« Arctic sea ice -- in pre-election perceptions | Main | PIOMAS November 2015 »


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

Pete Williamson


I dont know if the image is going to work but I thought it might be interesting to show the relationship between Sept ice area and ice gain during the following refreeze season. The image above shows the NSIDC Sept ice area vs piomas ice volume gain during the following refreeze season (apr vol - previous sept vol).

What it shows is that if a winter starts off with a relatively low ice area then more ice tends to form over the refreeze period. The simple physical explanation is that open water is the birth place of new ice and so more open water=more ice formation. In a sense this is acting as a negative feedback on ice loss.

The Sept ice area from the data set I'm using is 3.31km sq'd. If this year follows a similar pattern then we might expect that 17.7-18.7 km cubed of ice might form over the coming season and next years max be around 23.5-24.5 km cubed. So above the 2011-2014 lows and within the range of the other post-2007 years.

Of course there is a lot of weather to happen between now and then but I thought the relation shown in the graph was interesting.


Definitely an interesting graph, Pete. Thanks for posting.

In a sense this is acting as a negative feedback on ice loss.

Indeed, this ties in with the Slow Transition stuff that Chris Reynolds has been writing about for some time now.


crandles has also posted an interesting figure on the forum that indicates when autumn area is low and the ocean refreezes late in the season, most of the autumn snow falls into the ocean rather than on ice - meaning the ice has a thinner insulating snow cover and therefore grows thicker over winter than it normally would (then again it has also been suggested - by Peter Ellis, I think - that thicker snow cover might delay the onset of ice melt next spring, and that the overall effect on ice freeze/melt might be weather dependent). still it does seem quite plausible that there might be some kind of multiyear cycle where big crashes are followed by a couple of years of apparent recoveries before the next crash. negative feedbacks will almost certainly slow the transition to a seasonal ice free state, but it would be pretty surprising if 2012's record wasnt comprehensively beaten before 2020


Negatoive feedback is absolutely misused fot the Arctic refreezing no matter if of similar or more extent every Winter. It is just the result of the extreme coldness above certain latitudes during Winter, every Winter. It is not a response to global warming, that self-feeds further cooling. No evidence supports that, as opposite as Arctic amplification (more heat- snow and ice melting - more sun radiation absorbed - more heat). High-school stuff.

It limits the extent of summer melting because it requires the heat that it released during Winter. Howeve, every year is warmer and less heat is released and then required for melting in Summer. And, this is key imo, one bad year, and Arctic amplification can lead to such heat income that the whole Arctic melts in Summer no matter how much heat is required to melt the refrozen FYI extent. It can be abrupt, why not? Or it might be slow if that year doesn't come.


I do not disagree, navegante; even so, the rate of refreeze this year has been pretty amazing, if you are following the IJIS numbers.

I'm not sure what the explanation is. I doubt it can be tied to any sort of climate feedback. I hypothesize its related to the state of the water and ice at the end of the season being far different that was typical pre-2007. Unfortunately, I don't see any of our usual tools being able to quantify this difference. The changes happening are too great and too fast, and in fact the states of the ice we are seeing may be so transient and/or transitional we may never really understand what's happening.

Suffice to say, I'm watching the refreeze with great interest.


the rate of refreeze this year has been pretty amazing, if you are following the IJIS numbers.

You can say that again. 1 million km2 in just 6 days! That's pretty impressive.


not sure that I fully understand what you are trying to say, navegante, but there absolutely are possible negative feedback effects that will potentially slow down further ice loss, as the amount of ice gets smaller, and which might tend to cancel out positive albedo feedback. the concept of feedback here is just that as a continually increasing amount of heat is added, does the rate of ice loss speed up, slow down or continue in a linear fashion? greater venting of ocean heat in autumn due to less ice cover absolutely is an example of a negative feedback, and thinner layers of insulating snow could be another. understanding the balance of feedbacks is far more difficult than anything they teach in high school, but most ice models show slower melt rates as the amount of ice decreases, suggesting that negative effects win out. this has been discussed at some length on the forum, of course, including Chris's slow transition thread, and OSweetMrMath's ice modeling thread

Chris Reynolds


Thanks for that, it is a relationship I have observed and discussed before.

The definition of a negative feedback:
Negative feedback occurs when some function of the output of a system, process, or mechanism is fed back in a manner that tends to reduce the fluctuations in the output, whether caused by changes in the input or by other disturbances.

The feedback is that as open water area increases and late summer ice is thinner the volume produced over winter increases. This increased volume then carries through to the following summer and acts against volume loss by maintaining the volume of ice that needs to be melted the next year.

So it is a negative feedback as you correctly observe. However the effect is actually quite subtle and is summed up by Armour et al 2010, "Controls on Arctic Sea Ice from First-Year and Multiyear Ice Survivability".

Given the strong thickness–growth feedback of sea ice (Bitz and Roe 2004), where in a warming climate we can expect the thicker MY ice to thin at a greater rate than the thinner FY ice, and the fact that the ratio of MY to FY ice entering into the MY ice category each year is decreasing, it is likely that the difference between FY and MY ice survival ratios will decrease in a warming climate. If this occurs, the Arctic sea ice system would move toward a regime of decreased memory and decreased sensitivity to climate forcing...

So although it may seem it simply means more winter ice growth for more open water and thinner ice in summer, it actually has the power to reduce the impact of climate forcing. In that sense it is a moderating process as in a classic -ve feedback.

I have discussed this feedback here:
It is called the thickness growth feedback.

Note that in that post I show the change of winter warming as indicated by the measure 'Freezing Degree Days'. There has been no collapse in winter cold despite open water and heat venting in post 2007 autumns.


Point taken sofouuk and Chris, yet the nature of "carried thin ice throught to next Summer to act against ice loss" sounds more like a resistance mechanism. Simplistically in terms of linear eqs of forces,
- resistance is F = F_melt - F_new(X), being F_new larger as more ice is lost in previous summer but always smaller than F_melt, otherwise global warming would lead to ice gain.

- a negative feedback F = F_melt ( 1 - k* X) being X the anomalous ice loss extent, is also proportional to the "forces" needed for meltimg too. Note that the negative feedback must be proportional to F_melt, otherwise it is not feeding negatively back from the additional melting (the more melt, the more new ice, the less melt, etc. ..). Also the negative feedback might lead to oscillations more easily (supposing a linear approximation, which can be very inappropriate)

So this is too sim!plistic, probably a combimation of both and more factors

My 2 lire ;-)>


the rate of refreeze this year has been pretty amazing, if you are following the IJIS numbers.

You can say that again. 1 million km2 in just 6 days! That's pretty impressive.

While it is impressive it's not that unusual in October according to the NSIDC figures. In 2007, 2011, and 2012 it happened in 5 days. It happened in 6 days in 2006, 2008, 2009. In 2007 and 2012 a completely separate period lost 1M km^2 in 6 days. In 2007 and 2012 10 successive days started 6 day increases that amounted to more than a million Km^2.

According to the NSIDC figures it hasn't happened in either 5 or 6 days this year.

IJIS may have slightly different figures but they are rarely significantly different.

Pete Williamson

Chris, thanks for trying to explain it further but I'm too stupid to understand what you are saying. I tried the Armour paper but that too was beyond me.

It did strike me though that a large part of MYI loss in recent years has been to do with ice drift and export as much as to do with warming. IDK if Armour is unpicking the different factors, whether "forcing" in the paper is meant to cover all factors that lead to loss of MYI although it does look like the papers focus is on melting. As I said I dont really understand the paper so these comments are really questions rather than facts.

Pete Williamson

"been pretty amazing, if you are following the IJIS numbers.

You can say that again. 1 million km2 in just 6 days! That's pretty impressive.

While it is impressive it's not that unusual in October"

DavidR, Neven,
Just to continue the twaddle about how unusual (or not) the 2015 refreeze has been so far. I took the past 30days from this data set

Highest increase in this period is 2008 with 3.13million km sqd. Second is 2015 with 2.73, 2012 3rd with 2.70. The full list looks like this

2008 3.13m km sq'd
2015 2.73
2012 2.70
2013 2.69
2010 2.60
2014 2.45
2002 2.43
2004 2.27
2005 2.20
2003 2.11
2006 2.06
2007 1.88
2009 1.81
2011 1.71

post-2007 mean 2.41
2000's mean 2.30
1990's mean 2.15
1980's mean 1.70

So yeah impressive refreeze so far.

Cato Uticensis

In my opinion there are two main reasons for this quick ice recovery:

- The synoptic configuration which has been in place for several weeks with HP forming on the Beaufort Sea in response to lows prevailing on north Pacific and Alaska. This is preventing warm air from moving from Pacific to the Arctic. As a result, cold air is being conveyed to ESS and heavy snowfalls in Alaska.

- The heavy snowfalls that have affected Russia in the last few weeks, contributing to a significant cooling of Siberia.

As a result I would expect further freezing on Laptev and significant cooling of ESS in the next 7-10 days.

I am much less enthusiast of reflections over where the "heat is hiding" or "gone". We cannot simply quantitatively measure the heat contained in the arctic sea. I'm just an engineer. No one will convince me that ice is forming but heat is hiding below, like a thief, ready to jump out and say "here I am! now it's your problem!". If ice forms it means enough latent heat is being released for it to form. Viceversa, for melting. Cold leads to ice formation and heat to ice melt. Cannot figure out of heat leading to ice formation, or ice forming "in spite of heat". It's too much for a humble chemical engineer.

In my opinion we also tend to give too much importance to sea surface temperature anomalies. Obviously these temperatures are indicative of the status of the most superficial layers of water. Whoever is familiar with sea temperature, even at lower latitudes, knows that a few windy days and low temperatures can significantly affect the SST. And this is just what has happened in the Arctic where in a few days, huge positive anomalies have been turned into...ice.

Anyway the ice growth seems to be levelling off. There is still a long way to go before arguing that something is really changing in the arctic ice trend. One swallow does not make a spring, unfortunately.

Chris Reynolds


It is reasonable to view it as a resistive mechanism. It will be eroded as the winter cold lessens with further AGW. In the end there will be a situation where winter ice will grow, but there will not be enough to resist the summer melt and summers with practically no ice will become the norm.


Sorry, I'll try to do better, hopefully this will be more clear.

To start this, note that thicker older ice thickens by mechanical deforming of ice as floes are compressed together. Whereas younger ice, especially first year ice thickens as heat flux through the ice over winter causes extra ice to grow on the underside of the floes (thermodynamic thickening). The rapid extent increase at present is due to thermodynamic ice growth, it is a very rapid process when the weather is cold enough.

This all ultimately starts with something seen in the submarine and ICESat data and explained by Bitz and Roe.

What was seen was that thicker older multi year ice was thinning more than younger ice over the period of submarine data and the ICESat mission. PIOMAS shows something similar, all of the loss of volume has come from thicker ice. Bitz and Roe explained this odd behaviour by showing that the thicker older ice has a longer 'memory' then the thinner first year ice.

Take a mass of ice five years old. It takes five years to top up this ice with more five year old ice, but that means the ice sticks around for five years so it has a long memory.

Take a mass of first year ice, it takes just one winter to top up this first year ice. So it has a short memory.

This talk of memory is probably best explained by looking at what happens when a good melt summer takes a 'bite' out of the ice.

The mass of five year ice is hit by a summer of good melt weather, you'd need five years of normal summers to make up the lost ice, but with overall forcings acting against the survival of ice this 5 year old ice doesn't get the chance to recover. The five year ice 'remembers' the good melt summer and carries its impact on the ice loss into future years.

In that good melt year the first year ice is also hit. But the next winter is still cold enough such that the ice can grow back, by the start of the following summer there is as much first year ice as there was before the loss. The first year ice has in effect 'forgotten' the impact upon it.

What Armour et al are saying is that as the pack transitions to a mainly first year state the increased amount of first year ice reduces the average memory of the pack. This is because each winter the 'slate is wiped clean' by first year ice regrowing, and the ice 'forgets' what happened to it the year before. As you have shown, the more open water there is the more first year ice grows back, and the more the 'memory' is wiped.

The climate forcing can be something like the cold of the ice ages or warmth of the interglacials. But in the present situation the forcing causing the decline of sea ice is human driven global warming. At the regional scale this can be direct forcing from more CO2 (more back radiation of infra-red), indirectly though warmer moister air causes more infra-red to radiate to the ice from the sky. Also warmer moister air can increase clouds which may play a role, an unkown player is ocean warming - but this must play a role. However it is safe to skip over the actual complex roles and use the shorthand that the 'forcing' acting to reduce the ice is ultimately from human driven global warming (AGW).

Crucially this forcing acts upon old and young ice alike, it's just that the older ice has a longer memory than the younger so is influenced more.

Export of ice plays a role, but my understanding is that there isn't much of a trend in export through the Fram Strait. I'm not at all convinced by claims that export is driving the loss of ice, export plays a role, but the driver is AGW.


an interesting comment (from Nevens link),

    Chris Reynolds said...
    PS, should have added - I have read that more flat first year ice means more melt ponds other conditions equal. But can't remember where I read that.
    15 June 2015 at 11:34

Can anyone explain who may have first said this and if it is factual in anyway?


@ Cato,

    re: "Anyway the ice growth seems to be levelling off. There is still a long way to go before arguing that something is really changing in the arctic ice trend. One swallow does not make a spring, unfortunately."

Given the sea ice extents curious s-bends as of late I would have to agree that the story is far from being told!

John Christensen

AbbottisGone; On FYI and melt ponds,

Yes, there are more melt ponds on FYI due to:

1. FYI more often than second year or MYI has a flat structure, as it has not yet been compacted. However, far from all FYI is flat, as it is subject to the same forces compacting MYI and when the pack is pressing against e.g. the Laptev coast or some of the Arctic islands you will very quickly see the FYI gaining in thickness, as it is being compacted.
To what degree the FYI is flat therefore depends on weather and when the FYI was created.

2. Secondly, due to FYI being thinner, when melt ponds are created, the water on top of the ice will be more dense and therefore create a convex surface (bowl shape), so that more water can be contained on the ice. I have heard this argument from measurement of melt ponds, where larger melt ponds tend to be slightly deeper in the middle due to the pressure.

For these reasons, back in the 80s and 90s, when melt ponds developed during summer months, they were likely to be less extensive in size and volume compared to today (weather conditions being equal), but I have seen no actual comparison.

Pete Williamson

Thanks Chris it makes a lot of sense now. But I'm going have to disagree with you on the final sentiments.

There seem like plenty of reasons to think a variety of potentially natural processes that have contributed to ice loss along side AGW.

Fram Strait export appears to have a positive trend over the past couple of decades. http://the-cryosphere-discuss.net/9/4205/2015/tcd-9-4205-2015.pdf
Atlantic water pulses of warm water seem yet to be attributed and look unconvincing as arising from AGW http://icesjms.oxfordjournals.org/content/early/2012/04/18/icesjms.fss056.full.pdf
And atmospheric circulation changes since ~2000 have also favoured ice loss. http://orbi.ulg.be/bitstream/2268/175926/1/tc-9-53-2015.pdf

I understand these processes have themselves been linked with AGW but that too seems yet to be confirmed. Just to be clear I dont discount AGW just see it as part of a complex mix of natural and anthro that have contributed to ice decline. Honestly it surprises me that you have confidence in attributing all this ultimately to AGW.


Pete Williamon wrote:

While it is impressive it's not that unusual in October"


Meanwhile, the "laptev-bit" has been nearly filled, with as a result a considerble slown down in ice creation.

And there is still no ice bout 250 km North of Wrangel Island, which is unusual as well as impressive in the end of October. :-)

Chris Reynolds


As calculated by Wipneus (and as I have read in a paper), the volume export has declined.

Halvorsen et al find an increase in export of extent/area, this fits with greater mobility of the sea ice as it has thinned. However over the same time the ice has thinned so overall volume export has declined slightly.

Another factor here is the Arctic dipole, and this is related to the findings of Belleflamme et al. Zhang et al (Recent radical shifts of atmospheric circulations and rapid changes in Arctic climate system) find that since 2003 the AD has become more dominant. Overland finds that the AD plays a role in summer loss of ice since 2007 (The Recent Shift in Early Summer Arctic Atmospheric Circulation). The AD will play a role in export as the strong high pressure over Greenland propels ice through Fram, hence the statement by Belleflamme that "A significant increase in the summertime occurrence of a high pressure area over the Beaufort Sea, the Canadian Arctic Archipelago, and Greenland has been observed since the beginning of the 2000s, and particularly between 2007 and 2012."

That this also plays a role in the 1930s/40s ice loss event (Belleflamme) is interesting as although not shown in the literature, I continue to suspect strongly that the AD is a pattern emerging in response to ice loss. Bluthgen et al (Atmospheric response to the extreme Arctic sea ice conditions in 2007) state that:

"A pronounced low sea level pressure anomaly over the Eastern Arctic (200 Pa) reinforces a sea level pressure dipole over the Arctic that has been observed to become an increasingly important feature of the Arctic atmospheric circulation in summer. The anomalous pressure distribution contributes to sea ice transport from the Eastern Arctic and is likely to reinforce the original sea ice extent anomaly. The results thus support assessments of observational data over recent years that sea ice loss may feed back onto the atmospheric circulation in the northern hemisphere."

Indicating that what may be going on is a feedback. Blaine on this blog has suggested something similar.

So much of what you sugget may be natural seems to me to be a response to the loss of ice.

That leaves Beszczynska-Moller et al and Atlantic Water. AW is seperated from direct contact with the surface as it lies deep being more saline. Speculation continues as to the role it may play in ice loss, however the role remains unclear.

For me the strongest reason to point the finger at AGW for ice loss is this:

In every climate model, take away the anthropogenic forcing and the ice doesn't decline (grey), include the anthropogenic forcing and the ice declines (colours).

For more detail about what I think has driven sea ice loss see this post:
Sorry it is long.

Chris Reynolds


I agree with what John says, also you might find section 2.2. of the following paper informative.

Pete Williamson

Chris, a couple of points.

You say "That this also plays a role in the 1930s/40s ice loss event (Belleflamme) is interesting as although not shown in the literature, I continue to suspect strongly that the AD is a pattern emerging in response to ice loss."

It appears you are saying AD caused the ice loss and ice loss caused the AD. I guess you can have it both ways but it still looks less than an ideal explanation.

I think you also need to look a little more carefully at the Belleflamme graphs. The other time when the AD is dominant is the 1950s-1960s not 1930s-1940s as you say. This is not a period when ice losses were so great. In fact, from memory, any attempts to reconstruct ice extents in the 20th century have tended to suggest ice has been relatively stable. Past arguments have been that the 1930s-1940s warming of the NH was too short to have much impact on ice. But I think the Belleflamme adds something to that story. 1940s the NH was warming but atmospheric circulation didnt favour ice loss, outcome stable ice conditions. 1950-1960s Atmospheric circulation favours ice loss but NH temps are cooling, outcome stable ice conditions. Present conditions atmosperics and NH temps favour loss and we see significant loss. At no point do I rule out AGW but I also dont think the data supports the idea that other factors are ruled out, in fact NH warming appears necessary.

The model comparisons dont show what impact the dynamical processes have had on the ice because the models cant reproduce the specifc conditions that have occured in the real world. I think this is generally expressed that natural processes are an initial condition problem so remain untested. For example no CO2 runs that are 'forced' with Belleflamme's atmospheric circulation would be a better test. Put another way the natural runs work as a negative control for the anthro runs but they dont themselves work as good simulation of the actual natural conditions that we experienced so they dont really work as a test for natural impacts.

Chris Reynolds


It has been many months since I read that paper, so I have misremembered it slightly.

From the paper:

"Thus, at first glance, the SLP-based types 2 and 4 can both be associated with a positive DA phase, while the other types, particularly types 1 and 3, can be related to a negative DA phase. During positive DA phases, the sea ice export from the Arctic Basin through the Fram Strait and the Barents Sea is strongly enhanced, which is particularly effective for important sea ice loss during summer (Wang et al., 2009).

Fig 2, panel b, Type 2 shows a massive peak in ERA during the 1940s. Type 2 is far more like the post 2007 summer average. But I had remembered the ERA results for the 1920s and 1930s, so yes, the current situation is less like the 1940s regards the Arctic Dipole.

However note the final sentences of the paper.

"...some studies (e.g. Petoukhov and Semenov, 2010; Inoue et al., 2012; Bezeau et al., 2014) suggest
that atmospheric circulation changes can be induced by SIE anomalies. Therefore, the recent SIE decrease could be a trigger of the recent atmospheric circulation change inducing in turn a SIE decrease, suggesting a positive feedback."

The models generate a response to the anthro forcings in sea ice, atmosphere and ocean. whether they show a dipole anomaly is not relevant to the point that when anthro forcings are taken away the ice fails to decline. Note that Bluthgen finds their ice/atmosphere feedback in a GCM when studying the 2007 summer. However Zhang's finding that increased AD dominance after about 2003 may in part explain the more aggressive loss of extent and volume in reality as compared to GCMs.


Mapping the future expansion of Arctic open water

Sea ice impacts most of the Arctic environment, from ocean circulation and marine ecosystems to animal migration and marine transportation. Sea ice has thinned and decreased in age over the observational record. Ice extent has decreased. Reduced ice cover has warmed the surface ocean, accelerated coastal erosion and impacted biological productivity. Declines in Arctic sea-ice extent cannot be explained by internal climate variability alone and can be attributed to anthropogenic effects. However, extent is a poor measure of ice decline at specific locations as it integrates over the entire Arctic basin and thus contains no spatial information. The open water season, in contrast, is a metric that represents the duration of open water over a year at an individual location. Here we present maps of the open water season over the period 1920–2100 using daily output from a 30-member initial-condition ensemble of business-as-usual climate simulations that characterize the expansion of Arctic open water, determine when the open water season will move away from pre-industrial conditions (‘shift’ time) and identify when human forcing will take the Arctic sea-ice system outside its normal bounds (‘emergence’ time). The majority of the Arctic nearshore regions began shifting in 1990 and will begin leaving the range of internal variability in 2040. Models suggest that ice will cover coastal regions for only half of the year by 2070.


John Christensen

I have noticed a weak low pressure area around the Pole, which seems to stay with us a few days:


It seems odd to me that the LP is quite cold (-30 to -40C at 2 meters), so I wonder if this is a regular occurence that I just have not paid attention to?

The comments to this entry are closed.