Arctic sea ice free (extent)

[Juan C. García]

I will appreciate your comments about your answer 

[gfwellman]

If the question had been CT area, rather than NSIDC extent, I would have gone with 2017-2020, but I chose 2021-2030.  No deep science, just an eyeballed linear extrapolation.  Such extrapolation could be way off for longer time periods, but seems reasonable for 10 years or so.

[Juan C. García]

I agree with your forecast, gfwellman. It could be good to establish an equivalent “ice free” CT area and/or PIOMAS volume, before making a poll with these concepts.

[ChrisReynolds]

Actually I've just removed my last message and edit and retract my vote.

I missed the 'monthly' criteria for the September minimum. In which case, I'd say early 2020s.

[Neven]

I'm with gfwellman. But if I had to bet, I would say the first day that sees a sea ice area below 1 million km2 could very well be before 2020.

[crandles]

Some extrapolations just in case people want them for guidance:

Gompertz fit of extent first below 1m in 2023 (RMSE .436)
Exponential fit of extent first below 1m in 2019 (RMSE .435)
'Linear plus exponential' fit of extent first below 1m in 2018 (RMSE .434 but 4 parameters instead of 3)


Exponential fit of volume reaches 0 in 2015
Gompertz fit of volume gets down to 232 Km^3 by 2020


1m Km^2 extent and volume of 232 Km^3 so thickness of .232m is about the normal ratio of area to extent, perhaps that is just a little thinner than average. So, if you believe the gompertz extrapolation of volume then whether 2020&before or after 2020 looks rather borderline to me perhaps marginally in favour of being under 1m km^2 extent in 2020.

The trend looks slightly faster than gompertz fit of volume, indeed the 'linear plus exponential' fit suggests it is faster than the exponential fit.

This suggest I should go for 2017 to 2020 and yet there are the models suggesting a tail and I think I do believe there will be some sort of a tail. I doubt it will be as fat as the gompertz tail based on evidence so far.

I would be happier with a 2019 to 2022 range but I think it might just make it by 2020 so I will select 2017 to 2020.

[Dromicosuchus]

Based solely on extent, I would have probably guessed 2021-2030 or 2031-2040, but considering the PIOMAS volume estimates, and the confirmation of those estimates by Cryosat-2, I think it'll probably be mostly gone by 2020 at the very latest, and more likely will have sunk below 1,000,000 km^2 by 2015-2017.  Since that covers more of option A's range than option B, count me in the 2013-2016 camp--with the caveat that I think it's extremely unlikely that the sea ice'll hit <1,000,000 km^2 in 2013, and pretty improbable that 2014 will have that dubious honor.

[Jim]

I voted for 2017-2020, but I really want to see this years minimum value. If it is another record, then we could see ice free conditions very soon (2015-16), but if there is a 'rebound' (like 2008 and 2009) then the date could be some time away - we might see a pattern like...

Big Drop
Rebound
Rebound
Big Drop
Rebound
Rebound
..... and so on.

I really think this year will give us a very big clue as to the future course of the summer ice melt.

[Edheler]

I wish the time periods allowed for a bit more choice. I chose 2017 to 2020 but wouldn't be surprised by it happening in the early 2020's.

[FrankD]

Extent is a bit tricky because of the three major metrics its the one most dependent on weather conditions (aggregated over the season).

I seem to be on the seams of the time periods in the poll. Like crandles, if there'd been a 2019-2022, that would match my sweet spot.

My gut, which has more nerve endings that my brain (cite: Colbert, 2006), together with some volume curve-fitting, roughs out:
Daily minimum <1 M: ~2015-2017
30 day running min <1 M (ear Sep to early Oct): ~2018-2020
September mean (which was the question) <1 M: 2019-2022

Conservatively, opted for 2021-2030, but I think it will be right at the beginning of that range, and would not be at all surprised if it occurred in the period before that. I would be surprised it if occurred before 2017.

But the Arctic is just full of surprises these days...

[Artful Dodger]

2016, +/- 3 years.

That implies a 5% chance of SIE < 1M km² in 2013, a 50/50 chance in 2016, and 95% by 2019.

[DungeonMaster]

2013-2016
The Artic is mostly with thin ice on AARI now, and I fear that the last thick ice north of Greenland and Canada will start detaching from the coasts and moving as a whole towards warmer, ice-free areas - and start melting there those next years.

I'll closely watch the north-east coast of Greenland this summer to see how the last purple, thick areas resist - or die.

[birthmark]

A couple of years ago I said 2017. I don't remember what reason (if any) that I picked that year, but I'll stick with it.

[Jim Williams]

I'll agree with Lodger, though with extent you can have almost no ice and still use up a lot of sea surface.  I'll take the soonest category on the "aggression principle."

[anonymous]

Mmh, am I the only pessimist here? Remember how the big storm took out a million km² in a week? I would go with 2016 +/-1, but there is no category. Look, here's again PIOMAS raw plus anomaly, what might prolong that story after 2020?



I understand this poll is about extent, but how much extent is left given volume is zero?

[icebgone]

Ice melt reaches new records each year from 2013 through 2017.  Export of isolated and detached MYI by 2017 lowers remaining extent below 1,000,000.  Perennial ice free summer extent by 2030 and permanently ice free arctic before 2050.  Warmer water from Pacific and Atlantic plus albedo change trumps cold air and weather.  Another year like 2007 or a sudden release of co2 and/or methane in the arctic and all these dates could accelerate significantly.

[Carex]

I try, mainly for discussion with those with questions, to phrase my opinions in terms of climate, not weather.  So I like to say in the warmest year of the next ten or thirty.  Given that, I'll pick 2017-2020.  That would be within 10 years after the 2010 volume drop.  I'd say the chance is 90% to 95%.  And the chance of pre-2040 as 99.999%.  My reasoning is based completely on a visual review of photos and graphs.  A gut feeling that ocean warming could overwhelm this but do not have enough background in ocean currents or physics to assess my own opinions.

[Wipneus]

Given the choices I would guest about 2017-2018 frame.

That would be for a low straight average ice period of any 30 days.
The  NSIDC algorithm may compute a mean that is well above a straight average and the minimum may be in August.

[benjamin]

If one accepts the conclusion of Shepherd et al (http://xa.yimg.com/kq/groups/18383638/836588054/name/Science-2012-Shepherd-1183-9.pdf  link kindly provided by Chris Reynolds) then 2,700Gtons (+/-980Gt error bar) more than usual (well, more than deposited in precipitation actually) has been dumped into the Artic between 1992 and 2011 from the Greenland Sheet  and if one believes the CRYOSAT Volume as being similar scale (no more than 5,000Gt) to that then you could argue that pretty much all the "sea" ice has already gone and we are "only" left with the excess from Greenland which in the record breaking 2012 must have contributed another net 400 to 600Gt I would guess.
But I know that is not the intent of the question.
Nevertheless, the CO2, the Methane the temperature are all going one way - up - and all are accelerating so I think the PIOMAS  exponential volume is likely right of 2015 or 2016. But I think that the ever growing extra ice contribution from Greenland may push it back to 2017

[ChrisReynolds]

Whoa there Benjamin, 

A lot of the mass loss is actually from melt, and most of the calving is from the East and West Coasts, which are virtually sea ice free at the end of summer. We also know that ice growth is very vigorous in the Arctic over winter, and that the Drift Age Model (which tracks and ages parcels of sea ice) agrees pretty well with the ASCAT satellite system - without any input from glaciers.

[ChrisReynolds]

I don't know what the relationship between date of extent minimum and extent is. I only really do area. But in terms of area there is no relationship, there is no trend in date of minimum despite area going down markedly. However I'm assuming that this is also the case for extent.

I'm also assuming most loss and gain increase will occur around the period of maximum rate of change of extent, where length of ice front is higher than at minimum, and to a lesser degree around the minimum.

I've taken the JAXA extent series for 2003 to 2012, September. For each daily value I have subtracted the extent at minimum for that year. This has the effect of pulling down the extent so that each year hits zero for one day, the minimum criteria for which one might claim a daily extent sea ice free.

Then I've taken the average extent for each of those adjusted Septembers. Only one year 2003 then has an average extent below 1M km^2, that's 2003 at 0.944M. 3 years have an average of >2M km^2

Having assumed that the date of min extent is set by insolation and is invariant as is the case for area, I've set all extent values prior to the date of minimum to zero. In other words the ocean is completely ice free entering September and refreeze starts after the date of minimum for each year 2003 to 2012.

There are 10 years, 4 have an average of well below 1M, 2 have an average of just below 1M, 4 have a September average of well above 1M.

So I prefer a later rather than an earlier date after 2020 for sea ice extent average to be less than 1M km^2.

[gfwellman]

I was under the impression there was a slight trend to later minimums (driven by albedo feedback).  Of course, that mechanism can't push the minimum much past the fall equinox.  Presumably when we reach ice-free summers, the ice-out date will be before the equinox (and move progressively earlier) while the refreeze gets a little later due to thermal inertia.

I don't think there's any trend in the date of the maximum.  It might be getting more variable because the thinner pack can be pushed around more by winds.

Anybody have actual statistical results?

[ChrisReynolds]

Well I don't do extent, I consider it to be a virtually useless metric, and JAXA is too short to do this.

But a scatter plot of CT date of min vs CT area at min gives a slope of -0.0013, and R2 of 0.0001. Similarly the time series of dates of min has a slope of 0.038, R2 of 0.0029. Both based on linear trends. Therefore I conclude that the date of min is set by the solar cycle with some wiggle due to random weather.

[Artful Dodger]

...while the refreeze gets a little later due to thermal inertia.

Hi Greg,

Thermal inertia is an annual factor yes, but over decadal time-scales, increasing meridional heat transport will transform the Arctic Ocean into a lobe of the Atlantic, which also does not freeze during Winter.

[Gray-Wolf]

I think that we all suspect that "The end will come fast" so we should focus on the 'middle' i.e. the melt out of the bulk of the pack? With the thin ice ,this year, allowing for the scale of disruption we have seen so far I have to think that the bulk of the F.Y. ice will have gone by early Aug. Should we lose a section of the remaining multi-year over the same period (via transport through Fram) then we will have a lot of water to eat away at the remaining multi-year through Aug/early Sept?

Remember that some of that 'multi-year' also took a battering last year and so is really only a skim of older ice with a keel of FY ice. I believe that this factor was overlooked last year and allowed for a certain amount of 'solid' multi-year ice to retreat far faster than was expected?

I'd love to see a 'rebound' of ice levels this year but something keeps telling me that this is an unrealistic expectation this time around. In 08' and 09' part of the 'slowdown' of the melt season was the 'collapse and spread' of the remaining perennial ice (as prof Barber saw) and so a portion of that 'rebound' was really a collapse. Now we are without that perennial and so have no large floes to collapse and spread out (bolstering extent) only multi-year with FY ice below?

I think we wait until mid July and see just how much 'solid core' of the pack remains (previous years have the C.T. image of '100% ice' tracing out the pack that will remain come Sept?) I suspect that even this central core will prove less than 100% come Aug (Neven's 'Ice Islands' of Central Basin/Fram for the remaining ice come Sept?).

Any quirk of the weather over summer (high melt synoptics/high transport synoptics) may prove enough to finish off the remaining pack should they arise? The ice will continue to thin ,year on year, but another 'perfect storm' year arriving would finish off the ice over any of the years of the ice's final volume decline. In 07' we were told the 'perfect storm' years were cyclical around a 10 to 20yr time frame with the two previous of 07' being 10yrs apart. Does that favour 2017?

[john_mann]

Hi there all - sorry, new to this format.

I read a comment over the main articles about acoustic/seismic monitoring of ice sheet cracking. Can anyone elaborate on this?

[Nightvid Cole]

I picked 2017-2020, but I wouldn't be surprised if we have one last minimum slightly above 1 M km^2 again sometime in the early 2020s (and the deniers will proclaim it the beginning of a recovery). And the very next year it'll probably crash down to something like 500,000 km^2.

[Juan C. García]

John Mann:
I read a comment over the main articles about acoustic/seismic monitoring of ice sheet cracking. Can anyone elaborate on this?

Maybe they will help you on the topic: Arctic sea ice - Records and oddities.
http://forum.arctic-sea-ice.net/index.php/topic,35.50.html

[misfratz]

I picked 2017-2020, but only because the question specified monthly average extent. I'm fairly confident that the daily minimum will breach the 1 million square km mark before then.

The basis for this is the validation of PIOMAS with the Cryosat observations. Before then I'd been fairly sceptical of PIOMAS on the basis that it was a model with unknown errors and biases. The decline in sea ice volume is more consistent then the decline in extent, so I judge the possibility of a temporary plateau to be much less.

[Gray-Wolf]

I to did the 2017-2020 slot but that was before we saw this ongoing fragmentation event?

Last year GAC12 made a mess of the ice surrounding the central pack but did not get into the main body proper.

Should a continuation of this fragmentation lead to the whole basin becoming fractured then any storms this summer would disrupt the central ice much more and lead to a lot more inundation of the central area (and higher melt rates).

This fragmentation event has also raised the spectre of high losses into Fram from the 'rump ice' of the Canadian Archipelago. If we continue to see the forecast pushing ice along the trans Arctic current whilst the fragmentation itself pushes further into this 'rump ice' then we could well lose a good portion of our best ice into Fram before May's start to the melt season proper?

All in all I'm begining to wonder whether recent events have upped the odds for this year being the first to have the pack drop below 1 million?

[fred]

My guess is 2014. I don't think there is anything linear about this.

Not only does open water earlier have more heating, open water has more currents. I think it's like dropping a pat of butter in a soup: as long as the soup isn't stirred, the pat remains insulated by its melt a bit. Stir the pot and the melts almost instantly (i.e. relatively instantly: within margins of error for size of pat, etc.)

If melting is on the same scale more or less as last year, it seems that the chances of another storm dealing serious damage, if not a death blow, seem larger each year. I think the edge or tipping point for loss of summer ice could be well more than 1 million skms.

[Artful Dodger]

I recall these chilling words from a Century-old poem:

Turning and turning in the widening gyre
The falcon cannot hear the falconer;
Things fall apart; the centre cannot hold;
Mere anarchy is loosed upon the world,
The blood-dimmed tide is loosed, and everywhere
The ceremony of innocence is drowned;
The best lack all conviction, while the worst
Are full of passionate intensity.

William Butler Yeats, 1919, from The Second Coming.

Yeats was Not a Climate Scientist, but 2019 may not be far off...

[Bob Wallace]

There's no apparent recovery during this freezing season.  Monthly PIOMAS volume figures continue to track downward on an exponential curve.  The ice appears to be fracturing very early this year.  No melt-slowing factors have been observed.  Volume of transportation continues to hold steady, even with thinner ice being transported.  More heat is available to melt less ice.  The ice is thinner which allows mechanical forces to have a larger impact.

I can see nothing but a crash coming.  And, as it has been amply pointed out, one cannot have area or extent with zero volume.

I would have liked to have been presented with a 2013 to 2019 option, but having to pick I took the shortest.  About six times in the last twenty-five years we have seen annual volume losses that would take us to near zero at today's level.  It would not take an extremely unusual melt to make 2013 the year that the ice melted, only a "normal big melt".

[OldLeatherneck]

I picked 2017-2020 solely because it is the "engineering mentality" I still have which causes me to vote on the conservative side.  That being said, I wouldn't bet someone else's money on my vote, because I can envision the various scenarios that have the melt occur by 2016, if not sooner.  After all, when you look at the poll results of the first 40 voters, many of which are some of the 'most knowledgeable individuals, on planet earth, when it comes to all things arctic', 85% predict an ice-free summer by 2020.

[DavidR]

2013-16,. You can't really  go past the fact that the PIOMAS volume has been declining by about 800 cu Km per year over the past 9 years. AS of 2012 it was down to  3.4K in September.  As this rate is increasing there is no reason to  consider the decline anything but  exponential.

This suggests 4 years will easily  see off the 1 Million sq km extent,  and with just one 'warmer' year we could see the ice completely  gone. 

[Vergent]

In 2002 they said we would have an ice free arctic in 100 years.

If this alleged trend were to continue, the poles would be ice-free in summer just under 100 years from now. Imagine the summer cruise opportunities!

http://www.worldclimatereport.com/archive/previous_issues/vol7/v7n22/feature.htm


In 2006 it was 70 years.

Arctic sea ice could disappear within 70 years

http://www.greenpeace.org/international/en/campaigns/climate-change/impacts/habitat_loss/


late in 2007 it was 50 years.

According to the latest computer models, says Schauer, the Arctic could be ice free in less than 50 years...

http://www.ens-newswire.com/ens/sep2007/2007-09-14-03.asp

I figured they had a systemic error. A simple linear extrapolation of that error(in 2008) led me to the conclusion that we would be ice free in 2013.

Nothing since then has led me to doubt that conclusion.

[Vergent]

In 2011 it was 20-30 years, call it 25. . . Graph those points out and you get a graph that would make Wipneus blush.

[Artful Dodger]

I figured they had a systemic error. A simple linear extrapolation of that error(in 2008) led me to the conclusion that we would be ice free in 2013.

Nothing since then has led me to doubt that conclusion.

Thanks for sharing that, Vergent.

I also became aware of the trend in accelerated predictions during the Spring of 2006. At that time, I gave a talk to a lay audience on the validity of Arctic sea ice projections. I was met with blank stares and fidgeting. Crickets chirping.

Strong denial is another occurrence that has accelerated in the last 7 years. The most well-informed members of the public gather enough information to have a straw man for every reported scientific observation.

FWIW, in Spring 2006 I also predicted 2013 as the first sea ice free Summer, purely on my observations of the trend in decreasing predictions. Hah! Yet another 'Death Spiral'. My Audience was unimpressed.

I will return to speak to that group during the first sea ice-free September. I expect more blank stares and fidgeting. And the chorus of 'so what?' and 'it's a good thing'. However this time, I also expect a storm outside, as well as inside.

[Vergent]

x axis = year

y axis times 10 = "years until ice free"

It seems that the predictions are off by a factor of 10.

[Artful Dodger]

x axis = year

y axis times 10 = "years untill <sic> ice free"

It seems that the predictions are off by a factor of 10.

Hi Vergent,

There it is. The decline in Arctic see is IS linear. 

"Nothing to worry about. Move along. Resume SHOPPING!"

[gfwellman]

The decline in Arctic see is IS linear.

Actually, if you look at that graph closely, it hints at a Gompertz 

(Yes, I realize with 4 data points almost in a straight line, fitting anything other than a straight line isn't justified.  I'm just being a little funny because we like our Gompertz fits.)

[Artful Dodger]

The decline in Arctic see is IS linear.

Actually, if you look at that graph closely, it hints at a Gompertz 

Hi gfwellman,

Are you MAD?! 

This is CLEARLY a PARABOLA! The 3^rd Recovery happened so quickly, we're already moving onto a 5^th (of gin). 

[Vergent]

The decline in Arctic see is IS linear.

Actually, if you look at that graph closely, it hints at a Gompertz 

(Yes, I realize with 4 data points almost in a straight line, fitting anything other than a straight line isn't justified.  I'm just being a little funny because we like our Gompertz fits.)

Arctic sea ice could disappear within 10 years as global warming increases speed of melting

http://www.dailymail.co.uk/sciencetech/article-2187346/Arctic-sea-ice-disappear-10-years-global-warming-increases-speed-melting.html

Gompertz this, and Dodger, your parabola needs some Viagra.

[JMP]

Sometime around 2012's minimum I came to (the somewhat alarming) conclusion, just based on the graphs that I was seeing, that 2015 looked like the most likely date.  But, what do I know? just read what I can -- so of course it's merely an edumakayted guess. 

I see two wild cards in the deck: one is weather (which is perhaps the most understood) and the other is that AGW is completely new, and although it may parallel past warming events in some ways, there isn't a way to study by any exact method these things that have never happened before (the future cannot be observed etc).  This thought ties in with the comments above about "systemic error" etc.

So, I think before 2020 is a safer bet, but yes I do think it could even happen this summer, (and as long as I'm making guesses mind you) I do think that it probably wont dip below 2.0 and around 3.6 is 2.2 and around 4.0 might be a more likely minimum for 2013.  (It'd probably be better for PR if the minimum was lower this year but so-far nature plus anthropogenic input has not - afaik - moved in quite this linearly obvious way and the minimums have mostly been followed by rebounds (this way of thinking may be scientifically described but I would not describe it as scientific  just as luck would have it thinking etc. ))

I voted 2013-2016.       

[OldLeatherneck]

What I'm going to be looking for is the extent on or about the 1st of June, which is about the time the entire arctic begins to have sunlight 24 hours/day.  If by some chance the extent is 10.5 Million KM² or less and the loss during June significantly exceeds other previous years, there will be a chance to reach a minimum below 1.0 Million KM² this year, but I doubt it would occur soon enough to have an average below that for the entire month of September.

[Espen]

I think the vote should be: When do believe the Arctic Sea will be sea ice free below 1.000.000 km2 for the first time during the melt season?

[JMP]

I think the vote should be: When do believe the Arctic Sea will be sea ice free below 1.000.000 km2 for the first time during the melt season?

That was my understanding of what we're voting on, but just as measured by the NSIDC, and was thinking their standard measurement is a monthly average.   After checking I see now that their graphs are based on a 5 day average? and last years minimum of 3.41 million square kilometers was based on a daily average?  Is that right? -- Not that it would change my vote.

[crandles]

The monthly data is at

ftp://sidads.colorado.edu/DATASETS/NOAA/G02135/Sep/N_09_area.txt

3.61 for Sept 2012 average.

The minimum 5 day average was 3.36855
 
Links to both monthly and daily figures are on the graphs page:
https://sites.google.com/site/arcticseaicegraphs/

(Should we have a link to above graphs page somewhere in the header of this blog?)

[JMP]

Thanks for the clarification!   It would be helpful it there was an obvious link to the graphs from the forum.  On the home page of the forum might be sufficient if it's easier.  Some of us need all the help we can get.   

[Vergent]

Thanks for the clarification!   It would be helpful it there was an obvious link to the graphs from the forum.  On the home page of the forum might be sufficient if it's easier.  Some of us need all the help we can get.   

Just bookmark Neven's graph page.

https://sites.google.com/site/arcticseaicegraphs/