What will the Arctic resemble in 2050?

[Jim Hunt]

I accidentally discovered over my muesli this morning that a gala dinner will be held at The Fram Museum in Oslo next month. It is the prelude to "The Arctic Summit", where the topic under discussion will be "What will the Arctic resemble in 2050"

What would your advice be to the "150 policymakers, senior business leaders and influential commentators" who will apparently be in attendance?

[ChrisReynolds]

Most optimistic outlook:

Strongly seasonally ice free with increasing periods at the end of summer ice free, seasonal cycle still tied by March max and September min, but much more aggressive melt during the thaw season. Therefore a crash in most polar bear populations?

Warmer by several degrees C, substantial thawing of land permafrost with methane and CO2 emissions.

Substantial thawing of the sub sea marine clathrates, but no climate catastrophe from massive methane release.

GW will be posing an increasingly serious issue, with entrenched shifts in mid latitude climate on a seasonal basis, more intense flooding, more intense drought, cold outbreaks with less intense cold but still bringing persistent disruptive snow to Europe and Northern Eurasia.

Most pessimistic outlook:

Those of us who are sceptical about the speed of methane release are wrong. It is apparent that marine clathrates are reacting 'enthusiastically' to the ocean warming and that enhanced GW due to methane is underway.

Abbot & Tzipperman are correct*, and the sea ice has crossed the perennially sea ice free state (two separate  issues). This state is maintained by a combination of a winter cloud blanket* keeping the region warm, with enhanced heat transport* by both ocean and atmosphere adding to the warming; keeping the Arctic largely above zero degrees C, preventing winter ice formation. In response, mid latitude winter climate changes radically. EDIT-I should clarify - Abbot Tzipperman relates to cloud/heat transport, and I'm imagining the bifurcation after seasonally sea ice free found by Eisenman could have occurred by 2050.

Greenland responds to regional warming adding substantially to sea level rise.

Due to both GW and enhanced Arctic warming reducing the mid lat / Arctic temperature gradient - much more intense persistent rain events - increasing flooding, and more blocking highs - increasing drought. Very severe impacts on agriculture.

****

I know you were probably aiming at sea ice , but it's impossible to leave out the wider impacts.

[Jim Hunt]

Thanks for those "projections" Chris.

It's impossible to leave out the wider impacts

Quite so. As the Freeway Blogger put it recently:

What happens in the Arctic doesn't stay in the Arctic.

[crandles]

My reaction is forget about 2050, there is too much uncertainty. Start preparing for late 20teens' and 2020s' with seasonal ice free state.

Even if not all experts are convinced it will happen this decade, some preparation is a sensible precaution.

We should be running climateprediction.net / weather at home type simulations where ice melts out during summer to see what adverse weather become more common. Northern Hemisphere countries are running short of time if the conclusions suggest we should build more dams to cope with longer droughts because planning and building take time. Similar situation for flood defenses.

Not sure my reaction rates being listened to though. 

[OldLeatherneck]

I think Chris Reynolds has bounded the potentials correctly.  While policy makers will only want to hear about the "Best Case" scenario, I think it is important to watch the next 3-5 years very carefully.

If we see an ice-free September in 2016, or sooner, coupled with significantly increasing mass-loss of the GIS, we're probably headed for the worst-case scenario.  However, if by some chance we still have September ice in 2020 and the GIS mass-loss increase is only incremental, there is a chance that the best-case scenario will play out.

It needs to be stressed that all predictive models need to be re-aligned with current events as they occur.

[Artful Dodger]

Abbot & Tziperman was published in 2008. There's been substantial progress on the topic of Winter cloud forcings since then. For example:

Leibowicz, Benjamin D., et al. "Correlation between present-day model simulation of Arctic cloud radiative forcing and sea ice consistent with positive winter convective cloud feedback." Journal of Advances in Modeling Earth Systems 4.null (2012): M07002.

http://onlinelibrary.wiley.com/doi/10.1029/2012MS000153/abstract

[Artful Dodger]

Also relevant is:

Notz, Dirk, et al. "Arctic sea‐ice evolution as modeled by MPI‐ESM." Journal of Advances in Modeling Earth Systems (2013).

http://onlinelibrary.wiley.com/doi/10.1002/jame.20016/abstract

[Artful Dodger]

Bottom line from these two CCSM5 modelling efforts:
Summer sea ice disappears when C02 reaches 500 ppm

Winter sea ice disappears when C02 reaches 1100 ppm

These estimates represent the central estimates made by models that have shown themselves to be wildly optimistic.

YMMV.

[TerryM]

Bottom line from these two CCSM5 modelling efforts:
Summer sea ice disappears when C02 reaches 500 ppm

Winter sea ice disappears when C02 reaches 1100 ppm

These estimates represent the central estimates made by models that have shown themselves to be wildly optimistic.

YMMV.

Would those figures be for CO2 or CO2 equivalency?

Terry

[Artful Dodger]

Bottom line from these two CCSM5 modelling efforts:
Summer sea ice disappears when C02 reaches 500 ppm

Winter sea ice disappears when C02 reaches 1100 ppm

These estimates represent the central estimates made by models that have shown themselves to be wildly optimistic.

YMMV.

Would those figures be for CO2 or CO2 equivalency?

Terry

Good question, Terry. The papers don't give many details on the models or inputs to those models, just discuss their results.

However, it is generally accepted to use the Mauna Loa CO2 numbers for climate research, so I'd base predictions on them going forward.

If other forcings change substantially, the models will underestimate climate change. Sound familiar?

[Wipneus]

The papers don't give many details on the models or inputs to those models, just discuss their results.

They (Notz et al.), write that CMIP5 scenarios are used.
Start at http://cmip-pcmdi.llnl.gov/cmip5/ probably look at "forcing data" and possibly follow the link to the RCP database.

I will have to say more about the CMIP5 model results soon.

[gfwellman]

I think we can easily disagree with the 500 number (assuming it's a pure CO2 number - honestly I'm not sure where we are in equivalency) because we're still a hair below 400 and seemingly on a trajectory to ice-free summers pretty soon.  I say CO2 will be about 420 when we have an essentially ice-free summer, but there are other variables too - like if we eliminated all emissions tomorrow, we'd experience warming because the loss of cooling aerosols would take effect faster than any draw-down of CO2.  That alone could take us to an ice-free summer.  Conversely increased aerosols and natural variability swinging to "icier" could delay ice-free summers until CO2 is 440 or so.

The 1100 number, although a regime we know less about, feels pretty realistic.  It would take a heck of a lot of warming, insulation & heat transport to give us ice-free winters.  When the sun goes down for months, it gets *cold*. I don't see that happening by 2050 as a "reasonable worst case" but I suppose it's conceivable in some "black swan worst case" scenario where there is much faster & stronger permafrost and clathrate feedback then we currently think possible.  BAU gets us to 500 by 2050.  BAU + ugly feedbacks could be 600 ... which then implies really bad things for 2100, but not ice-free winters by 2050.

[Jim Hunt]

Crandles,

Not sure my reaction rates being listened to though.

You could raise your voice anyway.  I'm @jim_hunt on Twitter. How about retweeting my most recent musings, and/or crafting one of your own along similar lines?

We should be running climateprediction.net / weather at home type simulations where ice melts out during summer

Since our previous conversation along similar lines I've got CICE up and running quite nicely on my 6 core desktop. The supplied data is from 1997 however!

Since then Google have promised to give away lots of Raspberry Pi kits, so I've got CICE running on one of those too, albeit rather more slowly.  Do you suppose there might be any mileage in a distributed "Arctic Sea Ice Forum Model"? Alternatively are Myles et. al. up to the job, if you ask them nicely?

[crandles]

We should be running climateprediction.net / weather at home type simulations where ice melts out during summer

I've got CICE up and running quite nicely on my 6 core desktop. The supplied data is from 1997 however!

Do you suppose there might be any mileage in a distributed "Arctic Sea Ice Forum Model"? Alternatively are Myles et. al. up to the job, if you ask them nicely?

Re CICE, Is there any ability to manipulate the initial conditions to remove some sea ice?


To get an idea of weather event frequency, then models need to be run several hundred times so I expect we would struggle with an Arctic Sea Ice Forum Model unless there are lots of commentators with many computers that they can apply to such a task.

I have mentioned such a project a couple of times on the CPDN forum with little or no response. An email might be more likely to get a reply. The impression I have gained over there is that it seems to take months before a project gets anywhere close to beta testing.

It is almost certainly too big a job to get it set up appropriately when there are other priorities that they have grant funding for and need to concentrate on those.

I have little idea what would be involved with getting it 'set up appropriately' other than to say that we wouldn't want artifacts of the way the sea ice was removed to overwhelm the climate effects that should follow. Hence just removing some ice may not be enough, water temperatures may also need to be adjusted to keep the situation realistic.

Thanks for the encouragement. I like to see a bit more before suggesting it should be promoted perhaps as a Phd project that ought to get some funding. Anyone want to say whether they think that does or does not sound appropriate?

[Jim Williams]

The 1100 number, although a regime we know less about, feels pretty realistic.  It would take a heck of a lot of warming, insulation & heat transport to give us ice-free winters.  When the sun goes down for months, it gets *cold*.

There's plenty of *warm* up there already.  It's just separated from the cold by 100 meters or fresh water.  What the *fresh* does when there's no Summer ice is what will determine how long before we get ice free winters.

I'm betting the fresh water gets mixed up with the warm Atlantic water and we have warm cloudy Arctic in Winters shortly after we have little or no ice on August 1.

[ChrisReynolds]

Lodger,

Yes I know they're continuing to pursue the matter. But the first Abbott & Tzipperman paper seems to be the best one to refer to. i.e. 2008 Sea Ice, High Latitude Convection, and Equable Climates
http://geosci.uchicago.edu/~abbot/publications.html

As for what models (GCMs) show, as they're clearly underperforming and nobody has pinned down exactly why I do think there is reason to be sceptical about sea ice free in CO2 X 4. Consider the following two papers:

Kay et al, The Boundary Layer Response to Recent Arctic Sea Ice Loss and Implications
for High-Latitude Climate Feedbacks
http://nldr.library.ucar.edu/repository/assets/osgc/OSGC-000-000-001-728.pdf

Boe et al, Current GCMs’ Unrealistic Negative Feedback in the Arctic.
http://www.atmos.ucla.edu/csrl/publications/Hall/boe_et_al_published.pdf

Kay et al find that the September 2007 response of the atmosphere isn't unrealistic, due to a well mixed boundary layer there does seem to be a strong cloud response. But earlier in July the cloud response is too much because the boundary layer is more stratisfied. The problem being the global application of cloud parameterisation. Boe et al find "Most current climate models likely overestimate the climatological strength of the inversion, leading to excessive negative longwave feedback."

i.e.
Kay et al - too much summer cloud -> too little energy absorption
Boe et al - too strong a winter inversion -> too much energy loss

These are problems to be ironed out, and I await Wipneus's comments on CMIP5. But at present I'm not persuaded if CMIP4 and earlier show perennial sea ice free at 4 x CO2, because there are serious problems with them w.r.t. Arctic atmospheric processes.

It's also worth noting that in the Eocene the threshold for ice sheet formation was around 500 to 800ppm - Doria 2011 DECLINING ATMOSPHERIC CO2 DURING THE LATE MIDDLE
EOCENE CLIMATE TRANSITION
http://faculty.eas.ualberta.ca/wolfe/eprints/Doria%20et%20al%20AJS%202011.pdf
So deciduous forests (implying conditions during winter not conducive to an ice covered Arctic Ocean) could have occurred as part of a millennial scale response to increased CO2. The response of Arctic sea ice, unlike that of plant migration and ice sheet formation is proving to be far from slow. In my pessimistic scenario I implied a sea ice free Arctic in winter could be tenable with massive methane emissions and their attendant high radiative forcing within decades. But I stress that I don't think this is likely.

[Jim Hunt]

Re CICE, Is there any ability to manipulate the initial conditions to remove some sea ice?

I can't claim to have done a whole lot more as yet other than verify that all the right bits compile and run OK using the supplied test data, on my platforms of choice. Thus far ARM Debian and x86_64 Scientific Linuxes.

However you can certainly manipulate the initial conditions. By default things start with no ice! In normal use you supply a file containing the ice state at the end of the previous model run. I guess it's possible to manually set this file up any which way you choose, but at this juncture I haven't the faintest idea how to go about it.

I have mentioned such a project a couple of times on the CPDN forum with little or no response. An email might be more likely to get a reply. The impression I have gained over there is that it seems to take months before a project gets anywhere close to beta testing.

Doesn't sound terribly hopeful

Thanks for the encouragement. I like to see a bit more before suggesting it should be promoted perhaps as a Phd project that ought to get some funding. Anyone want to say whether they think that does or does not sound appropriate?

As I've been blathering about elsewhere on here, the need to hold out one's hand for funding from the powers that be before getting started seems to delay things for a few decades past their sell by date! I had in mind something more along the lines of rolling up sleeves and getting stuck in ASAP.  However at some point a large (albeit potentially distributed in many small chunks) pile of hardware would undoubtedly be required.

Microsoft Research appear to be supporting CPDN. In view of the urgency, maybe they'd help with marketing? Google might assist with the Android port? Amazon might offer otherwise unused cycles on their GPU cloud. Lego might part with some of their Mindstorms Marketing Money? (Perhaps fortuitously, CICE distributes stuff via MPI)

I'm a born optimist, but a "political pessimist" when it comes to addressing climate change. At any rate, it's early days for this particular potential project. Has anyone in the house actually worked with CICE/CCSM/CESM before?

If this conversation is worth pursuing, perhaps we should move it to a thread of its own?

[Wipneus]

... I await Wipneus's comments on CMIP5.

Unfortunately Quality Control has rejected the calculations. I was not aware that some models have grids with mixed ocean/land cells. Unlike eg. the PIOMAS grid. Difficult to test also.

That means the 300GB or so of data need to be reprocessed. Probably takes most of the coming week.

[Jim Hunt]

For some strange reason I just had a phone call from Dermot@TheEconomist. He wondered if I might like to buy a ticket to The Arctic Summit. I enquired about a press pass. He said he'd go away and look into it.

Should one be forthcoming, does anyone in the Oslo vicinity possess a spare sofa for a night or two?

[Laurent]

I don't know when the ice would be free in winter ! It is impossible to know from now when it will occur for that we have to see arctic ice free in summer and wait  around 5 years to have an idea !
What I am seeing is that the arctic ice free conditions in winter may happen much sooner than what it is said ! This is all the 9 Januarys from 2003 to 2013 :
http://diogene.net/infos-news/info-news3/january.jpg

[Jim Hunt]

I've been keeping my eye on that neck of the woods too Laurent. Back on the Arctic Sea Ice Blog, see also "a simplistic prediction of the 2013 melt season". Over on the other side of the ice cap:

We are 51 days ahead of last season

More here also.

[Jim Hunt]

I still haven't heard about that press pass, but The Economist have at least done me the honour of giving me their "Top Tweet" spot for The Arctic Summit:

https://twitter.com/search?q=%23ArcticSummit

[ChrisReynolds]

Wipneus,

I have experience with such frustrations, when it's ready it's ready.

[Pmt111500]

There's a simple answer to this that hasn't been mentioned yet, Arctic won't resemble anything we've seen before :-P.

To be more detailed, I think Arctic is still in the winter ice regimen, though the ice is mostly formed from falling snow and thus be a better insulator than ice formed in cloudless evernight of the arctic winter. I'm of the same opinion as C.Reynolds that the March-July will see mostly clear skies and thus enhanced melt in early season. I wouldn't be surprised of ice free Julys then but I do not believe the ocean heat transport can accelerate that much winter ice would be completely gone. The Gulf stream is one of the fastest currents in the world, but the speed slows down much when it crosses the Atlantic. One estimate for speed of water transport from Caribbean to Arctic was 4 months and I think the Greenland melt on summers will keep it slow during autumns producing eddies. This effect of course ends if or when the Greenland has melted away and then the NH will look very different from now.

[ChrisReynolds]

Pmt111500,

I think you've misunderstood, probably I wasn't too clear. Kay et al find that models are too cloudy in the summer. That doesn't mean the summer is unusually sunny (as far as I know).

However the transition to first year ice from multi year ice does mean a net gain of up to 1/3 energy for each region that changes from young to old ice.

[Jim Hunt]

For some strange reason my earlier "Top Tweet" seems to have completely vanished from the Economist's Arctic Summit feed. If at first you don't succeed......

https://twitter.com/jim_hunt

[Nightvid Cole]

I expect it to be roughly like Hudson Bay is now - mostly ice free for 6 months a year.

[Jim Hunt]

According to a recent academic paper on this very topic:

By midcentury, changing sea ice conditions enable expanded September navigability for common open-water ships crossing the Arctic along the Northern Sea Route over the Russian Federation, robust new routes for moderately ice-strengthened (Polar Class 6) ships over the North Pole, and new routes through the Northwest Passage for both vessel classes. Although numerous other nonclimatic factors also limit Arctic shipping potential, these findings have important economic, strategic, environmental, and governance implications for the region.

http://www.pnas.org/content/early/2013/02/27/1214212110.full.pdf+html

Readers of this forum might well go further than Smith & Stephenson, and suggest that all this will be happening long before 2050?

P.S. The Economist have now invited me to attend "The Arctic Summit", where this sort of thing will no doubt be high on "the agenda".

[fred]

The Arctic in 2050 will resemble Bar Harbor in the summer 2010.

(My instinctual response: "effin' Miami beach, man" was held up in committee and so a new response was drafted, which not only made it out of committee but managed to squeak by in both House and Senate, and consequentially you see it above.)

[Shared Humanity]

Just a lurker but having read all of these comments, I will weigh in with my nearly completely uninformed opinion.

With the Arctic essentially ice free for 6 months a year, significantly warmer SST especially in the shallower seas (although much colder than other oceans), increased infiltration by the north Atlantic and a jet stream that is increasing in amplitude and slowing down, we will have forgotten the GAC of 2012 as violent Arctic cyclones become the rule and the summer Arctic will resemble a stormy north Atlantic in the winter.

The shallower seas, particularly along the Siberian coast have depths.....

http://www.google.com/imgres?imgurl=http://geology.com/articles/arctic-ocean-features/arctic-ocean-seafloor-map.jpg&imgrefurl=http://geology.com/articles/arctic-ocean-features/&h=740&w=780&sz=207&tbnid=TYz2HJKmopKKpM:&tbnh=100&tbnw=105&zoom=1&usg=__MOKZ2n3Y3y4hy1rEJmMmp_Q9SmU=&docid=kkOhOM_3TiezqM&hl=en&sa=X&ei=7GQ7Ue_ZBcfg2QXD-oCAAg&ved=0CD8Q9QEwAQ&dur=897

.....resembling Lake Superior.

http://www.google.com/imgres?imgurl=http://www.glerl.noaa.gov/data/bathy/sup_bathy.gif&imgrefurl=http://www.glerl.noaa.gov/data/bathy/bathy.html&h=365&w=640&sz=20&tbnid=zH4SG1_92xxbhM:&tbnh=70&tbnw=123&zoom=1&usg=__AAEz_P6jxA-05wFtOz6TdGC_3k4=&docid=YMmnQIwIFBpemM&hl=en&sa=X&ei=a2U7UYvoPKS02wWIsYCQDg&ved=0CDIQ9QEwAA&dur=1580

In 1975, a winter storm with hurricane force winds and 11 meter seas in Lake Superior sunk one of the largest Great Lakes ships ever built (222 meters in length, 26,000 DWT) in 160 meter deep waters. With much longer fetches, I would think the coastal seas in the Arctic will have much higher waves.

http://en.wikipedia.org/wiki/SS_Edmund_Fitzgerald

Good luck with shipping in that.

[Jim Hunt]

@Fred - Presumably you have Hurricane Earl in mind?

@Shared - The consensus seems therefore to be that it will prove to be a bumpy crossing for any that attempt it!

[anonymous]

Jim, for a short period of time the Arctic Route - straight through the basin, without crossing the Russian maritime border - was open in 2012 for non-ice class vessels. IIRC, awi.de has maps.

[Artful Dodger]

Spring Break 2020, Playa de Nordenskioldbreen, Svalbard 

[OldLeatherneck]

Lodger,

I'd post pictures I took on the beaches in Mallorca, however, it might violate community standards!

I do have another question for you.  When you acheived the distinct honor of being a 'Full Member' on the Arctic Sea Ice Forum, did you get an coffee mug or a T-Shirt??

[Artful Dodger]

Lodger,

When you acheived the distinct honor of being a 'Full Member' on the Arctic Sea Ice Forum, did you get an coffee mug or a T-Shirt??

Hi, OLN

Nope. It's an ARCTIC UMBRELLA! 

[OldLeatherneck]

Lodger,

In my part of Texas, the only umbrellas that are allowed are camoflauge.

While it wouldn't be the first time I've been shot at, I'd rather it not happen because of my umbrella!!

[AndrewP]

The climate models and most experts like Meiers say 2030-2050 before we first see <1 million SIE. So I'd say we'll be seeing extent <1 million most summers by 2050.

I think it is easy to forget much of the volume and extent losses thus far have been driven by rapid warming (the 20002-2012 were incredibly warm compared to 1992-2002) and by ice export. The ice north of Canada and Greenland is much less susceptible to export, and it will require further rapid warming to produce continued volume/extent losses. I doubt 2012-2022 will warm nearly as much as 2002-2012 did compared to 1992-2002.

[Jim Hunt]

Good morning Andrew,

Have you seen the recent debate here in these hallowed halls (and next door) about the evident inadequacy of the current models, with the possible exception of PIOMAS?

If so, what do you make of them?  It seems most "local" experts here plump for <1 million most summers by 2020. What do you make of them?!

What might cause rapid "volume and extent losses" apart from mere "rapid warming"?

[Pmt111500]

Late December to March might look like this (possibly with less ice) (Northern Baltic river port 12.01.2010, c. 4 pm after föhn event (compare to SSW))

[fred]

@Jim Hunt

Bar Harbor because I see it becoming the new playground for the rich: better climate than the Carribean, losts of real estate, etc..   a playground for those who can afford to escape it all.

[Rick Aster]

If I were making the Arctic 2050 sci-fi movie and I wanted it to be super-realistic, I would be on my way to the Gulf of Alaska.

[AndrewP]

Good morning Andrew,

Have you seen the recent debate here in these hallowed halls (and next door) about the evident inadequacy of the current models, with the possible exception of PIOMAS?

If so, what do you make of them?  It seems most "local" experts here plump for <1 million most summers by 2020. What do you make of them?!

What might cause rapid "volume and extent losses" apart from mere "rapid warming"?

As far as I can tell the only argument is that some of the older models did not predict the recent decline in extent and volume. Newer models accurately simulate the decline in extent and volume.

If a model can accurately simulate the recent declines, then there should be reasonable confidence in its predictive power.


In addition, as I said above, the pace of arctic warming the last 10-15 years has been 5-10X the rate globally. In the long-run, the ratio is expected to be 2-3X. Thus I expect the arctic to warm a mere 0-.75C over the next 15 years, as opposed to the 1C it warmed over the last 15.

[Jim Hunt]

Good morning Andrew,

At the recent Arctic Summit in Oslo Rear Admiral Jon White from the "US Navy Pentagon" said that he was expecting "a largely ice free month" by 2023.

Here's my evidence - https://soundcloud.com/water-connects-us/ice-free-summer-arctic-numbers

What do you make of that?

He also said that he and his team needed "more data" and "ten years" before they'd be able to model what's already happening in the Arctic as we speak.

[Wipneus]

Newer models accurately simulate the decline in extent and volume.

If a model can accurately simulate the recent declines, then there should be reasonable confidence in its predictive power.

Andrew, accurately simulation of volume is complete non-sense.

Below is a graph of CMIP5-modeled volume trends in September using historical and projected rcp8.5 (=high) forcings.
(it is preliminary ATM, I want do some more checks. Else compare with the similar graph in http://www.cesm.ucar.edu/working_groups/Polar/presentations/2012/stroeve.pdf)

In the 1980's the models differ in volume from about 2 to 40 [1000km3]. So even for the historic part and without comparing with actual volume they cannot be called "accurate" in any reasonable way.
Do you see any other aspect in the graph that can be called an "accurate simulation"?

[Apocalypse4Real]

Jim are there other significant comments from Rear Adm Jon White?

Thanks in advance,

A4R

[AndrewP]

Newer models accurately simulate the decline in extent and volume.

If a model can accurately simulate the recent declines, then there should be reasonable confidence in its predictive power.

Andrew, accurately simulation of volume is complete non-sense.

Below is a graph of CMIP5-modeled volume trends in September using historical and projected rcp8.5 (=high) forcings.
(it is preliminary ATM, I want do some more checks. Else compare with the similar graph in http://www.cesm.ucar.edu/working_groups/Polar/presentations/2012/stroeve.pdf)

In the 1980's the models differ in volume from about 2 to 40 [1000km3]. So even for the historic part and without comparing with actual volume they cannot be called "accurate" in any reasonable way.
Do you see any other aspect in the graph that can be called an "accurate simulation"?

A lot of the CMIP5 models are actually pretty decent. One must remember that the models do not include internal variability. They only represent the forced portion of the trend. They don't have actual weather observations parameterized as PIOMAS does. As I've said above, much of the recent decline is likely not forced. Arctic warming has been much faster the last 15 years than it is likely to be the next 15 years.

As Zhang 2010 points out, models also considerably underestimate surface air temperature increase over the recent period as well. In addition, the subset of models chosen are able to accurately simulate the rapid loss of sea ice in 2007.

"... the extreme event of sea ice cover loss in summer 2007 also fall into the range of changes simulated by the subset."

Thus, newer models are able to accurately simulate the recent loss of sea ice. They just attribute a significant portion of it to natural variability. Arctic warming has simply outpaced simulated warming. Substantial spread remains in ice physics as well, but by selecting a subset of models which have an accurate sensitivity of ice parameters to air temperature, this spread is greatly reduced.

http://www.tellusa.net/index.php/tellusa/article/view/15690/17582

If PIOMAS is able to fairly accurately simulate ice volume when weather is parameterized, then it should be possible to simulate future ice conditions if arctic climate is able to be accurately simulated. This is essentially what Zhang has done in the study above. He has taken models which, when weather variables are parameterized, accurately simulate ice conditions (just as PIOMAS does). When these same models are put into future-cast, they predict an ice-free state between 2037-2065.

In essence, taking models as accurate as PIOMAS and feeding them with climate model weather parameters for the next century, yields an ice-free arctic no sooner than 2037.

[AndrewP]

The most significant figure from Zhang 2010 is the following. It shows the sensitivity of SIA to SAT for various models. As you can see, many of the models DO have an accurate sensitivity of SIA to SAT. The subset of models that was chosen is CNRM-CM3, GFDL-CM2, IPSL-CM4, MIROC3.2, ECHAM5/MPI-OM, CCSM3, and UKMO-HadGEM1. As you can see these models accurately simulate SIA when SAT is parameterized.

[ChrisReynolds]

Andrew P,

Actually the CMIP models do include internal variability, it is in the multi model mean that internal variability (weather) is averaged out to leave the forced trend.

Jinlun Zhang (not Xiandong) of the PIOMAS team produced a paper in 2010 using NCEP/NCAR data superimposed on a warming trend from GCMs to use PIOMAS to project. "Arctic sea ice response to atmospheric forcings with varying levels of anthropogenic warming and climate variability."
http://onlinelibrary.wiley.com/doi/10.1029/2010GL044988/abstract
In Zhang J 2010 there is an inflection at the point where the observed NCEP/NCAR forcing ends and the GCM + randomly shuffled NCEP/NCAR data takes over. This is implicit in the graphic Wipneus has posted - the observed sea ice extent may lie within the spaghetti of CMIP5 simulations, but it seems to me that its slope is steeper than any of the members of that group. That this behaviour happens in Zhang J 2010 is significant because it supports the idea that weather, via NCEP/NCAR, is a major player in the observed sea ice decline.

However I have serious doubts that what is going on is mere weather. The most obvious example of this is the summer dipole, which really seems to be a feedback from ice loss that started after the 2007 sea ice crash. Furthermore there is the issue of ice dynamics resulting from the transition to a mainly first year ice pack.

That noted, the graphic at hand shows sea ice volume for September, due to the stationary nature of the date of daily minimum, and rapid ice growth thereafter, a September (whole month) with no ice volume well after 2020 does not seem unreasonable. The 2040s is however stretching it in my opinion.

[AndrewP]

Chris,

You are right the graphic is of individual model runs. But the main picture you take away from looking at all those spaghetti lines is basically the ensemble mean. It's hard to see many of the individual runs, but if you look closely several of them do show precipitous drops at times.

So you agree that many of the models do accurately simulate the response of ice to SAT and circulation changes. But you believe the models are underestimating the likely SAT increase and changes in atmospheric circulation (dipole anomaly).

This is a significant improvement in the level of discussion over the off-hand dismissal of models I have seen in many other conversations here and elsewhere. If we select a subset of models that have a pretty good handle on ice physics and the response of ice to SAT changes, the question then becomes are climate models accurately projecting SAT and circulation changes? The ice physics are not perfect, but good enough to give estimates with reasonably small error bars, IF SAT can be accurately predicted. So is arctic SAT being accurately projected?

I believe the answer must be yes. The arctic cannot continue to warm at 5-10X the rate of the rest of the planet, as it has for the last 15 years. It would defy a common sense understanding of atmospheric heat transport. There is a likelihood of a slowing, stabilization or possibly even a correction lower in arctic temperatures.

I thus believe the first day of SIE <1 million sq km will not occur until the 2030s or 2040s.

[Jim Williams]

Seems to me projecting Arctic Sea Ice based upon Surface Atmosphere is leaving out half of what is going on -- maybe more than half given what we are seeing so far this year.  The ocean has a lot more heat than the air, and the only question is whether that heat can reach the ice.