When will Arctic Extent dip below 1,000,000 Km^2 (2022 poll)

[El Cid]

I also think that the Barents is of utmost importance, but I think its melting is not due to the positive feedback written down by gerontocrat but simply by more and more ocean heat transport from mid/equatorial latitudes as binntho said.
Nonetheless, as more and more heat is transported north, the Barents gets to open earlier and earlier making it possible to get a BOE. I think without a very open (during spring) Barents, Bering and Chukchi it is very hard to get a BOE

[The Walrus]

However much we may like to argue about the finer details of the chart, the general 40+ year trend is downward, with a linear trend of -0.1837 km3 per day (-297 km3 per year).

Am I going crazy, or does the slope in the related graph not say 0.8137. I just bring this up because it would have a large effect on how the average slope compares with the decadal slopes, and makes more sense to me, as the last decade seems lower in slope than the overall trend.

I noticed the same thing.  Which would mean that the last decades exhibits a much lower slope than the average.

[The Walrus]

The Walrus seems to be mixing things a bit here, but my statement about the maximums being lower and the mininums lower each decade certainly seems to hold for the 365 day averages (but the Walrus, if memory serves me correctly, insists on using the unreliable 42 datapoints which have no bearing on the problem).

I think you are mixing up maximum, minimum, and average.  When most of us talk of maximum or minimum, we are referring to the maximum or minimum amount of sea ice that year (either extent, area, or volume).  You appear to be using the terms to describe the maximum or minium in the averages over time.  Best to clarify your statements, lest they be misunderstood by the posters here (including me). 

A significant number of posts here frequently reference to what you call "unreliable datatpoints."  I do not think that I am alone in refuting your statement that they have "no bearing on the problem."  You seem to be going to extreme lengths to try and substantiate your claims.

[KenB]

However much we may like to argue about the finer details of the chart, the general 40+ year trend is downward, with a linear trend of -0.1837 km3 per day (-297 km3 per year).

Am I going crazy, or does the slope in the related graph not say 0.8137. I just bring this up because it would have a large effect on how the average slope compares with the decadal slopes, and makes more sense to me, as the last decade seems lower in slope than the overall trend.

I noticed the same thing.  Which would mean that the last decades exhibits a much lower slope than the average.

I read it as -0.8137, which is neither -0.1837 nor 0.8137.  I assume binntho transposed the 1 and 8 unintentionally. 

[The Walrus]

However much we may like to argue about the finer details of the chart, the general 40+ year trend is downward, with a linear trend of -0.1837 km3 per day (-297 km3 per year).

Am I going crazy, or does the slope in the related graph not say 0.8137. I just bring this up because it would have a large effect on how the average slope compares with the decadal slopes, and makes more sense to me, as the last decade seems lower in slope than the overall trend.

I noticed the same thing.  Which would mean that the last decades exhibits a much lower slope than the average.

I read it as -0.8137, which is neither -0.1837 nor 0.8137.  I assume binntho transposed the 1 and 8 unintentionally.

Hard to tell from his graph, but based on the four individual decadal trends, I do not think he transposed the numbers, as -0.8137 looks about right.  Averaging the four posted individual decadal values results in a trend of -0.703, which is within statistical error of the overall trend.

[gerontocrat]

Gero, I agree with you in general - but remember that the heat loss in the Barents is, to a large extent, from Atlantic waters carrying heat up from the mid latitudes and even the tropics.

It reminded me to see if I should update my ocean heat and temperatures file from NOAA data.

It is getting warmer in the North Atlantic

[binntho]

However much we may like to argue about the finer details of the chart, the general 40+ year trend is downward, with a linear trend of -0.1837 km3 per day (-297 km3 per year).

Am I going crazy, or does the slope in the related graph not say -0.8137. I just bring this up because it would have a large effect on how the average slope compares with the decadal slopes, and makes more sense to me, as the last decade seems lower in slope than the overall trend.

Hi Subline_Rime, you spotted a serious mistake which I have now corrected, in that post. Also in the following post I claimed that the rate of decline of the last decade was above the long-term rate (based on the error in the previous post), which should have raised flags as well. Anyway, I've corrected both.

[binntho]

As others have pointed out, I made a mistake in my posts of claming that the long-term decline was -0.1837 when in the graph (and the calculated 297km3 per year) was clearly 0.8137. A big difference there. Also, as the Walrus points out, the trend of the last decade, at -0.2507 or -0.2987 (depending on which year you start) is clearly lower than the long-term trend.

As to whether I confuse the maximum or minimum, I do not, and of course I knew that the Walrus was talking of the annual max and min, while I was talking about the general downward trend.

BUT my error does not change my main point in any way. There is a clear decadal (or, more correctly, multi-year) memory in the system. And anybody wanting to predict a flattening curve based on the last few years is betting against the odds.

Which leads to the following point: The very large variation from year to year (and significant variation from decade to decade), without any causal explanation, makes a mockery of anybody claiming to be able to see trends from a few years behaving oddly.

[binntho]

It is getting warmer in the North Atlantic

And polynomically as well! (Is that a word?)

But Gero, you say "North Atlantic" ... could you be a bit more specific?

[binntho]

Sorry about the monologue - I am in EAT and get up early in the morning, I guess most of my interlocutors are living in more civilized timezones! Anyhow, I and the Walrus have very different opinions on whether there is any reason to suspect a change in trend these last years, based on the various graphs that we keep seeing. The Walrus sees a flattening of the curve where I only see random variation. Only time will tell.

Anyway, the following graph shows deviation from trendline of my previous 365 day running average graph. Excel tells me that the standard deviation is 1133.4 which I've tentatively added to the graph. The current deviation is not the biggest in the positive direction (1998 takes the prize here) but it is certainly big.

It is also noteworthy that the downward anomalies are fewer but bigger than the upward anomalies. To my mind this implies the existence of strong melt-inducing factors that hit randomly and cause maximum damage if they happen at the same time. This could be a combination of two or three factors, such as early melt ponding, clear skies during maximum insolation and a strong storm or two. If two, expect strong melting, if all three hit together, trouble is afoot!

But all in all, there is still no other conclusion that can be reached than that the current anomaly in the graph, i.e., the apparent flatness of the last 4 years, is nothing but yet another wobble around the long-term downward trend.

AND THIS GOES DOUBLE (sorry about the shouting!) as long as nobody has any causal explanation for the variations in this graph. Any causal explanation has to offer a reasonable explanation of all the wiggles and wobbles, based on actual data. Not guesswork trying to explain only the one that you think is interesting.

[gerontocrat]

It is getting warmer in the North Atlantic

And polynomically as well! (Is that a word?)

But Gero, you say "North Atlantic" ... could you be a bit more specific?

NOAA divides the Atlantic into North & South. North is from the equator and includes the entire Arctic Ocean. There is no separate data for the Arctic Ocean on their website.

[El Cid]

AND THIS GOES DOUBLE (sorry about the shouting!) as long as nobody has any causal explanation for the variations in this graph. Any causal explanation has to offer a reasonable explanation of all the wiggles and wobbles, based on actual data. Not guesswork trying to explain only the one that you think is interesting.

Let's not go down this road again, we've been here many times before.

1) Very long, continuous periods of above and below "trendline" residuals show only that the trendline is wrong because the residuals are autocorrelated
2) The explanation by more learned man says that the trend is in reality a slow one but there was one acceleration when we lost multiyear ice and that led to a sort of system change ("hysteresis"). This means that the current "trendline" actually overestimates the speed of decline.

[binntho]

Let's not go down this road again, we've been here many times before.

1) Very long, continuous periods of above and below "trendline" residuals show only that the trendline is wrong because the residuals are autocorrelated

No

2) The explanation by more learned man says that the trend is in reality a slow one but there was one acceleration when we lost multiyear ice and that led to a sort of system change ("hysteresis"). This means that the current "trendline" actually overestimates the speed of decline.

That's actually quite funny. We are looking at real-world figures, and of course they are autocorrelated - this is a continuos process involving a very large and complex physical system. How can it not be autocorrelated? Your point about the number of points above and below trendline makes no difference whatsoever.

And "learned men" have said ... please! Nobody has put forward any real causal explanation based on observations or that fits the data. Very intelligent people have done a lot of thinking, but that doesn't really work in a scientific context and cannot be put forward as an argument for whether the moon is made of cheese or some mysterious forces causing the trendline to overestimate the decline - while the decline, at the same time, clearly follows the trendline to an amazing degree!

And hysteresis ... please stop using words you don't understand.

[El Cid]

That's actually quite funny. We are looking at real-world figures, and of course they are autocorrelated - this is a continuos process involving a very large and complex physical system. How can it not be autocorrelated? Your point about the number of points above and below trendline makes no difference whatsoever.

As I have clearly stated, it is not that the datapoints are autocorrelated but that the residuals of the trendline are autocorrelated, which is a problem you would understand if you had studied statistics. I have written about this many times on this site but everyone is like "blah-blah, who cares", which only shows that very few have taken statistics classes here.

some short descriptions that popped up after 1sec googling:

"The existence of autocorrelation in the residuals of a model is a sign that the model may be unsound. Autocorrelation is diagnosed using a correlogram (ACF plot) and can be tested using the Durbin-Watson test."

https://otexts.com/fpp2/regression-evaluation.html

https://www.displayr.com/autocorrelation/


(as for your remarks about my understanding of hysteresis I found it unnecessarily rude, not the first instance in your writings unfortunately)

[oren]

Folks, please continue discussing the very interesting subject, a really good discussion above, but avoid discussing the other posters, whether they studied something or not and whether the understand something or not.

[gerontocrat]

What matters?

How warm is it going to get?
Will warmth increase linearly or exponentially?

Do tipping points exist?

Can we still rely on historical data as a guide to the future? I hope not, as history tells us that homo sapiens will most likely burn fossil fuels at a high rate for a very long time and trash the natural environment to such an extent that the carbon sinks will reduce CO2 capture by a lot.

Technical arguments over historical statistical trends are a bit pointless when human behaviour may be the single most important parameter in determining the future.

I know nothing except in the end warmth will defeat cold. Perhaps that is why I spend most of the time collecting, analysing and presenting data; plus the questions the data raise in my mind for which I don't have the answer.

[Richard Rathbone]

In my opinion the key papers are:

Tietsche et al 2011, Recovery mechanisms of Arctic summer sea ice

https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2010GL045698

No tipping points/hysteresis in summer area/extent. Lots of other work showing the same thing, but I reckon Tietsche marks the point at which it became unreasonable to expect a tipping point in summer sea ice extent.

Ding et al 2018, Fingerprints of internal drivers of Arctic sea ice loss in observations and model simulations

https://www.nature.com/articles/s41561-018-0256-8 (paywalled, other routes to access may be available, but I'm not aware of anyone else thats looked at variability this thoroughly)

There is lots of internal variability in the summer sea ice extent, including multidecadal oscillations and the historic record is too short to characterise it from data alone. (and the vast majority of modelling studies don't run the models for long enough either) In the period around 2010 these oscillations lined up in a way that may well not repeat for 60 years.

Multidecadal oscillations are what generate autocorrelation in trend residuals. We have been in a "hiatus" since 2012 because the oscillations are in a recovery phase and this counters the GHG driven trend.

...

There were tipping points/hysteresis in volume/MYI. Summer extent got too small to maintain significant amounts of MYI and it is essentially gone and would take decades to thicken up again if GHGs dropped and area recovered. There may be a tipping point in winter sea ice area too, but thats not reached till after the normal state of summer sea ice is a blue ocean.

If global temperature keeps going up, or even accelerates in the way Hansen expects, then I expect to see a return to a clear negative trend in summer sea ice in the 2030s and it may well take the same sort of dramatic plunge in the 2050s and 60s that it did in the years leading up to 2012, (though I don't expect to be around to see it).

[crandles]

AND THIS GOES DOUBLE (sorry about the shouting!) as long as nobody has any causal explanation for the variations in this graph. Any causal explanation has to offer a reasonable explanation of all the wiggles and wobbles, based on actual data. Not guesswork trying to explain only the one that you think is interesting.

You obviously don't think much of
http://dosbat.blogspot.com/2014/03/what-caused-volume-loss-in-piomas.html
discussion of parts of
Lindsay and Zhang 2004
http://psc.apl.washington.edu/zhang/Pubs/tipping_point.pdf

my brief summary

The trigger for rapid loss:

MYI used to make it around the Beaufort gyre which gave the MYI time often lasting 10 years, time to be compressed and thicken.

Chukchi to Laptev started melting out each year so MYI didn't make it around the gyre. The area of 3+year MYI collapsed ensuring Chukchi to Laptev melts out each year.

Why doesn't it encroach further and further; Why does it slow down?

Basically when the thick MYI is down to a new much lower minimal level the period of rapid loss is over.

Thermodynamic growth of ice in winter. The thicker the ice the more insulation there is so heat is lost only slowly. Less heat loss=less ice formation. So there is a limit of around 2m of ice through thermodynamics (compression, slabbing and ridging obviously make some thicker). Whereas with very little ice thickness then the ice grows rapidly over winter.

Consequently when and where there is thick MYI losing its thickness down to 2m in summer, this ice is gone and it doesn't grow back. If instead we have 1.5m FYI then this all melts out but it  regrows back quickly in winter. This means much less net change from one year to the next.

There is still MYI left but it is now in relatively stable places near Greenland and CAA i.e. if it disappears from an area one summer it is likely to bounce back and return to cover the area again.

Expecting a reasonable explanation of all the wiggles and wobbles seems like it could be an unreasonable demand depending on what you mean. An explanation that is self consistent is certainly needed. What do I see this as having here: Does it
1. Explaining an acceleration in the rate of volume and extent loss then a slowdown in these measures that fits both maximum and minimum trends, and
2. If other measures like MYI is part of the explanation, these also need to show the right trends at appropriate times, and
3. The explanation is in the scientific literature without being seriously criticised. Being in the scientific literature before 2007 seems an added bonus.
4. Area related data fits the explanation.

I think it does and several people here including me seem to like it. There will always be some who don't like it, but do they have much of interest to say? We seem to keep going over the same ground said in slightly different ways on different threads.

[crandles]

There are also newer papers on it.
For example
https://people.earth.yale.edu/sites/default/files/files/Zhang_etal_BeaufortGyreJGR2016.pdf
The Beaufort Gyre intensification and stabilization: A modelobservation synthesis

[Sublime_Rime]

I agree with Crandles that the depletion of multiyear ice and the corresponding loss of the thickest ice occurring during the 1-2 decades of sharpest decline (2000s into early 2010s) go a long way in explaining some of the deviations from the long-term trend. I think this actually supports binntho's point about the decadal memory. Of course this only explains part of the variation we see. I think other short-to-midterm natural variations such as the ENSO cycle can explain a portion of the remaining variation (such as recent positive deviations in volume relative to the long-term trend). Finally, we have the shortest term variations (e.g. more seemlingly random weather fluctuations), which I'd expect make up most of the remaining variation.

Perhaps if some free time opens up for me this spring, I may take a stab at looking at how closely changes in multi-year ice and ENSO correlate to the PIOMAS data temporally, and if so, what proportion of the variation they explain.

[kassy]

In my opinion the key papers are:

Tietsche et al 2011, Recovery mechanisms of Arctic summer sea ice

https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2010GL045698

No tipping points/hysteresis in summer area/extent. Lots of other work showing the same thing, but I reckon Tietsche marks the point at which it became unreasonable to expect a tipping point in summer sea ice extent.

[7] The global AOGCM we use consists of the atmosphere component ECHAM5 [Roeckner et al., 2003] with a T31 horizontal resolution and 19 vertical levels, and the ocean component MPI-OM [Marsland et al., 2003] with a curvilinear grid that has a horizontal resolution of 50–200 km in the Arctic and 40 vertical levels. A dynamic–thermodynamic sea-ice model based on the work by Hibler [1979] is included. The model setup we use is a coarse-resolution version of the IPCC-AR4 model described by Jungclaus et al. [2006]. This higher-resolution model setup has been tested extensively and performs well in simulating Arctic climate [Chapman and Walsh, 2007].

Notice how relatively simple the model is.
Also atmosphere and ocean changes are modelled while land changes are not. As we know they are important and quite big compared to the changes in 2010.

We use ECHAM5/MPI-OM to perform a climate projection for the 21st century according to the IPCC-A1B emission scenario [Nakićenović et al., 2000]. In this reference run, annual mean surface air temperature in the Arctic rises from −14°C in the 1900s to −4°C in the 2090s. Arctic sea-ice extent declines, and the Arctic Ocean is typically ice-free by the end of summer from 2070 onward (see auxiliary material; we note that the sea-ice decline here is somewhat faster than in the higher-resolution version of the model). Between 2000 and 2040, when the rate of decline is maximal, Arctic summer sea-ice extent exhibits strong year-to-year fluctuations. As noted by Holland et al. [2008] and Notz [2009], this increase in variability is mostly due to changes in the ice thickness distribution and does not necessarily indicate proximity to some critical threshold.

So much more recent modelling moved that up to the 2050s already.

9] To examine the recovery mechanisms of Arctic summer sea ice, we simulate the consequences of an ice-free Arctic Ocean during summer. We set up experiments to start on 1st July from initial conditions that are taken from the reference run, but are perturbed by converting the entire Northern Hemisphere sea ice to water with the same properties as the sea surface water below the ice. Such conversion of relatively fresh sea ice to salty sea water has the advantage of leaving the properties of sea surface water unchanged. The start date is chosen such that the effect of the perturbation is maximal: starting from ice-free conditions earlier in the year leads to immediate re-freezing, and hence both earlier and later start dates imply shorter exposure of open water to sunlight, and a less pronounced ice–albedo effect.

In the real world there is enough heat below to really hamper ice regrowth. This is not represented here accurately. If you remove the ice from July the there is ample time to mix it up.

[11] Every 20 years between 1980 and 2060, three such experiments are started in consecutive years (e.g., 2019, 2020, 2021), so that we can analyze five different time slices with a three-member ensemble each. After the initial perturbation, we let the model run freely without any further manipulation.

So the outcomes are limited by the model. Better models and more observations have moved the dates forward.

I think problems will start when we get huge open sea areas early in places that were well protected such as the CAB so there is lots of time to mix up the heat from below. So the tipping point is not in the surface ice but in the ice/ocean system. The heat that is mixed up can prevent refreeze for a time and the ice that forms eventually is in warmer waters or has a thinner layer protecting it from the warmer water below allowing for earlier failure the next year.

I guess we will get to learn more in the next really bad year.
 

[dnem]

As I understand the paper, the quick recovery of the ice after the modeled perturbation (ice free on 7/1) is mainly due to a big increase in upward heat flux in the subsequent fall and winter, which is a key component of the slow transition theory.

Does this paper from 2010 account for increased cloudiness in the arctic in the fall? Would this limit the loss of heat to space and retard any recovery? Also, as Kassy mentions, there would be "lots of time to mix up the heat from below" after a 7/1 ice free start. Does the AOGCM they use account for the potential for increased storminess in the arctic?

I'm not entirely convinced by this modeling paper that the arctic will not exhibit "nonlinear threshold behavior." 

See also Bardian's post in What's new in the arctic:

"Climate models used by the UN's IPCC and others to project climate change are not accurately reflecting what the Arctic's future will be. Researchers at the University of Gothenburg argue that the rate of warming will be much faster than projected."

[The Walrus]

As I understand the paper, the quick recovery of the ice after the modeled perturbation (ice free on 7/1) is mainly due to a big increase in upward heat flux in the subsequent fall and winter, which is a key component of the slow transition theory.

Does this paper from 2010 account for increased cloudiness in the arctic in the fall? Would this limit the loss of heat to space and retard any recovery? Also, as Kassy mentions, there would be "lots of time to mix up the heat from below" after a 7/1 ice free start. Does the AOGCM they use account for the potential for increased storminess in the arctic?

I'm not entirely convinced by this modeling paper that the arctic will not exhibit "nonlinear threshold behavior." 

See also Bardian's post in What's new in the arctic:

"Climate models used by the UN's IPCC and others to project climate change are not accurately reflecting what the Arctic's future will be. Researchers at the University of Gothenburg argue that the rate of warming will be much faster than projected."

As Kassy stated, the beauty of the model was in its simplicity.  Nature is never quite so simplistic, but the best models are often the most simplistic.  Once we start complicating models with other parameters, they may overwhelm the model and give rather unusual results.  The timeframe in the paper seems reasonable, but since we still have a long way to go, a little tweek here or there (such as fall cloudiness) may move the needle significantly in the long run.

[be cause]

 .. were simple models able to predict recent heat domes that were beyond anything previously known ? I thought it took the latest most complex models to forecast well in advance what was coming ...

[jdallen]

As Kassy stated, the beauty of the model was in its simplicity.  Nature is never quite so simplistic, but the best models are often the most simplistic.  Once we start complicating models with other parameters, they may overwhelm the model and give rather unusual results.  The timeframe in the paper seems reasonable, but since we still have a long way to go, a little tweek here or there (such as fall cloudiness) may move the needle significantly in the long run.

The root mechanism determining how much ice we have and when *is* simple; it is the summary of  heat gain vs loss in the Arctic at the surface of the Arctic ocean.

The devil of course live in the details of how that heat is being moved around, and also embodied by the fact that year over year, the starting state of the system is a moving target - much like the inverse of Poe's "The Pit and the Pendulum".  Rather than the pendulum blade dropping lower, instead the "bench" (total system enthalpy) on which the Arctic ice rests is slowly being lifted towards it.  The amount of energy being applied seasonally remains essentially the same, and that fact is reflected in the relatively consistent volume of ice being removed season over season - going all the way back to when we started being able to effectively measure such things.  Jim Petit's volume graph is a great visualization of this:

http://iwantsomeproof.com/extimg/siv_annual_max_loss_and_ice_remaining.png

Consistently, that's 16-19,000 km2 of ice each season, going back as far as 1979. The annual variation is not a result of different amounts of heat being available; rather whether or not it gets captured by the system.

As someone else posted here or on another thread (Oren I believe) we currently are only a 2 sigma event from > 1,000,000 km2 of ice.  That could be cause by another "May Melt Ponds" event or GAC, or something new we haven't seen before.  We're just waiting on the dice to roll
- to give us clear May and June skies
- for another series of cyclones to tear up the Barents or Laptev or Beaufort
- for a rain even which imports massive amounts of latent heat and crushes albedo regionally
- or any of a dozen or more new events that we haven't seen the ramp up for

At the end of February, PIOMAS only reported about 19,400km3 of ice.  Between now and the end of April, we will typically *at* *best* pick up another 2000km2.   Another 2010 or 2012 melt season will put us uncomfortably close.

[The Walrus]

Yes, the root mechanism is heat gained vs lost, and the amount of energy applied as remained fairly constant (that may change with increased summer cloud cover and open water). 

The decrease in Arctic sea ice has resulted largely from a decrease in heat loss.  A combination of increased carbon dioxide and winter cloud cover has reduced the amount of heat lost to the atmosphere.  In recent years, larger areas of open water have been present much later into the start of the freezing season.  This has allowed more heat to be lost.  This can be seen in a graph comparing summer minima and subsequent freezing seasion ice growth.

[crandles]

Yes, the root mechanism is heat gained vs lost, and the amount of energy applied as remained fairly constant (that may change with increased summer cloud cover and open water). 

The decrease in Arctic sea ice has resulted largely from a decrease in heat loss.  A combination of increased carbon dioxide and winter cloud cover has reduced the amount of heat lost to the atmosphere.  In recent years, larger areas of open water have been present much later into the start of the freezing season.  This has allowed more heat to be lost.  This can be seen in a graph comparing summer minima and subsequent freezing seasion ice growth.

That last sentence seems to me to be jumping to a conclusion that seems wrong.

Larger areas of open water have certainly been present later. This has allowed extra heat absorbed into water during summer to escape. This is normal heat loss not from latent heat allowing ice to form.

So re subsequent freezing season ice growth, there are three trends here:

1. CO2 (perhaps also cloud cover) allowing less heat to escape which should result in less ice formation.
2. Freezing season starting later which allows less time for ice to form.
3. Ice volume is starting at a lower level so more ice can form.

The freeze volume is clearly increasing so 3 is clearly much more significant than 1 and 2.
1&2 should be viewed as tiny slow effects that are close to overwhelmed by changes in ice volume levels and inter-annual variations.

1 and 2 should result in a downward trend in ice maximum. Now the maximum ice volume levels are only slowly declining because we have reduced MYI to minimal volumes in safe locations, we can see only a small downward trend in maximum volume due to these causal factors.

That slow small downward trend in maximum volume should lead to a slightly faster decline in the minimum volume but this is still very small and slow compared to the 1998-2010 period when there were rapid losses of MYI extent, volume, age and thickness.

How fast are these small and slow trends? It seems hard to know, 0 slope seems well inside the uncertainty range.

[Richard Rathbone]

As someone else posted here or on another thread (Oren I believe) we currently are only a 2 sigma event from > 1,000,000 km2 of ice.  That could be cause by another "May Melt Ponds" event or GAC, or something new we haven't seen before.  We're just waiting on the dice to roll
- to give us clear May and June skies
- for another series of cyclones to tear up the Barents or Laptev or Beaufort
- for a rain even which imports massive amounts of latent heat and crushes albedo regionally
- or any of a dozen or more new events that we haven't seen the ramp up for

Ice loss is cyclical, the residuals autocorrelate, records happen in clusters. More 2-sigma events will happen, but the years to expect them are the 60s and 70s, not the 20s. With a bit of luck GHG concentrations will be dropping then and it'll show up as a hiatus in recovery, rather than the sequence of records being smashed that culminated in 2012.

Its not an accident that the English temperature record was set in mid afternoon in mid summer. The cycles have to line up too. Translating the arctic's multi-decade cycle to an annual or diurnal one, 2023 is the equivalent of evening or autumn. Just as evening is the wrong time of day and autumn is the wrong time of year for all time temperature highs, we are now in the wrong time of the half-century for ice melt records.

BAU on GHGs puts a downward pressure on the ice thats of a similar magnitude to the cyclical variation, so 2012 isn't unbeatable, but it won't be beaten regularly or by much for some time yet. The sort of plunge that brings a BOE in range won't happen till the cyclical variation lines up with the GHG trend again.  Maybe in 2050 there will be some forum in whatever medium there happens to be then, in which the same sort of excitement over the first BOE plays out in the way the excitement over record breaking did here in 2012. Maybe in 2070 the same sort of excitement over whether any ice forms during the winter will play out.

[crandles]

Ice loss is cyclical, the residuals autocorrelate, records happen in clusters. More 2-sigma events will happen, but the years to expect them are the 60s and 70s, not the 20s. With a bit of luck GHG concentrations will be dropping then and it'll show up as a hiatus in recovery, rather than the sequence of records being smashed that culminated in 2012.

There are ocean (multi) decadal oscillations and I don't deny there being some potential for some effects on arctic sea ice. But is there any good evidence for them being as significant an effect as you are making out here? I don't see it. I am more inclined to trust GHG as the cause of decline and the slow transition explanation for the acceleration and slow down in rates. There could be more things in play but I tend to think we are more likely to know about major causes such that it is less likely that decadal oscillations are a major factor. 

[The Walrus]

Yes, the root mechanism is heat gained vs lost, and the amount of energy applied as remained fairly constant (that may change with increased summer cloud cover and open water). 

The decrease in Arctic sea ice has resulted largely from a decrease in heat loss.  A combination of increased carbon dioxide and winter cloud cover has reduced the amount of heat lost to the atmosphere.  In recent years, larger areas of open water have been present much later into the start of the freezing season.  This has allowed more heat to be lost.  This can be seen in a graph comparing summer minima and subsequent freezing seasion ice growth.

That last sentence seems to me to be jumping to a conclusion that seems wrong.

Larger areas of open water have certainly been present later. This has allowed extra heat absorbed into water during summer to escape. This is normal heat loss not from latent heat allowing ice to form.

So re subsequent freezing season ice growth, there are three trends here:

1. CO2 (perhaps also cloud cover) allowing less heat to escape which should result in less ice formation.
2. Freezing season starting later which allows less time for ice to form.
3. Ice volume is starting at a lower level so more ice can form.

The freeze volume is clearly increasing so 3 is clearly much more significant than 1 and 2.
1&2 should be viewed as tiny slow effects that are close to overwhelmed by changes in ice volume levels and inter-annual variations.

1 and 2 should result in a downward trend in ice maximum. Now the maximum ice volume levels are only slowly declining because we have reduced MYI to minimal volumes in safe locations, we can see only a small downward trend in maximum volume due to these causal factors.

That slow small downward trend in maximum volume should lead to a slightly faster decline in the minimum volume but this is still very small and slow compared to the 1998-2010 period when there were rapid losses of MYI extent, volume, age and thickness.

How fast are these small and slow trends? It seems hard to know, 0 slope seems well inside the uncertainty range.

Why do you believe that less ice volume result in greater ice growth, but that less surface area does not?  Physically, the surface area should have agreater effect.

[Freegrass]

Good discussion here. But there's one factor that I haven't seen mentioned yet. And I don't know if it's a significant factor at all.

There's a lot more ice being made in the last decade or so, and when you're making more ice, you're also releasing more brine that sinks to the deep. I'm guessing that this must decrease salinity in the fresh water layer? Which in turn makes it easier to make more ice in less saline water? That ice is then also less saline? Or is this an insignificant factor?

Love this graph. Sea ice volume seems to be stuck above 4000 Km³ for the last decade. That's the equilibrium I was talking about. Curious to see if and when we'll start dropping below that in a consistent way. I'm ignoring the 2012 anomaly. 2019 and 2020 came close, but 2021 and 2022 were higher again...

[crandles]

Why do you believe that less ice volume result in greater ice growth, but that less surface area does not?  Physically, the surface area should have agreater effect.

When there is a low ice volume at start of freeze season, this means the ice is on average thin (with a lot at zero).

During freeze season, the ice growth approaches an equilibrium thickness. While ice is thin, energy can conduct through the thin ice quickly so it forms at a rapid rate, when it is thicker it takes longer to conduct the same amount of heat and the rate of ice formation slows down.

Consequently the later start to the freeze season doesn't have much effect.

The surface area of the arctic ocean is constant, there might be a slight retreat around edges of the maximum extent but this just isn't significant compared to what I am taking about.

[Richard Rathbone]

Ice loss is cyclical, the residuals autocorrelate, records happen in clusters. More 2-sigma events will happen, but the years to expect them are the 60s and 70s, not the 20s. With a bit of luck GHG concentrations will be dropping then and it'll show up as a hiatus in recovery, rather than the sequence of records being smashed that culminated in 2012.

There are ocean (multi) decadal oscillations and I don't deny there being some potential for some effects on arctic sea ice. But is there any good evidence for them being as significant an effect as you are making out here? I don't see it. I am more inclined to trust GHG as the cause of decline and the slow transition explanation for the acceleration and slow down in rates. There could be more things in play but I tend to think we are more likely to know about major causes such that it is less likely that decadal oscillations are a major factor.

Ding et al

Anomalies that were steadily building up decade by decade and aren't any more. June cliffs were  a big thing in 2012, and they aren't any more. The seasons in which ice has been preferentially lost have been following an oscillation of around the same sort of period that Ding at al found both in the data and in their long duration model runs. Give me a reference that says Ding has it wrong and its a 1-off forcing from MYI disappearance and it might convince me.

GCMs have been showing rapid ice loss events for a long time. Ding implies these are a consequence of multidecadal oscillations alternately lining up with the GHG trend and opposing it and the generally poor state of arctic representation in GCM results is due to these oscillations in the models being out of phase with the real world. They get the timing wrong because they aren't  spun up for long enough to get oscillations of this length of period in sync. They don't do all that well with syncing ENSO and this is an order of magnitude longer. The ensemble spread in GCMs is grotesque. They are utterly appalling at SIPN predictions when judged against their declared error margins. However the size of the ensemble spread and the degree to which they miss in SIPN is about right for failure to sync oscillations of a similar magnitude to the GHG trend.

I'm willing to accept that volume does respond to extent on a timescale that's long enough to produce a 10 or 20 year pulse in the September extent data, I think thats too fast for what the data shows, but the record really isn't long enough to be definitive about whether its being forced by a 20-year pulse or a 60 year oscillation.

Ding tied low extent in autumn to an arctic pressure pattern in the summer and a pacific temperature pattern in the spring. Convincing me that MYI loss drove pacific spring temperatures is going to take references of a similar thoroughness as Ding's study, but coming to a different conclusion.

[I’M IN LOVE WITH A RAGER]

That’s a lot of big picture and long term conclusions based on a single paper. Could you post it please so we can read it ourselves evaluate its postulates and conclusions? Not trying to be harsh, but what makes you so confident in this publication in particular compared so any of the other multitude of climate models/papers out there?

[crandles]

https://www.atmos.washington.edu/~david/Ding_etal_2017.pdf

How sea-ice variability and trends can impact the Arctic
atmospheric circulation is an area of vigorous research42. Studies
suggest numerous mechanisms in which sea-ice loss modulates
the large-scale circulation in the lower troposphere in winter43–45
.
This paper, instead, puts the spotlight on how the high-latitude
circulation impacts sea ice. Although positive feedbacks between
sea ice and the Arctic circulation exist, we find that these are small
during summer.

Sounds like this paper is going against most research?

Earlier work indicated that the decline of sea ice before the
1990s was in part owing to an upward trend in the North Atlantic
Oscillation (NAO) index20,21. However, since the early 1990s, the
apparent link between the NAO and sea ice has largely disappeared,
with Arctic sea ice declining further despite the reversal in the NAO
trend

Doesn't sound like a great track record for this sort of thing.

I am not expert enough to give a good critique of greater detail. Their conclusions on direction of causality were the opposite of my gut reaction expectation but that means very little indeed.

[Jim Hunt]

How sea-ice variability and trends can impact the Arctic atmospheric circulation

I covered this at some length at the time, since when Storify has gone down the proverbial pan!

https://GreatWhiteCon.info/2017/03/is-arctic-ice-loss-driven-by-natural-swings/

Needless to say Anthony Watts swiftly stepped up to the plate on the “skeptical” side of the “debate” with a guest article on his blog by David Middleton.

It can't be all bad if Axel Schweiger's name is on it?

[The Walrus]

Why do you believe that less ice volume result in greater ice growth, but that less surface area does not?  Physically, the surface area should have agreater effect.

When there is a low ice volume at start of freeze season, this means the ice is on average thin (with a lot at zero).

During freeze season, the ice growth approaches an equilibrium thickness. While ice is thin, energy can conduct through the thin ice quickly so it forms at a rapid rate, when it is thicker it takes longer to conduct the same amount of heat and the rate of ice formation slows down.

Consequently the later start to the freeze season doesn't have much effect.

The surface area of the arctic ocean is constant, there might be a slight retreat around edges of the maximum extent but this just isn't significant compared to what I am taking about.

I disagree.  Forty years ago, there was roughly 7 million sq km of open water at minimum.  In recent years, that has increased to about 10 million.  That disparity has continued throughout the autumn months, only narrowingly significantly nearing the winter solstice.  That is rather significant, and is a very large area of ocean with zero ice thickness compared to prior years.  As your graph shows, the ice growth rate at zero thickness (open water) is much greater than the growth rate thin ice, let alone thicker ice that was present in prior years.

[crandles]

and is a very large area of ocean with zero ice thickness compared to prior years.  As your graph shows, the ice growth rate at zero thickness (open water) is much greater than the growth rate thin ice, let alone thicker ice that was present in prior years.

That is exactly what I am arguing so I am not sure what you are disagreeing with.

[jdallen]

Ice loss is cyclical, the residuals autocorrelate, records happen in clusters. More 2-sigma events will happen, but the years to expect them are the 60s and 70s, not the 20s. With a bit of luck GHG concentrations will be dropping then and it'll show up as a hiatus in recovery, rather than the sequence of records being smashed that culminated in 2012.

There are ocean (multi) decadal oscillations and I don't deny there being some potential for some effects on arctic sea ice. But is there any good evidence for them being as significant an effect as you are making out here? I don't see it. I am more inclined to trust GHG as the cause of decline and the slow transition explanation for the acceleration and slow down in rates. There could be more things in play but I tend to think we are more likely to know about major causes such that it is less likely that decadal oscillations are a major factor.

(Leaving the previous comments intact for context)
I've struggled to find some sort of correlation between multi-decadal ocean oscillations and changes in sea ice for years, and failed to find any relationship between them and what we see year over year in the annual variation in sea ice.

As to the frequency of 2 sigma events, I'll point out that what *constitutes* one has changed, in as much as (referring to Jim Petit's graph), we have *4* events in the last 15 years where volume loss was above or near 19,000km3 in a single season - 2010, 2012, 2015 & 2021.  My conclusion here is, it actually will no longer take a 2 sigma melting event to take the sea ice under 1,000,000 km2 in extent. 

However unexciting it may seem, I concluded several years ago that the key to understanding and determining when this event will happen will be found in a closer examination of refreeze season conditions, not melt.

This discussion has already touched upon some of the key factors in play here, which are eroding the end-of-season volume numbers:

1) Net Arctic Ocean enthalpy, which is increasing dramatically not just at depth, but in upper portions of the water column (1-600m), as reflected in increased salinity from Atlantification and sea surface temperatures which hinder and have progressively pushed the end of the melt season further and further into fall.

2) Increased Northern Hemisphere atmospheric moisture.  This is a direct consequence of global increases in temperature (current pushing towards 1.5c), which increase overall atmospheric moisture carrying capacity by somewhere over 10%.  This both increases import of latent heat into the Arctic, and probably more importantly, amplifies the greenhouse effect of CO2 disproportionately at higher latitudes.

As an additional knock-on factor with the increased net atmospheric heat comes an additional amplification effect - more intense cyclonic storms at high latitude.  These do not in and of themselves decrease ice, but as we have seen with the high Fram export and slowed ice area and extent increases, they limit ice creation; first by simply importing replacement heat that leaves the atmosphere rather than that in the ocean, and secondly, disturbing the water column which pulls heat from the larger reservior at depth.

So, for my on going efforts to understand Arctic seasonal dynamics, I've begun to focus much more on two thing - first, variations in precipitable water over time above the Arctic circle, and secondly (cursing sparse data) trying to understand and quantify increases in and distribution of heat in Arctic seas water columns.

At this point, I suspect I won't have adequate data sets for that until after we see extent dip below 1,000,000 km2.  Nevertheless, it's these elements of the system I'll be focusing most of my attention on trying to understand their effect on long-term changes in the Arctic.

[The Walrus]

and is a very large area of ocean with zero ice thickness compared to prior years.  As your graph shows, the ice growth rate at zero thickness (open water) is much greater than the growth rate thin ice, let alone thicker ice that was present in prior years.

That is exactly what I am arguing so I am not sure what you are disagreeing with.

Perhaps I must understood.  I thought you were saying that lower extent of sea ice at and near minimum wasn't significant compared to the thickness of the ice.  My apologies.

In that case, we seem to be on the same page.

[crandles]

As to the frequency of 2 sigma events, I'll point out that what *constitutes* one has changed, in as much as (referring to Jim Petit's graph), we have *4* events in the last 15 years where volume loss was above or near 19,000km3 in a single season - 2010, 2012, 2015 & 2021.  My conclusion here is, it actually will no longer take a 2 sigma melting event to take the sea ice under 1,000,000 km2 in extent. 

Really? If the high sigma events are occurring too frequently then my conclusion might be that I at least need to consider if this means the model is wrong. A curve through the data might be better than a straight line? This would make the residuals smaller and the trend goes more horizontal at the end so this could well result in deciding that you need a much higher sigma event to cause a BOE.

Do you see this as a possibility? How long does the flattish trend have to continue until you reach that conclusion? Until the new trend is statistically significant? Think we have discussed that we have already reached that point a few years ago.

[Richard Rathbone]

That’s a lot of big picture and long term conclusions based on a single paper. Could you post it please so we can read it ourselves evaluate its postulates and conclusions? Not trying to be harsh, but what makes you so confident in this publication in particular compared so any of the other multitude of climate models/papers out there?

There aren't a multitude of papers on this topic. If you know of a more recent one with a credible claim to supercede Ding, then cite it. I don't think there is one, but it could easily have escaped my notice, I'm pretty reliant on the forum flagging papers of interest for my reading.

Stoat quoted by Jim on the GWC in the immediate aftermath of publication.

The flaw in this overall, without looking at the details, is that it’s hard to see a near-40-year trend and being so much natural variability. That seems to be asking for an awful lot of one-way variation.

I think this was fair comment then, but I did some digging plus the trend isn't so obviously one way now as it was then. Global temperature shows repeated "hiatuses" due to ENSO. September Arctic extent has been on hiatus for a decade. Stepwise patterns are what happens when a trend combines with an oscillation of similar magnitude.

If you took all of this at face value then they’d have solved one of the puzzles, that on the whole models show much less ice decline that reality.

Same point as I made, except that the passage of time makes me more convinced of it now than he was then.

[jdallen]

<snip>
 it actually will no longer take a 2 sigma melting event to take the sea ice under 1,000,000 km2 in extent. 

Really? If the high sigma events are occurring too frequently then my conclusion might be that I at least need to consider if this means the model is wrong. A curve through the data might be better than a straight line? This would make the residuals smaller and the trend goes more horizontal at the end so this could well result in deciding that you need a much higher sigma event to cause a BOE.

Do you see this as a possibility? How long does the flattish trend have to continue until you reach that conclusion? Until the new trend is statistically significant? Think we have discussed that we have already reached that point a few years ago.

I understand your reaction, but hear me out.

It's not about the melt, it's about how much ice is present at the start of the season.

A loss of 19,000km3 of ice during the melt season is no longer a low frequency/rare event.

In fact, we've been flirting with crossing 20,000km3 melt in a single season.

We're looking at not much more than about 22,000 KM3 of ice at the end of April if I'm estimating correctly.  We're not that that far away from losing most of that in what is a fairly typical melt season.

So again, it's not about what is becoming typical with our melt, rather it's much more about how much ice we start with - and here, volume is much more important than either area or extent.

Do I make more sense now?

[oren]

I think the volume at the beginning of the melting season ia crucial, in total and in distribution. What matters most of all is the volume in the High Arctic, and most especially the CAB and the Beaufort.
This has always been one of the tenets of the Slow Transition, and definitely bears watching very carefully.
I have recently begun to have doubts though about the accumulated accuracy of PIOMAS.  There are serious differences with CS-SMOS both in regards to absolute thickness and to the distribution of thick ice.
I have a strong suspicion that PIOMAS misses the open Nares and its associated ice export due to the low resolution of the PIOMAS grid and its reliance on NSIDC concentration data, and thus overestimates the thickness in the Lincoln Sea. This is a small but important location. Nares has nearly always been closed during winter so probably the model could not be calibrated around this phenomena, but recently this has become a more common occurrence.

[Freegrass]

What matters most of all is the volume in the High Arctic, and most especially the CAB and the Beaufort.

So true. And even more important is the distribution of the ice. If there's a lot of thick ice along the coast, it'll take a long time to melt, which delays the formation of open water. And it's open water that soaks up most of the energy from the sun. That's how we ended up with the ESS arm last year, because it took a long time before we saw open water there due to the thick ice in the ESS. That's why I always post those HYCOM comparison GIFS. They are a good indication of where open water is more likely to form early, or not. I know many here don't like HYCOM, but it's the best indicator we have.

If you compare 2020 with 2023, I can see another ESS arm forming this year, while in 2020 we had very early open water in the ESS and Laptev, and a very low minimum. But that all can change with the weather of course. Weather is the all important factor in a season, because it moves, melts, protects, or destroys the ice.



jdallen also mentioned that a warmer atmosphere holds more water

2) Increased Northern Hemisphere atmospheric moisture.  This is a direct consequence of global increases in temperature (current pushing towards 1.5c), which increase overall atmospheric moisture carrying capacity by somewhere over 10%.  This both increases import of latent heat into the Arctic, and probably more importantly, amplifies the greenhouse effect of CO2 disproportionately at higher latitudes.

While true, there's also a negative feedback from this; more clouds during peak insolation, like we had in 2021 and 2022.

To get to a BOE we need "perfect weather" during winter that blows the ice away from the coastline - like it did in 2020 - and then a strong melting season without too many clouds during peak insolation - like we also had in 2020 - topped off with a 2012 GAC, which we didn't get in 2020.

2020 could have broken all the record by far if we would have had a late August GAC, but for everything to line up perfectly is gonna take a very special year.

[crandles]

I understand your reaction, but hear me out.

It's not about the melt, it's about how much ice is present at the start of the season.

A loss of 19,000km3 of ice during the melt season is no longer a low frequency/rare event.

In fact, we've been flirting with crossing 20,000km3 melt in a single season.

We're looking at not much more than about 22,000 KM3 of ice at the end of April if I'm estimating correctly.  We're not that that far away from losing most of that in what is a fairly typical melt season.

So again, it's not about what is becoming typical with our melt, rather it's much more about how much ice we start with - and here, volume is much more important than either area or extent.

Do I make more sense now?

I follow your argument and back in 2007-2010 I would probably have agreed with what you are saying. As an additional argument to what I wrote previously:

April 2017 averaged only 20664 Km^3 and that lower amount should mean more open water formation and albedo feedback so if we are flirting with 20k km^3 melt then on your logic it should have all melted out.

It didn't it ended up rather normal. Were we just incredibly lucky or does this cause you to question your thinking?

Is it possible that some locations for volume like Beaufort/Chucki/East Siberian/Laptev are vulnerable and easily melt out while other locations against CAA and central arctic are safer locations? What could cause such a difference in vulnerability? Maybe the movement of ice towards CAA plays a large role in keeping open water formation down in those safe areas while it is promoted in other areas perhaps particularly in Laptev (re the Laptev bite). Open water formation allows albedo feedback to kick in.

I suggest it might be harder to get a BOE than you think.

 

[FishOutofWater]

There was very, very slow ice motion in 2017.

https://www.nature.com/articles/s41586-022-05686-x/figures/3

This article does a lot to help explain what we have been observing since 2005.

https://www.nature.com/articles/s41586-022-05686-x

Because of the complexity of the systems that led to the regime shift circa 2007, we can't use this article, or models, to predict the "blue ocean event" but it can help us understand what has been taking place.

[Richard Rathbone]

There was very, very slow ice motion in 2017.

https://www.nature.com/articles/s41586-022-05686-x/figures/3

This article does a lot to help explain what we have been observing since 2005.

https://www.nature.com/articles/s41586-022-05686-x

Because of the complexity of the systems that led to the regime shift circa 2007, we can't use this article, or models, to predict the "blue ocean event" but it can help us understand what has been taking place.

Nice paper, I like the way they've done the analysis and marshalled the evidence.

[Paul]

What matters most of all is the volume in the High Arctic, and most especially the CAB and the Beaufort.

2020 could have broken all the record by far if we would have had a late August GAC, but for everything to line up perfectly is gonna take a very special year.

I don't think it's quite that simple though, 2020 ice pack was quite compact(but thin) apart from over the Beaufort where it was more disperse. The reason why the 2012 GAC had such an impact is because it went through an area of very weak ice which separated a large chunk of ice from the main ice pack and slowly melted away and the after affects of the storm could of bought some upwelling of the open water which meant more ice melted away than it probably would of if the storm did not happen. I think 2012 was a freek in terms of just how low it eventually went whilst 2020 was a more true reflection in terms of what extent may look like in a warm summer.

And yet despite that, we were still a good 2.5 million square miles before any BOE so I agree with the poster above, a BOE is probably a longer way off than most are anticipating unless we get something unprecedented like a true ice free north pole for example which would be a significant and newsworthy event.

[oren]

Square kilometers, not miles.

 I think 2012 was a freek in terms of just how low it eventually went whilst 2020 was a more true reflection in terms of what extent may look like in a warm summer.

I tend to agree. But the first BOE will be a freak as well, on top of the long term trend.