« PIOMAS June 2017 | Main | Will Petermann calve another one this summer? »


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

Robert S

One (simplistic) way to think about this may be that May/June melt ponds equate to X cm more thickness loss, thereby causing thicker ice to melt out where there is extensive melt ponding. I have this suspicion that melt ponds will be less critical on thinner ice (all other things being equal... which they never are), because that ice will melt out in any case. So the question for 2017 may be how the combination of thinner ice and fewer melt ponds impacts the final extent. I'm tempted to think that melt ponding may be less critical this year, because that thinner ice may be going no matter what...

Robert S

Where some of the thick ice has gone: http://www.cbc.ca/news/canada/manitoba/climate-change-study-1.4157216


Another factor pointing to a possible rebound: concentration is looking better than the same time last year: https://sites.google.com/site/arcticseaicegraphs/concentration-maps/sic0613

I know this doesn't show thickness, but still... reason for hope that things might not go all runaway melt this year.

I'm tempted to think that melt ponding may be less critical this year, because that thinner ice may be going no matter what...

That could very well be true, Robert. And I'm definitely not saying that this is putting a spanner in the works of a new record.

Let me put it this way: If there was extensive melt ponding going on right now, building up lots of melting momentum for July and August, I'd be willing to bet (big) money that this year all records will be broken. As it is, everything is still open.

Melt ponding could still get going, or like you say, it may not matter all that much because so much of the ice is thin.

One other thing speaking for 2017 staying a very serious contender, is the current weather forecast. Although the spell of (relatively high) high pressure is now coming to an end, it's not like low pressure is taking over. In fact, it seems a weak dipole is going to form over the coming week.

SSTs are also on the rise. I'm waiting for June 15th to come around, as I have DMI SST anomaly maps for that date from 2012 and 2016.

Another factor pointing to a possible rebound: concentration is looking better than the same time last year:

Paddy, I think that this is another indication of low melt ponding. Normally, when the ice starts to turn bluish, that's also the start of large swathes of green/yellow/pink on the Uni Bremen SIC map. Because these swathes do not stay in the same place (most of the time), they're not really an indication sea ice concentration (ie open water vs ice), but rather of melt ponding.

Jim Hunt

Robert - Surely whether a particular area of sea ice "will melt out in any case" has no bearing on its albedo today?

If that area of ice is currently covered with snow instead of melt ponds it's reflecting most of the short wave radiation it receives, rather than absorbing most of it.

For those that follow such things, JAXA/IJIS/ADS Arctic sea ice extent for 2012 just dropped below 2017:



And then there was a huge hole in the ice around the border between the Chuchki and the CAB. Not much melt ponding, but, significantly, that area has been showing on off signs of melt ponds. Possibly that the ponds are forming and draining, all in one day, due to the fragmented nature of the ice.

They could, of course, also be forming and freezing all in one day too. I guess we won't know though unless someone goes and looks.

This, to me, is an indication of times to come. We saw this back in the autumn of the 2000's where the summer melt impacted the pack so badly that holes just opened up before the freeze set in again.

Now, though, it would appear it is happening in early summer. Which means we're seeing late summer ice conditions in early summer.

Not something which bodes well for the ice.


A thought for you Jim, 2017 crossed 2012 two days after 2007 and 0.5mkm^2 lower. 7 days later 2016 touches 2012 and is about the same difference, today, between 2017 as 2017 is from 2007.

I see it's 26F and has been snowing at Barrow. Yet the melt ponding is advancing rapidly and the landfast ice appears to be almost on the point of destruction. Might take a few more weeks though.

Sadly we lost the forecast and the data after 2011.


Susan Anderson

That CBC thing about Bell Isle icebreaking is interesting. Here was wunderground (Henson, including cite of Jim Pettit) a few days back.

I'm keeping an open mind, but observe in my most recent check of Nullschool that almost the entire Arctic Ocean surface temperature is above 0C. Since the melt point of saltwater is about -2C, doesn't that affect things as well? I realize warmth alone doesn't do the trick in a hurry, but it seems something that might affect the rest. Daylight will remain static until mid-July (yes, I know, cloud cover and weather ...).

Susan Anderson

Ah, for comparison I should have been looking at this.


Jim, Although 2012 has now passed 2017 it has also completed its' June dive and now declines at a very sedate rate for the next two weeks.

2011 crosses below 2012 in two days and it is likely that 2017 will do so within three. 2017 is only 76K behind 2012 and has the second greatest loss this month.

With the much thinner ice this year it is easy to envisage 2017 with a substantial gap below 2012 by the end of the month.


Talking about melting momentum and melt ponding, a huge area of ice went from ice to water and back to red/yellow on the concentration charts in two days on Bremen. Right on the Chuchki/CAB border.

I'm wondering if the ice is so thin now that it can melt pond and drain again in 24 hours. Which would invalidate the 3 day scenario and would produce much more unpredictable ice conditions.

It could, of course, just be freezing up again, but it must be pretty close to 24 hour sun there. Unless the weather closed in. But even then I'm seeing melt ponding at Barrow staying melt at 26F with snow.

I guess we'll know if someone decides to overfly or sail there and investigate.

Robert S

Jim Hunt: Your point about the albedo impacts is certainly accurate. I guess that the impact of melt ponds/more radiation absorption would be earlier melt, followed by more heat entering the arctic ocean, which certainly could be significant over the longer term, and would probably show up most during the refreeze.

The current fragmentation of the ice is also having some pretty interesting albedo impacts.

Wade Smith

[snip, too off-topic, too long, go to the ASIF; N.]

Wade Smith

[snip, see previous comment; N.]

Wade Smith

[snip, see previous comment; N.]

Hans Gunnstaddar

Ok, big prediction time:

Within the next 7-10 days there will be a sharp drop in extent below 2012 level. If this prediction is correct I'll explain later how it was forecast.

Rob Dekker

Why not explain your prediction beforehand ?
Is it PIOMAS record low volume ?

Also, 2012 doesn't decline that much over the next 10 days. Only 67 k/day if I recall. If you claim a significant decline over the next 10 days, why not go with 2010 ?

Hans Gunnstaddar

"Why not explain your prediction beforehand?"

Because it's unknown how accurate this predictive system is, but to put it to the test it has to be documented with a post (on 6/16). In fact, should this succeed the 1st time, 2nd and 3rd predictive posts will also be needed to establish increasing probability of accuracy. If it works for a 3rd time in a row, the method will be divulged.

John Christensen

Thank you very much for the melting momentum update Neven!

I would agree with Dr. Schroder that we should see slightly below average sea ice meeting this season due to lower melt pond fraction, seeming caused by lower than average high north temps (DMI 80N).

I am also very curious about the impact of Greenland, which often is given little space for consideration for the sea ice: The relentless storms last fall has caused much higher precipitation than usual (http://polarportal.dk/groenlands-indlandsis/nbsp/isens-overflade/ ), and the measured albedo of Greenland has improved as a consequence.
In 2012 we saw the heat dome over Greenland, which caused very warm air to flow down to surrounding sea surfaces, but this year the ice cap will help keep things temperate.

Back to lurking.


Thanks, John.

Since 2012, we haven't seen a melting season with a high amount of preconditioning through melt ponds. Maybe it's a sign of some negative feedback kicking in.

Last year still went low despite this lack of melting momentum, and this year the ice is supposedly thinner than it has been (although the gap between 2017 and 2012 has rapidly disappeared, according to the latest information on the ASIF). And there's already a substantial amount of open water in the Beaufort, Chukchi, East Siberian and Laptev Seas.

So, while there's a lack of melt pond formation, it may be that the influence thereof is becoming less of a factor. In this sense, and as always, it will be interesting to see how things play out this year.

Of course, once we get a year that is low volume-wise and heavy preconditioning at the start of the melting season...


Whilst the Beaufort and the Chuchki seemed to have slowed down (for now), the Kara continues to be low area and the ESS and Laptev have large melt areas just starting to grow. Also the Hudson Bay is now in serious terminal decline.

All of which leads to an accelerating melt for the next week or two. Not radical but it doesn't need to be.

I'm watching extent right now as it's much easier, for me, with chartic, to do comparisons with prior years.

For me it looks like 2017 will drop below 2016 and 2010, in about 7 days. It is then likely to continue on that track without excessive melting.

Whilst this is not volume, volume will be impacted by the significantly lower extent especially as there is a lot of broken ice, with melting going on, where there should be extremely thick ice and firm pack ice.

Barrow looks to be likely to lose it's landfast ice in the next week or two. Putting it firmly in the 2003/04/07 bracket.

The most interesting thing here is that, so far, the only thing which has really been exceptional is the prior winter ice conditions.

Which leads me to wonder just how much of the final demise of the summer ice pack will come from exceptional winters rather than exceptional summers.

All melting seasons are different and they are drastically impacted by the weather. What is changing is the position from which they start.

I still believe we will get a more similar season to 2007, than any of the others. Whether it causes a new minimum with the same kind of impact remains to be seen.

My future forecast remains the same as it did last year. 2022 is my view for truly game changing events. Although I do expect a fairly dramatic melting season eventually in 2017.


I've been following the chat for a long time and eventually I decided to jump in.

There's a strong cyclonic circulation in the central Arctic going on since a few days ago:


a direct consequence of the low pressure over the area:


I'm curious about what effect will that have on the ice at the edge in the Kara and Barents Seas.

Jim Hunt

One intriguing recent development is the expansion of this year's "Laptev bite" towards 80N:

At 4 days out this is by no means a certainty yet, but ECMWF is currently forecasting that the low pressure will develop into a sub 970 hPa cyclone:


I'm curious about what effect that will have, if it comes to pass!

Jim Hunt

P.S. See also this post by Neven on the Arctic sea ice forum:


A comparison between the ECMWF deterministic and ensemble forecasts of the potential cyclone.


Thanks, Jim.

And next week I'll be showing how both these forecasts fared (for June 24-29), when compared to what happened from day to day. It's high time I learned some more about weather forecasting. ;-)



Jim , that Low is just about right indicating early stages of disintegration of the main pack ice, a tad earlier than I anticipated, I don't think it will persist as long as it will be in later August. Mean time I would trade some of my prediction skills for persuasion abilities, if I could, things are looking grim, many people don't know it, these are the voters of interest.


In the vast area of Arctic Ocean If there is more sea ice snow than 2016, as Neven's melt pond graphs suggest, the collapse of sea ice will appear unnerving, sudden, unexplainable, but if we study a smaller area, such as Baffin Bay, Hudson Bay and Barrow Strait:


the larger Arctic Ocean Pack becomes easily predictable by the machinations of macro regions. Is a fact of physics you can learn on such wonderful science shows such as PBS NOVA (one of USA's greatest jewels, to be cut off if dumb dumb Trump gets his way).
Water effects can be literally modelled on a very small scale. And so the Arctic Ocean mini models abound in nature, but having a lab would help further replicate.

http://www.pbs.org/wgbh/nova/earth/megafloods-of-the-ice-age.html .

John Bilsky

Just took a 12 second plunge in the Beaufort Sea at Prudhoe Bay. The air temperature is around 60ºF and I've been told that the sun won't set until Aug. 1st at this latitude. The shoreline is essentially ice free but not for far. However the ice looks horribly rotten. I suspect the majority of the melt ponds are highly disguised by snow slush and in my uninformed opinion may *appear* to be "solid ice" instead of what they likely are.... slush ponds. How far this can be applied to the entire Arctic is anyone's guess. Back to lurking and waiting.


John, the sat images since a few days ago show large extensions of the Beaufort sea ice no longer as pristine white floes but as ice clearly tinted in blue. In agreement with your in-situ observations. Mo more snow disguising melt, well, perhaps further North closer to the CAB

Bill Fothergill

@ John Bilsky "Just took a 12 second plunge in the Beaufort Sea at Prudhoe Bay ..."

If these "in-situ observations" are being supported by the NSF, I really think that you should significantly extend the plunge duration.


John Bilsky

Thanks guys.

@ Navegante.... We will see soon enough huh?

@ Bill F. :-) You funny guy. :-) The water was cold but looking back at the experience it was certainly possible to stay in longer. The initial shock was a bit much for the cranium. Surprisingly it was quite comfy after getting out.


It looks like the Barrow Landfast ice just started to exit. Whilst not all the point has gone, a significant portion in front of the camera has gone and, mostly, it seems to have melted out rather than been pushed away.

If the rest goes today, it will equal 2007.

Jim Hunt

Neil - Actually it would seem to have been "pushed away" by the recent easterly winds?


Meanwhile O-Buoy 14 has captured the onset of surface melt in the heart of the Northwest Passage:

Amongst other things!



Jim, I watch it daily over multiple times in the day. The near shore ice went first to melt. Then the ice outside of the ridges went second and melted out. It was only after the significant melting in the outer ice that the wind took over.

It's not like the 2016 events where the wind quite literally shunted the ice all up the coast.

I watched the last 24 hour video. Yes there was some drift from the wind, but no real solid push of all the ice like we've seen before.


I just looked again. It's pretty clear, that is melt.



Very nice photos Jim

If there was normal surface snow thickness the ice would appear vivid blue green, there is enormous amount of snow on sea ice in that area,
it shows, therefore its all going to be gone sooner perhaps than ever. Great to see the action nearly live!


And, as of the latest image, at the time of writing, the temp is 52F and that is bright sunshine.

Hans Gunnstaddar


NeilT, just got back from business trip to Michigan and was watching Bremen on a mobile phone while away. Not sure if you've noticed, but some serious assault on CAB taking shape the past few days. May be on the cusp of some serious extent decline.


I did Hans. Looks like all that snow is suffering serious melt now. When it's gone we'll see what the shape of the ice is like for the rest of the melting season.

Barrow ice pretty much vanished in front of the camera yesterday, it's still lingering up to the point but wont last long with clear skies and high temps.

What is most interesting to me, right now, is the state of the ice from the Kara right into the pole. Last year there was a barrier of 2M ice blocking all intrusions from the Atlantic. It doesn't seem to be the same in 2017.

Still, as is being said on the forum, the fact that melt has really kicked in only at the end of June will impact what happens at the end of August.

Of course, assuming the thickness estimates are correct. Volume is still at an all time low and the ice is in a seriously degraded state given the slow start.

Then again I still see 2017 tracking roughly, in extent, with 2007 and still well below. We shall see.

I see three key points in the historical extent chart I'm following. First week in July, Mid July and the first week in August.

More watching.

Hans Gunnstaddar

Yes, definitely agree on situation in Kara and will watch for those extent points to come.

You may be right about extent this year tracking similar to 2007. My view remains a new record to exceed 2012 and the reason of course is low volume and poor condition of ice accelerating melt moving forward, similar in context to 2016, i.e. not looking like it will necessarily have a big extent drop, but will.



The drift indicates your assessment right Hans. What happens here
is a false extent signal due to mixing, is common now a years. For 2017, we will get a very strong something, either cyclone storm or pervasive high pressure, sea ice remainders will settle on one side or another revealing the already open water a lot more.

Is it possible to have melt ponds graph on Arctic Sea Ice graphs Neven? I find it interesting for revealing the thickest snow layers.

Hans Gunnstaddar

"...false extent signal due to mixing..."

Agreed, Wayne and like you posit, once something occurs to push the slush to one side or the other, open water will pervade.

Below is a new post by Scribbler; 'Arctic Sea Ice Melt Analysis: The Concerning Development of a Beaufort Warm Pool During Late June'


"Now, the development of a pool of warm water in the Beaufort Sea even as a strengthening ridge is poised to inject more heat into this key region threatens to increase sea ice melt pressure as we enter mid-summer.

During the coming days, a ridge extending over Northern Canada is predicted to drag a surge of warm air across both the Canadian Archipelago and the Beaufort Sea. Over-ocean temperatures in the Beaufort are expected to considerably vary above the norm — hitting as high as 48.6 degrees Fahrenheit (9.2 C) in some regions. An over-ocean value that is around 7-9 C above average for this time of year."


Well the barrow ice is not quite gone and, still, is not going anywhere on the wind. It does seem to me that some of it is protected by the layer of snow that it got.

Which presents a question. A thick layer of snow will protect the ice, both from freezing and melting. However when it melts that presents an absolutely huge heat transfer in water, both to the ice below and the water below that.

I wonder if anyone has actually calculated the heat transfer of that much liquid water (sitting on top of the ice) and what it's impact would be as it drains away.

Does snow absorb the heat in melting, only to release it like a bomb later?

Because, in that case, things could get a little rapid.

Hans Gunnstaddar

Well NeilT, even if there aren't any definitive answers to those questions, this melt season should provide a great case study for comparison in the future. I for one am very curious to see if it throws a monkey wrench into the 5 year repeating cycle.


Hans, talking about comparisons gave me a thought.

What are we really talking about here if we are talking about an emergence of a 5 year cycle?

Given that this should be the culmination of another 5 year cycle, what would extent and area look like if it were an event of the same magnitude?

To define magnitude, I think we need to look at the long term average extent and average area, at its September minimum, then the divergence from that average at each cycle peak melt.

For 2007, average 1979 to 2006 the average September minimum is:

Extent 6.86
Area 4.84

2007 actuals

Extent 4.32 divergence from the average September minimum 2.54
Area 3.12 divergence from the average September minimum 1.69

For 2012, average 1979 to 2011 the average September minimum is:

Extent 6.54
Area 4.63

2012 actuals

Extent 3.63 divergence from the average September minimum 2.91
Area 2.82 divergence from the average September minimum 1.81

Now if we assume the cycle is continuing, then we would expect to see, at least, the same change and, in a declining arctic, perhaps more.

Rate of change in the divergence from the peak melt average September minimum between 2007 and 2012

Extent 0.38
Area 0.12

In essence that is all that requires to change, as an increase from the divergence from the satellite record average September minimum, for this to be a continuation of the 5 year cycle.

In real terms I make that the following divergence from the satellite record average September minimum in 2017.

Extent 3.01
Area 2.57.

Also the average September minimum has declined, of course, including 2012 and now stands at:

1979 - 2016 average September minimum

Extent 6.31
Area 4.49.

So, if my spreadsheet is correct, if we see a minimum of 3.01 in extent and 2.57 in area or something very close, then this is a continuation of the trend. For me.

All figures from the NOOA monthly data for September


Hans Gunnstaddar

"So, if my spreadsheet is correct, if we see a minimum of 3.01 in extent and 2.57 in area or something very close, then this is a continuation of the trend. For me."

I would say your numbers are good. I have 2.58 for extent but didn't run numbers on area, so let's say for extent a range of 2.5-3 is what's expected for this 5th year in the repeating cycle.

Rob Dekker

Sea Ice Prediction Network finally published the June report (based on May data)
The median Outlook value for September 2017 sea ice extent is 4.43 million square kilometers.

Looks like an "in-the-middle" prediction, possibly balancing low volume data with low melting pond data.

Hans Gunnstaddar

Could be an 'in the middle', Rob, however there's still approx. 75 days of melt to go and we're just now in the meaty curve of ice melt, with 2017 closely paralleling 2012 extent. It's like a horse between Secretariat and Sham at the Belmont Stakes in 1972 (as seen on YouTube). Who knows which one is which, i.e. if 2017 falls back like Sham or races ahead like Secretariat.

Hans Gunnstaddar

Correction: 1973


In those extent and area stats, the differences between


1.63 extent, 1.19 area


1 extent, 0.81 area.

I have an issue with the last stat for 2016 as chartic says it was 4.13 and this csv says it was 4.72. If we see a 2007 style drop between 2016 and 2017, then we're well into new territory. If we see a 2011/12 style drop then we are still continuing the trend. If we see the same reduction in drop, from 2007 to 2016, we don't get a new record and the cycle stalls.

Onwards to September. This is not an exceptional melting season, so far. The most exceptional thing about the melting season is how fast it is vanishing (tracking 2012), with an average input of heat and insolation.

Which is, in the end, what global warming is all about. Step changes where the environment has changed so much that smaller input is required to tip it.

Hans Gunnstaddar

NeilT, From Neven's latest Piomass update is the following:

"As Jim says in the PIOMAS thread on the Arctic Sea Ice Forum: The average July-through-minimum melt of the last ten seasons would render a 2017 September minimum of 3.18 thousand km3, or nearly 500 km3 below the 2012 record."


Thus continuing the 5 year cycle as far as I'm concerned.

Of course for that to truly continue we'll need to see ice re-growth and slow summer melt in 2018/19.

There is also the opportunity for some more than average melt in July and August, therefore driving 2017 even lower. I believe, if I remember correctly, that 2012 was not an exceptional season until the GAC.

I believe my focus on Arctic melt has changed. I still watch it daily to see what is going on but my anticipation levels have changed to annual rather than weekly or even monthly.

I no longer obsess over the next few days or weeks. I am looking months forwards and backwards.

Whilst not as instantly gratifying, the overall experience is more positive.

Bill Fothergill

@ NeilT "I have an issue with the last stat for 2016 as chartic says it was 4.13 and this csv says it was 4.72"

Neil, that Charctic value of 4.137 million sq kms is the lowest running 5-day average figure, whereas the value quoted in the monthly summation .csv file represents the monthly average.

NSIDC use a legacy averaging technique which many people find unsatisfactory. This is predicated upon applying the 15% cut-off rule temporally, rather than just spatially. In other words, if a particular grid-cell was seen as being ice-covered for 15% of the swathe-path passes during the month, the monthly averaging technique would count this towards the monthly total.

In other words, if a region of, say, 2 million sq kms was seen as ice covered for just the first 2 and the last 3 days of September, then that 2 million sq kms would be ice-covered for >15% of the measurement period, and would therefore count towards the monthly average. On the other hand, if the same grid cells had been covered for all 30 days of September, NSIDC's monthly averaging technique would report these two scenarios as having the same extent.

Personally, I think that averaging the daily values over the month would give a more realistic summation.

For what its worth, here are the 1979-2016 September averages (in millions of sq kms)

Col 1 = year, Col 2 = published, Col 3 average of Sept dailies, Col 4 = difference.

1979 7.22 7.051 -0.169
1980 7.86 7.667 -0.193
1981 7.25 7.138 -0.112
1982 7.45 7.302 -0.148
1983 7.54 7.395 -0.145
1984 7.11 6.805 -0.305
1985 6.94 6.698 -0.242
1986 7.55 7.411 -0.139
1987 7.51 7.279 -0.231
1988 7.53 7.369 -0.161
1989 7.08 7.008 -0.072
1990 6.27 6.143 -0.127
1991 6.59 6.473 -0.117
1992 7.59 7.474 -0.116
1993 6.54 6.397 -0.143
1994 7.24 7.138 -0.102
1995 6.18 6.080 -0.100
1996 7.91 7.583 -0.327
1997 6.78 6.686 -0.094
1998 6.62 6.536 -0.084
1999 6.29 6.117 -0.173
2000 6.36 6.246 -0.114
2001 6.78 6.732 -0.048
2002 5.98 5.827 -0.153
2003 6.18 6.116 -0.064
2004 6.08 5.985 -0.096
2005 5.59 5.504 -0.086
2006 5.95 5.862 -0.088
2007 4.32 4.267 -0.053
2008 4.74 4.687 -0.053
2009 5.39 5.262 -0.128
2010 4.93 4.865 -0.065
2011 4.63 4.561 -0.069
2012 3.63 3.566 -0.064
2013 5.35 5.208 -0.142
2014 5.29 5.220 -0.070
2015 4.68 4.616 -0.064
2016 4.72 4.505 -0.215

As one would expect, the average of the daily values gives a lower monthly figure - in every year - than the value published by the NSIDC. In almost all cases, the rank either remains unchanged, or varies by a single position if one employs averaging of the daily values. A notable exception is 2016, which goes from 5th to 3rd lowest.

Regarding the putative 5-year cycle which you and Hans tend to favour, I personally still remain unconvinced about this. As we have discussed on an earlier thread, whilst the rolling annual average also supports your hypothesis, I don't yet feel the forensic trail is long enough to be certain one way or the other. I also have this horrible feeling that 2018 or 2019 could end up worse than this year.


Thanks Bill, I knew about the rolling daily average but wasn't quite so aware that the values in the spreadsheet were monthly averages rather than lows. That makes a lot of sense.

The problem with the 5 year cycle is that we won't know, even at a low level of certainty, until 2020. Because 2018 and 2019 need to show some re-growth from a low in 2017 and then the cycle of dropping melt needs to begin again.

For me the real evidence won't come up until 2021/22. If we see second iteration of 2006/7 this year followed by a re-growth and then a third one in 2021/22, then I'd say that it's fairly certain that a cycle is going on there.

This month is a busy month in the extent records. Years drop into and out of the record lows this month. 2010 and 2006 drop out. 2011 drops in and out again, 2012 gets set to smash all records in August/Sept.

Much more waiting to go before we have any idea how it's going to end up, but even then one GAC and anything could change. Some of this ice is very thin and other ice is thinning without vanishing. I recall the ice in the Chuchki/ESS which just looked like smoke on the water. It was counted fully as extent and area but when the GAC came it just vanished.

So I'm making no predictions right now, but the fact is that 2017 is right at the bottom of the charts right now and the only reason it is there is the very sorry state of the ice from the 2016 melt and the very low winter ice growth into 2017. Which leaves the ice very vulnerable.

Rob Dekker

I seem to be one of the few with an rather optimistic outlook for September minimum.

A month again, my entry (based on May data) for SIPN was 5.4 M km^2 for the NSIDC average September :

That was mostly caused by a very large land snow anomaly (4 million km^2) in land snow cover, and moderate 'area' and 'ice concentration' numbers over May.

Now the June numbers are in, and the land snow cover anomaly was still there. Here is the Rutger's snow lab data :

The anomaly is still huge : some 3.5 M km^2 more than last year and a larger amount of land snow cover than since 2004.
This anomaly has to have an impact on the albedo feedback in June, and thus the amount of heat that warms the Arctic never made it this year.

Most of us know that ice 'extent' is not a very good predictor, and 'area' is better. There NSIDC's 'area' number for June came in pretty high (8.51 which is 8.54 after the 'pole hole' adjustment) is higher than most years over the past decade.

As a result, the formula I use for June prediction of Sept average (explained here) :
suggests still 5.41 M km^2 for September 2017.

The difference with May numbers is that the standard deviation of the (hindcast) prediction is 342 k km^2, which is considerably better than the 550 k km^2 SD for a linear trend.

This year appears to be a serious test for my (mostly land-snow cover based) prediction method.

Rob Dekker

Let me add that in June, there were two periods of a week each that created a dipole with some serious ice compaction. That brought the extent numbers down considerably.

The data on 'area' and land snow cover and ice concentration suggest that once the weather turns more neutral or cyclonic, 'extent' will start to stall, and there is not enough heat (melting momentum) to continue the rapid extent decline that we have seen so far.


I guess time will tell Rob. 2017 is still at the lowest on volume, almost at the lowest on extent and not there yet on area. As you say snow has had a big impact but, also, snow has insulated the ice over the winter from growth and, in many areas, is 0.5m thinner than it should be.

I know very well that area is the figure to follow, but unless I build my own charting model from the data, chartic is the easiest tool for me to use right now, given that CT area is now gone. So I'm watching extent.

The Beaufort is almost open to the Pacific, the Laptev, ESS and Chuchki are all continuing to draw back. The Laptev bite is well in progress and the big unknown is what will happen with the Atlantic side.

We have 3 weeks of July left and all of August. Whilst extent may be stalling, there is a lot of peripheral ice out there which is vulnerable and will likely melt out.

Whatever happens, it will be extremely interesting. More so because of the lack of melting input, compared to the state of the ice.


suggests still 5.41 M km^2 for September 2017

Wow, Rob, that's even higher than David Schröder's 5.1 (4.6 to 5.6) million km2. Higher than every ear since 2006, and higher even than rebound years 2013 and 2014.

But this year so far hasn't been nearly as cold /cyclonic as 2013 or 2014, and according to PIOMAS volume is still lowest on record, after the mildest winter on record. And then there's all that early, open water on the Siberian side of the Arctic and SSTs starting to respond strongly (not that far behind 2012 and 2016).

This year appears to be a serious test for my (mostly land-snow cover based) prediction method.

Yes, it's a test for the influence of snow, and its potential as a negative feedback.

Bill Fothergill

@ NeilT & Hans

Despite my frequent protestations that I am unconvinced by the idea of a 5-year cycle, I nonetheless feel it is incumbent upon me to point out when others - especially those far more knowledgeable than I - might beg to differ.

Whilst responding to a comment on the ASIF IJIS thread earlier today, I cited a GRL article from way back in 2003. As it had been some time since I had last looked at that paper, it seemed worthwhile to refresh my memory of what was being said therein.

Just above Fig 1, there is a sentence which reads ... "The 365-day running mean of the daily anomalies suggests a dominant interannual variability of about 5 years".

Bill Fothergill

I realise that there are some seriously polarised views, both here and, even more particularly, on the ASIF, concerning the rapidity of Arctic Sea Ice decline over the next few years.

Those with such views - of whichever persuasion - might benefit from a quick perusal (followed by a slower, more considered perusal) of the following GRL publication. The title of the paper is...

"False alarms: How early warning signals falsely predict abrupt sea ice loss"


Make of it what you will.


Thanks for the time and insight Bill. I won't even discuss this in the forum any more, the discussion is too toxic.

I was doing no more than setting out my stall for why I believed that 2016 was going to stall, why it was not going to make a record and why I believed it was going to set the scene for a 2017 which could be a record breaker.

In contrast to the "false alarms" and predicting abrupt sea ice loss, I believe Hans and I have been doing the opposite. Where we are predicting another 2 year pause on the journey to destruction starting this freezing season and rolling back again in 2020. Then leading up to another sharp drop again in 2022.

I have been very careful not to promote huge new records for 2017, but have talked more about unexpected levels of melt for the year consistent with ending the 5 year cycle.

If we look at 2017, with high snow cover, low volume and indifferent weather, the levels of reduction in all three metrics are, in truth, unexpected. 2017 should not be at, or around, the bottom of the extent and volume tables given the atmospherics.

One other point. I was looking back to 2002 on the AMSRE archive on the Bremen site when I realised that, looking at the images, we are in the 3rd round of a 5 year cycle.

Have a look at 2002, the re-growth in 2003/4 and the path to 2007 with 2005/6 leading the way. It may not be so obvious from the numbers but it is quite obvious from the shape and pattern of the ice.

In the end, there is nothing left to do but wait. We won't know, for sure, until Autumn 2018 whether this is going on, still, or not.

Rob Dekker

Neven said

suggests still 5.41 M km^2 for September 2017

Wow, Rob, that's even higher than David Schröder's 5.1 (4.6 to 5.6) million km2. Higher than every ear since 2006, and higher even than rebound years 2013 and 2014.

I know. It's a pretty outlandish projection, but that is what comes out if you run the numbers.
Very important to note is that my model does not include "ice thickness" (or "ice volume").
Since we had a pretty unprecedented warm winter, ice may be 10% thinner than previous years, which is is not captured in my model.
So we may end up in the "happy middle" as was posted before, where volume and albedo effects even out. Something in the mid 4's.

Bill Fothergill

@ Neven & Rob D "Higher than every ear since 2006"

Would that represent an extremely close shave?



I don't know, Bill. What would Van Gogh say? :-P


Zelfs met één oor hoor ik het ijs smelten.

Bill Fothergill

Ja, dat klopt.

Alternatively, he might have said "Owww! That hurt, I must remember not to try that again!"

Or, if around today, he might have wondered if he'd ever get to see another Elfstedentocht


He wouldn't, Bill. The first Elfstedentocht was organised nineteen years after his death.

Bill Fothergill

Ah, but you have forgotten my weasel-words caveat "... if around today, he might have ...".

I was working in Amsterdam at the time of the last (or at least, the most recent) Elfstedentocht, and remember how it briefly dominated the news.


Rob, the slater probabilistic model has predicted way less than that for the end of August: do you not like that model for any specific reasons?

Rob Dekker

I am a great admirer of (the late) Drew Slater's probabilistic model.

Also note that his model tends to bounce around a bit on the short term, since it is based on ice concentration only, which varies rapidly.
For example, it just bounced back sharply to 4.93 for August 29.

Let us see where the ice takes us.


Thanks, Rob: I didn't know that but I suppose I got confused by how many comments there were on his model! I knew there was a level of appreciation for it hence Jim having it on his blog resource pages, cheers!

Hans Gunnstaddar


"Scientists announced Wednesday that a much anticipated break at the Larsen C ice shelf in Antarctica has occurred, unleashing a massive iceberg that is more than 2,200 square miles in area and weighs a trillion tons."

“Its volume is twice that of Lake Erie, one of the Great Lakes,” wrote researchers.

Hans Gunnstaddar


'Massive Wildfires Burn From California to the Arctic Ocean as Temperature Records Shatter'

Rob Dekker

2017 is still dropping rapidly, keeping pace with 2007 and 2012.
If this continues much longer then my hypothesis that June land snow cover and ice concentration are a good predictor for September SIE will be debunked.

Hans Gunnstaddar

Let's see how it plays out, Rob. I've been laying low lately because there's nothing like just watching the daily changes, seeing how it progresses over time. When 2017 started to make a move recently to more extent than 2012, it suddenly reversed course to drop more in line with 2012, as we can see from this graph:


It's a work in progress...

Rob Dekker

Hi Bill,
On July 5 you posted the September SIE numbers using the 'daily average' method as opposed to the standard NSIDC method of calculation. On the Forum you asked me to run these numbers through my method to see if the 'daily average' method is better 'predictable'.
I just posted the results on the forum here :

Result is that no, the 'daily average' method is not better predictable than the standard NSIDC SIE numbers.
However, NSIDC 'area' numbers ARE better predictable, obtaining a beautiful 270 k km^2 SD of the residuals.
That method still predicts that 2017 will end up where 2013 and 2014 were, so it still predicts a 'rebound' year.

But, as noted above, we really should be starting to see a major 'stall' in SIE decline really soon now for that prediction to remain somewhat believable...

The comments to this entry are closed.