The 2017/2018 freezing season

[Neven]

There are two of interest, one called Melt Day Onset which seems to be a static file made in late spring 2017 and another called standard deviation of sea ice concentration which may be related to error assessment.

The former marks up ice pack locations according to the date when melt was first observed. That's of interest per se but only 2016 appears provided as a comparison year. This archive is something to be watched (along with ESRL's daily melt ponds) in spring 2018, presuming melt ponds are a helpful leading indicator for summer outcome.

This might be part of research involving Julienne Stroeve as a co-author (found it: 2016 Mortin et al.). I asked her if she could keep me informed on when melt onset was happening and how it related to previous years a couple of times, but she's very busy, of course.

It would be great to have this kind of info at our disposal. As I've understood it, melt onset doesn't mean melt ponds will form right away, but it does precondition the snow on the ice, so that when solar radiation comes into play, melt ponds may form faster. This preconditioning mostly takes place when it's cloudy!

[A-Team]

Oren mentioned a quick way to make animations a while back: just make a list of the urls needed and ask ImageJ to make Stack from List, crop as needed, export at the animation speed wished, load to forum. That works well for ESRL graphics of the current day using the attached list for their fig.3 (grep to something else if needed),

Their model is seeing very little net melt the next few days. The winds, pressure temperatures are not notable above Svalbard for the next few days considering sea water at the exposed salinity freezes at about -1.8ºC (if not mixed or agitated).

Nullschool animations are most easily made saving whole-window screenshots to Gimp while scrolling through the forecast with shift-k. After image duplication, the text annotations can be cut out, made smaller, and layered over the tiled-up resized crop of the Arctic Ocean prior to flattening, slicing into frames and saving those layers as a forum animation.

[A-Team]

This might be part of research involving Julienne Stroeve as a co-author  ... she's very busy, of course. great to have this kind of info ... melt onset doesn't mean melt ponds will form right away, but it does precondition the snow on the ice, so that when solar radiation comes into play, melt ponds may form faster.   mostly takes place when it's cloudy!

Right all around. Clouds can have counter-intuitive effects in the off-season. It is really an upwelling versus downwelling radiation balance story where common sense can flatline into not even wrong. ESRL has some good-looking daily products in that department that we should do more with. Even if not spot-on, they're physically based and the only thing currently out there; by playing with those, we'd be ready if something more accurate comes along.

Stroeve is maybe still on sabbatical in London? She was a PI on that fancy plane landing transect above Alert too, the best thing going for snow samples and real ice thickness. She is listed in the melt date HDF5 above along with:

:contributor_name = "Walt Meier, Ruth Duerr, Florence Fetterer, Julienne Stroeve, Matt Savoie, Sean Mallory";
:contributor_role = "PrincipalInvestigator,author,author,author,author,author";

[Niall Dollard]

Quick, one year round the sun, snapshots of the US NIC snow and ice charts. (Only 2d of course).

Not much difference really in coverage between the 2 years. But difference in ice location. Back in 2016 we had the Wrangel Arm, this year the East Beaufort Tongue. Both years show a bite above the ESS which was more pronounced in 2016. Ice edge is a little more south at the top of Kara Sea in 2017.

On first viewing you would think the late Sept 2016 ice was making some early progress towards the Chukchi - but as we know subsequently, max ice development on the Pacific side was very low in the end.

 

[Pavel]

Interesting pattern of the Jet stream is going to set in several days on September 30. The wave from subtropics to the North Pole through the Greenland. Brings more heat to the Arctic, cold anomalies in Syberia and Greenland ice sheet growth

[jai mitchell]

Excellent summary of the 2017 season here:https://news.mongabay.com/2017/09/at-2017-minimum-scientists-ask-is-arctic-entering-the-thin-ice-age/

[Sterks]

Starting to get less and less interested on stories that fall on the side of greatest drama. Like a bell I have been hearing for years and is not alarming or thrilling anymore, but a bit bothering.
What about "Volume rebounds 2000 km3 from spring, but statisticaly continues its downward trend just as area and extent do"
Maybe  "A relatively cold spring and summer at high latitudes helps the Arctic end in slightly better conditions than 2016 and 2015"
Or perhaps "The least drama side (or cold-facts side) of several scientists expecting seasonally ice-free Arctic by somewhere mid-century, while not dramatic, may have some base after all"
The last one is an inconvenient statement for some with apocalyptic thoughts who want to see the blue Arctic in their lifetimes  preferably next year.

[Neven]

Starting to get less and less interested on stories that fall on the side of greatest drama. Like a bell I have been hearing for years and is not alarming or thrilling anymore, but a bit bothering.
What about "Volume rebounds 2000 km3 from spring, but statisticaly continues its downward trend just as area and extent do"
Maybe  "A relatively cold spring and summer at high latitudes helps the Arctic end in slightly better conditions than 2016 and 2015"
Or perhaps "The least drama side (or cold-facts side) of several scientists expecting seasonally ice-free Arctic by somewhere mid-century, while not dramatic, may have some base after all"
The last one is an inconvenient statement for some with apocalyptic thoughts who want to see the blue Arctic in their lifetimes  preferably next year.

Because the consequences of Arctic sea ice loss do not start once the Arctic is practically ice-free. They have already started, we don't know what the impacts will be exactly, and so, as long as things do not show any signs of returning to the pre-2005 mean (at the very least), the drama is great.

A seasonally ice-free Arctic by somewhere mid-century is dramatic. At the end of the century is dramatic. Do you have any idea how fast that is on geological time scales?

[gerontocrat]

Having had my daily dose of NHC.NOAA.GOV and looked at weather-forecast.com, methinks there is a chance of stormy weather arriving in the Arctic next week. Specifically, remnants of Lee in about 7+ days and Maria in 9+ days ?
Much depends on the persistence or otherwise of the high over NW Europe currently sending Atlantic lows North instead of hitting the UK.

Just looked at Pavel's post above re jet stream. This should increase chance of Lee/Maria remnants arrival in the Arctic happening?

[meddoc]

Yesterday snowfalls in some parts of Siberia are month earlier than usual. Snow depth in north-west Yakutia is 10cm what is the biggest ever for 27 of August, according to a russian wheather TV

The Arctic Refridgerator Door is open, still. And will only get worse.

[A-Team]

the consequences of Arctic sea ice loss do not start once the Arctic is practically ice-free. They have started already ... seasonally ice-free Arctic mid-century

The perennial confusion on the forums over what 'seasonally ice-free' means allows just about any futuristic belief system to flourish.

If you think there there must be a perfect low albedo match between open water and a cloud-free peak solar isolation season (6 weeks on each side of on the June 21st solstice) for it to 'count', that's a long ways off because melt hardly gets underway in early May now and doesn't peak until mid-September. Thermal lag isn't going away any time soon.

If you think there there must be a fairly good match between open water in regional seas and peak isolation for it to 'count', that's already here for the peripheral seas such as Chukchi, Barents, Beaufort, Bering, ESS, Laptev. The Chukchi did not freeze over until mid-December 2016, opened already in early June 2017 and has been totally ice-free now for months.

If you think that Arctic amplification of global warming is all about albedo, you're completely wrong. It's primarily a fall and winter effect of clouds and radiation imbalance over reduced sea ice. And what's happening in the Arctic today isn't staying in the Arctic.

The animations below revisit SMOS, the preferred satellite instrument for measuring thin ice (less than a meter thick). UH considers it fairly worthless in melt season and stops updating their image archive in mid-April. They've got a new paper out on it however, https://www.the-cryosphere.net/11/1607/2017/tc-11-1607-2017.pdf

Meanwhile UB continues to post their daily thin ice report. The first 25 days of this month are shown below, along with a comparison to UH sea ice concentration for the 25th. UB is showing a lot of thin ice poleward of Wrangel whereas AMSR2 has somewhat reduced ice concentration. Only ice less than 0.5 m is colored according to the legend. SMOS isn't able to locate 0.5 to 1.0 m ice this time of year. The agreement with ESRL ice thickness isn't stellar.

UB SMOS might also be worth comparing later on to ESRL thickness products which do not use this satellite. However, while UH provides their netCDF files, UB does not. That, plus a very small UB graphic in a peculiar color scheme, puts quantitative comparison with ESRL out of reach.

[Thomas Barlow]

Starting to get less and less interested on stories that fall on the side of greatest drama. Like a bell I have been hearing for years and is not alarming or thrilling anymore, but a bit bothering.
What about "Volume rebounds 2000 km3 from spring, but statisticaly continues its downward trend just as area and extent do"
Maybe  "A relatively cold spring and summer at high latitudes helps the Arctic end in slightly better conditions than 2016 and 2015"
Or perhaps "The least drama side (or cold-facts side) of several scientists expecting seasonally ice-free Arctic by somewhere mid-century, while not dramatic, may have some base after all"
The last one is an inconvenient statement for some with apocalyptic thoughts who want to see the blue Arctic in their lifetimes  preferably next year.

With all due respect, I may have misread you, but I think you are looking at it in a very black and white way: Ie. "Is it at the lowest, or is it not?", as are many people on the internet today.
I'm sorry, but that is just a very primitive way to think about the Arctic.

First off, to me (and I think this is what Wadhams thinks too based on e-mails and knowing how old Oxford boffins from a bygone era, tend to talk), is that a "Blue Ocean Event" doesn't really mean a full blown open ocean like the Mediterranean or something. It means a ton of icebergs floating around all over the place, with enough room to sail ships through them in most places, and a load of ice clinging to land masses.
ie. Icebergs everywhere, but no discernible (or very small) "ice-pack". Huge portions of the Arctic Ocean were getting close to that this year.

The Arctic Ocean itself is close to, or at, the worst shape (volume) on record, apart from thick ice crushed against land-masses. And this is the answer you should give to climate-science deniers who are all over the Net now, braying about how the "Arctic is growing".

North Atlantic waters seeping into the Arctic are warmer than ever, the Nares was open all year since last year, the SSTs all around are warmer, the overall Arctic Ocean volume (ignoring thick ice crushed against land) is close to the worst state on record at start of freezing season, if not the worst, the ice itself is said to be poor quality, the fire seasons are pouring more soot over the Arctic than ever, and the fire seasons are longer than before, the seabeds are thawing, the tundra is warming.

I try really hard to see the good news, and the idea that 'we dodged a bullet' being floated all over the internet now. I don't see it. “Extent” is not telling the true state the overall Arctic Ocean.
These general extent graphs are deceptive. I do think volume is the most important factor, and that volume in the overall Arctic Ocean to be the most important indicator. There could be more snow in winter, but that's about it (which is what climate-scientists predicted for decades now).

So "dodging a bullet",  or "7th worst on record", it is not.
But wait for the science-deniers across the right-wing news to start shouting that ""the Arctic ice is in the best shape in years, and getting better.""

I am very concerned that ice-aficionados are going to let the climate-science-deniers get away with it, because these ice-observers themselves are saying "things are not that bad, because extent (of sea-ice in the N. Hemisphere) is better than expected". I think it's dangerous, and neglectful.

(I know this is largely OT, just responding to what may be an OT comment - could put it somewhere else.)
.

[Niall Dollard]

I am very concerned that ice-aficionados are going to let the climate-science-deniers get away with it, because these ice-observers themselves are saying "things are not that bad, because extent (of sea-ice in the N. Hemisphere) is better than expected". I think it's dangerous, and neglectful.

(I know this is largely OT, just responding to what may be an OT comment - could put it somewhere else.)
.

I was wondering about the general focus from some of your previous posts, Thomas. Particularly in relation to comparing this year to previous. I have always assumed that the ASI Forum members are all pretty well knowledgeable folk and understand many of the intricacies and importance of the Arctic Sea Ice. I didn't think that deniers would ever be really interested in this forum, that is bar the odd troll.

So unless I am way off track here, I think we are all on the same hymn sheet and have great concern for the Arctic.

In your paragraph quoted, I believe you have explained the reasons for, let's say your extra sensitivity. Complacency. It is a valid point. We would not want to give any impression to others outside of this forum that the Arctic is somehow turning a corner.

However inside the forum I would hope that members here are free to highlight discuss debate the intricacies of the ice. Good or bad. If I make a comment that volume is gone up - it is no way intended as to give a boost to the deniers. There are plenty of places on the net where they can gather their spurious facts - I doubt if they are bothered about what you or I are saying.

I would be concerned that healthy discussion would be limited if we all had to hold our tongue typing for fear that we are somehow letting the side down.

But I doubt if anyone on here is being complacent. There are things I would say here for example that I would be careful saying outside of this forum.

[Thomas Barlow]

I am very concerned that ice-aficionados are going to let the climate-science-deniers get away with it, because these ice-observers themselves are saying "things are not that bad, because extent (of sea-ice in the N. Hemisphere) is better than expected". I think it's dangerous, and neglectful.

I didn't think that deniers would ever be really interested in this forum, that is bar the odd troll.
So unless I am way off track here, I think we are all on the same hymn sheet and have great concern for the Arctic.

Yes, you are WAY off track to what I posted.
Try again. Go back, read it again, without prejudice. It's English, so if English is your second language, I'm sorry, let me know and I will re-explain it.
I said NOTHNG about people here having to hold their tongue. Your prejudice is palpable, and you are wasting cyber-space with erroneous interpretations.

And why didn't you pick up on HIS statements about people on this forum? : eg. ""The last one is an inconvenient statement for some with apocalyptic thoughts who want to see the blue Arctic in their lifetimes ;-) preferably next year."" His whole post was a critique/smear of people on this forum (not of me, because I am not a doomer, but a rationalist). whereas mine wasn't. His was CLEARLY a veiled comment on some people's doomerism here. There's nothing really wrong with his comment by the way (except that it seems OT to me), and it stimulates conversation - about the topic at hand, not whole posts about personality judgements, as yours is -  but you deciding that I am  the one 'gaslighting' instead of the other, is just you projecting your own prejudices, and wasting space.

Unless English is your second language, in which case I apologize.

Stop wasting space with personality attacks, and deal with the substance of my post. You are just deflecting from the discussion and substance of the post with these  personality cliques.

(If you delete your post, I'll delete this one, because this personality BS is a waste of space. You can try critiquing the substance of my post instead - about the Arctic. Maybe it needs the be on a different thread though. This is 'The 2017 Freezing Season', not some place to critique people's concerns or lack thereof about the Arctic, as Sterks started with. I'm quite happy to have this discussion moved elsewhere. Deal with the substance of my post next time.)

[gerontocrat]

I look at this thread for data.

[Forest Dweller]

The perennial confusion on the forums over what 'seasonally ice-free' means allows just about any futuristic belief system to flourish.

If you think there there must be a perfect low albedo match between open water and a cloud-free peak solar isolation season (6 weeks on each side of on the June 21st solstice) for it to 'count', that's a long ways off because melt hardly gets underway in early May now and doesn't peak until mid-September. Thermal lag isn't going away any time soon.

If you think there there must be a fairly good match between open water in regional seas and peak isolation for it to 'count', that's already here for the peripheral seas such as Chukchi, Barents, Beaufort, Bering, ESS, Laptev. The Chukchi did not freeze over until mid-December 2016, opened already in early June 2017 and has been totally ice-free now for months.

If you think that Arctic amplification of global warming is all about albedo, you're completely wrong. It's primarily a fall and winter effect of clouds and radiation imbalance over reduced sea ice. And what's happening in the Arctic today isn't staying in the Arctic.

The animations below revisit SMOS, the preferred satellite instrument for measuring thin ice (less than a meter thick). UH considers it fairly worthless in melt season and stops updating their image archive in mid-April. They've got a new paper out on it however, https://www.the-cryosphere.net/11/1607/2017/tc-11-1607-2017.pdf

Meanwhile UB continues to post their daily thin ice report. The first 25 days of this month are shown below, along with a comparison to UH sea ice concentration for the 25th. UB is showing a lot of thin ice poleward of Wrangel whereas AMSR2 has somewhat reduced ice concentration. Only ice less than 0.5 m is colored according to the legend. SMOS isn't able to locate 0.5 to 1.0 m ice this time of year. The agreement with ESRL ice thickness isn't stellar.

UB SMOS might also be worth comparing later on to ESRL thickness products which do not use this satellite. However, while UH provides their netCDF files, UB does not. That, plus a very small UB graphic in a peculiar color scheme, puts quantitative comparison with ESRL out of reach.
[/quote]

Thank you A-Team,

that explains some things even to a novice like myself.
Any thoughts on that polyanna which seems quite big?
Or is that a regular pehenomenon?

[Neven]

I look at this thread for data.

I apologize, I should've (re)moved Sterks' comment, instead of replying to it.

[Sterks]

Yes, Neven, removable, OT, but there was a Jai Mitchell's off topic just above mine to which I reacted. As I would have, if the headline he was bringing had been ""the Arctic ice is in the best shape in years, and getting better".
All the same, letting politics and/or interests one side or other twist the truth or plainly lie.
Back to topic.

[A-Team]

FD asks: Any thoughts on what caused that big polynya this year? Or is it a regular phenomenon?

Its origin and persistence remain a mystery but it finally appears to be getting sheared apart. We put our heads together on causation but nothing persuasive ever surfaced that made physical sense for that location.

A look-back for 2012-17, displayed as a CW hexagonal kaleidoscopic animation below, indicates polynyas developed at a similar (but not identical) location in some but not all years, unhelpful. I meant to email a Russian scientist who specializes in that region's ice for comment but never got to it. We'll have to get back on it if it re-forms again in 2018.

Origin and persistence should probably be separated. The ice pack, viewed from far way, deforms somewhat like bread dough. Just as the hole in a doughnut survives on a hockey rink despite getting knocked around, once formed, the polynya could be somewhat protected by rigid ice around it.

The 2017 summer melt season saw very little change weather patterns or ice pack movement. The staples of Arctic ice motion -- Beaufort Gyre, Transpolar Drift, Fram export, Nares export and CAA garlic press -- were basically inoperative. The bottom animation shows the polynya over the 18-26 Sep 17 time frame. Strong CCW rotational shear can be seen on the 22nd and 23rd, presumably from the low pressure system discussed up-forum.

The origin: who knows, maybe there was a week there with a break in the clouds. Happenstance with no deeper meaning is the null hypothesis and it is not easy to find grounds to reject it.

See https://forum.arctic-sea-ice.net/index.php/topic,1834.msg125949.html#msg125949 for more discussion and imagery.

[Pavel]

Cold anomalies in Siberia and the Laptev sea cools quickly. I was expecting slow freeze up of the Laptev but now it looks it will freeze like in previus years. Chuckchi and ESS still look terrible, that may lead to low November SIE but we'll see

[Niall Dollard]

We had two 100K AMSR2 Area gains in the past two days, finally getting a bit of freezing momentum going. Much of it was from the Russian side. Currently air surface temps there are circa -10 C near the ice edge.

Teasing through today's ECMWF 12Z surface temp forecast for the Arctic, shows that cold will hold for a while yet near the Russian edge. But warm southerlies will be moving north over Svalbard to the Pole.

This anticyclonic ridge over the Pole sinks slowly south and east. By the end of the run (out to end first week October) the anticyclone is now centred in a more typical Siberian position. The Arctic on the Russian side will have lost its cold surface conditions at that stage - which should put a slow down then on ice development.

Meanwhile the American side becomes more dominated by low pressure (by end of run pressure falls over much of Arctic Basin).  Some cold temps and likely ice expansion through the CAA then next week and later into the Beaufort. Nothing much to be expected from the Atlantic side (more in retreat mode). Chukchi is very slow this autumn.

 

[A-Team]

Here is the next ten days of surface temperatures of the Arctic ice pack itself, not the air. The animation of ESRL 6 hr x 10 day forecast uses the orthographic projection (view from space) as described in the new PanoplyCL javascript manual. This shows the earth rotating with its sun shadow as the colors change for the surface temp as time precedes.

The Arctic Ocean is too small for the whole-globe view to be effective. It may be feasible however to fill in the northern oceans with nullschool weather displays as its orthographic settings can easily be set to match those of PanoplyCL.

Here the viewpoint is directly above an initial lat,lon. In this instance, the first frame is directly below Little Diomede in the Bering Strait (65.7º, -168.9) but it worked better to drop down to 45º,-170º and hold the latitude fixed while longitude was incremented by 360º/40 ESRL frames = 9º per time series advance.

That series of lat,lon can be replicated in calling up the corresponding view in nullschool's Earth. Those are set in the url. However Earth uses a 3rd scaling parameter (zoom or height of the view above the surface); its correspondence needs to be established with the 'Visible Radius' parameter of PanoplyCL which is 90º below.

Because PanoplyCL makes clean colored maps (fixed palette, no dithering) whereas nullschool does not, the overlay is best done by removing land and water from the Panoply display, letting nullschool display show through for non-ice regions (still image below).

Once all the ducks are in a row, the whole animation just takes 2-3 clicks of the mouse to make  and load onto the forum.

[slow wing]

  A-Team, thanks for your (typically) wonderful graphics above #118 concerning the persistent hole observed this year in the ice north of the Laptev Sea.

  You conclude "Happenstance with no deeper meaning is the null hypothesis" and say there is no reason to reject it.

  Your graphics show, as you say "polynyas developed at a similar (but not identical) location in some but not all years" from 2012 onwards. In my view that would be surprising under the null hypothesis. The persistence of the 2017 hole was already surprising to me.

  While I agree the null hypothesis cannot yet be rejected, in my view there are other hypotheses that are reasonable and worth exploring further.

  In particular, the polynya was over the Gakkel Ridge - a prominent baryometric feature and with known geothermal vents.

  An issue that needs to be addressed here is that underwater volcanism has been abused by some as a counter to climate change concerns. Here is an example where a polynya was observed over the Gakkel Ridge and all sorts of non sequiturs were drawn from that: http://climatechangedispatch.com/heat-from-deep-ocean-fault-punches-hole-in-arctic-ice-sheet/.

  The intuition of some climate scientists has been that the volcanism probably doesn't have much affect on the sea ice: https://dotearth.blogs.nytimes.com/2008/07/01/whats-up-with-volcanoes-under-arctic-sea-ice/.

  However, that seems somewhat equivocal. The argument is that the heat certainly effects the local ocean depths but is diluted before it gets to the surface.

  The amount of volcanic heat averaged over the Arctic basin is said to be of order 0.1 W/m^2. As it takes around 334 kJ/ kg to melt ice, that is about enough to melt around 3 mm of ice over 4 months (10^7 seconds), averaged over the Arctic basin.

   But what if much of the heat from the Gakkel Ridge volcanoes effectively reaches the surface over the much smaller region where the polynya formed? That much heat over ~1/300th of the Arctic Basin (say 20,000 km^2 which, eye-balling, looks of order the size of the polynya) could melt of order an extra metre of ice rather than just 3 mm and so could punch a polynya in the ice.

  To me the polynya was a notable and recurring feature and some sort of mechanism for getting volcanic vent heat to the surface to cause that is not ruled out as far as I know.

   Are there measured temperature and salinity profiles in that region? (I seem to recall there being a string of Russian tethered buoys but that may be closer to shore.)

UPDATE: there is some instrumentation, see http://research.iarc.uaf.edu/NABOS2/technology.php:

[Rob Dekker]

If you think there there must be a perfect low albedo match between open water and a cloud-free peak solar isolation season (6 weeks on each side of on the June 21st solstice) for it to 'count', that's a long ways off because melt hardly gets underway in early May now and doesn't peak until mid-September. Thermal lag isn't going away any time soon.

While I don't think anyone claimed that there must be a perfect low albedo match between open water and a cloud-free peak solar isolation season, PIOMAS does seem to suggest that the highest rate of melt is around summer solstice (June 21) :



So while it may intuitively feel like thermal lag should be a major player, volume development suggests that ice melt is mostly dominated directly by insolation, especially during recent years.

[Niall Dollard]

Here is the next ten days of surface temperatures of the Arctic ice pack itself, not the air.

Yes. Nice animation. It concurs very much with the ECMWF, showing the pack warming right up the middle of the Arctic, the colder temps leaving the Russian side and a cold plunge through the CAA.

If the forecast holds through, the conditions by Oct 7 are mild over much of the Arctic and we could see freezing plateau for a time, at that stage.

[A-Team]

mild stretch coming

Good. Unlike windyty, we don't have the ECMWF option available for ESRL or nullschool. Here is that same data for surface temperature of the ice pack, now in polar stereographic. The forum at first refused to load it, claiming an ordinary gif "failed security checks, contact admin" but now it seems to pass muster.

peak rate of volume loss

That would be the slope (first derivative) of the Piomas volume graphs. Since that is determined primarily as a first-of-the-month monthly average, those graphs are curve-fit to a very sparse set of points, only 3-4, so the nuances are illusory. As the graph stands, the rate of volume loss is flat (2nd derivative indistinguishable from zero) for most of the summer. Some products have seasonally varying error (for example SMOS thin ice thickness) which complicates assessment.

People do not swim much in Lake Superior in June. Mid-August, yes.

UPDATE: there is some instrumentation, see http://research.iarc.uaf.edu/NABOS2/technology.php:

Thanks for relocating that. Those would be the people to ask about the persistent polynya. The late Sept view is updated through the 27th below. It would be most impressive if the polynya were able to restore itself after days of significant shear and translation.

The only documented Gaekel Ridge volcanism is a good thousand km away where it is widening much faster; short-lived magmatic events at 4000 m depth cannot plausibly bring thermal upwelling to the surface. You can see that just from the minimal effect of Kilauea Iki lava flow into the ocean on youtube. It skins over quickly, forming a barrier to diffusive heat flow. The energy is just not there to heat cubic kms of very cold water fast enough, a density gradient would dissipate rapidly via other processes.

[oren]

peak rate of volume loss

That would be the slope (first derivative) of the Piomas volume graphs. Since that is determined primarily as a first-of-the-month monthly average, those graphs are curve-fit to a very sparse set of points, only 3-4, so the nuances are illusory. As the graph stands, the rate of volume loss is flat (2nd derivative indistinguishable from zero) for most of the summer.

I believe Wipneus plots his PIOMAS graph from the daily PIOMAS data, not from the monthly, and from my experience with the data there are differences in slopes during the month. Of course it doesn't change the fact that the smoothed 2nd derivative is near zero during most summers.

[A-Team]

The daily is not available for earlier years, even though smooth graph lines are shown for them. Only recently have we gotten mid-month. Wip wrote at one point the daily was too noisy to work with. (This could be quantitated statistically by variance etc. but that hasn't surfaced.) Some sort of smoothing, like a rolling window, may have been applied. The algorithm is effective volume averaged over grid cells. This doesn't really show on the daily animation of the month because the grid cell is far too sparse.

It would be very difficult to take derivatives on this sort of data as error analysis would be very central to that. To compare derivatives at specific dates by subtraction of very similar numbers at a statistically significant level brings in another whole new array of problems.

While ESRL only serves volume implicitly as a thickness grid (only from 15 Aug 17 on), anyone here could javascript the daily differences using a stock manual script and default 'Combine Plot' which is subtraction. It's unusual though for good things to emerge from subtracting two large numbers that are very similar, so it might be better to try weekly.

But even that runs afoul of the very small change in thicknesses involved relative to model precision. You can see that just by watching minimalist squiggles in the zonal average plot (area-under-the-curve is effectively volume, after adjusting to the number of cells). The animation is edge-cropped so volume measurement and its change are just a single click with a contiguous color pixel counter (available in any post-1998 imaging software). But just from comparing the change from first to last frame (10 days), daily differencing will prove delusional (2nd image). It's asking too much of a model. There's better action watching paint dry.

No one has been able to get Piomas out of its funky format into standard climate science formats like geolocated gridded HDF5 (or at least there seems no online archive of it). That would enable easy comparisons with other estimates of daily volume and indeed that is a primary purpose of these formats.

In the larger picture of global climate change, there's no central archive, so computers crawl the internet to find machine-readable data for maybe 25-30 variables to collect the input for to run ever increasingly complex models. It is unworkable to have every archive using a different ad hoc storage scheme. That's silo science; we don't have time for that anymore.

[Blizzard92]

Hi all,

As we are now moving forward into the freeze season, I have updated my FDD monitoring plots for 2017-2018 at http://sites.uci.edu/zlabe/arctic-temperatures/

[Shared Humanity]

Hi all,

As we are now moving forward into the freeze season, I have updated my FDD monitoring plots for 2017-2018 at http://sites.uci.edu/zlabe/arctic-temperatures/

Time will tell but it would appear that 2017 is tracking with 2016.

[Daniel B.]

Hi all,

As we are now moving forward into the freeze season, I have updated my FDD monitoring plots for 2017-2018 at http://sites.uci.edu/zlabe/arctic-temperatures/

Time will tell but it would appear that 2017 is tracking with 2016.

To early to make too much of a pronouncement.  One could also say that it is tracking with 2015 or 2011.

[Ice Shieldz]

A-Team:

"It is unworkable to have every archive using a different ad hoc storage scheme. That's silo science; we don't have time for that anymore."

Absolutely, that is the most salient truth in all this, and to my understanding it's a key reason why ASIF exists. So thank you for all your work in reaching across the silos and bringing the science together in a visual and accessible way- especially now when we need it most!

[A-Team]

bringing the science together in a visual and accessible way

What I am doing, one-off prototyping, is not really sustainable. The real excitement is over at Dev Corner where Dryland is building a master controller that can inject data sources into sequential scripts of retrieval, html templating, Panoply, ImageMagick, gimp, ncgen and others.

When there is massive daily updating of multiple resources, task automation becomes essential, lest there be no time left over for interpretation and commenting as to what it all means.

In some cases, this might even enable innovation: for example just as UH merges Cryosat2 and SMOS to better get at thin ice seasonally, we might take their product and overwrite RASM-ESRL ice thickness on its periphery where a model will be suboptimal relative to observational data..

The sea ice thickness animation below is one that automation could roll over once a day. It joins 39 daily initial states to ten days of forecast. The ESRL archive of D0's begins on Aug 15th and runs to Sept 28th at which point daily forecasts take over. (There is some sort of normalization glitch prior to Sept 16th.)

Technical note: automating conflict resolution between files 'too big to load' and maximal size of display that will run without a click is fairly easy: a montage of all the frames determines safe autocrop boundaries that won't clip off ice in any of the frames. Three clicks in Gimp.

Technical note: forum software won't accept png animations though web browsers do. For overlays, say Hycom over ESRL thickness, the number of colors will double the gif limit of 8-bit. Each frame has to be cut down to 256 colors in total; that process forces loss of smoothness (gross dithering of minor colors). However palettes like UH AMRS2 get by with 100 colors. Hycom can then be allocated  the remaining 156 (via mode shift to indexed color in Gimp). That is plenty for both to display colors smoothly in the final gif.

https://forum.arctic-sea-ice.net/index.php/topic,1259.msg130182.html#msg130182

[Niall Dollard]

Thanks A Team.

I remember asking about strange ice thicknesses near the ice edge in the Beaufort earlier in the month. So it was a normalisation glitch. Comparing these before and after glitch still images attached, the glitch was especially pronounced in the magenta highlighted areas, with thicknesses stepping from dark red to blue.

But after Sept 16th, the thicknesses do look on a par with what you would expect this year - before that date, thicknesses are not trustworthy. 

[Sterks]

Thanks A Team.

I remember asking about strange ice thicknesses near the ice edge in the Beaufort earlier in the month. So it was a normalisation glitch. Comparing these before and after glitch still images attached, the glitch was especially pronounced in the magenta highlighted areas, with thicknesses stepping from dark red to blue.

But after Sept 16th, the thicknesses do look on a par with what you would expect this year - before that date, thicknesses are not trustworthy.

The ice model was forced to assimilate a PIOMAS update. (?)

[echoughton]

Daniel B And others...uh...the freezing season is tracking like every season ever...it's freezing

[Daniel B.]

Daniel B And others...uh...the freezing season is tracking like every season ever...it's freezing

lol.

Water does what is always does when the temperature drops?

[gerontocrat]

Yes, it's freezing - but how quickly? DMI 80+ degrees north graph.

[oren]

After the initial refreeze of the.central CAB, the refreezing very much stalled, but in the last few days it picked up vigorously. Initially I would have bet on another lame freezing season, but now I am back on the fence.

[A-Team]

after Sept 16th, the thicknesses do look on a par with what you would expect this year - before that date, thicknesses are not trustworthy.

No, it's just a scaling visualization issue. Here the ESRL program is experimental, PanoplyCL is in beta, and I am making novel forum-adapted graphics out of raw data files like RASM-ESRL_2017-09-28-00_t048.nc rather than using 'official' ESRL web products (from which glitches may have vanished in their better-informed processing). There's no methods paper out yet.

Here's how scaling works: the ice thickness numbers modeled for a given day fall within a certain range. If PanoplyCL uses the default 'fit to data' setting, the palette colors will be assigned evenly over the full range of data values, even if some are sparsely occupied or altogether unrepresented.

However maybe 0.5% are artifactual outliers due to some internal problem in the calculation, for example near the coastline. These outliers can fluctuate wildly from day to day. If they are not filtered out somehow, because palette color assignments follow the range in 'fit to data' mode, the colors of the map won't be consistent from day to day (ie frame to frame in the animation).

Outlying thicknesses will get palette colors on the far right, intermediate colors won't be used at all, and core range thicknesses will be crammed into what's still available in the palette.

Meanwhile, the core range that ice thicknesses take on is much smaller and varies seasonally but very little from day to day. After skimming through the files, I replaced 'fit to data' with a fixed 3.6 m for the outer limit of thickness for all 49 days to get consistent coloration. Ice at higher thickness, some reasonable like 3.65 m and some obviously erroneous but rare outliers like 17.5 m get collapsed into the final palette color, just like in WorldView.

Around 16 Sept, ESRL seems to have refined the internal algorithmic glitch that was creating the outliers. From that date on, the ranges varied little and 'fit to data' barely differed from 'fit to 3.6'. That doesn't imply that the earlier data was wrong, only that 'fit to 3.6' wasn't the right scale for synching the palette scale to actual thicknesses or for smooth color transitions over the change.

Since the early parts of the animation were well synched to each other up until the glitch, they likely just need a constant rescale, say 5.0 instead of 3.6 m, to make the whole series harmonious. To determine that rescale, reload the the 15th and 16th in Panoply's linear grayscale and divide in Combine Arrays.

Some computer mishap seems to have occurred on Sept 10th. There are data gaps on 8th and then again for the 11th and 12th which have not been filled. On the 13th (and subsequently) when the archive resumed, they dropped the t024.nc time slot, going right to forecast day two t048.nc. Elsewhere a file would be described by its start and stop hours, eg t000 to t024.

These mishaps occur in all the archives. Yesterday, on another forum, a pacman data gap in AMSR2 drew comment. That's not been fixed and cannot be fixed if the responsible satellite or receiving ground station had a glitch.

In the bigger picture of netCDF climate files, pre-calculated statistical properties of the data distribution should be part of the file, not sit implicit in machine-readable binary. That would allow consistent automated definition of outliers as well as 'histogram equalization', both discussed at length over at Dev Corner. Panoply does provide thicknesses averaged over latitude (with land masked out) which is a start.

RASM-ESRL ice thickness has commonality with sparsely gridded Hycom. Both of these are in serious conflict with the less-nuanced ringed display of Piomas. A round of Cryosat2 flight lines takes a month to get close to full coverage; it's challenging to compare experiment with model.

The only thickness reset available is new ice forming each fall from open water. The older ice will wander off farther and farther into uncertainty as time goes on, whatever the modeling system. Ice thickness is a derived product in RASM-ESRL; they have 49 outputs overall.

From the perspective of differential equations, where did the initial conditions come from on the start date? It's time evolution won't be any better than that. Daily satellite inputs might reset attributes like ice boundary or melt pond fraction but thickness is known only on the open water (zero).

The model is initialized with the NOAA Global Forecast System (GFS) analyses and the Advanced Microwave Scanning Radiometer 2 (AMSR2) sea ice concentrations. The model is forced at the lateral boundaries by 3-hourly GFS forecasts of winds, temperature, and water vapor.  Weather Research and Forecasting (WRF3.5.1; run with 40 vertical levels) atmospheric model; the Parallel Ocean Program (POP2) model; the Los Alamos Community Ice Model (CICE5.1; and the NCAR Community Land Model (CLM4.5). All components, run at 10 km horizontal resolution, are coupled using a regionalized version of the CESM flux coupler (CPL7), which includes modifications (Roberts et al. 2014) important for resolving the sea ice pack response to weather events. Other model optimizations include: a bulk double-moment cloud microphysics scheme for droplets and frozen hydrometeors, running ensemble forecasts initialized with GEFS ensemble members, and extending the model domain to include the Bering Strait and Svalbard.

The most striking feature in RASM-ESRL thickness animation (passing over glitch) is how little the ice has moved about this summer (or grown or shrunk). The patches of enhanced thickness jostle about indecisively. There's no indication of textbook Transpolar Drift or Beaufort Gyre. If that persists, the plan for the frozen-in Polarstern might need serious revision.

The animation below shows the dramatic effects in map coloration using a fixed palette to display the same data in core vs outlier range settings. Scientific visualization has a great many issues; we can explore them only when the data archiving site provides the underlying netCDF file. Many do, but others just post their favorite depiction (which may use an awful palette and bury data under text and grid lines).

[Pavel]

Yes, it's freezing - but how quickly? DMI 80+ degrees north graph.

Looks slightly colder than 2016 but persistant heat advection forecasted in next 10 days. The freeze up of Laptev/North Kara may happen a bit earlier than last year but in general it tracks like a quite mild (not extremely mild)

[numerobis]

There’s a rather decent Arctic cyclone over Hudson Strait and south Baffin today. YFB barometer reads 985 hPa and falling. It’s dumping 10+ cm of snow, which means insolation down here is getting bumped back up from now on.

Edit: 981 in Iqaluit, 972 min pressure on this storm.

We could be making so much electricity right now!

[Niall Dollard]

No, it's just a scaling visualization issue.

That's no problem. No I don't have any issue with the ESRL data and I appreciate all the development work. 

Recent snow cover image from US Natice shows cover extending through northern Russia. However not a lot yet to the northeast of Yakutsk.

[A-Team]

Here is the official ESRL ice thickness product from Aug 22nd to Sept 28th, Alaska down, non-stereographic projection. Or rather thinness. They set a palette cut-off at 1.6 m or less. Everything thicker is given the same color as the 1.6 m bin.

Someone was complaining about this up-forum, why not use a higher cut-off like 2.6 m (or more) and display the thicker ice too. That would have required 10 more palette colors. Right now, ESRL is using 16 colors already in a so-called paired palette. It's quite effective but pushing the upper limits already on workable number of discrete colors.

Whoever made this map, their interest was in displaying thinner ice and wind. These home page graphics are archived ok as REB_plots but as final gifs, not netCDF files. Otherwise we could have easily redone the graphic to a higher cut-off in an augmented or wholly different palette, in Greenland down stereographic without the wind vectors and lat lon lines.

Note prior to this date, the ESRL map did not cover the entire Arctic Ocean. They are unlikely to revisit earlier dates so the extensibility question comes up, how hard is it to add subsequent days to the current calendar coverage? Not hard at all. The file is at 3.5 MB for 34 days; it could be brought out to ~100 days before hitting forum limits.

They offer a D5 forecast to Oct 3rd, though not D1-D4. It has to be resized as its dimensions are off. That's attached separately rather than add to the initial state series. ESRL does not currently update those later as improved hindcasts.

It would be feasible, if we had the underlying netCDF, to see how they're doing by comparing old D5's with their later initial D0 for that date. However it's very hard to demonstrate forecast skill during long stretches of uneventful weather as the baseline forecast-to-beat  (each day just like the previous) does so well.

The ESRL map does manage to pick up the persistent polynya on some days (as flashes of white), even though the resolution is nothing like the 3.125 km of UH AMSR2 that displays it clearly.

I looked around to see if this was a known matlab, ncr, gmt, panoply, or colorbrewer palette. It wasn't. Methods for capturing a nice palette like this and saving it into Panoply are described over at Dev Corner. It could be used to color re-color the time series above but it would be difficult to match ESRL's projection without more information.

[Niall Dollard]

Some Russian coastal freezing extending into the Laptev (bottom centre) , showing up on Worldview today. 

[A-Team]

Clouds in the summer play two important roles by reflecting incident sunlight and intervening in upwelling and downwelling radiative transfer processes. By September, the sunlight is effectively gone and the interest shifts to the insulating properties of clouds, keeping heat in and so delaying open water refreezing and opposing the thickening of ice.

However the quantitative details are quite complicated and we will need to follow energy fluxes this fall, the best/only source for that apparently being certain ESRL products.

The less-than-great animation below shows the cloud deck for the month of Sept in channels 7-2-1 of Terra Modis. Because the satellite pulls together pie-shaped wedges as it runs through a full cycle of its near-polar orbit over the course of a day, the image is a composite of different times.

Further, as the season progresses, the 'pole hole' of Arctic darkness gets larger every day, making cloud monitoring impractical already by Oct 1st. The blue line marks the open water boundary as derived from UH AMSR2. Outside the line, we are looking at the effect of clouds on sea water freezing; inside the line, on the effect on ice thickening.

Relevant to this, ESRL provides daily surface air temperature, air velocity ,sea surface temperature, ice surface temperature, salinity, snow thickness, ice thickness, bottom and lateral freezing rates, and ice edge evolutionn (newly fixed, freezehour). Salinity affects the freezing temperature of sea water, lowering it in the Arctic to -1.8ºC, though following full brine exclusion 0.1ºC is enough for re-melt.

From these and cloud satellite inputs, the RASM_ESRL model derives various energy fluxes of interest to us this season: net surface energy flux (Arctic24, lower left, below), net energy flux at surface (Arctic23, UL left), longwave flux (Arctic22 LL), GFS net shortwave flux (Arctic 22 UR), snow/ice/ocean absorbed solar flux (Arctic 22 LR, flatlining now), 310K potential vorticity and surface theta (Arctic20), heat flux ice to ocean at ice edge (Arctic15), net energy flux at surface yet again (Arctic14), longwave net, shortwave through ice to ocean, snow/ice broadband albedo (all in Arctic11), precip (Arctic9 and 4), ice thickness thermodynamic tendency and snow melt (Arctic3), and snow depth (Arctic1 LL), the bottom line being refreeze season history and forecasts.

These will take some time to go through to see which holds the most interest for us. The net surface energy flux, below, is entirely in the negative over the forecast period. The map is one of their older formats emphasizing an Alaska perspective rather than the whole Arctic Ocean we want. Since no netCDF file is provided, only the final 28-frame gif, there's no role for Panoply here, just Gimp. It's also quite a nuisance to extract only the D0 from each many separate animations in order to construct the daily repository without the forecasts.

Unfortunately no one here follows the three main cloud satellites, two of which are in the A-train constellation. These may have data archives that we can work with directly in Panoply.

-- CloudSat, a cooperative effort between NASA and the Canadian Space Agency, runs 2 minutes and 30 seconds behind Aqua, launched with CALIPSO on April 28, 2006.

-- CALIPSO, a joint effort of CNES and NASA, follows CloudSat by no more than 15 seconds, launched on April 28, 2006

[Pavel]

The early freeze up along the Taimyr peninsula this year

[A-Team]

Here is the multi-year comparison, both visual and boolean, for 2017 vs 2012-2016. This is the last day of the year for which 2012 is in the UH AMSR2 archive. This year had quite respectable melt except when compared to 2012.

The lower images look at the peripheral freezing forecast from ESRL, called freezehour.gif on their web page archive. This is a difficult scientific visualization problem because so little of the image is ice margin.

Initially, they had the rest of the ice set off from open water (white) but re-used a palette color that had been set aside for hours of forecast. Also they had far too many colors using the same small space. Next they reduced the palette but this time used white for both a middle forecast hour and open water.

The underlying problem across ESRL's graphics is the difference between 0 and NaN. They need a consistent open water resp. ice mask. (For a variable like ice surface temperature, open water has no applicable number, not the same as zero in the palette.) Here the notion of replacing a gif animation using colors for successive frames isn't effective or accurate, the latter because there's back and forth on freezing as well as ice movement.

The top animation, not provided at ESRL, uses a sequence of 24 consecutive static forecast images. It shows the ever-widening periphery of the ice pack quite well. The way they have the archive structured, it is necessary to download a large compressed Reb_plot bundle for each day and then fish through all this folders for instances of the graphic, even if as here only one of the 44 graphic products is wanted. We are in the process of fixing that.

The bottom animation, also not provided at ESRL, restores time to their static image by cycling through the palette. This is better but still doesn't accommodate ice motion which is weird since they have this forecast elsewhere. I don't know what software they are using but it is not matlab, NCAR's NCL, Panoply, Photoshop, ImageJ, or Gimp.

What we are after here, come freeze-over in Jan 18, is a map of first year ice colored by its freeze-up date. In other words, open water that forms ice in early October has 3 extra months of thickening available relative to Chukchi ice that first freezes in late December. Other things being equal, the ice last to freeze will be the first to melt.

As more of the Arctic Ocean ice pack becomes first or second year ice, there's a benefit to age class refinement. If the lack of significant weather persists, compaction, dispersion and export will not interfere with this mark-up and tracking newly formed ice.

[Thomas Barlow]

Peter Wadhams interview on Radio NZ - Oct. 2 2017.
""You can't measure ice thickness from space...well, only with extreme difficulty...so the best way to measure thickness is to sail underneath the ice, and use an upward-looking echo-sounder.""

http://www.radionz.co.nz/national/programmes/afternoons/audio/201860805/peter-wadhams-preparing-for-an-arctic-without-ice

[Niall Dollard]

I like that freeze progression image, A Team.

Current forecast models are suggesting the Arctic is near to its peak temperature anomaly (for the next 10 days) around now. The recent Scandanavian anticyclone has directed a surge of warmth north but at least the Arctic has been spared the remnants of tropical storms Maria and Lee, which instead have moved in an easterly path across the Atlantic.

Instead of high pressure, the Arctic is set to be dominated by a series of slack depressions. 850hPa temperature anomalies are still showing high temps over the Beaufort and later through the CAA and northern Greenland. But temperature anomalies drop again over on the Russian side  for the Laptev, ESS and at last there are signs of progress being made towards the Chukchi.

Given the still warm SST anomalies especially north of Alaska, it is difficult to see how the Arctic can cool to a level that would bring the DMI N80 graph close to the 1958-2002 mean. Very unlikely, until the ocean's ice cover north of Alaska/Russia is complete.