Just look at 2012. Conditions for large losses were much better (and they happened), but 2012 ended at 3.63 million km2 (off the top of my head, I'm too lazy to scroll up). So, how can 2016, being behind 2012 so much already on the last day of the poll, after the melting season so far (clouds, clouds, clouds) and given PIOMAS numbers, satellite images, surface air temps, etc, etc, etc, possibly go below 3 million for the NSIDC September average?
I didn't vote below 3M - but I was in the 3.5-3.25 bin, which would be a record and is lower than most - so playing devils advocate, a few examples of unlikely but possible factors which could take it below 3M:
- the melt season could end two weeks late. (it did after all start two weeks early, so who is to say that's impossible....)
- The unprecedented level of fragmentation might permit rough seas in the CAB for the first time in memory, which would invalidate existing model assumptions re: surface temperature & salinity, and hence mis-predict the magnitude of end-of-season bottom melt.
- PIOMAS volumes for the last ~1M of modeled thickness may have always been systematically off by a few decimeters at this point in the season, or density & melt-rate for thin ice may be poorly understood, in a way that has been obscured in the past by forcing modeled thickness to zero for observed open water. (I'd venture that none of the very few buoys out there were initially placed on new or partially melted-out ice, which has accounted for much of the extent in recent years, and apart from those, there's radar, which is finicky and tells us nothing about density, satellite images, which tell us nothing about thickness, or even much about the difference at fine granularity between extent and area, and surface air temps, which tell us nothing more than "all the ice hasn't melted yet").
... In other words a less-than-perfect volume model would be undetectable until after the fact, and would make uniformly thin ice less predictable, paving the way for a potential late-season surprise)
- the widely-distributed open water and unusually (unprecedentedly) warm southern latitudes might engender more severe weather than heretofore observed (which to some extent seems to possibly be happening right now).
...In this context it's worth noting that the surface area of the Northern Hemisphere is ~255 million sq. km, much of which is the warmest it has been in recorded history. So although the likelihood of a change in the weather, encouraging unusual melt due to mixing between ~1% of the surface which is covered with very thin ice and the anomalous energy accumulated over the summer across the other 99% might be unquantifiable, it hardly seems "impossible".
I could go on, but hopefully the above is enough to help to reinforce something that is self-evident from the day-to-day accuracy of, for example, the ECMWF 5-day forecast: Science is quantifiable, but reality is chaotic. IMO we have for several years been in the regime where the difference between ice and no-ice quite plausibly comes down to chaos. If you have ever said at this time of year "it all depends on the weather" then at some level you know this to be true.
It's already a highly interesting melting season that might go really low even though weather conditions weren't conducive to melting most of the time when it mattered (June, July). Why isn't this worrisome enough? Why does it have to be some Hollywood scenario nobody foresaw, except for the guys who foresee it every year? [...]
IMO the most interesting part of the melt season will always be the end, because that is the time in the year at which the science of the models can most visibly come into conflict with the chaos of reality. If my own predictions have so far tended to come in on the low side it is simply a reflection of this... Hollywood scenarios are not required or expected!
Poll