PIOMAS September 2016

Another month has passed and so here is the updated Arctic sea ice volume graph as calculated by the Pan-Arctic Ice Ocean Modeling and Assimilation System (PIOMAS) at the Polar Science Center:

Weather conditions in the past month - the (Great?) Arctic Cyclone followed by the Mega-Dipole to be precise - have left a mark on volume decrease. During August 2016 more sea ice volume was lost than during any other August in the past decade, with the exception of 2008, even more than during August 2012 (2909 km3 vs 2744 km3). Just like on the area and extent graphs, 2016 is almost in second position here too, as the difference with 2011 shrunk from 621 km3 to 84 km3.

Here's how the differences with previous years have evolved from last month:

Last month I wrote:

There is a lot of melting potential in the East Siberian Sea region, extending into the Central Arctic, that could make this year end up close to the 2011 minimum, which is second lowest after the 2012 record lowest minimum.

We now know that melting has been massive in this region, as can be seen on this Uni Bremen SIC maps animation, showing the spectacular changes from August 2nd to September 2nd, the main reason for the significant volume drop:

Wipneus' version of the PIOMAS volume graph also clearly shows that this year's trend line dropped a bit faster, breezed past 2010 and is now almost at a par with 2011:

And so there's a small downtick again on the PIOMAS sea ice volume anomaly graph, which shows how much actual sea ice volume deviates from the linear trend, but not down to 2 standard deviation territory:

As for average thickness, here's the PIJAMAS graph based on my crude calculation of PIOMAS volume numbers divided by total JAXA sea ice extent:

2016 perfectly follows the new volume regime that started in 2010 and was interrupted during 2014 and 2015. The exact same can be seen on the thickness plot from the Polar Science Center:

So, can 2011 still be overtaken in the one or two weeks left in this melting season? 2011 reached the minimum on the earliest date in the past decade (September 10th) at 4302 km2. Sea ice volume is currently 4638 km3, so second place is definitely possible, but not easy, as in the past decade only 2008 and 2014 managed to lose the 336 km3 required until the minimum was hit. Added to that: the weather forecasts look like this melting season could end somewhat earlier, also given all that dispersed ice on the Pacific side of the Arctic, in what is now known as the Wrangel Arm.

But this also depends on how much longer bottom melt continues. Sea surface temperatures are really high, and this year is ending low, regardless of weather conditions it seems. Remember, during June and July very little sunshine reached the Arctic sea ice pack to create melt ponds and build up momentum for the remainder of the melting season. This was the kind of weather that made 2013 and 2014 stall, and prevented 2015 from going lower than it did.

Not this year, though. This year another kind of momentum carried things forward despite the weather. A mild winter, early opening up, extreme low snow cover, probably caused the Arctic to soak up enough heat to not care about the June and July sun. And who knows, maybe a pulse of warm water - extremely difficult to measure - from the Atlantic and Pacific continued the long-term process of complete Arctic sea ice loss.

The world hasn't experienced the warmest average global temperatures on record for three years in a row for nothing. This heat eventually ends up in the Arctic.

Comments

[Hans Gunnstaddar]

That's a good question, viddaloo. Definitely a strong reaction, but what the mechanics of it are would be very interesting to know. Maybe someone on the board will hit on something.

[NeilT]

Sorry being dense again.

Yep I'm totally in agreement. If we see a huge melt event in 2017, followed by a bounce back in 2018 (most likely, it seems to be doing that now); we're seeing a pattern. Albeit an early one.

[navegante]

Really sad to hear this about Dr. Slater.

[Bill Fothergill]

RE: Andrew Slater - Shocking news. I remember feeling the same disbelief when Katherine Giles (UCL) was killed in a bicycle versus lorry accident 3 years ago.

@ viddaloo & Hans

Could a possible "rebound" driver involve a sort of flip-side to the ice/albedo effect?

As the Arctic heads into the long polar night, the absence of sea ice allows a greater energy transfer rate from the ocean to the atmosphere, and eventually, back out to the cosmos. I don't know the magnitude of any such effect, but one might reasonably expect this to act as a - possibly brief - negative feedback mechanism.

On the other hand, this could just be a chimera. In such a noisy data set, it's difficult to claim with any confidence that a 5-year cycle has unambiguously emerged out of the noise after just 2-and-a-bit cycles.

I'm sitting on the fence on this one, but I'll be watching the 2017 developments with more interest than normal.

[Hans Gunnstaddar]

"As the Arctic heads into the long polar night, the absence of sea ice allows a greater energy transfer rate from the ocean to the atmosphere, and eventually, back out to the cosmos."

Plausible and interesting idea, Bill.

"I'm sitting on the fence on this one, but I'll be watching the 2017 developments with more interest than normal."

Sounds good & should be very exciting to see what happens in 2017!

[Hans Gunnstaddar]

"Yep I'm totally in agreement. If we see a huge melt event in 2017, followed by a bounce back in 2018 (most likely, it seems to be doing that now); we're seeing a pattern. Albeit an early one."

Alright, good to have you on board NeilT! We shall see how it plays out.

[NeilT]

Bill, I had a look through the data sets back to the beginning of the satellite record. It seems to be showing a 10 year pattern but hard to work it out without matching all the factors.

You'd need to factor in solar cycle strength, Nino strength, AO strength and weather events to try and match it all. Also there is the fact that a cycle might not clearly represent itself given our measuring metrics when there is such a huge differential in actual ice volume and coverage. It could easily be that this cycle was always there but invisible based on the metrics we currently use, but made visible by the sheer lack of ice and the fragility of ice today.

I'm sure, 30 years form now, we'll know as much as anyone could know, about how the cycles and how they manifest themselves.

The critical point now, as Hans and I believe, is whether 2016 is a main event which just missed (2007/2012), or whether it's the precursor (2006/2011), and whether 2017 is going to be like 2007 or 2012. Because if they cycles Hans and I believe we are seeing are actually there, then we'll see a 2012 style drop next year. Then we can forget about cycles because there won't be enough ice left, in summer, to measure them. Well once the rebound of 18/19 has worked out of the system.

The only good part of this is that Hans and I won't need to wait much longer to see the result of what we think we're seeing. Because if it is a 5 year cycle then 2022 is going to be a revelation in terms of loss of ice, duration of the ice free arctic and the delay of the freezing season into the Autumn...

Of course if it isn't then we'll know next year.

[Jim Hunt]

Michael reports on the forum that according to a provisional interim release of daily gridded data this years minimum PIOMAS Arctic sea ice volume was 4.3882 thousand km³ on September 6th.

[viddaloo]

Jim, my PIOMAS estimate app goes lower, but only slightly lower and still higher than #2, 2011.

PS: Is PIOMAS based on Jap/German concentration maps or NSIDC?

[Jim Hunt]

Quoting from the PIOMAS overview:

For the ice volume simulations shown here, sea ice concentration information from the NSIDC near-real time product are assimilated into the model to improve ice thickness estimates and SST data from the NCEP/NCAR Reanalysis are assimilated in the ice-free areas.

[Rudmop]

It is very interesting to plot the past 37 year average PIOMAS volumes in the Arctic, from April to September against the same time period surface area data from NSIDC. PIOMAS has an error of +/- 750 km^3 and NSIDC has an error of +/- 1,000,000 km^2. To fix the error, perform a linear regression, with the y intercept at 0. Now do the same for the 2016 data. The value of the slope is the average thickness of the ice. Take this thickness and multiply it by the surface area for each April and September. This of course will give you the volume. Take the difference in volumes from the April to September extremes for each graph. It shows that the past 37 year average had a melt volume of 16200 km^3 and this past 2016 had a melt volume of 14,000 km^3.

[viddaloo]

Rudmop, this is interesting! But what conclusions do you yourself draw from these calculations? As you said, both NSIDC and PIOMAS have huge error margins.

[John Christensen]

Thank you for providing these numbers Rudmop, which summarize the melting season quite well.

Another metric could be the PIOMAS volume reduction in km3 from beginning of the summer (E.g. day 151) to the latest date in the data (Day 244).

As you see, the Arctic sea ice melting of 2016 is third lowest since 2006, due to the overall ice preserving weather pattern this summer - even with the late season ice crunching cyclone:

2012 15,672
2015 15,547
2010 15,429
2009 15,307
2007 15,254
2008 15,177
2013 14,953
2011 14,892
2016 14,683
2006 13,807
2014 13,113

[Rudmop]

The question I wished to answer from my investigation was : how much more energy today is melting the ice during the Arctic summer melt season. My hypothesis was at least as much of a difference as Carbon dioxide is trapping in. I was surprised to find less volume melted this past year than the last 37 average. Maybe low sunspot activity this year is also a factor.

[John Christensen]

Rudmop,

Just consider that arctic sea ice is being reduced both due to reduced winter time freezing as well as enhanced melting in summer time.

In the past 12 months the lack of freezing has been the dominant aspect, whereas the melting has not been spectacular at all.

[John Christensen]

In other words, the cloudiness/cyclonic weather pattern of this summer reduced radiative heating sufficiently to cause overall sea ice melt to be less than normal.

Given the significant importance of clouds in a 24-hour daylight environment, the weather pattern can dominate the overall melting increase trend.

[Rudmop]

As to the extent of cloudiness, I often check out weather stations such as the ones on Coffee Club Island, Longyear Beyen. Thule Airbase, etc. Albedo of course is another factor to consider. If we are taking all those years as an average, then the model is going to be based on average as a standard. If this year was grossly more different, concerning average albedo, cloudiness, advection, sunspot activity, etc. then perhaps it can't be compared to the past 37 year average melting. I think it would be interesting to perform the volume of ice produced from September 2015 to the peak ice volume in the following April, 2016, preform a linear regression, and compare that to the same analysis of the past 37 year average. Knowing the difference in ice volume produced could at least allow us to calculate a net energy loss difference. CO2 will randomly release photons of IR as it cools in the Arctic winter. This will help cool the warmer currents brought into the Arctic by the heat engine/thermohaline ocean conveyor belt. The Carnot efficiency is calculated from the difference in temperature of the hot and cold reserviors. Assuming this difference is due to differences in energy loss during the Arctic winter , past versus present, we can gain a better idea of how much heat is being lost. This year we know there was less mass of ice melted, so we know less energy was absorbed and trapped.

[viddaloo]

This year we know there was less mass of ice melted

Rudmop, this is exactly why I asked. With the huge uncertainties of modelled volume, how do you or we *know* this year melted less ice? Models could equally well be confused by the extreme and record big dispersion of the ice, leading to much thinner ice in real life, but perhaps fooling the models into guessing there's a much bigger area than there actually is, and that this area is thick as, well, multi–year ice.

Bottom line is we just don't know. I've spear–headed an initiative to collect pocket–change from ice bloggers and commenters to buy buoys for the North American universities to deploy, so we can know what happens to the ice. Your support could mean make or break! :D

[Rudmop]

Plotting the PIOMAS volume on the Y axis and NSIDC area on the X axis gives a nice visual of how the surface volume changes relate through the freeze/thaw cycles. Performing a linear regression, smooths out the errors in the data. The slope also is the average thickness. Using the volumes obtained from the linear regression will allow for a determination of the ice volume gained in a winter or lost in a summer, by subtracting the final volume for the period in question from the starting volume of said period. This analysis removes errors in the data, as I stated. I just completed that analysis for the volume of ice gained last winter in the arctic. The past 30 year arctic winter volume increase of 15700 km^3 versus an winter increase for 2015-2016 of 26500 km^3. That is a whopper of a difference as compared to the differences taken solely from PIOMAS. The 37 year past volume differences, taken from the PIOMAS data only, yield 16800 km^3 of ice added over the arctic winter, versus 16400 km^3 of ice added over the 2015-2016 arctic winter.The numbers are what they are; I cannot change them. I am simply on a quest for the amount of energy used to melt the sea ice in the arctic. I am looking for differences in energy that are required to melt ice in the arctic. In this way I can relate that difference back to the coefficient of heating for carbon dioxide. So far the amountil of energy difference does not measure up against the difference that I have projected by the coefficient of heating for CO2. I have calculated this coefficient and will have my book out on it soon. I, like all climate scientists, am searching for a climate model that can make accurate predictions.

[Rudmop]

Whoops, 2015-2016 winter ice build up was 14000 km^3.

[Rudmop]

....by the linear regression, as opposed to 16400km^3 by the PIOMAS data solely.

[viddaloo]

Keep us posted when you have a graph 'proving' last winter put on 26500 km^3, will you? I've found gallows humour is really something I chose very early on in life, I guess in my teens when watching British satire, so not at all a sudden change for me after calculating what's ahead of us. Like your style!

[Rudmop]

I'm guessing the 12 minute time difference between my whoops post and the British satire comment can be attributed to lag time in server speeds and time zones, or perhaps you read my whoops comment at posted the reply because you could not resist, or perhaps you used the 12 minutes to comment sometime between 22:14 and 22:59. I don't know which of those possibilities induce your affection for my style. If catching my errors and correcting them is the reason for the appeal, I'm human too. I'll quit using my phone and lunch to post so as to avoid those mistakes.

[Rudmop]

But after correcting my spreadsheet error, I get an astonishing value of 1680 km^3 less ice volume added thoughout the arctic winter last winter, than the arctic ice volume added over the past 37 year winter average. If you take the difference in only PIOMAS data, not comparing to NSIDC, the volume differences are only 400 km^3. Using the linear regression for the PIOMAS/NSIDC graph, the average thickness of the 37 year average is .0017 km and it is .0014 for last year. During the winter of 2015, the arctic added 1.1 times less volume of ice c

[viddaloo]

I don

[viddaloo]

As I was about to say, I don't know about other people, but after 8 comments talking about a plot, I grow curious to see the actual plot itself. Any chance this could be posted or linked?

[Rudmop]

Plot these PIOMAS Volumes for the past 37 year average melting (~April-Sept), from PIOMAS against NSIDC Ice extent over the arctic melting period. The x axis is the left column, the y axis is the right column. Then I have the 2016 melt season below this. Put both the data sets in a spreadsheet and add the trendline, set intercept to 0, display function. Slope is the thickness. I'm sure you know how to do this, but I am only explaining to you what I did in my mind; simple enough I know. I like to be specific on my thought, so there is no guess work for someone else, in trying to replicate what I did.

15500000 28000

14200000 25560

12700000 22860

11000000 19800

8200000 14760

6500000 11700

9000000 16300

2016 Melt season

14500000 20300

13000000 18200

11000000 15400

9200000 12880

6750000 9450

4500000 6300

[Rudmop]

The graphs above will be perfect because they are the actual data that is a result of the regression. I was not clear on that. So the way I got that data is from the raw data obtained off the PIOMAS/NSIDC Graphs. Here is that data: Same format as above, from April to September.

37 year average data during melting season.

15500000 28000

14200000 27000

12700000 23500

11000000 17100

8200000 12300

6500000 11300

September 2016 melt season; april-september

14500000 22500

13000000 21000

11000000 16600

9200000 10100

6750000 5900

4500000 4100

I made the graphs by plotting the NSIDC ice extent (area) on the x axis and the PIOMAS volume on the y axis. I set intercept to 0 and the slope was the depth of the ice on average. From the equation we can find the April and September ice Volume. Taking a difference in the volumes gives us the amount melted. There was less melted this year than the past 37 year average. I will need to add the freeze data from last sept 2015 to April 2016 if you want me to; however, you may want to collect that data.

[Clivepmitchell]

Thoughts in response to Rudmop's comments:

Arctic sea ice is not a single system reacting as a unit to heating. The heating is not applied uniformly.

The amount of ice in an area cannot drop below zero. If heat is applied to one area until the ice in that area is gone, further heating will result in rapid temperature rise but not further loss of ice unless the heat from the hot spot is transported to an area of remaining ice and transferred effectively.

Conclusion:

As sea ice thins and disappears in places, the link between heating and ice volume loss may weaken.

[viddaloo]

The amount of ice in an area cannot drop below zero.

Clive, while most of what you say is true, I also find it puzzling that at the very moment when excessive heating and positive feedback really take off, ie when the last bit of ice is melted, that's the exact same moment when slowists say full stop, the ice cannot drop any more, as cannot go below zero. Isn't this a bit like saying a painter can never finish his wall, because the latest spot painted already changed colour and thus cannot change colour even more?

In two following days of clear skies, the same amount of sunshine that finishes off the last sea ice in a given area by taking it from 0C ice to 0C water on the first day, can potentially bring that same mass of H2O on the second day from 0C to 360C because of increased absorption multiplied by enthalpy(?). So saying the ice cannot go below zero is only telling half of the truth, IMO.

[Clivepmitchell]

viddaloo, Thanks for the reply, though I'm not sure what a "slowist" is. I agree with what you said (apart from 360C - but only because I haven't calculated it myself, yet).

I had been assuming that sea ice would keep melting and absorbing excess heat down to zero volume then temperatures would skyrocket, producing more water vapour to dump heat onto Greenland. So measuring ice volume loss was a proxy for measuring net heat. (It averaged around 10^21J per annum for over 10 years).

I'm now thinking that perhaps progress from a mostly ice/water to a mostly water/vapour condition will not be a sudden switchover, but rather a drawn-out scenario with some parts of the Arctic slowly melting ice and other parts exporting extreme heat via routes which avoid the remaining ice. In this way the Arctic could undergo fierce heating but still have random years of comparatively low ice volume loss.

What I'm trying to say is that I can no longer regard PIOMAS sea ice volume minima alone as a reasonably reliable way of measuring Arctic net heating - too much is already being absorbed by evaporation in ice-free areas with no guarantee that this heat will end up in sea ice.

Arctic heating is definitely not slowing down!

[viddaloo]

viddaloo, Thanks for the reply, though I'm not sure what a "slowist" is. I agree with what you said (apart from 360C - but only because I haven't calculated it myself, yet).

I think your own words can serve as an on the fly definition:

I'm now thinking that perhaps progress from a mostly ice/water to a mostly water/vapour condition will not be a sudden switchover, but rather a drawn-out scenario with some parts of the Arctic slowly melting ice and other parts exporting extreme heat via routes which avoid the remaining ice.

This is slowism, and in a rather extreme specimen: Of all the parts of the Arctic, some will melt slowly (A) and others will route the heat so that it avoids the ice (B). If what you wrote is your opinion, then according to you the Arctic sea ice will disappear not just slowly (A), but considerably slower than "slowly" because of the significant part of the Arctic described in (B), slowing the "slowly" down further.

I disagree with you based on everything I've seen and read about the topic of Climate Change, I think the collapse will be abrupt, and those who bother may call me an abruptist.

(Setting rationality aside for a moment, if asked what I hoped for I'd maybe say "huge regrowth of ice for decades on end", but I don't believe the 0–360C feedback effect allows for such massive increases in annual sea ice without a significant element of magical / wishful thinking.)

[Clivepmitchell]

For what it's worth, I studied thermodynamics/fluid mechanics, and taking potential effects of jet stream changes on global agriculture into account I fully expect civilization to be over within 10 years. To keep myself sane, I only read this blog and other sites on Saturdays.

So until next week at the earliest, good-night!