So I've done some number crunching using the NSIDC links Wipneus was kind to provide (I should have had them already), in order to arrive at some better extrapolation of Antarctic SIE at the minimum, based on regional historical data. I am using NSIDC data for Jan 26 as current date.
Regions: bham: Bellingshausen&Amundsen Seas, wedd: Weddel Sea, indo: Indian Ocean, wpac: West Pacific Ocean, ross: Ross Sea
On current date 1997 had a total extent of 3279k, declining to 2264k at minimum on Feb 27. Regional minima occurred within several days of each other, notably wedd started growing about a week before the general minimum. Had all minima improbably occurred on the same day, the total minimum would have been 2125k.
Currently, for Jan 26, 2017 has a total extent of 2702k, or about 576k less than 1997. Extrapolating the total naively using the 1997 decline to minimum would give 1688k for this year.
Looking at each region separately:
bham: now 504k, was 462k on 1997 same date. Fell to 182k at 1997 minimum, declining 280k
wedd: now 1149k, was 1299k on 1997 same date. Fell to 1184k at 1997 minimum, declining 115k
indo: now 277k, was 287k on 1997 same date. Fell to 124k at 1997 minimum, declining 163k
wpac: now 641k, was 569k on 1997 same date. Fell to 426k at 1997 minimum, declining 144k
ross: now 132k, was 661k on 1997 same date. Fell to 348k at 1997 minimum, declining 313k
The real shocker is ross, which by itself explains the whole difference between the years and is totally unprecedented. Previous lowest for the date was 424k.
Using the following guesswork:
bham declines 280k to minimum, same as 1997 though it currently has more ice
wedd declines 120k to minimum, or 220k *
indo declines 163k to minimum, same as 1997
wpac declines 144k to minimum, same as 1997 though it currently has more ice
ross declines 30k to minimum, or 100k **
* wedd – looking at the five lowest years for the current date with 919k to 1,130k, three declined 223k-246k to Feb 27, one declined 126k, and one actually grew by 101k (but declined 120k to that year's minimum which came about much earlier). 1997 also declined 115k. Note: timing of wedd minimum vs. total minimum can make a major difference in the outcome.
** ross – any-date minimum was 124k and we are already at 132k. 1997's 313k decline is impossible. Looking at the ten lowest ross years for this date (424k-712k), averaging their decline to Feb 27 gives us about 200k, still impossible. So it's either almost nothing 30k or almost everything 100k, anyone who can analyze actual ross ice situation is warmly welcome to comment. Can it go to zero?
*** The above can be improved for wedd and ross by checking each year's decline to its own minimum date, instead of using Feb 27 for all. The above can be improved for all regions by looking at other similar years and averaging their declines to minimum instead of just using 1997.
The bottom line gives us the following NSIDC SIE minimum based on regional extrapolation:
Low decline scenario - 737k, down to 1,965k
High decline scenario – 907k, down to 1,795k
Midpoint decline – 822k, down to 1,880k
Following this statistical guesswork, I think <2m has a much higher probability than >2m.