« 2011/2012 Winter Analysis | Main | PIOMAS May 2012 »


Feed You can follow this conversation by subscribing to the comment feed for this post.

Artful Dodger

These results make it even more important that there is a successful launch of NASA's new Orbital Carbon Observatory (OCO-2).

Artful Dodger

Looks like at least a 27-month delay after NASA cancels the Taurus XL launch contract for Feb 2013:

The timetable means OCO 2 will likely be grounded until at least mid-2014, more than a year after its previous target launch date.
That is, 27 months after a new Contractor is found...


IMHO it is inexcusable!!

Chris Reynolds

Interesting paper. Thanks for posting Neven. My thoughts are in a new post at my blog.

You'll find the Johanessen paper linked to over there and a striking graphic from it is at this link.

This paper's going to draw a lot of flack from the denialists, it'll be another nail in the coffin for the idea that they're open to reason.


Odd how they seemed to ignore repeated short term weather and wind patterns, observer bias maybe? How easily they blamed the sea-ice dynamics for increased ice extent in the Antarctic but could not use the same logic for the decreasing ice in some parts of the Arctic and increasing in others, unlike the NSIDC who seem to be able to actual see what is there


Eventually, folk might accept that short term repeated events can exist within much longer overall patterns that can seem to be driving much faster towards a certain outcome yet can just as quickly turn around and move in another direction. Time will tell on that one.

So in my opinion a rather limited review of their own beliefs rather than a full investigation of all the causes of ice dynamics. But it is good that they have posted it, at least their views can be held against actual developments in the future.
I see we also have to bear the fallacy of thinking we know how thick ice was more than a few decades ago, we don't, more's the pity, so any argument concerning actual ice volumes is limited.
The researcher in me would love to see the ice not melt, just so that we could escape this CO2 paradigm and get back to real research. But then again a pragmatic part of me hopes it does carry on decreasing as a switch to increasing ice would be a very bad thing, as that would indicate a move to a new ice age and that would be far, far worse than the negligible warming we have had recently.
Sorry if that upsets the CO2ists out there, but there are some of us out here that think climate is far more complicated and messy than you might have us believe and this adherence to the almighty carbon god is just restricting so much other research.

Daniel Bailey

"as that would indicate a move to a new ice age and that would be far, far worse than the negligible warming we have had recently"

(Is this a Poe?) You propagate 2 memes in one fragment: the impending ice age (http://www.skepticalscience.com/heading-into-new-little-ice-age.htm) and a variant of the CO2 has a negligible effect (http://www.skepticalscience.com/empirical-evidence-for-co2-enhanced-greenhouse-effect.htm)

First meme: This has been looked at before by climate scientists. Per Tzedakis et al 2012 (http://junksciencecom.files.wordpress.com/2012/01/nature-geoscience-ice-age.pdf), “glacial inception would require CO2 concentrations below preindustrial levels of 280 ppmv” (for reference, we are at about 395 right now…and climbing). Earlier, Tyrrell et al 2007 (http://plankt.oxfordjournals.org/content/30/2/141.full.pdf+html) examined this, concluding that we have already skipped the next glacial epoch. Furthermore, Tyrrell concludes that if we continue our present fossil fuel consumption, we will skip the next 5 glacial epochs. So no glacial epochs the next million years…

Sleep well on the cold-that’s-not-coming. The climate changes wrought by the heat on its way…yeah, that’s worth losing sleep over.

As for the second meme, even this Denier-par-excellence has this to say about it: "The serious skeptical scientists have always agreed with the government climate scientists about the direct effect of CO2.

Or this, even bigger denier: "Is there a greenhouse effect? Concedo. Does it warm the Earth? Concedo. Is carbon dioxide a greenhouse gas? Concedo. If carbon dioxide be added to the atmosphere, will warming result? Concedo."

(Oh, I got more, baby)

You know, you can look these things up ahead of time. Here's a good place to start (numbered for convenience):
Global Warming & Climate Change Myths

Any questions can be posted on the relevant threads there. Tell 'em I sent ya.

(Typepad still won't let me hyperlink anything. Takes half the fun out of debunking myths)

R. Gates

This is an interesting an important paper, and certainly I concur with the general finding...the downwelling LW from increased greenhouse gases in the atmosphere is the reason for the Arctic sea ice retreat over the long-term. However, the mechanism whereby downwelling LW is melting the ice is something I think is a bit more complicated, and actually may not be as direct as simply warmer tropospheric temperatures. Certainly warmer tropospheric temperatures do play a role(i.e. increased downwelling of LW)but warmer currenta and warmer water in general in the Arctic also play a big role. As the majority of excess energy from the imbalance created by additional greenhouse gases have gone into the oceans themselves, and the bulk of sea ice is actually under the water, and sea ice volume loss has been readily observed and is quite dramatic, and warmer water has been observed entering the Arctic from both the Pacific and Atlantic, the only reasonable conclusion is that the melting of Arctic sea ice from warmer water is at least equal to, and likely exceeds the actually melting from direct increased downwelling of LW. In essence, the sea ice is being melted by air and sea, with perhaps even more melting from the warming ocean given that the bulk of the ice exists under water. Certainly more open water only exacerbates this melting from the bottom as more SW radiation penetrates the Arctic water for longer and longer periods each year, adding to the heat content of Arctic water and melting the ice even faster.

Account Deleted

adherence to the almighty carbon god is just restricting so much other research.
Which other research? You would think that if this was the case - the rest of us scientist would be jumping on the denailists band-wagon.

Bob Wallace

Which other research?

Well, there's the question of how far one can sail into the oceans before tumbling over the edge....


I love it whem folk tell me I can't ponder on something because 'other' scientists have looked at that before, sorry I don't buy into that argument.
Why use the word meme when idea would have done? At least you didn't use the word of the moment, concept. Sounds like one of my students trying to impress me with their knowledge rather than their understanding.
As for the lack of other research, well you must be an outside observer not to realise how a prevailing paradigm, like CO2 forcing, can bias many research proposals, especially original research, but maybe that's a meme too far for some folk to comprehend.

Yvan Dutil

Karl, CO2 is a greenhouse gas and is bound to increase temperature. This not a theory, not a paradigm, this is a fact. Anybody who have seen a IR spectra of the atmosphere known it first hand. Only the magnitude of the effect can be argued.

And, yes the Sun effect has been checked. Indeed, it has been examined for 200 years. Climate science is born form this study. And, yes, the effect exist but it is small.


Guys, let's not go too much into the AGW theory per se. There are plenty other blogs for that and this one is about Arctic sea ice.

Of course, one would expect that as total ice volume keeps getting smaller this should have something to do with warming. A lot of warming. As R. Gates says this warming is probably coming from the oceans.

Ocean heat content has increased considerably since the Industrial Revolution, so it really looks like a case of 1+1 ('how could it NOT be caused by global warming'?. But like I say at the start of this blog post: to prove it scientifically is another matter.

So one conclusion I'd like to draw straight away is: kudos to Notz and Marotzke for trying. If the ice keeps disappearing I'm expecting a lot more papers on this topic (and there already have been some, like the Johanessen paper Chris R mentions).

One question I have is: why did they not look for a correlation with the AMO? I'll go and read the paper tonight.


R Gates,

I agree that more of the energy is coming from water than air. I would advance different reasons though. Just because more of the ice is underwater does not seem very convincing to me. However, if you look where the ice edge has retreated there is a huge retreat in Barents which receives atlantic water and little retreat elsewhere. This seems to me to be a strong signal that the extra energy is coming more from water than air.

AFAICS that doesn't contradict paper saying

"Greenhouse gases increase the downwelling thermal radiation. This radiation, in turn, is the major player in the heat budget of Arctic sea ice."

Surely ocean absorbing more energy through albedo feedback is still part of 'increasing the downwelling thermal radiation'. So no contradiction?

L. Hamilton

Although it is true that Arctic ice is melting from below, in contact with warmer water, my understanding from Mary-Louise Timmerman's talk at last week's Montreal IPY meetings is that much of this involves albedo-heated near-surface water from adjacent open areas, rather than Atlantic or Pacific water that tends to remain stratified (with exceptions for upwelling or turbulence) tens to hundreds of meters deeper.


I've never been convinced that co2 was the whole problem, thinking that the massive increase in radio/radar/microwave transmissions must play a part too, and they are absorbed in water easier than ice. That said the whole point of 'climate' and 'weather' is to equalise energy potentials in line with the second[?] law of thermodynamics thus if polar ice is melting [as a trend] then we have global warming, it is we do. That is the proof. When I was working on the house looking into thermal stores i was very suprised to find that water is one of the best you could get, far better than dense concrete, of equal volume, so it must transport far more heat than air.
Its also melting at the south pole but since they could hardly be more different even honest people could doubt the effect in antarctica. At least on the surface. http://www.climatecentral.org/news/antarctic-ice-shelves-are-melting-from-below-study-finds/
I used to think the great changes would happen faster in the south but now with the growing likelyhood of an ice free arctic ocean I anticipate a chaotic shift of climate dynamics in the north.

Daniel Bailey

"thinking that the massive increase in radio/radar/microwave transmissions must play a part too"

Believe it or not, Skeptical Science actually has a discussion thread on that, too: It's satellite microwave transmissions

Pete Williamson

Just for clarification is the term "self-acceleration" used in this paper the same thing as 'polar amplification'.


Chris Reynolds

Pete Williamson,

"Just for clarification is the term "self-acceleration" used in this paper the same thing as 'polar amplification'."

I assume that was a question? So I hope I'm not answering futilely. No. The two are different. Notz & Marotzke find that the loss of Arctic sea ice is not due to self-acceleration because record years of ice loss are followed by rebounds - in other words - records aren't immediately followed by new records.

Whereas polar amplification is the amplification of polar (specifically Arctic) warming trends over and above global warming trends.

I have wondered whether the loss of volume implies a possible mechanism for self-acceleration. In that the volume loss carries the signal of ice loss on from year to year. N&M mention the Tietsche paper and a paper by Eisenman (finding no tipping point) in support of their contention. Note that as Kevin O'Neill pointed out to Dr Schweiger over at RC recently - the Tietsche paper doesn't show the decline of volume seen from models like PIOMAS since 2007.

What would self-acceleration look like?

Since 2007 there's been a stabilisation of ice ages up to 3-4 years, with continued loss of older ice. This is probably because of the failure of the Beaufort Gyre flywheel - Beaufort is now a killing ground instaed of a loop where ice is aged and returned to the CAA and N Greenland. So the typical maximum age of ice is now around 4 years, and ice falling out of the sub 4 year old category into the older category is not surviving. Hence the decline of the older ice.

We're now 4 full melt/freeze cycles after 2007. But we've not seen a further crash. Volume has continued to decline, but otherwise what followed 2007 looks more like a period of adjustment after the effects of 2007.

Surely in a process of SA after 2007 we'd see further crashes as the massive gains in energy fromm ice albedo take effect. Instead we've had a return to the preceding volume trend in 2008 & 2009 (PIOMAS) with a decline in Spring 2010 due to weather and a continuation of the effects of that in 2011, likely to be followed this Spring.

So I don't think the volume losses can be viewed as part of a process of SA. But am willing to be persuaded.

Account Deleted

Okay Neven will change topic.
It is just from my experience with the UK funding councils it is the quality of the science and track record of the researcher/group that determines which projects get funding, not whether is is CO2 related. That said I'm not 100% sure how they decide which of the projects that are rated Alpha 4 get funding - our last one didn't :(

R. Gates

"One if by land, two if by sea..."

Here's the thing, Dirk Notz and Jochem Marotzke simply found the best correlation between increasing CO2 and decreasing sea ice among all possible external forcing candidates, and only posited the likely mechanism to be increased downwelling LW. This of course makes sense, but of course correlation is not causation as we all well know. It remains to be discovered the exact mechanism(s)whereby the increase in external forcing (i.e. the shift in the Earth's energy balance) is actually resulting in a gradual decline in ice, namely, what are the exact pathways whereby additional Joules of energy retained in the Earth system are making their way to the ice causing the change of state of enough mass of sea ice so as to result in the steady decline of Arctic sea ice volume? Though global climate models have predicted this decline, they have done a rather poor job at it, having rather underestimated how fast the ice might indeed decline. The exact mechanisms of the decline are also not completely well defined in the models, and obviously the underestimation means those mechanisms are incomplete in the models.

My guess as to the mechanisms whereby increased Joules in the Earth energy system from increased CO2 downwelling of LW is leading to decreases in Arctic sea ice: (in ascending order of importance)
1) Direct melting of ice above the water from increased LW downwelling
2) Direct melting from increased water temperatures at the more southern Arctic latitudes. During periods of ice divergence, especially late in the summer, we see ice from more northerly Arctic areas diverge and drift into this warmer southerly Arctic waters and melt.
3) Increased melting warmer water being transported from outside the Arctic (especially from the N. Atlantic side)

Certainly then, IMO, warmer water does play a role, and it remains to be proven as to the exact mechanism whereby increased downwelling is causing oceans to warm. My guess is, the mechanism is related to alteration of the thermal gradient across the ocean skin layer. Increased LW downwelling penetrates only the very top of the ocean skin layer, but this is exactly what is required to make the thermal gradient across the skin layer less steep, and thus heat flows less readily from ocean back to the troposphere, and thus, it is not that the oceans are being warmed directly by downwelling LW, but rather it is causing the oceans to lose heat less rapidly, and thus globally, ocean heat content rises. The net result is that, if you take all three variables: Increasing ocean heat content, increasing CO2, and decreasing Arctic sea ice (inversely related), you'll see a very high correspondence and even, as described above, a plausible mechanism.

Rob Dekker

R Gates, thanks for your notes.
Interestingly enough, the overall loss in sea ice over the past couple of decades amounts to only about 1 W/m^2 in heat accumulation :

which causes approximately a 10 cm/yr loss in ice thickness.
Note that AGW forcing is of the same order of magnitude.

Now, in the recent PIOMAS thread here at Neven's, I suggested that that heat came mostly from below (ocean heat flux increased due to increased Atlantic Water temperature ; which can easily be explained by AGW at lower latitudes).
After all, basic thermodynamic properties of ice suggest that ocean flux increase most directly affects the equilibrium thickness of thick ice. Even very minor (single W/m^2 flux increase) will cause a dramatic reduction in the maximum thickness that sea ice can obtain even with atmospheric temperatures through winter remaining the same.
If that is true, then your point (2) and (3) seem plausible, although the equilibrium thickness of sea ice in the presence of ocean heat flux has it's limits : The first 1 W/m^2 ocean heat flux will reduce equilibrium ice thickness to a few meters, but it takes exponentially more ocean heat flux to reduce equilibrium ice thickness further.

So basic physics of dynamic ice growth dictate that a minor increase (single W/m^2 forcing; in line with what AGW forcing can do) in ocean heat flux will significantly decrease the maximum thickness that MYI can achieve, and it seems that this effect has been observed. However, beyond that, to 'knock out' the remaining FYI, your point (1) "Direct melting of ice above the water from increased LW downwelling" would need to come into play, or polar amplification thereof.

To get an idea of what "direct melting of ice due to increased LW downwelling" can do, a simple calculation shows that FYI will reduce in thickness from 1.6 meter down to 1.4 meter during the winter season, if direct atmospheric forcing of a few C higher winter temperatures materializes. Note that this does not automatically mean that all Arctic sea ice will disappear. It just means that the extent of Arctic sea ice in summer will probably be smaller, but it depends on the 'positive feedbacks" that operate during summer (such as albedo feedback) on weather that will cause a collapse of summer sea ice or not...


I would tend to suggest the main drivers of thickness of ice to be as follows:

For FYI, 1) winter air temperatures and 2) upward heat flux.

For ice being crushed, the strength of the crushing forces which may depend on 3) mass of ice and 4) strength (& 5? consistency?) of winds.

Yes, I accept that a given upward heat flux will tend to melt thick ice more than thin ice. But this tendancy seems to be overwhelmed by crushing forces as MYI tends to get thicker over time. I suppose it is possible that this historic situation has changed with increased upward heat flux and less mass of ice to do the crushing. Unless you want to put forward evidence for this change, it seems unsupported to concentrate of an overwhelmed effect.

1) winter air temperatures apply everywhere.
2) upward heat flux may only apply in local areas where water flows under nearby ice.

3) mass of ice is declining. I wonder if models get the full effects of this? There is more crushing as ice is thinner but less crushing due to less mass. Crushing creates leads where more ice can form. I have no idea whether models include these effects or how well it is done.


>"a simple calculation shows that FYI will reduce in thickness from 1.6 meter down to 1.4 meter during the winter season, if direct atmospheric forcing of a few C higher winter temperatures materializes. Note that this does not automatically mean that all Arctic sea ice will disappear. It just means that the extent of Arctic sea ice in summer will probably be smaller"

Well, I would say we are already a few C higher in Arctic in winter per

and ice volume is down not just 1/8 but from 33 to 22 over 1979 to 2011 per

So far all the sea ice hasn't disappeared. So far so good.

2011 melted 18k out of 22k km^3. Continue increasing the forcing for 4/11 of 32 years i.e. about 12 years and it seems likely? that maximum volume will be down to around 18k km^3 by then. If/when maximum volume falls to 18k km^3, is loss of all(/95%+) the arctic sea ice then automatic?


Meaning of self amplification?

>"Arctic sea ice is not due to self-acceleration because record years of ice loss are followed by rebounds - in other words - records aren't immediately followed by new records."

Does self amplification mean the same as there is positive feedback?

Is the paper jumping to a conclusion with this way of testing for self amplification? Is is possible that after a record year particularly like 2007, you get an adjustment period of one or two years. Could this adjustment tend to overwhelm an acceleration effect that tends to occur rather slowly over a longer timeframes?

Does the curve indicating accelerating loss of ice while the major forcing is rather steady in its increase represent evidence that there is 'self amplification'?

Which is stronger evidence (and why)?

Chris Reynolds

Rob Dekker,

"However, beyond that, to 'knock out' the remaining FYI, your point (1) "Direct melting of ice above the water from increased LW downwelling" would need to come into play, or polar amplification thereof."

I think that to a degree it's helpful to seperate the forces driving the long term trend in loss of sea ice from the forces involved in the loss of ice during the melt season.

The loss of ice is due to increase in sensible heat budget in the Arctic, from the bottom of the ice there's ocean heat fluxes and gain due to open water ice albedo during the melt season. These are partially offset by heat loss due to open water and thin ice in the early freeze season. From the top there is increase in downwelling LW (clouds and GHGs) and influx of warmer air from lower lattitudes (increased storm tracks into the Arctic and the Dipole Anomaly). These reduce the growth of sea ice and hence reduce the volume/thickness.

During the melt season the key issue is open water formation efficiency (OWFE). i.e. ice of say 2m thick might lose 1m during the melt season so doesn't create open water, with the same thickness loss ice of 1m thick leaves open water with the same thickness loss. We are in a process where the OWFE is increasing because the factors identified in the preceding paragraph are thinning the ice. During the melt season the dominant driver of the melt season changes, certainly during the post 2007 era, is OWFE leading to increased open water and increased amplification by ice-albedo feedback.

So what is needed to knock out the remaining ice is thinning, the ice albedo effect and OWFE increases will take care of the dominant dynamics in a given season - weather allowing.

Wang et al 2008 shows that major episodes of ice loss are associated with the Arctic Dipole:
How much of this is due to ocean, Bering Strait, and how much atmosphere?

I'm starting to be persuaded that ocean heat flux is a bigger player than atmospheric heat. Indeed Boe et al 2009 note that whilst CMIP3 models show Arctic Amplification in the atmosphere, they do not show this in the Ocean (fig 1 and text) - could this be one of the reasons for model under performance? However as Polyakov et al 2011 (link at end of this post) note thinning of Siberian fast ice shows that there is certainly an atmospheric component. And with regards Atlantic Water (AW) - the persistence of the ice edge in the Atlantic sector (which has hardly moved apart from recession away from the Siberian coast) shows that most of the AW influx drops to deeper layers due to its salinity (density) as it moves over the lip of the Nansen Basin - so is deep and may not directly affect the surface. Polyakov et al figure 4 shows that whilst double diffusive heat fluxes may be in play (noted as up to 150m deep) the AW deep warming is remaining below 100m. So I think that a more likely candidate is Bering Strait influx particularly in Summer, NSIDC have noted that summers since 2007 have been dominated by the Dipole Anomaly (+ve mode) which increases Bering Strait influx as well as atmospheric heat fluxes.

To wrap up - ocean and atmosperhic changes are having an effect, but the atmosphere is clearly a player, and so is Fram Strait export. I see what Crandles is saying wrt mechanical crushing - just not sure how great a factor this is.

Polyakov et al 2011, Recent Changes of Arctic Multiyear Sea Ice Coverage and the Likely Causes.

Chris Reynolds


> "Does self amplification mean the same as there is positive feedback?"

I really don't think it does, the authors note that without ice-albedo (+ve feedback) the seasonal cycle would be markedly different. Self amplification is surely when the feedback is large enough to over-ride other factors and force the change.

> "Could this adjustment tend to overwhelm an acceleration effect that tends to occur rather slowly over a longer timeframes?"

Which is basically what I was wondering - but longer periods of time would need a memory to hold the state between years when the acceleration 'retards' - which is why I suggested sea ice volume losses.

> "Does the curve indicating accelerating loss of ice while the major forcing is rather steady in its increase represent evidence that there is 'self amplification'?"

The Johanessen scatter plot only runs up to 2007.
Plate b suggests an acceleration from 2003 to 2007, as an aside the Arctic Dipole becomes dominant after 2003, however if you carry the scatter plot on with data to 2011 then there's a marked bounce back to levels around 2005/6 before a descent to near 2007s low. So it is far from clear to me that the scatter plot is not conseratively explained as forced trend with weather noise.

I Ballantinegray1

I feel the thinning of the pack (with the 'aid' of AGW) has initiated a death spiral due to the fragility of that 'thin pack'. The ice is able to travel better than before with the loss of 'ice dams/bridges/arches'. With no obstacle to this export now central basin areas lose ice over the winter season and new FY ice replaces those loses (as it used to with 'leads') meaning more 'early' open water in late spring. The longer ice sits in water busy absorbing sunlight the faster it melts.

I also believe that 'perfect storm ' synoptics will be the way we see our first 'seasonal pack'.
After 07' it was assessed that this was a 10 to 20yr cycle with the two 'perfect storms' prior to 07' being 10yrs apart.
Once we dip below 1 million then the F.Y. ice will find it difficult to 'age', in any significant amounts, before the next perfect storm arrives?
If we keep in mind that we also have an ongoing 'methane' issue (on land and at sea) then the Arctic temps may also warm faster than is currently observed making ice retention doubly difficult?
We shall see ,over this years melt season, just how 'Swiss Cheese' the central pack is due to 'stretching/transport' over winter (and the late growth F.Y. ice infill this caused)


>"I really don't think it does, the authors note that without ice-albedo (+ve feedback) the seasonal cycle would be markedly different. Self amplification is surely when the feedback is large enough to over-ride other factors and force the change."

The trouble with this is with that you would expect ice changes to be monotonic and it is easier to see this isn't the case by looking at all years than to have to look up what happens after record years. So, if that is the definition, why are they using an inferior method that looks at too few years?

A definition where 'after entering new territory, the feedback is (likely to be?)great enough to over-ride other factors' might fit the usage better but seems a little odd?

This also opens the door to the possibility that earlier records, though in unexplored territory were not really different territory but 2007 was so large a change and the ice volume low enough that it was different territory. While there was recovery in 2008 and 2009 in area/extent terms, there was further volume decline and there may well be self acceleration with volume decline possibly accelerating. Probably too soon to tell yet.


I think the methane issue is far bigger than is currently being admitted. I think that there is a huge amount of methane currently being released which is directly being consumed by bacteria and directly heating the sea.both on the Siberian continental shelf and also in the Hudson bay area. the Kara sea this last winter strongly suggests this. or some other thing has changed significantly this last winter.

look at this graph. almost the same every winter for 30 years until this last winter.


Chris Reynolds


I strongly disagree that SA would imply a monotonic trend, it seems to me it would imply a series with an increasing rate of change. The problem is that the change of CO2 fits this profile, so even if the authors are correct then the profile of sea ice loss alone could still be taken to support SA. The authors don't look at only a few years of record minima, they "find a significant negative 1-year lag auto-correlation throughout summer for the year-to-year changes of both the satellite and the presatellite record (see auxiliary material)."

So I think the argument for SA in terms of area is still incorrect. However this leaves PIOMAS volume. I'd overlooked what the authors say about this, but here it is: "Note that we here only deal with sea-ice extent. Model reconstructions of sea-ice volume show a more persistent downward trend..."

If we can trust PIOMAS then it seems to me that it is potentially the most significant index in terms of SA because the losses can be taken as indicative of system energy gains. I am beginning to suspect that if the paper has a weak point it is on this issue of SA. But I need to think more about this, if you or anyone else has further comments I'd be interested to read them.


Most of the ESAS emissions are under the Laptev & East Siberian Seas (Shakhova et al "Extensive Methane Venting to the Atmosphere from Sediments of the East Siberian Arctic Shelf."), yet these don't show the same behaviour as Kara. The Beaufort has experienced cold weather this winter, is that due to lack of methane? Up until March this winter has shown the usual patterns of CH4 - large anomalies over land permafrost of Siberia. Only March shows a large anomaly over E Siberian / Laptev Sea - which could be due to the same weather pattern responsible for the ice behaviour in Kara. What I'm saying is that I don't see strong evidence for a pivotal role of CH4 in the ice behaviour this winter.


A series with an increasing rate of change is more specific than monotonic but I think it also is monotonic. Anyway I suspect you agree it is not 'feedback is definitely greater than other factors' but something more like 'feedback is more likely than not to be greater than other factors'.

>"the change of CO2 fits this profile"
huh? Only barely and that is over longer timeframes than involved here. So I am lost with that second sentence.

I should read the paper again more carefully. Thanks for pointing out these extra bits I had forgotten.

The negative one year lag seems to me to be a strong point against highlighting what happens in the year following each new record.

I would suggest that if there is "significant negative 1-year lag auto-correlation" and Tietsche et al found recovery after a shock in two years that a sensible way to test for positive feedback (and perhaps also self acceleration depending on its definition) is to look at trends over periods of at least 2*2 years so say 5 years.

Looking at periods immediately after 5 year averages have reached new lows might be useful if the 5 year trends are not monotonically becoming more negative (ie steeper).

I agree that volume seems most appropriate data to use, if it can be trusted. But just because I agree, doesn't mean everybody would agree.

Arguing that there isn't self acceleration seems like a contentious thing to argue. Perhaps the opposite view needs to also be argued.


Can anyone else get that auxiliary material


>"Which is basically what I was wondering - but longer periods of time would need a memory to hold the state between years when the acceleration 'retards' - which is why I suggested sea ice volume losses."

Is ice thickness distribution a possibility?

Reduced area increases heat through albedo feedback which tends to reduce volume. But the ice melted tends to be that on the edges which tends to be thin ice. The remaining ice therefore tends to be thick while the equilibrium distribution of that volume of ice should include more thin ice and less thick ice. So the adjustment that tends to happen in the next year tends to thin some of the thicker ice i.e. effectively spreading it so there is an area recovery.

Once this recovery has happened, there is more thin ice and the scene is set for another area minimum next time there is good summer melt weather. Before the adjustment has happened, there is little chance of another area minimum.

Not sure I have a good grasp of all the forces leading to there being an equilibrium ice thickness distribution. However some possibilities include:

1) Less mass of ice for mechanical crushing.
2) With some time, more bottom melting can do some thinning of thick ice.
3) Thick ice if cracked in thiner slices that it is thick can simply fall on its side.

Are there altenative explanations for significant one year lag autocorrelation?

Chris Reynolds


Monotonic; given what you regard as monotonic then I agree. By CO2 fitting the profile I mean CO2 shows an exponential increase, the inverse of the decline of sea ice. So what I meant was that both CO2 and SA would imply an ice loss series with an increasing rate of change. However there is another point I made to Arcticio just now in the comments at my blog: Notz & Martotzke in the section on forced response find that the SD of the satellite era (detrended) is the same as that for the pre-satellite era. This doesn't seem to me to support SA because I would expect SA to change the statistics (Variance and SD), although I don't know enough to prove this mathematically. So I suspect that this is another factore in support of the interpretation of forced response + natural variability.

One year lag negative autocorrelation - no I can't download the additional materials.

I've been looking at this, using extent and area detrended using first differences. The 1 year lag autocorrelation I get is -0.574 for extent and -0.363 for area. Interestingly the extent series shows that a record loss year isn't followed by a new record, but this rule does not hold for area.

Could part of the explanation be the Tietsche effect - after a record low there's more open water and then thin ice, this leads to greater heat loss from the surface layer of the ocean which means more vigorous ice growth over the following winter.


>"Could part of the explanation be the Tietsche effect - after a record low there's more open water and then thin ice, this leads to greater heat loss from the surface layer of the ocean which means more vigorous ice growth over the following winter."

Possibly, however if that vigorous ice growth occurs in the first year following a new low to prevent a record following immediately, why wouldn't this also prevent a new low just a couple of years later? If that is the explanation then it appears to me that remains supportive of no self acceleration. If that is what the data points to, so be it. It doesn't solve the puzzle marked * below.

>" I would expect SA to change the statistics (Variance and SD)"

I don't see why it would have to change variance/SD. If it changes the trend rate and you view the variance is just 'weather noise', why would it have to change?

The increasing rate of change between 1979 and 2010 for Co2 looks very slight to me and the forcing from CO2 is likely to be weaker than linear with CO2.

PIOMAS volume data clearly shows substantial acceleration in the downward trend. I see this as a strong signal that there is self acceleration while the 1 year lag negative auto-correlation is only a weak signal that there isn't self acceleration. It should be necessary to look 3 and 4 years ahead not just one year in order to make this a strong signal.

My view is that this acceleration could come from:
1) Accelerating forcing with little or no self acceleration, or
2) A near linear forcing (like CO2) and self acceleration, or
3) It could come from some past one off forcing plus self acceleration, or
4) Some combination of 2) and 3) or
5) A combination of different forcings ...

i.e. it isn't very useful for attribution just using observations without models.

Explaining why it is (or isn't) sensible to believe in self acceleration despite the 1 year lag negative auto-correlation (PIOMAS volume acceleration) seems like it could be a useful thing to investigate and explain.


Self-acceleration… very interesting, but it occurs to me that you may be mistaking the CO2 forcing effects. You suggest the CO2 concentration is in an almost linear increase, not showing the Gompertsz or whatever decline in sea ice volume. The thing that really gets to me is that CO2 rise in the not so distant past must have a lagged and accelerating effect in the very near future. Around 1992 it was clear that there was a regime change in the overall picture of weather and climate. Around 2010 most of us (who are awake) are aware that extremes have become commonplace on global scale. In that theatre we are discussing weather over the Arctic sea ice. It is like SC Barcelona or the Yankees losing a game on their path to win a competition…


I am taking the Tietsche et al to give a lag of at most 2 years. As the trend over 2008 to 2011 seems steeper than the trend up to 2007, then it appears to me that yes there is self acceleration. Over 2 or 3 years noise might make it appear there is an acceleration when there isn't. However when you can do the split at any year providing there is at least 4 years before and after the break, then I think it is unlikely to be noise.

I admit to liking the Gompertsz fit and yes that does mean that there could be self acceleration that later in the process seems to disappear.

>"you may be mistaking the CO2 forcing effects"

How does an almost linear CO2 forcing create acceleration? (I am thinking that a faster linear rate of CO2 increase would increase the slope of ice decline but would not cause acceleration unless it is self acceleration. But perhaps I am silly to think that?)

Chris Reynolds


Sorry to pick on one issue, but I think it's key with regards a potential misunderstanding about definitions.

>"A near linear forcing (like CO2) and self acceleration."

What we have IMO is a near linear forcing with amplification. Amplification is not the same as self-acceleration. As I understand the paper - SA is a state in which positive feedbacks are driving the changes with little / no help from CO2 or other forcings.

I agree that key to this issue is the volume / extent dichotomy and how these seem to paint different pictures, i.e. SA / No SA. It seems to me that the extent argument of negative 1 year lag autocorrelation is void if we can accept PIOMAS (and Maslowski's NPS?) volume

. This is because volume can 'bypass' any argument about extent behaviour and offer the system memory needed to propagate the impetus of SA driven changes.

I'm really stuck on how to move this issue on. I've been adding more complexity to my sea ice analysis spreadsheet - trying to compare the gain period (from minimum to maximum) and the loss period (maximum to minimum) After 2006 both show a substantial jump up, this suggests that there is a negative feedback in the gain part of the annual cycle to counter the greater ice-albedo effect after 2007, I still thinkk this is probably largely the Tietsche effect.

From 1980 to 2006 the slopes are 0.0105 (gain) and 0.1196 (loss) however they aren't statistically significant. I had been wondering if most of the loss of volume was actually due to amplification (i.e. ice-albedo) during the loss season. Gain minus loss shows a similar small trend this time downward, the running total of gain minus trend shows a substantial and accelerating drop in volume down to -13Mkm^3, which is of the same order as 2011 September anomalies (-11Mkm^3).

As I think this has turned out to be a blind alley I'm not posting the graphs, but if anyone wants to see them I will post - just ask.

With regards SA changing the statistics. Consider the two periods; pre-satellite (A) and satellite (B). During A there is little change in extent or volume, during B both decline strongly. If we assume SA is responsible for the sea ice decline in B then we're tacitly assuming that positive feedbacks are dominant, in such a situation the positive feedbacks will amplify any reduction in sea ice, this should result in wilder swings and an increase in the standard deviation. So still (despite your objections) I would expect the SD of period B to be greater than period A, if the process was driven by SA. Amplification can be involved, and will change these statistics, as the Crysophere Today anomaly series implies but this doesn't necessarily mean SA as the increase in both Gain and Loss parts of the annual cycle after 2007 implies interplay of both positive and negative feedbacks.

In closing I offer one bit of circumstantial evidence; the last time sea ice was in a worse state than now was during the holocene thermal maximum, and this was during a period driven by external forcing - insolation.


>"What we have IMO is a near linear forcing with amplification."

I tend to agree.

>"It seems to me that the extent argument of negative 1 year lag autocorrelation is void if we can accept PIOMAS (and Maslowski's NPS?) volume"

Again I tend agree.

>"I had been wondering if most of the loss of volume was actually due to amplification (i.e. ice-albedo) during the loss season."

Given a near linear CO2 forcing, I am inclined towards slope of 1979-1995ish being CO2 forcing with acceleration down from trend being amplification diven. Is that what you meant?

(My tendancy to agree can be quashed by appropriate data.)

>"I'm really stuck on how to move this issue on."

I'm also stuck without more data so have emailed Dr Schweiger to see if more thickness distribution information from PIOMAS could be made available. I need to work out a method to test my return to equilibrium thickness distribution idea. I don't see any huge problem in working out an objective test but if anyone knows any standard tests that are appropriate to the situation do feel free to tell me what to google.


How does an almost linear rise in CO2-content produce accelerating effects?
I’ve always been intrigued by the phenomenon of buffer-systems. Or, if you like, rubber band effects. In my field, plants and vegetation, I read a lot on acid rain during the eighties. I never forgot the analyses that were made on plant cell uptake of minerals through osmosis. The lower acidity of surrounding groundwater wouldn’t be a problem until the stage that dissolved aluminium particles could enter the cell membrane, thus progressively diminishing the plants ability to maintain its immunity system.
BTW I was never convinced by the BAU lobby that acid rain was a hype. Its continuing its silent devastation.
I wrote this to have an analogy for the way linear CO2 rise is actually having an accelerating impact on a lot of ecological cycles, in which geophysical processes like the demise of Arctic sea ice are embedded. All seems to be okay looking at the tree, while at the same time on hair root level every sign is flashing red… Bet you get what I mean.
Now I know several alumni have eloquently dismissed high probability of thresholds, irreversible change, in the state of the sea ice. FI the Tietsche effect is often mentioned as a possibly strong negative feedback. Now if OLR could be so strong as to venting excess heat right out to space from the Arctic, how could global heat forcing continue? Instead it does, and a lot of excess heat is stored in deeper ocean layers, while our main eyecatcher is SST’s. And especially during La Nina years global mean SST’s don’t look to be alarming. It does secondly through increased evaporated fresh water; a lot of excess heat is stored in that loop, too. It does third in the middle and upper troposphere. Read Davidson; he’s the guy measuring it in the sun disks. It reflects in an ever colder stratosphere. It does fourth in creation of rotten ice, globally, storing excess heat in warmer ice. Not necessarily melting it, but making it easier to flow and crack.
The Arctic sea ice cover is one of the most dramatic battlefields in all of these processes. The consequences for NH weather are obvious, and they are already occurring. The first year of more or less complete loss of summer ice cover wouldn’t suddenly unleash these weather consequences. It would be a next stage in what we are witnessing. And we do for enough time to know we got to do something. The first year of almost complete loss is special because it could be the wake-up call.
While most sources of info have failed us, only MODIS remains. On day 128 2012 it pictures an almost appeasing sight. Close to the white void in the mid twentieth century. But it is like the tree I used as a parable. We know it is at the end of the rubber band, like if it’s just centimetres from the surface of Kawarau River while doing a bungy jump. But because usually that band isn’t failing, I get back to the tree. Does it have more time, while there isn’t much lethal heavy metal in the groundwater at any particular moment?
Thus we have a dilemma called chance. Crump might be right, there is a plume of chance like in the projected course of a hurricane. But to be perfectly clear, a first ice free late summer isn’t unique for who cares to look around. Damaging changes in the biosphere are haunting us everywhere…

Daniel Bailey

Wayne Davidson's blog is here:

R. Gates


Your post May 9, 2012 at 3:00...Excellent, right on target, and very insightful.

The thing about SST's that many don't realize is they are really measuring heat flux from ocean to atmosphere, and are overall very poor indicators of gains in ocean heat content. Of course, SST's get all the headlines, especially during El Nino years, as we know soon after we get high SST's, all that heat leaving the ocean will enter the troposphere and show up as globally higher tropospheric temperatures.

The more important, and difficult to grasp dynamics of ocean-atmosphere heat exchange is of course the role that increasing greenhouse gases play in warming the ocean, not by putting more heat into the ocean, but letting less heat out. The alteration of the thermal gradient between ocean and atmosphere that greenhouse gases create is the hard to grasp dynamic that is really so crucial to what is happening. During La Nina years, and of course, cool phases of the PDO, we are seeing less net heat flow from ocean to atmosphere, and of course, some AGW skeptics will point at the relatively flat temperatures of the troposphere (from less heat flux from the ocean) during these La Nina dominant/cool PDO periods as a sign that AGW theory must be wrong, etc. They of course, don't want to consider that 2011 for example, was the warmest La Nina year on record...though this bit of data is critical.

When the next warm phase of the PDO begins and El Nino once more dominates the ENSO cycle we will once more see monstrous gains in tropospheric temperatures. The fact that we've not cooled more during this cool PDO, La Nina dominant period is a critical sign of the power of anthropogenic warming. The fact that we are not seeing record SST's right now, but rather, huge gains in ocean heat content should be clear indication of the tropospheric warming that is to come. Of course, the Arctic sea ice is getting hit from both sides, as it melts more from higher ocean heat content as well as higher tropospheric temperatures, which are of course magnified in the higher latitudes.

Daniel Bailey

To further illustrate the rise in temps of the La Nina and El Nino years:


R. Gates

Nice graph Daniel, thanks for that. Being so handy with graphs, if you get the chance, you ought to plot SST's versus global ocean heat content. This can be very illustrative for those wanting to see the pulse of energy flow on the planet and the true gains in the Earth's energy system over the past few decades. Given that the oceans contain the vast majority of the non-tectonic energy of the planet, for those suggesting that the Earth has "cooled" over the past few decades, this can be proven as the nonsense that it is by looking at ocean heat content. Also, graphing SST's versus ocean heat content is very illustrative at showing how SST's are best at simply being indicators of the rate of heat flux from ocean to atmosphere, slowing down during La Nina dominant/cool PDO periods, and ramping up during El Nino dominant/warm PDO periods.


Very nice graph, Daniel!

Thanks to all of you for the various discussions here and in other threads. I have finally made some time and brain power to read all of them and it's very interesting/useful.

Chris Reynolds


> Given a near linear CO2 forcing, I am inclined towards slope of 1979-1995ish being CO2 forcing with acceleration down from trend being amplification diven. Is that what you meant?

No it's not. I was figuring that the major positive feedback is ice-albedo, this is by definition active from March to September. So I was interested to see if the PIOMAS volume loss was mainly due to an increase in the spring/summer loss. I don't think it is.

I've been pondering this all day...

The following two scatter plots are for annual ice area; a) to make them compatible with Johanessen, b) as the CO2 forcing is year round. Forcing is calculated by Myhre's formula (IPCC 2007). The data are for 1979 to 2011, covering a wide range of ice loss slope. Yet in both cases a linear fit is good.

CO2 concentration.
CO2 forcing.

CO2 RF isn't strictly applicable as it's TOA RF, but if we can use that formula to offset the effects of band saturation, then the near linearity of both graphs shows that in a qualitative sense we can use CO2 over that period without much of a problem.

Then here are some graphs of 1979 to 2011 sea ice area and volume.
Sea Ice area.
Sea Ice volume.

Here's what I'm thinking. The area minima is set within the Arctic basin during Summer so is subject to amplification (ice albedo), the area maxima is set outside the Arctic basin during winter so is not affected by amplification to the same degree. It's hard to see how it can even be affected directly by volume changes within the basin because its volume is zero in the summer, so it has no memory to carry the signal of SA. Yet, apart from a post 2007 jump there has been only a small increase in the annual area range. Small when compared to the decreases in both annual average and summer minimum, which track each other quite well. It would probably have been better for me to include winter maximum, but annual range implicitly includes that.

If the 'forcing' driving the changes was due to Self Amplification (SA) then surely the annual range of area would increase - because during winter when the maxima is set this is being set in waters away from the Arctic basin, waters which contain no ice during the Spring/Summer, therefore volume cannot carry the signal of SA in those regions. But under the SA paradigm the minima is set within the Arctic basin where volume losses are carrying the SA signal forward. So under SA the seasonal range increases.

Instead we are seeing little change in the seasonal cycle range, as would be expected for a continually acting external forcing, because whilst there is amplification within the Arctic basin, both the remote and central regions are subject to that forcing. After 2007 the thinner ice has a higher Open Water Formation Efficiency (OWFE), so is a prime candidate for albedo-amplification, which reveals itself as a minima-driven increase in the seasonal cycle.

So I think the SA paradigm doesn't fit the available data, what we have in terms of both volume and area looks more like changes forced by a factor that applies in winter, as well as in summer (where it is amplified).

Chris Reynolds


Forgot to add:

I'm not comfortable with the idea of using time periods of four years or more to try to examine SA. Have you read Blanchard-Wrigglesworth (great name) et al, Influence of Initial Conditions and Climate Forcing on Predicting Arctic Sea Ice.
They find that four years and over and you're entering the realm where initial conditions have less impact and forcing is dominant.


I have received a prompt reply from Dr Schweiger. On a quick look at the data, the thickness distribution does look as if it is being regulated by the system with the distribution passing through trend more frequently that I would expect if more random variations were occuring.

However, the area or extent minimums are not lining up with larger deviations from the trend in the way I was expecting - they seem to occur at any stage in the cycles.

So I think the data does kill off that ice distribution idea.

So it probably is the ice growth feedback as not only you suggested but also Dr Schweiger suggested that others had found this.



Most of what you say (May 9 22:37) sounds sensible. But just wondering if you have adequately considered whether there are competing feedbacks that offset to hide their effects.

1. Albedo feedback in summer amplifying the reduction of ice during melt season.

2. Ice growth feedback in winter. This might partially hide the self acceleration in the summer minimum.

The presence of both near the end of the series is revealed by the increase in the range.

Chris Reynolds

I am assuming competing feedbacks, with your 1 and 2 being the major +ve and -ve feedbacks.

My reasoning all along has been that if self-acceleration (SA) has been at play it suggests a dominance of positive feedbacks - with ice-albedo being the prime candidate for the major player. So from my understanding - if negative feedbacks are hiding (offsetting) positive feedbacks this isn't a state of SA.

I'm intrigued: What has Dr Schweiger sent you?


I would like to share but don't feel I should without permission to do so. Sorry about that.

Chris Reynolds

I understand, no problem.

Kevin McKinney

Another good thing about this blog is the newsfeed. Here's an item from the sidebar that got me thinking:


Never thought of potential phenological mismatches arising from changes to the annual melt patterns before!


And something else from Don Perovich:

In the Arctic Ocean, the blanket of permanent sea ice is being progressively replaced by a transient winter cover. In recent years the extent of the northern ocean's ice cover has declined. The summer melt season is starting earlier, the winter freeze is happening later, the areal extent of the ice has decreased, and more ice is failing to last through the summer. A key uncertainty in this ongoing climate transformation is how seasonal sea ice affects and responds to climate dynamics as compared to the traditional multiyear sea ice. Tackling an important branch of this issue, Perovich and Polashenski analyze how the albedo of seasonal sea ice changes throughout the summer melt season. The ice's albedo affects how much sunlight enters the system and hence influences biological productivity, ice extent, and future rates of warming.

For four years, the authors measured the albedo every 2.5 meters (8 feet) along a 200-m (656-ft) stretch of seasonal ice off the northern coast of Alaska. They find that though the albedo of snow-covered winter seasonal ice is the same as that of multiyear ice, the equivalence fades rapidly with the summer thaw. They find that seasonal sea ice albedos experience seven distinct phases: cold snow, melting snow, pond formation, pond drainage, pond evolution, open water, and refreezing. Though the albedos of seasonal and multiyear ice experience similar transitions, the rate and extent for the two types of ice vary drastically with the potential for a large effect on the Arctic Ocean energy budget. The authors find that over the course of one melt season nearly 40 percent more energy would enter an ocean system with seasonal sea ice cover than one with multiyear ice.

L. Hamilton

"So I think the SA paradigm doesn't fit the available data, what we have in terms of both volume and area looks more like changes forced by a factor that applies in winter, as well as in summer (where it is amplified)."

This reminds me of an observation from Mary-Louise Timmermans' talk at the Montreal IPY meetings last month: bottom melting of sea ice can occur in winter as well as summer. If so then winter ice volume would reflect a balance between top freezing and characteristics of water just beneath the ice.

Mike Constable

I have been trying to understand how sea-ice melts. Images from the cameras at the north pole last year showed the sequence quoted by Neven above, but the edge of the floe became undercut. To me this indicates that the greatest melt is around the water-line, with the extra absorbed sunlight warming the ocean and promoting melting there. Any reflected radiation and warmed air/vapour from the sea surface will also tend to attack the ice immediately above the surface, increasing the tendency for the ice to be undercut.

On a larger scale icebergs can show this even more clearly, as they melt the wave-cut platforms rise above the waterline indicating that the bulk of the berg providing flotation below is melting relatively slowly compared with the parts at and above the surface.

From this I would understand that albedo is the driving force for melting more ice and the thickness of the ice determines how long the higher values of albedo last. Thus any increase in sea temperatures that delay freezing/promote melting will weaken the polar ice-cap. But the rate of freezing on the bottom of thick ice will be slower compared with thin ice so this will give some increase in heat-loss from the polar seas during winter as ice thickness reduces?

Seke Rob

For those also keeping a [3rd] eye out on the Antarctic and Global conditions, the CT data in an alternate format, with 8 months of added data since last update:


This year I've dubbed: "(Leap)Year for the Non-Green Conscious?

US polls indicate that the well-being of the planet does have a bearing on their voting behavior, a majority now. Let's hope ESA's CRYOSAT-2 http://www.esa.int/esaLP/LPcryosat.html gets lots more press, as it was so hailed by the denialista that PIOMAS was another poor model. Seems they´re close enough to debunk that notion.

-- Rob,

P.S. Where's that red line going in CTGB01?

Piotr Djaków

Meanwhile there was very warm in Kara Sea during 2011.


Winter 2011/12 anomalies:

The comments to this entry are closed.