Latest PIOMAS Volume update

[gerontocrat]

Who expected any Fram export anyway?

Well, not me.

Qu. (there always is). Is sort of zero Fram export July to late fall the normal since way back when, or is it a newish thing since MYI sort of evaporated?

1) the grey line in the background is the average (2001-2017) of the monthly export. 2018 is well below even that.

2) last year I posted an animation of fram export just to answer your question (with 2017 showing very little export). Rather than looking up the port, I attach the animation again. "Zero" summer export happened before, but is certainly not the rule.

Muchas gracias - looks like always low in that period and getting less and less over time. Low area and extent in the Greenland Sea the logical consequence?

[Richard Rathbone]

I had expected a slightly larger drop, but 2018 does creep closer to the top 3.

July looks middle of the road to me.

A strong August will put 2018 in the 2nd-6th mix, a weak one and 7th could be threatened, but there's very little difference from 2nd to 6th in September, so the most probable placing looks to be 6th to me.

[RikW]

I think PIOMAS overestimates the ice at the ESS, it looks really fractured at worldview on the 31st, and it doesn’t improve last days...

[gerontocrat]

Having at last find a really good unzipper -  a freebie as well, and realised how well Excel now handles input from ftp and txt files, I thought I would apply the methodology I use for JAXA and NSIDC extent and area data to daily volume data.

So here it is, and the first graph attached looks very much like the area and extent graphs.

Caveat : My first go at this - so submitted on an "errors and omissions excepted" basis.

EDIT: Using the decades averages on the graph seems to make it clear volume loss is accelerating.

[A-Team]

think PIOMAS overestimates the ice at the ESS,

On the Fram, it might be worth looking at seasonal bias of measurement error. While Piomas looks at volume, it must first determine floe motion across a flux gate. That is not done from models but rather satellites. Most of these struggle mightily with artifacts during the summer.

The Fram has had undersea moorings for many years with upward-looking sonar that determines ice volume fairly accurately from freeboard and area of floes passing overhead. That data is not real-time but has been summarized in recent journal articles. It would be instructive to compare that with Piomas.

Note the Polarstern is out there right now looking at strengthening eddies off East Greenland arising from opposing but adjacent strong currents.

Motion uncertainty is shown by purple arrows in daily OSI-SAF, perhaps the most advanced monitor of ice motion (semi-daily). Note areas of lowish concentration are not even attempted as there can be rapid melt, dispersion and compaction.

This spring saw a strong SSW that completely cut off the previously prevailing wind pattern giving CW circulation of the ice pack that favors Fram export. This abruptly halted export of the extruded Kara tongue ice which came to occupy about two-thirds of the Fram intake funnel. These SSW do not occur with regularity across the years, nor does the Kara tongue.

Somewhere I posted enhanced Ascats of Fram (and Nares) export for the last 9 years as a single multiplexed animation. Ascat is not effective during the summer but does give a very good visualization of flows during the rest of the year. It is very rare to see a persistent back-flow; the one this year eventually did get re-exported. Thus a fair amount of volume just goes back and forth across the flux gate, melting out and breaking up as it goes, but entirely forth in the end. No exceptions in nine years: the Fram is a ratchet.

http://osisaf.met.no/p/osisaf_hlprod_qlook.php?year=2018&month=02&day=09&action=Today&prod=LR-Drift&area=NH&size=100%25

The attached png superimposes Fram export for the years furnished, to produce daily intensity averages.

[Shared Humanity]

I would like to thank all of you for posting such expert analysis. This is why I come here

You know who you are....

[Tom]

EDIT: Using the decades averages on the graph seems to make it clear volume loss is accelerating.

I find it very interesting that despite the lower actual volume your decade trend figures seem to show the "melt from now to min" actually steadily decreasing decade by decade.

Any idea why this would be, or have I misunderstood the figures?

[Tom]

Re my previous post.  Could this just be down to less ice being available at (presumably warmer) lower latitudes at the end of the season?

If I read correctly there was less volume loss from this point in 2012 than the 80's average, despite the final volume being so much lower.

[AmbiValent]

Re my previous post.  Could this just be down to less ice being available at (presumably warmer) lower latitudes at the end of the season?

If I read correctly there was less volume loss from this point in 2012 than the 80's average, despite the final volume being so much lower.

I think you've got it. The earlier ice melt in the periphery has gotten faster because peripheral ice is increasingly preconditioned to be thin and broken up - because that was how it ended up in the previous melt season. And such preconditioned ice, once broken up again, offers a much larger surface to the melt, and so melts faster than the thick ice there used to melt in the 80s.

And this means that later in the melt season when the insolation for melting is decreasing, there's much less easy ice on the periphery, and the ice in the Central Arctic Basin, where the insolation is decreasing fastest, is available but much harder to melt.

[Richard Rathbone]

Re my previous post.  Could this just be down to less ice being available at (presumably warmer) lower latitudes at the end of the season?

If I read correctly there was less volume loss from this point in 2012 than the 80's average, despite the final volume being so much lower.

Yes, some insolation that previously went into ice and ice-covered water and melted ice in the summer, now goes into open water and delays the refreeze in the autumn instead. I think there's also a climate oscillation at play too, that moves around the season which is most favorable for melt over the decades.

[Tom]

Thanks both for your insights.  I'm finding the knowledge and information available from the contributors to this forum fascinating.

@gerontocrat if you have the time and inclination I'd be interested to see the piomass loss by month over time for the melting season.

[gerontocrat]

EDIT: Using the decades averages on the graph seems to make it clear volume loss is accelerating.

I find it very interesting that despite the lower actual volume your decade trend figures seem to show the "melt from now to min" actually steadily decreasing decade by decade.

Any idea why this would be, or have I misunderstood the figures?

Not a clue - unless it is that the melt in the peripheral seas happens earlier while the CAB is being stubborn?
Graph attached with total gain and total loss year by year. Doesn't answer your question ... makes it more confusing as values of gain and melt seem to be increasing over time

[gerontocrat]

Thanks both for your insights.  I'm finding the knowledge and information available from the contributors to this forum fascinating.

@gerontocrat if you have the time and inclination I'd be interested to see the piomass loss by month over time for the melting season.

Send me a message tomorrow - numbers or graph?

[FishOutofWater]

The location of this years relatively thick ice is a problem. There's relatively thick ice in the ESS which both the ECMWF and GFS models predict will be hit by strong warm air advection over the next ten days. The other pile up of ice is on the way out of the open and moving main channel of the CAA. The location of the relatively thick ice combined with the weather forecasts indicates that there will be high volume loss over the next ten days.

2012 is out of reach but there's going to be an atypical surge in volume loss in the first half of August.

[oren]

I'd be interested to see the piomas loss by month over time for the melting season.

I'd like to step in with some charts regarding this question. Volume has been declining over time, basically in all months but with a somewhat different shape of trend. Here are some charts for total PIOMAS volume in day 135 (mid-May) to day 212 (end-July) and day 260 (~Sep. minimum).
The second chart shows the half-month volume loss (for example the "day 135" volume loss shows the loss from day 120 to day 135). Some trends are obvious:
* early volume loss has been on the increase since ~2009
* mid-season loss has been on the increase since 2000, with a maximum reached in 2012
* late-season loss has been on the decrease since ~2009

It's interesting that early and mid season loss have been increasing despite having less volume to being with. Its only natural that late season losses are decreasing as less ice is available to melt.

More interesting insights can be gained by focusing in the next posts on different geographical regions:
* "Outer" regions, those that typically have zero volume at minimum. All regions excluding "Inner+".
* "Inner+" regions, those that typically participate in the minimum, including the CAB, Kara, Laptev, ESS, Chukchi, Beaufort, the CAA and the Greenland Sea.
* Only the CAB, the most protected and northerly region.

[oren]

The outer regions starting volume is decreasing over time, while the end result is typically near-zero. Interestingly, the trends for volume loss are more or less constant in early and mid-season, with late-season losses shrinking due to the obvious fact that no volume is available to lose.
The main study of these regions should be the length of open season, as volume is reaching near-zero earlier.

[oren]

The "Inner+" regions have been decreasing over time, with a possible leveling-off since 2011 (the trend of day 135 makes this claim dubious). Losses have been on the rise in early season, rather steady in mid-season, with only a very slight decrease in late-season. While the difference between 2000 and 2018 is around 6-7 (1000 km3) in the beginning of the season, it is around 8 at the end of the season.
IMHO this high variability plus rising losses mean that low winter max volumes of recent years will eventually translate to new record minimum volumes.

[oren]

CAB volume has been decreasing, with a possible leveling-off since 2010, theoretically due to loss of MYI until little MYI remained, with FYI growing to the same thickness every year (the slow transition theory).
Interestingly, losses seem to have been on the rise at all times (except day 135 which is showing no loss and no trend), though with lots of variability.
IMHO, this means a near-certainty that the leveling-off has an element of luck and will not last.

[oren]

To sum up this series of posts, I think it would be fair to say when drilling down into the various regions that volume losses are slowing down only when there is no more ice to melt in that particular geography. At all other times, the global warming signal is quite prevalent in the rise of volume loss during most parts of the melting season.
In addition, volume losses seem highly variable, with 2012 seeming to score several "lucky" losses in a row. With rising loss trends across the season such losses will need less luck and eventually the 2012 record will be broken.
What seems like slowdown of losses in the late season really comes from the inclusion of the outer regions in the statistics.
OTOH, it seems the CAB specifically still has a way to go before becoming "ice-free" in volume terms.

[Juan C. García]

Using the decades averages on the graph seems to make it clear volume loss is accelerating.

I like to do my PIOMAS analysis on a decadal basis.

From my point of view, it is not important anymore how we end up on 2018 or 2019.
The important thing is to talk about climate, and a decadal change is a measure of climate (not a particular year).

So, the important point is that looking at September, we lost 32.7% of the volume on 2000-09, but we lost almost the double (63.6%) on 2010-2017.

[Stephan]

Thank you Juan for your work.
It clearly comes out that the major change in ASI happened in 2007. Since then only minor changes have occurred, expressed mainly in further reduction of volume, especially in OND due to later freezing / intrusion of warmth through the Atlantic side.

[litesong]

..... the major change in ASI happened in 2007. Since then only minor changes have occurred....... further reduction of volume....

It appears that the solar TSI, which has been at a low level for the past 12 years (& for 3+ years setting a 100 year record low) is temporarily affecting Global warming.

[Dry_Land_Is_Not_A_Myth]

Juan C. García,
That is an absolutely wonderful representation of PIOMAS. You have a gift. You're also spot on about assessing decade trends versus yearly. Thanks for the reminder to look at the big picture.

[Juan C. García]

Thank you Juan for your work.

Juan C. García,
That is an absolutely wonderful representation of PIOMAS. You have a gift. You're also spot on about assessing decade trends versus yearly. Thanks for the reminder to look at the big picture.

Thank you for your comments. 

I corrected the title "Months 2005-17" to "2005-18" because the 163 months go from Jan 2005 to July 2018.

[A-Team]

the major change in ASI happened in 2007. Since then only minor changes have occurred, expressed mainly in further reduction of volume, especially in OND due to later freezing / intrusion of warmth through the Atlantic side.

Yes, 2007 in some ways had a more dramatic downstream impact than 2012, pinching out MYI classes as well as subsequent volume by the mechanism discussed below. The NSIDC account of Sept-Oct 2007 is at the link below; the overlay of the 2007 minimum on 07 Aug 18 AMSR2 is shown in the attached png.

It looks like most of the 'extra ice' this year will be midway between the NP and Wrangel/NSI. GFS surface winds suggest the peculiar lift-off in the Lincoln Sea will continue a few more days; however a few days of opposing winds later in the month could flatten it against the coast, making the final outline of 2018 ice look even more like 2007.

Delayed freeze-up in the fall prolongs the local melt season in warm-water kill zones and otherwise offer a chance for consolidating winds. This in turn restricts the seed pack to the CAB (recalling that new ice mainly forms on the pack periphery), resulting in an ever higher proportion of FYI relative to SYI.

The second animation shows that ice loss between early August and mid-September is quite variable, with GAC2012 bringing about the most extreme situation in just a few days. Looking at this time series makes me dis-believe the hastily written journal articles post-GAC saying 'oh that ice would have melted out anyway'.

This animation advances in triple frames: first the August 5th for the earliest year, then its September 10th, then the half-overlay of the minimum on the August, pause, then the next triple, up to the 2018 whose minimum and overlay frames are conjectural.

It's also worthwhile to compare each melt season end to its beginning (notably Sept 2017 to Sept 2018) if you take the view that the CAB largely just sits there from year to year (possibly thinning slightly) and melt season primarily consists of undoing undoing the freeze season's new peripheral ice. In that perspective, the year is evaluated by its departure from time-reversibility: did the melt season do less or more than undo its freeze season.

However this year has seen significant inroads into the CAB along the Atlantification corridor, in the Laptev and possibly the Lincoln Sea (where Atlantic Waters eventually exit out the Nares at depth).

While lift-offs are as common and in correspondence with sustained CW rotation of the ice pack (ie anti-cyclonic pressure systems), this one could be unveiling a severely mechanically weakened -- and possibly thinned -- local ice pack.

Recall though the winter Ascat series that showed almost half the very thickest east Lincoln ice getting pushed down the Nares, so quite a bit there is just refrozen matrix, now breaking up into free floes.

Recall too that the interface of the CAB with the outer islands of the CAA was in motion for much of the fall and winter, with huge blocks tumbling end over end and some exiting into a Beaufort-Chukchi stringer. Even though this area might see some of the coldest mean temperatures and historically have been land-fast, in recent years with a smaller rotating rigid body, it is moving fast on the outside of the merry-go-round and coming up against these immovable islands. Much of the very thickest and oldest MYI has been lost in this way.

http://nsidc.org/arcticseaicenews/2007/10/589/#1October

[oren]

Wonderful animation, and I agree GACs do more damage than they are credited for.

[A-Team]

GACs do much damage

Right. The 2012 UH AMSR2 animation below shows very little happened to the ice after the great cyclone had passed by. Most of the stray lower concentration ice was mopped up by the 3rd week of August, with very little happening in the following three weeks to minimum (red boundary), just consolidation to the core CAB.

The question is, what would have happened to volume had the GAC not happened, what weather ensemble should have been used? The AMSR2 archive does not have earlier years. Taking what happened over the same time frame for 2013-2018 as weathers, a lot less consolidation is seen (previous post and 3rd animation below).

We could wonder too how well the Piomas volume algorithm performs on middling concentration dispersed peripheral ice vs big blocks of solid ice, given that its resolution (cell size) is quite coarse relative to features of low concentration ice.

The other issue is whether it takes a late season GAC to end the season with such a flourish: possibly a mere late AC or just a persistent weather pattern with weaker but consistent winds over a longer time frame -- like we are seeing now: 12+ days of a slow moving high off the NSI -- could significantly affect end-season outcome. Here the mechanism would be different, with winds dispersive, perhaps transporting peripheral ice that otherwise might not have melted into warmer waters.

The first 8 days of August are showing significant and possibly unprecedented ice retreat in the Laptev and SV-FJL-SZ corridor but of a scale too small to move the volume needle notably. A much larger area in the Beaufort-Chukchi is in play; it's not clear enough time remains in the season for that to resolved out as melt given moderate winds and temperatures.

As Fridtjof Nansen observed early on, Arctic sea ice actually moves at a 20-40º angle to the right of the direction of persistent wind blowing across the surface, rather than the 45º expected from pure Ekman theory for surface currents. This means ice movement has to be observed (eg from AI on satellite pairs: OSI-SAF), that it does not follow from meteorology, and that it cannot be inverted to recover meteorology (even though it is pressure-gradient induced).

In early August 2018, ice moved in a fairly consistent pattern (4th animation) due to the stable anti-cyclonic winds. It is not easy to distinguish perimeter melt from ice simply being pushed away since the uncertainty soars (purple arrows) in regions of low sea ice concentrations.

[Sterks]

Yes although strong storms under 980 like it hapenned in late August 2015 over Beaufort (after a series of smaller but yet dispersion and melt-promoting storms during a relatively cold August) also make a big difference. Notice the 2015 evolution above.
ECMWF predicts a storm like this around 15-17, over ESS-CAB, not a GAC, but can keep steady melt rates for the rest of August

[gerontocrat]

How come there is not a PIOMAS Volume poll as in the JAXA and NSIDC extent polls. After all volume is what will matter most in the end, n'est-ce pas ?

[oren]

Great idea!

[Richard Rathbone]

How come there is not a PIOMAS Volume poll as in the JAXA and NSIDC extent polls. After all volume is what will matter most in the end, n'est-ce pas ?

Non. The major effect on heat transfer is whether or not there is ice coverage, not how thick it is. The major way the Arctic influences the rest of the world's climate is in the heat transfers.

Volume is a measure of how long ice coverage can survive an energy imbalance so it has some predictive value for when area will go, but the important factor for general influence on climate is the area.

[Shared Humanity]

Ultimately, volume is the single best metric. We should have a poll. This will be one more poll in which I will never participate

[gerontocrat]

Ultimately, volume is the single best metric. We should have a poll. This will be one more poll in which I will never participate

Huh! he said in reply. You want a poll to be designed and implemented but never to use it. Weel, weel. But I have bashed out a graph that will probably never be used, so may as well post it just the once.

Having the daily volume measurements and having done similar for JAXA extent, I made a graph of average volume for the previous 365 days over time and added a couple of trend lines for good luck.

[uniquorn]

Worldview, north of CAA, aug9. A gap in the clouds, is that 3m ice?
edit: added url. https://tinyurl.com/y7qllc3h
PIOMAS, jul31.

[Greenbelt]

Ultimately, volume is the single best metric. We should have a poll. This will be one more poll in which I will never participate

Whether volume, extent, or area, I think these 365 day trailing averages are the most useful statistics for explaining sea ice loss.

[CameraMan]

is that 3m ice?

Doesn't look to be.   Lots of rubble and some open water visible yesterday.

[Neven]

I'm a bit late with this, but I hadn't had a closer look at the data until yesterday and today, for the latest ASIB PIOMAS update.

First of all, Wip, I believe your PIOMAS graph hasn't been updated to the end of July yet (I used the one you posted based on thickness data).

But second of all:

I had expected a slightly larger drop, but 2018 does creep closer to the top 3.

This may explain why I expected that. From the PIOMAS update:

Something funny has happened on the PIJAMAS graph, showing average thickness for the entire Arctic ocean (a crude calculation where you divide PIOMAS volume with JAXA extent). It seems that extent has gone down relatively faster than volume, and so the average thickness has flatlined instead of going down. Basically, it means that more volume is spread over a smaller expansion of sea ice. This too, perhaps, explains why I expected volume to have gone down more during July. I expected volume to follow extent, and thus thickness to continue to go down some more:



We see the exact same thing happening on the Polar Science Centre thickness graph:

[Wipneus]

PIOMAS gridded thickness was updated (last day 15 August). Computed volume is about 6.19 [1000 km³], sixth lowest behind 2010-2012, 2016, and 2017.

Attached is the animation for August so far.

[binntho]

Is it just me or is there a strange disconnect between the Piomas on the 15th and the image below, also from the 15th, from AMSR2? Piomas seems to show ice significantly closer to Canada, in the ESS and also further towards south along the Atlantic front.

Is it the lack of resolution on Piomas? Or is the last frame some sort of average of the fist 15 days? I had a look at the two images side by side and it's actually quite a lot of extra extent being shown in Piomas compared to AMSR2.

[Wipneus]

Fram Strait ice volume export temporary halted.

[Wipneus]

PIOMAS volume and volume anomaly, updated with volume calculated from the gridded thickness.

[oren]

Is it just me or is there a strange disconnect between the Piomas on the 15th and the image below, also from the 15th, from AMSR2? Piomas seems to show ice significantly closer to Canada, in the ESS and also further towards south along the Atlantic front.

Is it the lack of resolution on Piomas? Or is the last frame some sort of average of the fist 15 days? I had a look at the two images side by side and it's actually quite a lot of extra extent being shown in Piomas compared to AMSR2.

As PIOMAS is calibrated to NSIDC sea ice concentration, with its coarse resolution and larger reported ice area, you should compare the last image of the animation to the NSIDC image of the 15th.

I should add, I expected the first half of August to bring about a larger volume drop, due to the warmth all around the CAB, but I am guessing the very low rate of area loss may have prevented a larger volume drop.

[RikW]

To be honest, I think PIOMAS is overestimating the ice this year/ more than other years. It has lot's of ice on place where there isn't any ice on worldview of where the ice is really fractured. Unfortunately, because it is one of the few sources for this information.

[Richard Rathbone]

To be honest, I think PIOMAS is overestimating the ice this year/ more than other years. It has lot's of ice on place where there isn't any ice on worldview of where the ice is really fractured. Unfortunately, because it is one of the few sources for this information.

PIOMAS is primarily a measure of volume, rather than where that volume is. There's a much greater error margin on an individual pixel than there is on the total. You should expect to see these sort of trivial differences, and if you want to make anything out of them, you need to quantify them and compare them against the expected accuracy of PIOMAS.

[magnamentis]

To be honest, I think PIOMAS is overestimating the ice this year/ more than other years. It has lot's of ice on place where there isn't any ice on worldview of where the ice is really fractured. Unfortunately, because it is one of the few sources for this information.

glad for the increasing discussion on the matter and if you leave aways "this year" i'd second that while i'd say instead. increasingly each year.

the worse the source data for their calculation represents the real state of the ice the farther their results are from the real volume if we were able to measure it instead of just calculating it.

resolution of the underlaying data is way to high to represent the current volume iMO, they should start to use data from UH but U.S. scientist refuse to use non-U.S. data most of the times. sorry if that's not a qualified statement but as an observer in various fields of science and technology one can get the impression. ready to stand corrected if that's an entirely incorrect assumption

[oren]

They should start to use data from UH but U.S. scientist refuse to use non-U.S. data most of the times. sorry if that's not a qualified statement but as an observer in various fields of science and technology one can get the impression. ready to stand corrected if that's an entirely incorrect assumption

Of course that's a wrong assumption, their model was built in 2003 long before AMSR2 was launched. NSIDC is integrated into the model. Overhauling models never comes easy. In addition, there is no historical data for AMSR2 before 2012, while PIOMAS has monthly volume calculations since 1979 and daily calculations since 2000.
I assume recalibrating PIOMAS for AMSR2 data so that it would fit the historical data calculated using NSIDC is doable, but surely very difficult. I am certain this has nothing to do with such U.S/non-U.S politics as you suggested.
BTW, Dr. Zhang is extremely nice. They used to publish PIOMAS data once a month. At some point I wrote to him, told him how much the data was valued at the ASIF, and asked whether they could publish it twice a month. He simply said yes... and they have been doing it ever since.

[jdallen]

Fram Strait ice volume export temporary halted.

Not surprising when there is almost no ice in the Fram or even within a couple hundred KM of it, LOL!

[magnamentis]

They should start to use data from UH but U.S. scientist refuse to use non-U.S. data most of the times. sorry if that's not a qualified statement but as an observer in various fields of science and technology one can get the impression. ready to stand corrected if that's an entirely incorrect assumption

Of course that's a wrong assumption, their model was built in 2003 long before AMSR2 was launched. NSIDC is integrated into the model. Overhauling models never comes easy. In addition, there is no historical data for AMSR2 before 2012, while PIOMAS has monthly volume calculations since 1979 and daily calculations since 2000.
I assume recalibrating PIOMAS for AMSR2 data so that it would fit the historical data calculated using NSIDC is doable, but surely very difficult. I am certain this has nothing to do with such U.S/non-U.S politics as you suggested.
BTW, Dr. Zhang is extremely nice. They used to publish PIOMAS data once a month. At some point I wrote to him, told him how much the data was valued at the ASIF, and asked whether they could publish it twice a month. He simply said yes... and they have been doing it ever since.

hey, i expected that, which is why i quasi invited such a reply.

in the aftermath i admit that to simply find out what you correctly stated in your reply, i could simply have asked, instead of choosing the kind of provoking approach.

however that is, thanks for the reply. it is helpful to bring order to ideas that are based on things that i know they are true and then others that are not true but based on first, hence that are kind of overly pessimistic.

 

[Wipneus]

PIOMAS has updated the gridded thickness data (official volume update not yet at this time). Volume calculated from thickness is 5.24 [1000km³] at 31 August, sixth lowest for the date.

Attached is the animation for August.

[Wipneus]

Latest thickness map, compared with previous years and their differences.