to zero ice" dates, then one or both had to be in error.

That seemed an astonishing failure to grasp basic maths: if the area is dropping, and the thickness is dropping, then it is axiomatic that the volume should be dropping even faster. Obviously it is physically impossible for matters to continue such that there is zero thickness whilst there is still area present, and vice-versa.

What will happen is that there will be a discontinuity (the rapidity of which will be interesting to learn) in the decline rates of the various metrics. I know you are familiar with all this, but I have appended three simple charts which I knocked up a while ago to demonstrate this effect to someone I know.

In each (extremely simple) case, I have assumed that the length, breadth and thickness of an ice mass each decrease in a perfect linear fashion.

Case (1) has these three values declining such that they would all go to zero in 10 years.

Cases (2) and (3) have the length & breadth (and, consequently, the area as well) decreasing at an unchanged rate. However, in Case (2), the rate of loss of thickness has been decreased, but has been increased for Case (3).

The format of the graphs may look somewhat unusual to some, so a quick explanation may be in order.

The primary Y-axis shows Area, whilst the secondary Y-axis displays volume. The X-axis is used to show thickness. However, as each data point on every data-series is one year apart, the separation of each data point can be considered as being a proxy for the passage of time.

In Case (1), there is no abrupt discontinuity, and the curves for each metric smoothly coalesce at the zero point. However, Case (2) demonstrates a discontinuity in the thickness metric, whilst Case (3) has a similar discontinuity for area.

As mentioned earlier, the physics of the actual ice loss will vary enormously from this simplistic mathematical representation. The losses will be anything but linear - and that's without introducing any noise into the situation.