More catching up of sea ice....

Although warm for the first 10 days of Feb ECM/EPS forecast, it looks more favorable in the middle of February with recent model runs.

Given the abnormal drop in Feb 2025 it looks less likely now that we will end up below 2025 at least for the March mean, and perhaps closer to the 2017, 2018 which would imply around 13.7 million km^2. Although, who knows what might happen in March....

The PV split that may/may not happen after the strat. warming in the next 2-5 days is still uncertain but I'm leaning more towards a stretched vortex. The GFS analysis on stratobserve shows the upper levels of the PV being split while the lower levels keeping coherence, but its still very far out.. GEPS seems less bullish on a PV split from what I can observe. Similarly perhaps for ECMWF and the ensembles as far as I can tell...

@parhizj
The 5 nearest years to the value on 3 Feb 2026 13.29 which were
2018 13.28
2017 13.37
2016 13.37
2023 13.43
2019 13.43
The March average for each of these year ended up 13.68 to 13.97.
Perhaps suggesting that narrow range is likely.

However the 6th nearest 2012 ended up at 14.22.
Perhaps suggesting there is still considerable uncertainty. Should it also be taken to suggest the risk is on the high end?

@ChristopherRandles The subseasonal models I recall are fairly good at forecasting sea ice extent one month in advance... The Feb. CFS run for March’s SIE is near ~ 14.2-14.3 which still puts it around ~14 to 14.1 for ViSHOP mean based on my earlier eyeballing.... (hasn't changed much at all)

I also went through the 500mb anomaly maps for February from 1991-2025 using PSL's tool, and compared them to some of the subseasonal 500mb anomaly maps for February from tropicaltidbits (CANSIPS, CFS monthly and the more recent CFS weeklies also) . The closest analog seems to be 2007:

(I'm generally ignoring the difference in datasets and just focusing on the pattern)

The weekly CFS shows more lows across Northern Russia than the earlier monthlies and a stronger Aleutian high.....

Looking at Feb 1 to Mar 1 daily 2007 from ViSHOP:

(14.11 - 13.79) = 0.32

Looking at your reference years and calculating the same daily differences from Feb 1 to Mar 1..

0.62, 0.50, 0.60, 0.48, 0.76

The mean from these five is then 0.59.

Given that we have had a steady strong rise for the first few days of February and with the conditions for more ice melt last year per the reasons you mentioned below (with faster reversion in this case) I think then 13.2 + 0.32 = 13.52 is probably a fairly good but weak lowerbound for what is reasonable to start off March 1 daily value. The mean of March is usually around 0.04 less then March 1 value, so I think 13.47 then would be a decent monthly lower bound for the mean. If we take the 0.59 value + 13.2 = 13.79 - 0.04 = 13.75 for a middle point prediction.

So yes, I think there is a bit more room for upside given the we still have a high CFS prediction, but subjectively 13.7-13.8 seems like a good bet then if we discount the CFS prediction and rely on these ad hoc methods...

Looks like we quickly caught up to roughly where we were last year based on daily data.

Jan 1, 2025 was 11.91

Dec 31, 2025 was 11.88

Mar. 2025 monthly mean was 13.54 for reference, so based on that alone looks like no update is required to this market since the point prediction is 13.53...

ENSO however looks like it will return to neutral by ~ February, but given that last year we were in still in an El Nino for those months, I think based on that alone we should expect greater freeze up in the coming months for 2026 compared to 2025...

UFS is still quite a bit higher and has dropped only marginally since I last checked -- after subtracting about 0.535 or so for the difference in datasets, eyeballing it yields around 13.9 for VISHOP:

@parhizj ENSO is a long way from the Arctic. There might be some effect (hard to tell?) but if there is an effect I would suggest it may well be delayed quite a lot?

Referencing NSIDC since it has a nice easy chart we are still fairly low.

I think its still too far out to say whether we will get a polar vortex split from the medium range models, but if the rough timing from the last few runs verifies we will probably start to see a resulting plateau in the end of February .... (probably a trend similar to 2025 then but delayed a week ?). Though, there should be a bit more confidence in predicting it this time given how stretched it is already (at the very least it looks like a pair of anticyclones develop from the upper stratosphere downwards this time) ...

upper strat:

10 hPa winds from a pair of recent runs near the end of the medium range:

At the very least the timing of it will be a bit variable...

I noticed polymarket also now has a related market (but for NSIDC and for the daily max though) (the daily in NSIDC is technically a 5 day trailing mean). It is a different dataset, but at the moment it still seems a contender for the mean low this year even just referencing the daily max market....

https://polymarket.com/event/max-arctic-sea-ice-extent-this-winter

@ChristopherRandles I was surprised by your initial predictions being on the high end and ended up browsing data…

I only had a vague guess that it would be on the low end based on looking at some simple statistical calculations and the recent anomalies i mentioned on the gistemp market, but looking at a similar webpage that references a different dataset, some simple statistical indicators to date would seem to indicate we might be headed towards even a record low…

https://forum.arctic-sea-ice.net/index.php/topic,4329.msg434830.html#msg43483

At the very least the area sea ice extent matched my expectations from seeing the anomalies from gistemp (https://forum.arctic-sea-ice.net/index.php/topic,4329.msg434711.html#msg434711)

I looked at CFS (https://www.cpc.ncep.noaa.gov/products/CFSv2/CFSv2seasonal.shtml and it shows more towards the high end but accounting for some cold winter bias which it is known to have and examining the last few years of Nov init SIE predictions, not the highest bin (the bias by eye between the two datasets looks to be roughly in the +0.2 to +0.3 range compared to vishop from a quick glance), which would put the mean in the third or second highest bin? This was a bit surprising also…

I glanced at the NMME products also to figure out what’s going on

Aside from some below average temps north of the east Siberian sea… (because elsewhere it still looks like quite anomalous temps)

I can only speculate that perhaps above average precipitation in Dec and March will bring it up? I don’t know 🤷

Climate Prediction Center - Outlooks: CFS Forecast of Seasonal Climate AnomaliesCFSv2 Seasonal Climate Forecasts

@parhizj There is strong reversion to the trend. If there is low ice extent as winter approaches then there is lots of water that gives off heat easily whereas ice insulates the warm water so much less heat loss. The ice formation depends on the amount of heat loss.

Above is pretty solidly based on thermodynamics physics and the data shows it.

What is much less clear is whether after a year with a strong downward movement, do you tend to get a rebound the next year? Graph probably suggests no (1 yes 2 no?) and I may have overdone my expectation for a relatively high level this year.

An upward trend since 2015 I would suspect is spurious. On the other hand, there was a period 2002-2012ish where the arctic changed to much thinner less multiyear ice on the Russian side. Once that thick and MYI was gone, the thinner remaining ice melts out each year, and new slower trends have been apparent since then. So a latest trend line might plausibly be only getting down to about 13.9 rather than the ~13.7 if I used a straight line through March data from 2008.

The data for above graphs

Year, March, Prev Nov, Gain from Nov
2008-03-01 14.62376506 9.418130429 5.205634636
2009-03-01 14.45991381 10.0384542 4.421459606

2010-03-01 14.55340665 9.539251367 5.014155278

2011-03-01 13.97838974 9.339794233 4.638595509

2012-03-01 14.59582716 9.3974318 5.198395361

2013-03-01 14.38029497 9.046578 5.333716968

2014-03-01 14.17024394 9.615223233 4.555020702

2015-03-01 13.72900245 9.7491209 3.979881552

2016-03-01 13.83592787 9.5349185 4.301009371

2017-03-01 13.67942894 8.384798967 5.294629969

2018-03-01 13.72431094 9.1714044 4.552906535

2019-03-01 13.96533181 9.492540759 4.472791048

2020-03-01 14.10312587 9.006902233 5.096223638

2021-03-01 14.03425487 8.678768433 5.355486438

2022-03-01 13.95167955 9.5020577 4.449621848

2023-03-01 13.85361284 9.388955833 4.464657005

2024-03-01 14.21931497 9.331787833 4.887527134

2025-03-01 13.54402413 8.8062687 4.737755429

@parhizj Precipitation adds snow to ice already formed increasing insulation and reducing ice formation though this is more about thickening of ice already present than ice formation. Snow or rain on water doesn't form ice, while solid ice pellets might sound the right state to be helpful for ice formation, I would suspect that mixing of the water might be more of a hindrance to ice formation in at least some conditions. Further the clouds needed to produce precipitation also insulates and reduces outgoing radiation so less ice formation.

@ChristopherRandles thanks for notes…

I skimmed the arctic-rcc (which just got designated by the wmo last year) presentation slides (that happened last week) and their outlook (with the last 9 years reference period) is very coarse but it appears they are roughly predicting the same categorically below normal sea ice for winter 2026 as 2025 albeit with lower confidence, although it appears to be based on forecasts from a bit earlier on (early October at the latest).

So given the vishop data has the March mean for the last 9 years as 13.897 (if I did that correctly) then it would seem they are predicting it should be below this.

They don’t have a statement released yet but it should come out soon given that their conference was last week. The presentation is below that has their outlook on page 115

https://arctic-rcc.org/sites/arctic-prcc/files/2025-11/ACF16_presentations_day_2.pdf#page115

~

Page 42 also has a sea ice concentration chart, and the slide notes the quasi cycles of 2-6 years and it does look like it should bounce back up, but will it be this year as part of the trend?

@parhizj

That is better forecast than trusting me, and I have reduced my holdings.
Also
https://climate.rutgers.edu/snowcover/chart_daily.php?ui_year=2024&ui_day=314&ui_set=2
and
https://climate.rutgers.edu/snowcover/chart_daily.php?ui_year=2025&ui_day=314&ui_set=2
are perhaps a bit similar probably worse particularly re Northern Europe and maybe that area is big enough to persist.

Maybe I should reverse holdings rather than just reduce them.

@ChristopherRandles plenty of time to update at least in the months to come based on other models

I found the experimental UFS/NOAA model I believe that the slide deck i showed you referenced below:

https://www.cpc.ncep.noaa.gov/products/people/jszhu/seaice_seasonal/index.html

Should be superior to cfs per the paper mentioned on the page (skimmed it)

No product for the Canadian yet afaik.

Also has the archive of the netcdf files as far as I can tell, so you should be able to be a bit more accurate than eyeballing it, but just visually alone their 1991-2020 reference period is about 15-14.465 ~= 0.535 above the vishop mean for March for the same reference period, which eyeballing the chart below is between 14.5 and 14.6, so adjusting for the difference in datasets would bring it down to roughly 14.0, which was basically also my guess from visually adjusting down from CFS (from what I took as the bias visually from looking at the last few years… i.e. 14.2-0.2=14 was my guess last night)