« Mission possible | Main | Research for a novel »


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


Best wishes with your presentation later today Neven, I am delighted you have been invited to participate.

Here we are just poised right at the top of the roller coaster ride of the 2014 melt season....
Which was a timely reminder to myself - and I hope others out in ASIB-land too - to put something in the Tip Jar. My contributions to it by necessity are v modest but at least if we lurkers did from time to time it would add up & help!

Nightvid Cole

It seems to me that the snow cover in June does a good job of foreshadowing the ice cover in September. Ice without snow by June 20th does not survive the season, almost without exception, and ice that still has snow cover on July 1st of a given year is likely to make it to the next season.

Jai Mitchell

I know it is probably nothing but the only 2 other years that had comparably warm arctic winters at the DMI were 2012 and 2007. If this is some kind of indicator, then the intensity of this years temperature anomalies indicate that the 2014 melt will blow through the 2012 year by mid/late august.

Rick Aster

I remember getting the impression in July 2012 that the weather no longer mattered to ice melt (and writing about that thought at the time). For months the extent graphs did not seem to respond to changes in the weather. Perhaps, though, it was just that the location and movement of small low pressure systems did not matter. The lesson from June and July 2013, I thought (by August of that year), was that cloud cover does still matter, if there is enough of it. I still believe there must be ways of improving two- and three-month predictions of sea ice without predicting the weather.

Jai Mitchell

bad link to the webcast.


Congrats to the Sea Ice Prediction Workshop for recognising the importance of having your voice in the discussion. I'll vicariously share in your glory ;>)

The most disturbing thing in the analysis is that
"ensemble predictions do not improve as the season evolves".
It seems as though we should be zeroing in on the solution as more data becomes available.

Every year since I've been following the ice I've predicted that it would be the lowest ever, and this year is no exception.
I watch as CO2 & CH4 percentages increase and think that the Arctic Sea Ice will follow in lockstep. Even when faced with the blockage of Nares Strait I assumed that openings through the Canadian Arctic Archipelago would render it mute.

It appears that I'm a hopeless pessimist incapable of learning from my own mistakes. With this in mind I'll refrain from adding my 2 cents to this year's predictions other than to say again that we're going to break all previous records.

Shared Humanity

I'd also like to congratulate you for the well deserved recognition.

It makes me feel important just cause I hang out here a bit.

Dan Carter

As a botanist that's followed this blog for a couple of years, I tip my hat.


Thanks, everyone. Hopefully the webcast and my presentation work out. I've never done one before.

bad link to the webcast.

Thanks, Jai. Fixed now.

Hans Gunnstaddar

Congrats on your acquisition of a 1960 Topps Yogi Berra, the only year they had a horizontal format. Just kidding about the card, but not about congrats on your invite to give your two cents worth on sea ice predicting. Feel free to throw the collective average of predictions from this website during last melt season under the bus as well.


Ha Ha, very funny.




ASIB guest blogger Larry Hamilton speaking live now.

What time are you on, Neven? If poss, can you leave a message here just before you start?


Will do, idunno. If all goes well, I'll be on in about 30-45 minutes, after the coffee break.

Jai Mitchell

They just asked for you on the phone comm.


And I answered (but I guess that wasn't audible). There's a huge delay, so here's to hoping all goes well.

Crozet Dutchie

I heard your answers clearly Neven, listening and watching on my computer screen. waiting for the coffee break to end… :-)

Crozet Dutchie

The biggest problem is the audience questions: inaudible most of the time. Hold the microphone like a rock star, indeed!!


No, don't throw my presentation in the recycle bin!

Jai Mitchell

That was excellent but my stream cut out midway through, any chance this was captured and can be put up on youtube?


The greater the precision accuracy the greater the mastery of the subject. I think that those who don't project lack confidence
and fear showing that they lack understanding. Those who try
eventually learn from their mistakes and become masters only when the ultimate peer reviewer, the future, agrees.

This year should be easier to be accurate given that El-Nino appears to be here. Again the lesson of 2013 was compaction plays a role, the best ENSO combination should be a strong El-Nino over the winter and a La-Nina appearing at about End of May (a la 2010).

But a look back at 2012 it was:

La-Nina like in January:


trending El-Nino Mid april


a small El-Nino in July


lasting a while and we know what happened in 2012.

This year we will have likely a stronger El-Nino, if so it is bad news for sea ice. Arctic past winter was warm same as 2012,
the continents were colder in Europe 2012, north America in 2014, again similar. The main difference so far is that there is a colder zone at about the Archipelago. However up to date Sun disk observations have picked up warmish signals from a cold start, the atmosphere is already rapidly warming. This because it was mainly clear weather. Again not so good for sea ice because anticyclones , clear weather generators, these exacerbate melting in the summer. Until my data is more complete, tentative projection 2014 melt will surpass 2012.


Well done Neven, and thanks to Larry for suggesting that other participants might like to write guestposts here. Warmly seconded.

If you are party to a group e-mail of participants, Neven, you might like to write to confirm this invitation?


Phew, that was quite exciting for me. I felt great until my computer crashed, after that I felt horrible. :-)

I believe the webcast can be downloaded later, but I'm not sure.

L. Hamilton

I understand that today's webcast will be archived and available for later viewing. I'll post the link here when I have one.

If you are party to a group e-mail of participants, Neven, you might like to write to confirm this invitation?

I did that once when starting the ASIF, and going by some of the mail addresses there are cryospheric scientists who've become (lurking) members.


BTW, can you believe that there's a guy speaking now called Neven as well (not uncommon in former Yugoslavia, but not that common either), and one of his colleagues is called Asif.

Chris Reynolds

Jai, there may be something to your idea of warm winters preceding crashes, but it's far from the only factor so it's taken a bit of fiddling to bring out a link.


To make that plot I've taken the interannual differences for NSIDC Extent September average extent. I've also calculated NCEP/NCAR Jan/Feb (JF) average temperature anomaly north of 80degN, and Jan/Feb/Mar (JFM) average temperature anomaly north of 80degN. Anomalies are relative to the GISS baseline of 1951 to 1980.

The horizontal axis of the plot is the sequence of values for each year's extent difference. The vertical axis is of temperature anomaly. The most negative interannual losses - typically associated with new record years (left of the plot) are associated with large temperature anomalies, but not all the time. Hence the range of temperature increases for more negative interannual extent losses. Whereas at the right side of the graph - gains from the previous year the variance is more restrained.

As I say, other factors are clearly at play, 4 out of 5 of the largest peaks in late winter temperature are associated with negative interannual extent changes - but this may be at least partly because of the recent warming associated with a period of net ice loss. To help the interested decide if there's anything to it, here's a time series:


With regards prediction.

I think it's helpful to use the trend, with what is deemed an appropriate function (exponential, quadratic, etc) as the baseline and see if that can be improved upon.

The initial thickness in April, whether for regions or the whole PIOMAS domain is, in my experience, little better a predictor than the best-fit trend. i.e. the initial thickness mainly carries information about the trend, not the deviation from the trend. Initial thickness do not seem to deviate in any given year such that they strongly influence deviation from the trend. I tend to agree with Dr Stroeve that melt season weather is a major (the major?) factor in determining the deviation from the trend.

Chris Reynolds

Just caught a bit of Dr Zhang talking about using NCEP CFS to make PIOMAS predict - interesting.


Chris - If my notes are to be believed also using April 1st thickness from IceBridge + EM improved matters.

Jai Mitchell


I have been thinking about this and I would try to do the following.

1. Apply the following function as a weight to daily temperature anomaly series data for days 40-150

2. Sum the daily weighted values.

3 Apply the following function as a weight to daily temp anomaly series data for days 150-270.

4. Sum the daily weighted values.

5. Sum the outputs of the two series to get full season weighted temperature attribution.

6. Apply as a 40% attribution factor and include AO and ocean heat content to the other 60% (note, the temp functions may be changing significantly with AO and OHC from bearing straight influx and/or runoff. Not sure if land anomalies (river runoff increases) make it stronger or weaker over the course of last decade.

It would be interesting if you can extract a weighted annual function that produces a correlation.

Jai Mitchell

hmmm it seems the second function drops off too quickly, should extend out to about 250 before dropping much below 2.

Jai Mitchell

This is better

Jai Mitchell


commenting on your data, the second graph shows the correlation I was noticing starting around 2002. I think that prior to that, the average ice thickness was too great for this effect to stand out.

Liam Baker

It would makes sense to me that the ice extent becomes more dependant on weather as the ice thins.

As a feedback system, one can imagine the ice have a proportional and integral (PI) response to weather. For thick ice, the integral term is dominant as it takes a lot of thermal energy to melt/create it. As the ice thins, the proportional term takes over as the first-year ice is more responsive to short-term weather conditions.

Liam Baker

Slightly off topic, it appears an El Nino is on the cards later this year. http://www.bom.gov.au/climate/enso/

Colorado Bob
Quoting 202. ColoradoBob1:
NOVOSIBIRSK, Russia, April 2 (RIA Novosti) – Record-setting temperatures topping 71 degrees Fahrenheit (22 Celsius) were recorded Tuesday in several cities in Russia’s Siberia, a representative of the Novosibirsk meteorological service told RIA Novosti Wednesday.

“It was the hottest April 1 on record for several western Siberian cities, including Novosibirsk, Tomsk, Kemerovo, Barnaul and Gorno-Altaysk,” Renad Yagudin said.

“The average temperature in Russia has increased 0.4 degrees every ten years. Overall, the temperature in the area is 6.5-16.2 degrees Fahrenheit (2-9 Celsius) higher than the record set in 1989,” he added.

Wednesday was also expected to be a record-breaking day for western Siberia.

According to Aleksander Frolov, the head of Russia’s Agency on Hydrometeorology and Environmental Monitoring, global warming will have a greater impact on Russia than any other region in the world.


In the last 35 years, the average temperature in Russia increased by 1.5 degrees Celsius (2.7 Fahrenheit), while the average figure worldwide is 0.8 degrees Celsius (1.4 Fahrenheit), government meteorologists have said.

Some parts of Russia have shown even more extreme warming. In the Arctic, south Chukotka and Kamchatka regions temperatures have risen 150 to 200 percent more than in the rest of the country.

In October last year, Norwegian and Russian scientists said that surface water in the Barents Sea was 5 degrees Celsius warmer than normal. They linked the peak-temperatures with the unusually warm summer in the northernmost parts of mainland Norway and on Russia's northern Kola Peninsula.


It's a bit hard to be sure at this distance from Boulder, and without reading the (in press) Schroeder et al. paper, but it looked as though "spring melt pond fraction" was in fact the best predictor of September ice extent. There was a brief discussion about a previous CPOM "melt pond" paper over in the Developer's Corner:


Ghoti Of Lod

Chris and Jim I also noted that Dr Zhang very clearly showed that his technique, though good at predicting 2012, was completely off for 2013. I'd say it doesn't account for the weather sufficiently.


Ghoti - I fear predicting "the weather" in the Arctic four months ahead is a non trivial problem in itself. The melt pond paper I referred to above relied on reanalysis data for atmospheric forcing, which is a bit hard to come by several months in advance!


Phil Jones' summary just now suggested "our blogging friends" want to see "lots o' plots".

I thought Neven made it clear yesterday that actually we'd like to be able to download "lots o' data"?!


They may of course be looking in the wrong ocean.

The following graduate paper from 2010 found, astoundingly, that from 1979-2010 the best predictor of ice extent in Beaufort in October is the SST in the Carribean in May...


...and ruled out an awful lot of the usual suspects, from ENSO to SIE in the same area earlier the same year.

And now, in a most exciting development, the sainted David Sanger has only gone and asked her about this...


More to follow, I hope, from people like Stone or Dr Tom Murphree, who appear to know what they're talking about, which excludes me, obvs.

Ideally, I'd love to see them do a SEARCH prediction based on the Carribean data in May 2014.


Thanks for the mention Larry!


Thanks Neven for your presentation yesterday. I'm a sea ice model scientist and I'm attending the meeting. For some reason the free signup was not available yesterday, but was today. Not sure if econnexus saw the last discussion after Larry's mention, but the consensus seemed to be that Neven said both "lots of plots" and "lots of data". I look forward to some fun discussion.


Hi Dave, and welcome!

I'm Jim by the way, but currently using a nom de plume to try and fool the TypePad patent pending "spam" filter. I have to say I wasn't anticipating that my comment above would be read out loud in Boulder!

Nonetheless, despite my somewhat dodgy feed here in the UK, I'm pretty sure Phil didn't mention the "lots of data" bit, and so made the point.

As a professional scientist yourself, what did you make of Larry's comments yesterday about "citizen scientists"?

Chris Reynolds


2013 was the perfect example of weather. PIOMAS volume 'jumped' relative to 2012 in May due to weather and didn't fall back until this February - due to weather. Up until May 2013 sea ice thickness was such that with a re-run of 2012's weather it's hard to see how we wouldn't have had a repeat of 2012.


Sorry, you've lost me with that stuff you're doing. There are mathematical techniques for extracting relative roles of various factors - but it looks like rather a lot of work to me at present.


Agreed, or perhaps more intuitively - for a given average thinning from April to September (say 2m), the closer average April thickness gets to that average thinning the easier open water is formed, and the greater weather impacts that alter seasonal thinning can have on open water formation.

Anyone interested...

I've been playing around with PIOMAS main daily volume series. Taking the interannual volume changes from year to year for the daily volume minimum from 2000 to 2013 I can make a series of yearly volume changes. If I make a guestimate of 2014 daily minimum volume of say 4.120k km^3 (might be lower, might be higher) I can then take random choices from the list of interannual volume changes from 2000 to 2013, and use them to make random series of volume loss from 2015 to 2050.

So I now have a series of artificial volume loss trajectories, based on the past volume loss behaviour of recent years. I can then work out when in each of the 25,000 artificial series the first occurence of zero volume occurs.

So I tally how often this happens for a single year, to get an occurrence distribution of when the first time volume reaches zero. This is shown in the graph below.

Of course this tells us nothing about whether future year on year volume losses will be typically greater, or less, than in the period from 2000 to 2013.

What it is intended to do is to answer the following question:

If the interannual volume losses can be taken as indicative of the net energy gain of the ice/ocean system, if this net gain persists into the future - how long until the volume falls to zero?

In the spirit of that question, if I leave out 2007, 2010, and 2012 as not being indicative of the net energy gain, but outliers due to special circumstances then the occurrence distribution of when the first time volume reaches zero is as follows:
With a slower decline to zero in the 2020s, rather than a peak around 2020.

For what it's worth I find the length of the tail unconvincing due to my interpretation of volume loss implying net energy gain in the ice/ocean system.

Chris Reynolds

Jim (econnexus),

I've got similar references in papers. PIOMAS data can help, for example I still think the extensive late survival of ice in Chukchi/ESS in 2010 was due to the large export of MYI earlier that year from the central Arctic region. But I've been unable to get further than such post hoc interpretations. When I try numeric methods to improve predictions I find the improvements are not substantial, e.g. an R2 going from (IIRC) roughly 0.8 to 0.85.

I thought Neven made it clear yesterday that actually we'd like to be able to download "lots o' data"?!

Like DaveB said (welcome BTW), lots of everything. Obviously I prefer graphs and maps because I'm not good with data.

I'm really glad that they're putting all of the talks online, because I was out all day, and yesterday was really fascinating.

it looked as though "spring melt pond fraction" was in fact the best predictor of September ice extent.

Personally, I've got a hunch that the amount of melt ponding during the first half of the melting season is very important for the season's outcome.

I wrote about it two years ago: New Data, Melt Ponds on Arctic Sea Ice.

I've been trying to get the folks behind this to publish more data, but due to circumstances (lack of funding) they're not pursuing it further. :-(


Jim, I understand now. Phil did leave out the lots of data part. I think the message has come across now that it's great to have everything out there. This could lead to a whole new discussion, but the availability and serving of data is a big challenge. From the model side, we prefer to use a binary format known as netCDF, which is self-describing with coordinate information, etc. and numerous software tools can read; while the observational community often prefers ascii/text formats. Larry mentioned the comma separated format.

Neven, the melt ponds as a predictor was new work to me. I need to read the paper. I suspect there is something else going on in the prediction framework, but I can't say for sure.

Jai Mitchell


I was attempting to come up with an anomaly-derived function that would show the temperature contribution to SIE.

The graphs I posted showed a first-run comparative value of the anomaly effect. This effect changes as the anomalies in winter are much higher than summer and the effects of the anomalies in winter are much less than in summer.

If you could point me to the daily anomaly data for the last 20 years, I could play around with it.

Helen Wiggins

Hello Neven and Arctic Sea Ice Blog community I am a member of the Sea Ice Prediction Network (SIPN) project. Neven - thank you so much for your excellent presentation yesterday at the Sea Ice workshop. I loved it. In the workshop breakout groups today that were focused on the Sea Ice Outlook, your talk inspired good discussion and agreement that the sea ice researcher community needs to be more active on this and other key blogs and this great community of citizen-scientists, and of course we need to do a better job at providing data. Perhaps we can have a couple more SIPN project members, in addition to Larry Hamilton, guest post here a few times during this melt season?

Chris Reynolds


NCEP/NCAR temperature series by month can be obtained here:
NCEP NCAR temperature is a reasonable reflection of the actual temperature as measured by GISS for example.

Air Temp Surface
Latitude? 70 to 90 Longitude? 0 to 240
* Monthly
Area weight grids - YES
* Raw Data

You can choose a different region, like north of 80 degN. I chose 0 to 240 because that leaves out the land temperatures of northern Greenland and the CAA.

Give it a few seconds and that should give you tabulated data by month.

I assume you know how to calculate the anomalies, just in case: Say you want to use the GISS baseline of 1951 to 1980. for each month calculate the average temperature for the period 1951 to 1980, then each months temperature is an anomaly of temperature when the monthly average is subtracted from the figure for each year, i.e. Anomaly = MonthValue - BaselineAverage.

Chris Reynolds


Sorry forgot to add - I know of no easily useable daily temperature data, unless you want to use netcdf.


Welcome, Helen!

and of course we need to do a better job at providing data.

Like I said, a great job is already being done, but it's never enough! :-)

Perhaps we can have a couple more SIPN project members, in addition to Larry Hamilton, guest post here a few times during this melt season?

That, of course, would be great. If I'd have more time, I'd like to do interviews as well every once in a while. Perhaps even Skype interviews like Peter Sinclair does, and put those on Youtube. Or something like a weather (hindcast and) forecast. I made a couple of Youtube videos to go along with the written sea ice updates, but it was just too much work, and as a non-English/American speaker it's difficult to do things improvised.

But I'm very satisfied with how the blog is right now too, so no one should feel obliged to do anything. I might be able to start doing a bit more once our house is finished.

Jai Mitchell

well, I went to it and didn't really get much out of it. The Feb/March summed weighted average anomaly values (with march weighted as 2.3 times feb average values) doesn't show any real correlation to extent anomalies (shown as positive values to correlate well with temp anomaly curve)

anomalies are in thousands of Km.



NSIDC has called the(5th lowest ever) max...



NSIDC source for the above has a lot more, including an update on multiyear ice, mention of SEARCH and news of a new resource...



The new resource...


apparently has maps for SAT, sea ice, permafrost extent, vegetation, snowcover, and more.

Thanks to the scientific community for this speedy response to the request for more data. That's probably enough for one day;)

Colorado Bob

Why Arctic ice is disappearing more rapidly than expected: River ice reveals new twist on Arctic melt


April 2, 2014


Simon Fraser University


A new study has discovered unexpected climate-driven changes in the mighty Mackenzie River's ice breakup. This discovery may help resolve the complex puzzle underlying why Arctic ice is disappearing more rapidly than expected.


Chris Reynolds


Why not try removing the trend in sea ice extent? We know that the trend in sea ice extent/area is strongly related to the trend in volume, as indicated by the relationship between volume and extent at minimum:
This is to be expected because due to open water formation efficiency it's the volume loss that is the driver of sea ice loss.

So surely what you need to be doing is looking at the residuals from the trend in sea ice loss. i.e. what causes the variations from the trend, not what causes the trend itself. As I understand it, that's where to look for improving seasonal prediction.


From the NSIDC writeup:

'This winter the multiyear ice makes up 43% of the icepack compared to only 30% in 2013. While this is a large increase, and may portend a more extensive September ice cover this year compared to last year, the fraction of the Arctic Ocean consisting of multiyear ice remains less than that at the beginning of the 2007 melt season (46%) when a large amount of the multiyear ice melted. The percentage of the Arctic Ocean consisting of ice at least five years or older remains at only 7%, half of what it was in February 2007. Moreover, a large area of the multiyear ice has drifted to the southern Beaufort Sea and East Siberian Sea (north of Alaska and the Lena River delta), where warm conditions are likely to exist later in the year.'


sofouuk - It will certainly be interesting to see how the older ice now in the Beaufort Sea holds up as the melting season progresses. There's much discussion about that sort of thing over on the forum at the moment:


By way of example, anomalously warm conditions already exist in the Beaufort and East Siberian Seas, and have done for several months. Here's the surface air temperature anomaly for the first three months of 2014:

Chris Reynolds


I've been watching this because I think it could be comparable to 2010. The similar export of MYI in 2010 was a large (possibly the main) factor in the 2010 volume loss event.


Chris , Ice wise perhaps a 2010, weather wise no, 2010 had a particular explosion of sun disk vertical diameters similar to 2005.


which made me predict it be warmest year in history in March.

This was because El-Nino peaked in January, warmth and clouds from it spread all over the world. So I'd expect 2014 to be more like 2012 weather wise, but the ice field is different than at the beginning of 2012.

Artful Dodger

Neven wrote: | April 02, 2014 at 21:55

"I wrote about it two years ago: New Data, Melt Ponds on Arctic Sea Ice."

Hi Neven,

Happy Spring to you, old friend! Just a quick note that the blog link to this paper is now stale: (here's a new one)

Rösel, A., Kaleschke, L., & Birnbaum, G. (2012). Melt ponds on Arctic sea ice determined from MODIS satellite data using an artificial neural network. The cryosphere, 6(2), 431-446.

Cheers, and have fun!

Jai Mitchell


Yes, the residuals would have a better time of it I think. I am more specifically looking at extent which goes to zero residuals at max. Therefore the trend is isolated seasonally.

Speaking of which, I went through the process. Developed the weight function above, took the monthly integral to get the monthly average weight value (klunky I know! Wish I had good daily data!)

multiplied the monthly weight values to the monthly average temperature anomalies values you directed me to (thanks! oh by the way I did 80-90 degrees to duplicate DMI as best as I could)

compared to sea ice extent values from SMMR and SSM/I-SSMIS

going through all of the ice melt seasons, the only time periods in 2 and 3 month intervals leading up to max melt was (of course!) the August-September value. All others (even entire melt season) had negligible correlation.

The monthly weight functions for temperature anomaly (degree K above/below mean) are

feb 5.5
mar 16.8
apr 37.9
may 107.2
jun 112.3
jul 97.6
aug 87.6
sep 38.3

Here is the only function that returned good correlation:


Here is the scatterplot


(note I plotted annual sea ice extent anomalies as positive to prevent the temp/extent negative correlation.


Jai Mitchell

oh, as a note, the correlation to August-Sept weighted temp anomalies did not exist prior to 2004. interesting. . .

Jai Mitchell

what do we want?


when do we want it?

Whenever you feel ok with putting it up online!!!


Rick Aster


Just a hunch, I think August-September temperature might have minimal impact on ice 2 meters thick or thicker. As ice trends thinner the weather at the end of the melt season could have a greater impact on extent. This might explain a correlation emerging in recent years.

Nightvid Cole

I'm pretty sure that the correlation for September has the causation mostly the other way: Open water at high latitudes leads to a warm September.

Chris Reynolds


So in this plot:
Am I correct in interpreting the blue line as your method's 'prediction'?

BTW - you are careful to remove trends before obtaining correlations aren't you? If you don't do that you'll get false high correlations due to trends in synch and counter synch. Some chap at the forum was getting really high correlations for predictions using volume, and he was just matching the extent and volume declining together.

This graph shows PIOMAS volume changes in ten day increments throughout the year.
Neglecting transport for a moment and considering it as energy - it's apparent why the volume cycle changes as it does, it's almost in lockstep with insolation.
The lag is slight, with sea ice volume loss happening only a couple of weeks after peak insolation.

Looking more closely at each of the ten day increments on that graph each year can be seen. What can also be appreciated is that mid season the interannual variation is small compared to the overall heat flux going into melting ice.

It seems somehow relevant to me...

Nightvid, I Agree.

Jai Mitchell


no, these are historic values, there is no prediction involved. It is the sum of the average august temp anomaly (K) multiplied by a factor of 87.6 summed with the average September temp anomaly multiplied by a factor of 38.3.

Sea ice extents are historic. They trend together nicely, but do not trend together at all (they actually show some negative correlation!) prior to 2004.

I would like to get your piomas ten-day variance and try to plot it against the same period temperature anomaly if I could find temp data at that level. I would guess there is a fairly strong correlation.

Yes, I developed the weight function based on the insolation curve, coupled with a land-warming feedback in the summer months. That is why the curve I guestimated looks the way it does.

Chris Reynolds


I should have said 'hindcast' - forecasting past years to test the method.

The ten day PIOMAS differences is simply calculated from the main PIOMAS daily series. It's just V(n) - V(n-9) for ten day increments in n. You could work with the difference between current day and previous and it will produce a noisier version of the same curve. Working with monthly data is fine, it has the advantage of reducing short term noise.

Daily temperature is available from NCEP/NCAR but it's in netcdf format. I use Panoply, from which csv copies of data can be obtained. But daily data means that's not a very efficient way of working. I don't know which programming langauges have ready made netcdf extraction routines.

Now I'm not as tired as last night I've put my finger on why those difference plots seemed relevant. It's the relatively small size of interannual variation compared to the bulk of energy going into the ice. Also various factors go into those small changes: Temperature of the atmosphere and ocean, cloud cover, ice albedo. Bringing ice movement back into the equation you've got convergence/divergence of the pack, and ice export (Fram/Bering but also transport by the Beafort Gyre).

Hans Gunnstaddar

"So I'd expect 2014 to be more like 2012 weather wise, but the ice field is different than at the beginning of 2012."

Wayne, that's why I'm predicting 2014 to be close to the same as 2011, then in 2015 it will break 2012's record. 2014 will be a melt season exemplified by playing catch up due to 2013's increase in ice volume.

Last melt season I correctly predicted a rebound. Ok, 4.35 wasn't as much as it ended up, but it did rebound, so now I put my prediction record on the line with the above prediction. Drum roll...


Hans, I do not agree that we need another year to play catch up, as we have been playing that game all winter now and consequently have caught up with previous years (PIOMAS).

Snow cover extent is currently much lower than last year, as well as most other years, and in my opinion snow cover is perhaps the most important indicator of what the summer melt will look like (much more than ENSO). The summer cyclones are of course the jokers, as I got to learn last year, they can easily screw up any minimum predictions.


DCS - I await the PIOMAS numbers for March with bated breath, but as Chris pointed out previously catch up was just about achieved in February:

Do you recall Rob Dekker's snow cover based predictions from last summer?



Hans, 2011 has some similarities with 2014 to date, but there was no strong La-Nina in December 2013 . 2011 was right after the warmest year in history. 2013 was 4th. Ice wise 2013-14 had a very cloudy winter. 2011-12 was equally cloudy. The only determinant factor left is El-Nino, which if strong will change the circulation dynamics and cloud coverage. I can say with a great deal of confidence that sea ice is much thicker in the Lower North American Arctic, not so during the same period in 2011. This thicker ice favors anticyclone genesis creating a near certain greater influx of solar heat right where the thickest ice can melt during the right period (May till end of July).

Doomcomessoon , yes cyclones are jokers indeed, they reverse seasonal trends, making winter warmer and summers colder. But 2014 has no semblance of 2013 which had cyclonic, or pervasive surface adiabatic presence from spring all the way to February 2014.
The jokers are taking a small break it seems.

Hans Gunnstaddar

All points noted and considered, I still predict 2014 ending up about where 2011 did, with 2015 surpassing 2012's record.

Remember, it took a 5th season after the 2007 record to get to 2012's. Going to a 3rd season after 2012, which will be 2015, seems like the minimum reasonable time frame for a new record.


I'm not sure how far the melt will go this year yet, at all, but in view of the comparisons with 2011, there are some other factors I think we need to ponder.

First is, 2011 I believe actually had a positive snow cover anomaly as compared to the present.

Second is, net ocean heat content has risen considerably, which in particular is demonstrated in significantly reduced ice coverage in the Barents.

Both of these, given supportive lack of cloud cover could give the melt season a significant "running start". Most concerning in my view is the lack of ice coverage at high latitude, along the line from the ESS to the Fram. Energy picked up there might have a disproportionate impact.

My perception over all of the system is, that there is an energy symmetry point around which the ice coverage fluctuates, which shifts based on the residual energy increases or losses year over year. Ice coverage seems less an influencer of this point of symmetry, than it is an effect, though ice does buffer the movement of energy in the system over all, via a combination of insulation, albedo and phase change energy.

Since 2011, I think that symmetry point has steadily shifted in the direction dictated by higher over-all heat content in the system. Thus, given mostly equal seasonal conditions during melt, the effect now on reduction of ice would be greater than in the past, due to the net over-all increase in energy in the system.

I think this may throw off your prediction, Hans, though the rest of your logic seems sound. The base problem is, those sneaky net system enthalpy gnomes have been moving the goal posts on us, and not in a directly we like.


*DIRECTION we like... I hate autocorrect...


The PIOMAS numbers will be most interesting indeed, and I've got a feeling that the maxium this year will be second lowest, perhaps lowest.

My SIE minimum estimate is 2,5E6 Km^2 give or take 2 million. Though, I agree with you Wayne that the pattern this year looks very different from last year, thus another year like 2013 looks quite unlikely to me.

Hans Gunnstaddar

"Thus, given mostly equal seasonal conditions during melt, the effect now on reduction of ice would be greater than in the past, due to the net over-all increase in energy in the system.

I think this may throw off your prediction, Hans, though the rest of your logic seems sound. The base problem is, those sneaky net system enthalpy gnomes have been moving the goal posts on us, and not in a directly we like."

True enough, and we will see how it plays out. Very exciting to be on the eve of another melt season knowing Neven will probably once again open the flood gates to many other posters predictions as well.

Did anyone else watch the TV special tonight, I think it was on NBC on climate change? So lacking on details for theories and explanations regarding AGW. For example they talk to a woman scientist about her 'theory' of increased amplification of the jet stream causing weather to remain in place longer than when the jet stream had less amplification, which also includes the idea of blocking, but never explain that it is the reduced temperature gradient between the Arctic and the tropics that is considered the root cause.

Not one idea went beyond a sort of grade school kind of simpleton explanation. How can people ascertain if an idea has merit if they don't know the underlying suggested cause? What was said instead was I 'believe' global warming is causing greater weather extremes. That isn't an explanation.

They also had a consultant to the government on this subject that said any data not covering at least 50 years had no merit. 50 years?! I guess we better hope things don't change very fast!


Here's a genuinely SCARY forecast:
My reaction: "Wow - this is the end of the warning phase, and the beginning of the action phase."
Ken Rushton.

Nick Naylor

Here's the recorded NASA press announcement with Q&A's:


Quite a lot of insight into the ice sheet, as well as how the press receives this sort of info.

The comments to this entry are closed.