Arctic Sea Ice Extent, Area, and Volume

[donsutherland1]

On 7/30/2021 at 8:49 PM, Weatherdude88 said:

Slater Probabilistic Sea Ice extent showing 5.15 million square kilometers on September 18th. It's very possible NSIDC sea ice extent minimum will be greater than 5 million square kilometers, given current NSIDC area and compaction at high latitudes. Remember, we are ahead of 2013 in area for the date. 2013 had a minimum of 5.10 million square kilometers of extent.

Just as had been the case last year, the Slater model performed horribly. Had one taken the mean 2010-2020 decline from the date the model forecast was generated, one would have had a vastly better idea of minimum extent.

That figure was 4.4 million square kilometers (JAXA). The SIA had an implied minimum of 5.15 million square kilometers.

Now that Arctic sea ice extent has increased for 3 consecutive days, it is likely that the 9/12 figure of 4,612,915 square kilometers will be the 2021 minimum.

That will be the highest minimum extent since 2014 when the minimum extent was 4,884,120 square kilometers. The last figure at or above 5 million square kilometers was more than a decade ago in 2009.

 

[bluewave]

We now have a new AI based sea ice forecasting system.The paper was just published in late August. So their first official forecast was issued a few weeks ago on the ARCUS site. The September daily minimum was very close to their September average monthly forecast. The September average extent will come in a little higher which is always the case due to the higher average extent in early and late September. 

Name of contributor or name of contributing organization:
IceNet1

https://www.arcus.org/files/sio/32360/icenet1_2021_sio_august_individual_report.pdf

a) Pan-Arctic September extent prediction in million square kilometers.
4.75

IceNet is a sea ice forecasting AI system which predicts monthly-averaged sea ice probability (SIP; probability of sea ice concentration > 15%) up to 6 months ahead at 25 km resolution on an EASE2 grid. IceNet is based on a deep learning U-Net architecture, and has been trained on climate simulations (CMIP6) covering 1850-2100 and observational data (OSI-SAF SIC and ERA5) from 1979-2011. IceNet’s monthly-averaged inputs comprise SIC, 11 climate variables, statistical SIC forecasts, and metadata. IceNet is introduced in the following pre-print, with the study soon to be published in Nature Communications: https://doi.org/10.31223/X5430P. IceNet was also presented at the Oxford ML and Physics Seminar Series: https://youtu.be/JAKWhEU09Xo.
Brief explanation of Outlook method (using 300 words or less).
At each 25x25 km ocean grid cell in the Arctic and at each forecast lead time from 1 to 6 months ahead, IceNet produces a probability that the SIC will be less than 15% (no ice), between 15% and 80% (marginal ice), or above 80% (full ice). To compute the SIP map for this SIO submission, we sum the probability of the two ice classes to obtain P(SIC > 15%). IceNet comprises 25 different U-Net models, whose output SIPs are averaged to produce the final SIP forecast. To compute the SIE, we sum the area of grid cells whose SIP > 0.5.

 

 

[taylorsweather]

On 9/15/2021 at 6:57 AM, Taylorsweather said:

It may take several days longer to be sure, but Jaxa extent minimum may have been reached two days ago at 4.612 million km2.  We currently stand 54,000 km2 above that number after two straight days of increase.

Extent is now 92,000 km2 above its low three days ago.  So minimum may have occurred on 9/13.

[Typhoon Tip]

17 hours ago, Taylorsweather said:

Extent is now 92,000 km2 above its low three days ago.  So minimum may have occurred on 9/13.

I'm curious how much of these seasonal nadir stops are guided by the vagaries of the AO index leading... 

That's obviously a very complex question.  Multi-faceted factoring, most likely, determines.   Such as momentum under air - as in, SST and ice thickness, be it solar sensitivity ... cloud permutations... and holy hell.

But, perhaps the general negative versus positive index mode may encapsulate and thus correlate in general.  A season's stop-loss takes place whether higher or lower AO.  

Positive modes of the AO are warm for mid latitudes, but are cold inside the arctic domain.  Vice versa for negative index modes.   We all know that...  Those that don't want the Earth to die from GW ... would rather the summer be a +AO mode, and the winter to be oscillating between the ( - ) and ( + ) states...and on and sardonic so on.. lol.

But more seriously, when using the CPC history the curve of the AO spent the ballast of the summer positive. It did briefly go negative around the first of September, but only briefly so... If then considering momentum and transience amid the index, it may not be enough to significantly deviate from an established seasonal vector to keep the domain conserved.. I mean comparing over recent years .. perhaps 'offsetting' the bottoming momentum of the physical ice from realizing how far it 'could have' plumbed.  

 

[ORH_wxman]

On 9/16/2021 at 10:02 AM, donsutherland1 said:

Just as had been the case last year, the Slater model performed horribly. Had one taken the mean 2010-2020 decline from the date the model forecast was generated, one would have had a vastly better idea of minimum extent.

That figure was 4.4 million square kilometers (JAXA). The SIA had an implied minimum of 5.15 million square kilometers.

Now that Arctic sea ice extent has increased for 3 consecutive days, it is likely that the 9/12 figure of 4,612,915 square kilometers will be the 2021 minimum.

That will be the highest minimum extent since 2014 when the minimum extent was 4,884,120 square kilometers. The last figure at or above 5 million square kilometers was more than a decade ago in 2009.

 

FWIW, the slater model is predicting NSIDC extent, not JAXA. NSIDC tends to be a bit higher than JAXA....often 100k to 200k higher on the min.

It still performed poorly this year, though. Just not quite as bad as when compared to JAXA figures.

[LongBeachSurfFreak]

On 9/16/2021 at 11:33 AM, bluewave said:

We now have a new AI based sea ice forecasting system.The paper was just published in late August. So their first official forecast was issued a few weeks ago on the ARCUS site. The September daily minimum was very close to their September average monthly forecast. The September average extent will come in a little higher which is always the case due to the higher average extent in early and late September. 

Name of contributor or name of contributing organization:
IceNet1

https://www.arcus.org/files/sio/32360/icenet1_2021_sio_august_individual_report.pdf

a) Pan-Arctic September extent prediction in million square kilometers.
4.75

IceNet is a sea ice forecasting AI system which predicts monthly-averaged sea ice probability (SIP; probability of sea ice concentration > 15%) up to 6 months ahead at 25 km resolution on an EASE2 grid. IceNet is based on a deep learning U-Net architecture, and has been trained on climate simulations (CMIP6) covering 1850-2100 and observational data (OSI-SAF SIC and ERA5) from 1979-2011. IceNet’s monthly-averaged inputs comprise SIC, 11 climate variables, statistical SIC forecasts, and metadata. IceNet is introduced in the following pre-print, with the study soon to be published in Nature Communications: https://doi.org/10.31223/X5430P. IceNet was also presented at the Oxford ML and Physics Seminar Series: https://youtu.be/JAKWhEU09Xo.
Brief explanation of Outlook method (using 300 words or less).
At each 25x25 km ocean grid cell in the Arctic and at each forecast lead time from 1 to 6 months ahead, IceNet produces a probability that the SIC will be less than 15% (no ice), between 15% and 80% (marginal ice), or above 80% (full ice). To compute the SIP map for this SIO submission, we sum the probability of the two ice classes to obtain P(SIC > 15%). IceNet comprises 25 different U-Net models, whose output SIPs are averaged to produce the final SIP forecast. To compute the SIE, we sum the area of grid cells whose SIP > 0.5.

 

 

Great find Bluewave. Tremendous work by Tom Anderson. With continued learning accuracy will increase further. I’m not sure there is much hope for indigenous communities in regards to sea ice however. Villages are literally falling into the sea as a result of erosion caused by the lack of sea ice. Talk about disparity’s in climate change effects on given community’s.  

[donsutherland1]

4 hours ago, ORH_wxman said:

FWIW, the slater model is predicting NSIDC extent, not JAXA. NSIDC tends to be a bit higher than JAXA....often 100k to 200k higher on the min.

It still performed poorly this year, though. Just not quite as bad as when compared to JAXA figures.

Thanks. It still missed badly on NSIDC both last year and this year.

[LibertyBell]

3 hours ago, LongBeachSurfFreak said:

Great find Bluewave. Tremendous work by Tom Anderson. With continued learning accuracy will increase further. I’m not sure there is much hope for indigenous communities in regards to sea ice however. Villages are literally falling into the sea as a result of erosion caused by the lack of sea ice. Talk about disparity’s in climate change effects on given community’s.  

it's on governments to give land to these communities.  I also believe that people living in oceanic nations need to be taken in by other nations and given land to live on

 

[ORH_wxman]

On 7/2/2021 at 9:24 AM, ORH_wxman said:

Given we're now past 7/1, it is time for the annual forecast.

Quick update on standings:

2020: -430k

2019: -380k

2018: +390k

2017: +70k

2016: -180k

2015: +190k

2014: +320k

2013: +390k

2012: -590k

2011: -140k

2010: -310k

2009: +710k

2008: +280k

2007: -330k

 

 

You can see last year's forecast here:

 

And here is last year's verification:

 

As for 2021.....below is what the final area would be if we followed the path of every other year from here on out. An example is that if we followed the path of 2012 from here on out and lost another 4.3 million sqkm of ice like that year did post-7/1, then we would finish at 2.7 million sq km. The reason the forecasts are pretty accurate is that area loss post-7/1 is pretty stable over time going back to 1979. There has been a bias toward a little bit higher in the post-2007 world, but it is much smaller than the differences we see prior to July 1st (i.e., most of the area loss can be explained by what happens prior to 7/1.)

 

 

 

Two things stick out on the graph....one, is that any minimum above 4.00 million sq km is basically impossible no matter which path we follow. Nevermind that we haven't had an area min above 4.0 million sq km since 2006 anyway, but there isn't even a realistic shot at one this year like perhaps was plausible in years like 2014.

Secondly, the chance at a new record minimum is basically impossible too. The two largest melt years post-7/1 were 1989 and 2016. You can see following both of their paths still only produces a min around 2.5 million sq km....well above the record-low 2.23 million sq km minimum in 2012.

 

Given that information, I will use the post-2007 average as my baseline. I might even skew a little on the high side of that unlike last year because the forecast is quite stormy over the arctic basin for the foreseeable future. The EPS have a pretty strong vortex over the pole through mid-July which is wasting what's left of prime insolation season up there.

 

Taking a post-2007 average result post-7/1 purely at face value would yield a minimum of 2.9 sq km in 2021. I'll skew just a touch higher than that based on the forecast and current distribution of the ice. So for an area minimum, I'll forecast 3.0 million sqkm +/- 300k. That would place 2021 around 8th lowest.

Sea ice extent is a bit more fickle since compaction plays a larger part unlike area. But I'll go 4.3 million sq km on NSIDC extent +/- 500k. I have a larger range on the extent min since it has a much higher standard deviation than area.

 

I believe the extent minimum has been reached now that we are almost 140k higher on NSIDC than the 9/13 min of 4.7 million sq km...and the area minimum was reached on 9/1.

So time for verification based on the predictions from June NSIDC area.....

The minimum area was 3.17 million sq km on 9/1. This places 2021 as the 11th lowest area minimum. The more unique aspect of the area minimum was that it was the 2nd earliest area min on record. Only 1992 had an earlier min than the 9/1 date this year....the min that year was 8/31. The forecast above predicted 3 million sq km so I was a little low on my prediction but well within the error bars. I would consider this a pretty strong forecast.

The minimum NSIDC extent was 4.7 million sq km on 9/13. This ranks 12th lowest on record. My prediction of 4.3 million sq km was too low but still within the wider 500k error bars. Extent is notoriously harder to predict than area. The melt conditions in May/June correctly predicted this wouldn't be a top 5 melt season, but somewhat underestimated how much ice would survive. The favorable conditions in late July and August likely aided in some of that extra retention not able to be foreseen at the end of June. I'd still consider it a decent forecast but not as good as the area forecast was.

 

These figures may be revised slightly by NSIDC in the future as is often the case, but the overall standing of 2021 isn't likely to change much at all from any revisions. They are usually very minor.

[forkyfork]

almost time for 88 to come back and post about record gains

[bluewave]

Multiyear ice finishes at 2nd lowest on record with extent in 12th place.
 

 

 

[bluewave]

8 hours ago, Weatherdude88 said:

2020 was the second lowest sea ice extent minimum value on record. This is why 2021 northern hemisphere multi year ice ranked as second lowest at minimum (0 - 1 year).

Now that the freezing season has started, the remaining ice is now all multi year ice (greater than 1 year old). So if we finish next melting season in a similar place, multi year ice may be in approximately 12th place. 

The above statement, highlights the recovery from the 2020 melting season VS. the 2021 melting season.

The record winds during the winter transported the MYI into the Beaufort giving it more chance to melt out during the summer. 
 

https://nsidc.org/arcticseaicenews/

Despite September total ice extent being high compared to recent years, the amount of multiyear ice as assessed from ice age (Figure 5e) reached a near-record low, with an extent of only 1.29 million square kilometers (498,000 square miles), just slightly above the value of 1.27 million square kilometers (490,000 square miles) at the end of the 2012 melt season.

[Typhoon Tip]

I'm not sure what the significance of this may be ... how situational/conditional it is, just to this season, but that is a big difference in a single year to year/by date comparison of sea ice coverage.

The SSTs of the expose Beaufort Sea are naturally very cold, even in the summer... so it is "conditionally" set up for rapid re-freeze. But in order to do so, still requires the pattern in the air-ocean coupled environment to support the phase transition to ice. 

Obviously there are a lot of textured complexities in the total manifold of forcing mechanisms ... some that included Time as a function ... Comparing one season to the next is risky.   But that's a huge difference there between these years.   Eye-ballin', larger than TX

 

[StormchaserChuck!]

It certainly looks like -NAO/-AO, which we will have been in 45 days... other than that, much drier. 

[Typhoon Tip]

I recommend “Arctic Drift”

GBH

 

[NEOH]

On 10/17/2021 at 12:38 PM, Typhoon Tip said:

I'm not sure what the significance of this may be ... how situational/conditional it is, just to this season, but that is a big difference in a single year to year/by date comparison of sea ice coverage.

The SSTs of the expose Beaufort Sea are naturally very cold, even in the summer... so it is "conditionally" set up for rapid re-freeze. But in order to do so, still requires the pattern in the air-ocean coupled environment to support the phase transition to ice. 

Obviously there are a lot of textured complexities in the total manifold of forcing mechanisms ... some that included Time as a function ... Comparing one season to the next is risky.   But that's a huge difference there between these years.   Eye-ballin', larger than TX

 

Definitely a huge difference year over year.

 

[so_whats_happening]

Heyy about right on time to start showing the positives. Seems like data is always cherry picked anymore since you were focusing on how high Antarctica was over summer now at some of its lowest levels in satellite era and the fact that the Arctic was essentially on the road to recovery. Ice just getting obliterated down there as they head into summer. But hey the Arctic seems to be doing ok right! Never mind CAA is doing horrendous for this time of year which is usually the last place to really get hit in intense summers and its close by regions of Baffin Bay, Hudson Bay, Greenland Sea and Beaufort Sea just struggling hard right now.

So annoying to see when folks only post the good and not the bad but this stuff happens when the mid latitudes roast and get into some wild patterns of persistence. Anyway hope it stays this way cause one decent SSW or storms coming into Bering Sea or Atlantic front could ruin the day real quick. We may see a rather extensive halt in sea ice expansion come mid December if things play out as many seem to think it will. Lets watch and see.

[NEOH]

On 11/23/2021 at 7:01 AM, so_whats_happening said:

Heyy about right on time to start showing the positives. Seems like data is always cherry picked anymore since you were focusing on how high Antarctica was over summer now at some of its lowest levels in satellite era and the fact that the Arctic was essentially on the road to recovery. Ice just getting obliterated down there as they head into summer. But hey the Arctic seems to be doing ok right! Never mind CAA is doing horrendous for this time of year which is usually the last place to really get hit in intense summers and its close by regions of Baffin Bay, Hudson Bay, Greenland Sea and Beaufort Sea just struggling hard right now.

So annoying to see when folks only post the good and not the bad but this stuff happens when the mid latitudes roast and get into some wild patterns of persistence. Anyway hope it stays this way cause one decent SSW or storms coming into Bering Sea or Atlantic front could ruin the day real quick. We may see a rather extensive halt in sea ice expansion come mid December if things play out as many seem to think it will. Lets watch and see.

It's pretty remarkable the ice extent is closer to the 1990's average than 2016. Obviously a lot can change in the arctic quickly. 

[ORH_wxman]

Chukchi sea froze over faster than any year since after the 2012 monster melt. The difference is 2021 also has refrozen on the Atlantic side very fast unlike that 2012 autumn. That's why we're more like the early 2000s right now for extent

[so_whats_happening]

6 hours ago, NEOH said:

It's pretty remarkable the ice extent is closer to the 1990's average than 2016. Obviously a lot change in the arctic quickly. 

Its great to see of course. Although im not sure remarkable is appropriate yet. If we can hold an average season without any big shake ups and push into this summer holding strong I will start using remarkable (I know sounds pushy). For now it is good to see that the Arctic is holding it's own. I still worry as we go into mid December signs of ridging poking into the Bering and the Atlantic front, good thing for mid latitudes bad up there.

[FPizz]

17 hours ago, ORH_wxman said:

Chukchi sea froze over faster than any year since after the 2012 monster melt. The difference is 2021 also has refrozen on the Atlantic side very fast unlike that 2012 autumn. That's why we're more like the early 2000s right now for extent

I might be wrong, but it looks like Hudson Bay is still mostly unfrozen?  In past years, I think it is pretty iced up by now and that counts to the area #.  Area is pretty high considering the Bay has pretty much no ice.  

[ORH_wxman]

12 minutes ago, FPizz said:

I might be wrong, but it looks like Hudson Bay is still mostly unfrozen?  In past years, I think it is pretty iced up by now and that counts to the area #.  Area is pretty high considering the Bay has pretty much no ice.  

Yes Hudson Bay is lagging badly....the arctic ocean though and adjacent seas are way ahead of recent years.

Hudson Bay had no ice at this point in 2016, 2012, 2010, 2006, and 2003 too, so while rare, it isn't unheard of.

[taylorsweather]

35 minutes ago, ORH_wxman said:

Yes Hudson Bay is lagging badly....the arctic ocean though and adjacent seas are way ahead of recent years.

Hudson Bay had no ice at this point in 2016, 2012, 2010, 2006, and 2003 too, so while rare, it isn't unheard of.

Yes, Hudson Bay is noticeably lagging but it's northern reaches are beginning to freeze over as permanent night begins.   It won't take long to catch up.  

 

[bluewave]

The higher extent than recent years around the Chukchi Sea is a result of the record +EPO vortex since the late summer.

 

 

 

[IUsedToHateCold]

In terms of sea ice this year is a lot like 2010. It's also like 2010 in terms of IOD, La Nina, relatively similar SST anomalies. 

 

 

 

[LibertyBell]

Chris, haven't we been in the negative part of the PDO cycle for a few years now?

 

[so_whats_happening]

Im not Chris but yes with the exception of the stretch from about 2015 to 2017 (wonder what could have enhanced that, hmmm lol). If we go by the idea of about 30 years on average for signals to flip we should be close by about mid 2020's and by 2030 area we should be on a more positive regime. Not to say it follows it exactly though.

[so_whats_happening]

Arctic taking hits now extent and area have slumped for now but with prospects of ridging pushing into the Bering sea and ESS/ Chuchki as well as along the Atlantic front we will likely continue the stall or have ever slight gains over the next couple weeks its still cold up there just not as cold. The cold gets displaced over land most notably the North American side could see quite the cool down take shape outside of just Alaska and Western Canada. We have dropped back down again to the 2010's average so still keep running along that path since end of melt season. 

Globally though wow at the Antarctic right now. Some of its lowest values since 2016 taking hold which is dropping extent and area globally to almost the edge of the 2SD region. Still not as bad as 2016 as the Arctic has not taken a major hit this season, yet.

Im sure ill get the lil hotdog thing everyone seems to cherish so much on this site.

[stadiumwave]

 

[forkyfork]

1 hour ago, stadiumwave said:

 

climate change is cancelled!!!