The 2017 melting season

[A-Team]

THANKS!

No problem. I should like to thank a very generous donor in Greece for a new computer -- the fastest Mac ever built (no hard drive etc) -- which makes more complex animations possible, the Evans and Sperl Foundations for extensive logistical support, countless others for all the open source code and initial imagery, and of course Neven and the regulars for years of most excellent hosting and posting.

This new ESRL site, while experimental, appears a real breakthrough for both forecast and reanalysis products. We need to somehow assimilate what they are doing to stay in the  game for near-real time.

It does not work just to gesture at their url and say it shows such-and-such because they overwrite the online graphics every day. It's not possible to link to individual gifs because they're in a bundled -up archive compressed. Nor is copy-and-paste off the web page effective because their imagery is over-sized.

Yet we'd like to be like wikipedia on these forums and have every posted claim linked to a supporting source lol.

Quite a few people here are now doing respectable stills and animations though cropping, contrast, and resizing skills remain elusive. It's possible but not easy to restructure ESRL products to 700x700 pixel forum size with full retention of information. Panoply can potentially do this better without adding another layer of complexity -- provided it comes to you with configured defaults.

So for now we need to look at exactly which of the 37 ESRL products are most useful going into the refreeze season and how to get a dozen or more members reprocessing up to forum standards. It may be feasible to mirror the ESRL site in a Panoply post-processed sense which would the bar enough to let everyone in.

Right now the ESRL archive is up to 10.1 GB with Aug 31st upload taking the form of three file types summing to 330 MB unpacked; the netCDF can be opened in Panoply remotely.

RASM-ESRL_4NIC_2017-08-31.tar.gz   37.4 MB REB.2017-08-31.nc   232 MB REB_plots.2017-07-27.tar.gz 60.8 MB

Panoply itself offers not a single word towards a v4.8 help manual despite a 2002 initial release but there's a 2012 German tutorial that works through your choice of Pangaea data set that illustrates Panoply's capabilities.

https://www.geo.uni-bremen.de/Interdynamik/images/stories/pdf/visualizing_netcdf_panoply.pdf
http://www.meteor.iastate.edu/classes/mt452/EdGCM/Documentation/EdGCM_Panoply.pdf version 1.5 (ancient)

Below is the same ESRL sea ice motion as above re-animated using different Panoply features and options. This is a vast improvement over Hycom's forecast motion both in scientific accuracy and graphical communication of it.

[magnamentis]

so, instead of repeating words i thought to illustrate where i see this season end:

2n lowest in the second half of the range of dates that minimums occured earlier.

reason is that there is a lot of ice widely spread in parts even peripheral and it's so thin
and fragmented that bottom melt and every kind of wind and wave action will have an impact on that ice in parts till october and i'm sure we shall sooner or later see heavy winds and waves
hit different regions so that the ice will remain in a fragile and relatively thin state so that bottom melt can continue to do it's job quite late into the year. even an october minimum would be possible even though i more tend to see a minimum between 17th and 23rd of september which of course is an arbitrary choice more of a gut feeling, let's see.

and now i'm curious and i did not say it will, only it could and that's my guess !!!

what @oren wrote somewhere would be the opposite stance which is very well possible as well while i believe that bottom melt due to the heat in the system and wind and wave action due to the bad consistency of the ice will prevail this year but again @oren's theory makes sense as well and it will be interesting to see how things go.

[cesium62]

The weather conditions still warm enough to continue melting futhermore in September. (The pics below will update). There's not so much open water north of 80 latitude. The ice should continue to retreat on the edges and most likely the minimum extent should occur relatively late in September

I think the opposite.  All that slushy ice between the ESS and the pole will be easy to refreeze should the low temps come, as the water around it have not had a chance to mix. I think the balance of probability is more towards an early refeeze before Sept 15th, rather than a late one.

That's not the opposite.  Whether or not there is an early refreeze says nothing about whether or not there is a late minimum extent.

[Cid_Yama]

That's ok.  Imagine being a young person with children.  loss of sea ice has been equated with the end of everything they hope to continue.

It would be really hard to face the truth.

If I were a young person with a family, I would fight tooth and nail against that conclusion.

Have you ever seen By Dawn's Early Light where the electronics weapon specialist on the B-52, can't grasp that the base behind them, including his wife, has just been destroyed? 
 

[greatdying2]

The first two animations below ...

Terrific, thanks A-Team!

Interesting idea to use the geometric mean for filtering. Seems to work quite well.

[Rob Dekker]

The weather conditions still warm enough to continue melting futhermore in September. (The pics below will update). There's not so much open water north of 80 latitude. The ice should continue to retreat on the edges and most likely the minimum extent should occur relatively late in September

I think the opposite.  All that slushy ice between the ESS and the pole will be easy to refreeze should the low temps come, as the water around it have not had a chance to mix. I think the balance of probability is more towards an early refeeze before Sept 15th, rather than a late one.

That's not the opposite.  Whether or not there is an early refreeze says nothing about whether or not there is a late minimum extent.

I'm with oren on this one.
If an early refreeze occurs (and the slush on the ice edge suggests so) then the minimum will be early, not late.

[pauldry600]

At just below 4.63m now.

Another week to 10 days and wel be at min id say.

May flatten out for a while though.

[Tigertown]

Or a monster storm could come through at some point during the winter and wipe out half the ice, making a new minimum.  Maybe. The only thing that I am confident in knowing anymore, is that I don't know. Let's enjoy the tranquility while we have it, though.

[oren]

... (quote removed)

I'd rather you take this back. It's a science forum, not a personality forum.
I wish I could do what A-Team calls for, I lack the computing power, the technical know-how, the time, and mostly the discipline to acquire what I lack. But I still commend it. And I do my utmost best to avoid ad-hominem statements.

[Thomas Barlow]

Tropical air on train-tracks to the Arctic.

[cesium62]

The weather conditions still warm enough to continue melting futhermore in September. (The pics below will update). There's not so much open water north of 80 latitude. The ice should continue to retreat on the edges and most likely the minimum extent should occur relatively late in September

I think the opposite.  All that slushy ice between the ESS and the pole will be easy to refreeze should the low temps come, as the water around it have not had a chance to mix. I think the balance of probability is more towards an early refeeze before Sept 15th, rather than a late one.

That's not the opposite.  Whether or not there is an early refreeze says nothing about whether or not there is a late minimum extent.

I'm with oren on this one.
If an early refreeze occurs (and the slush on the ice edge suggests so) then the minimum will be early, not late.

It's nice that you think that if event A occurs, unrelated event B will also occur.  It's more interesting if you explain a causal link between the two events.

Oren suggests that slush refreezing north of 80-degrees, which will not change the extent (it will change concentration), will somehow prevent continued loss of extent further south.

I'm pretty sure the two of you can make the point you are trying to make more clearly and accurately.  The quick model that Pavel sketched allows for the ice north of 80 to refreeze early while still providing a late minimum extent.  Saying that you think the ice north of 80 will refreeze early is not "the opposite".

<Don't include so many nested comments; it clutters the forum, and makes replies difficult to follow. usually only the most recent comment in a sub-thread is enough to convey what you want.: JP>

[miki]

Tropical air on train-tracks to the Arctic.

And that storm western Hudson Bay, that yesterday wasn't there...

[A-Team]

Here is a hybrid of ESRL zonal ice location (aggregation by latitude, REB plot aice_h.nc) comparing 6 hour intervals of 22 Aug (purple line) and 01 Sep (blue)  with three weeks of UH AMSR2 ending the same date and 30 days of Hycom ending 09 Sep 17.

The ESRL products are still in design flux; they don't move from their Alaska focus to whole Arctic Ocean until Aug 22nd in their archive. In the ten day animation of ice thickness below, the palette itself is in day to day flux. Since ESRL's focus is D5 prediction, it's not likely that they will fix archival plots.

[Ice Shieldz]

Tropical air on train-tracks to the Arctic.

Well the path for tropical air to make it from the tropics to the arctic only begins with surface winds and then through the mid-upper level winds 500hpa+ in higher latitudes.  So yes it can get there but the path is more circuitous.  Below is GFS Sept. 6th forecast for Precipitable Water and 500hpa winds - note hurricane Irma in lower left.

We can see the moisture mostly from Harvey spanning across northern Atlantic.

[Archimid]

To me, if weather variations are completely ignored the Arctic seems primed for an early minimum. However, any warm air intrusions might  take this year to second place very fast.

[Sterks]

Really, that nobody stands up to you and your arrogance...

I'd rather you take this back. It's a science forum, not a personality forum.
I wish I could do what A-Team calls for, I lack the computing power, the technical know-how, the time, and mostly the discipline to acquire what I lack. But I still commend it. And I do my utmost best to avoid ad-hominem statements.

I think you are right (and the moderator). I have removed the post, I dislike ad-hominems as well.
I apologize for that.

[Rob Dekker]

I'm with oren on this one.
If an early refreeze occurs (and the slush on the ice edge suggests so) then the minimum will be early, not late.

It's nice that you think that if event A occurs, unrelated event B will also occur.  It's more interesting if you explain a causal link between the two events.

Fair enough, cesium.
The thought was that there are many pixels in the north-of-ESS area that are ice free, and these pixels will freeze over quickly once re-freezing starts. If that's early, there will be an early minimum, and if that starts late, the minimum will be late. But I admit that the causal link is not that strong.

[Sterks]

Perhaps some got caught up in day to day details and misleading forecasts ---myself about a week ago---.
It is safe to say, no matter if the minimum is today or by end of the month, that the amount of ice, and more importantly the spatial distribution of it, is similar to last year's end of season, with more and thicker ice in the Atlantic side, and sparser and thinner ice in the Pacific side, although the way the ice pack has reached to it is really different. There will be ---most probably--- more volume, extent and area at minimum, but the differences will not be that significant. If the Arctic experiences another winter as 2015/2016 or 2016/2017, it will have to dodge another bullet in 2018. If.
Really interested on how this winter is going to be.

[Pavel]

Another signficant point is the sea water temps. It looks that SSTs are warmer in ESS/Beaufort/Chukchi than previous years. The Hudson Bay is warm also. The warmer Sea of Okhotsk increase the probability of the new lowest annual maximum SIE

[oren]

It is safe to say, no matter if the minimum is today or by end of the month, that the amount of ice, and more importantly the spatial distribution of it, is similar to last year's end of season, with more and thicker ice in the Atlantic side, and sparser and thinner ice in the Pacific side, although the way the ice pack has reached to it is really different.

Interesting observation. Indeed the path to this distribution was quite different, but the end result similar. I still wonder myself what the end result will be, the thin slushy ice is in competition with the looming refreeze.

[Jim Hunt]

I have ex Prof. Judy to thank for the heads up on this open access paper!

"Extreme cyclone events in the Arctic: Wintertime variability and trends"

Using a combination of station data from Ny-Ålesund and ERA-I, we find a positive trend in extreme cyclone events over the Arctic North Atlantic in winter. This trend is closely associated with the significant winter warming of the region. Composite analyses of extreme storm events at Ny-Ålesund reveal the importance of associated anomalous heat and moisture transports from mid-latitudes for Arctic amplification, in accordance with previous studies. From the observational point of view, we find that Svalbard is located in a key region of climate change, and long-term observations from the Ny-Ålesund station are representative for the broader Arctic North Atlantic region.

Whilst we're on the topic of cyclones, see also:

http://GreatWhiteCon.info/2017/09/the-2017-arctic-sea-ice-metric-minima/

[A-Team]

The first animation shows how the location of open water has changed relative to its position as fixed on 13 Aug 17. To keep the colors scheme simple, non-zero sea ice concentrations are blacked out,

The second shows the snow thickness forecast from ESRL, REB.2017-09-01.nc, via Panoply in 40 time increments out to Sept 11th. Note the snow is typically ankle-deep and at most mid-calf; the forecast has it thinning. While this seems a decent product, the ESRL archive doesn't start up until mid-August which leaves us without data for spring and summer.

The thicknesses could be summed etc in Panoply but more informative is where thick snow lies over thin ice etc; 9 bin palettes for both gives 81 thickness combinations for each animation frame. I'll add the ice thickness counterpart animation in a bit.

The netCDF sea ice concentration file from UH AMSR2 hasn't worked out yet: it provides concentrations in a generic raster '2D' file rather than as the geolocated 'Geo2D' that Panoply needs,  although UH provides linked '1D' linked lat,lon files.

The role of snow changes with the seasons, from being a reflective surface in early summer inhibiting melt pond formation to being an insulating blanket in winter keeping frigid Arctic air away from the ice.

However 'snow' doesn't really describe the range of variability of physical properties for a snow pack. The Sami language of northern Scandinavia has some 300 words for snow types and Eskimo-Inuit over a hundred. There's been a lot of back-and-forth on this since F Boaz's writings in 1911; for an update, see:

https://en.wikipedia.org/wiki/Eskimo_words_for_snow
http://ontology.buffalo.edu/smith/varia/snow.html
https://www.princeton.edu/~browning/snow.html

Neither ESRL or anyone else is offering a rain product. That's unfortunate because we know for certain that it rained at the North Pole on 05 Aug 17, that lightning detection products for the Beaufort implies summer rains there too, and more warm vapor swept in this fall from the North Atlantic. The issue here is both latent heat release from the rain and associated humidity and major optical and thermal changes in the properties of the surviving snow.

[A-Team]

The first animation shows ESRL-PSD's forecasted snow thickness (orange) in six hour increments for the next ten days overlying forecasted sea ice thickness (blue). The second shows their ice thickness by itself.

The third shows contours for the thicker snow classes over the same ice thickness series. The fourth shows the various arithmetic means over this period, again with contours of thicker mean snow over mean ice thickness.. This time of year, there's little incoming solar energy; depending on its melt history, thicker snow might insulate the ice underneath from the cold.

These are produced from ESRL's netCDF Geo2D files, as visualized with Panoply, as post-processed in Gimp, via thousands of discarded intermediate files. It might barely be possible to keep up with ESRL's daily output depending on which of the 37 products in which repurposed combinations prove most informative here.

[VeliAlbertKallio]

Around Antarctica scattering of pulverized and transported sea ice helps its rapid re-growth. In this sense Arctic has become not too dissimilar. So, I expect easy re-joining and growth from the loose pieces of ice...

I'm with oren on this one.
If an early refreeze occurs (and the slush on the ice edge suggests so) then the minimum will be early, not late.

It's nice that you think that if event A occurs, unrelated event B will also occur.  It's more interesting if you explain a causal link between the two events.

Fair enough, cesium.
The thought was that there are many pixels in the north-of-ESS area that are ice free, and these pixels will freeze over quickly once re-freezing starts. If that's early, there will be an early minimum, and if that starts late, the minimum will be late. But I admit that the causal link is not that strong.

[FishOutofWater]

DMI 80ºN to the pole Temperature data graphed by Zach Labe show that the central Arctic core had a very cold summer. Yes, there are some issues with the DMI data but they actually reflect the weather this summer which was warm around the Arctic coast on the Pacific side and cold near the pole. It was a strong negative Arctic Oscillation summer.

The cool weather is why the melt season was slow.

[magnamentis]

It is safe to say, no matter if the minimum is today or by end of the month, that the amount of ice, and more importantly the spatial distribution of it, is similar to last year's end of season, with more and thicker ice in the Atlantic side, and sparser and thinner ice in the Pacific side, although the way the ice pack has reached to it is really different.

Interesting observation. Indeed the path to this distribution was quite different, but the end result similar. I still wonder myself what the end result will be, the thin slushy ice is in competition with the looming refreeze.

currently all speaks for your theory that the relative widely spread slushy ice could make for an early relative jump in extent and area gains, also similar to last year. IMO it depends how much of the reminder is 10, 20 and 30cm thick. in case of a lot of 10cm ice my theory still has a chance, if all remaining ice is 20+ cm thick i think your views will prevail, i give it another 5-10 days max to declare defeat LOL

[crandles]

The cool weather is why the melt season was slow.

Are you sure it isn't a case of:

Area was low so more area had air temp pegged to Sea freezing temp as to normally being more ice so air temp pegged to FYI/melt pond melting point?

Then the melt season might well have been slow due to low energy flux perhaps because of clouds?

If these are dominant effects, then I am unsure, but perhaps there doesn't seem much reason to think high temperature is a good measure of high heat flux into the ice. But I am no expert and maybe others think differently?

If the ice is all in small pieces, would that mean that heat flux into ice can occur efficiently without temperature being raised much?


To avoid doubt, I agree melt season appears to have been slow, I am just not sure about attributing this to air temperatures.

[Sterks]

i give it another 5-10 days to admit defeat LOL

The strong warmer winds from the Atlantic are more certain to materialize, so indeed I would not admit defeat in another 5 - 10 days, despite the ice pack to start closing in the most broken areas.
Hey, it's been cold up there!

This is how hycoms sees ice drift in five days, conditions lasting for two or three days.

[FishOutofWater]

I don't understand the dynamics. Maybe vorticity has increased around the margins of the Arctic ocean with the loss of sea ice and the storms spin clouds and low geopotential heights towards the pole. Maybe it's a matter hemispheric or global dynamics. Someone smarter than me at atmospheric physics is going to have to figure it out.
 
What I do know is that the Arctic Oscillation was negative almost all of the summer.

[Niall Dollard]

Looking at the latest JAXA figures, it does look like we are nearing the end now of the melt season so maybe a good time to do the post mortem. This may not be the right thread and mods feel free to move it. 

Dividing it into two, not necessarily independent, states : Air & Ocean.

Air

The extensive late spring snow conditions especially on the Eurasian side was important. There were significant departures from normal of air temp at 925hPa level during May. By the time the snow cover had vanished we were already at peak insolation. Yes air melt then started to pick up during late July/August period but the Arctic had bought some time. It was a heavy snow year also on Greenland. The melt season on the ice cap only got going also in late July and overall NSIDC reckon reduced melting and heavy early springtime snowfall may result in a net increase in Greenland’s ice mass this year for the first time this century.

Ocean

On the Pacific side, warmer sea temps helped an early melt out of the Chukchi Sea. On the Atlantic side I don't know. Is there any change in the northern advancement of Atlantic Water ? Is there any indication of AMOC slowdown ?
   

 

[A-Team]

The first animation shows "snow/ice surface temperatures" from today to Sept 13th, from today's ESRL. The DMI 80ºN number has limited value in describing conditions but there's a multi-year record that could equally be relevant or misleading for melt seasons in the 'new Arctic'.

The second shows ice thickness predicted for the same date range: the ice pack morphs to new shapes with little net change in volume.

[cesium62]

the relative widely spread slushy ice could make for an early relative jump in extent and area gains,

I still don't understand why refreeze of existing extent will lead to a jump in extent

[oren]

the relative widely spread slushy ice could make for an early relative jump in extent and area gains,

I still don't understand why refreeze of existing extent will lead to a jump in extent

Courtesy of Wipneus, here are the maps from 2 days ago for both NSIDC at 25 km resolution and Uni Hamburg at 3.125 km resolution. The area from ESS to the pole counts as low concentration but full extent in NSIDC, therefore an early refreeze of the "slush" will not result in an extent uptick. But on UH there are a lot of "open water" pixels mixed with the slush, should that region refreeze there will be a marked increase in AMSR2 extent.

[Andyph]

First post - Hi Everyone.

One thing that strikes me about this season is there appears to have been less export - certainly over the summer.
How much volume was exported this year?
How much is typical?

The other thought that strikes me is that watching the ice disappear in the Arctic is like watching the tide come in. You get a few big waves that make real progress then a lot smaller ones. Substitute years for waves. This year looked like it was a big breaker when it was out at sea but run out of puff before it go to the beach. Still, I would not stand there as you will get wet soon!

[Niall Dollard]

@A Team

Those ESRL ice thickness forecast plots look odd, especially in the vicinity of the "Beaufort Bite". If I am correct, in the thickness plot in reply #4022, dark blue indicates very thick ice, over 4m. There is a line of this right at the edge of the Beaufort Bite. I would expect a far more gradual increase in in thickness as you move away from the ice edge, like the attached Hycom image. 

[A-Team]

expect a far more gradual increase in in thickness  away from the ice edge, like the  Hycom image

Good question, Niall. ESRL does provide sections on validation and forecast skill assessment that might offer more details. JimH notes up-forum they are using v5.1 whereas Hycom is on v4.6 of ACFNS. A version change history might help but it won't yield a quantitative description of improvements applicable to this specific instance. The ESRL model is much more up to date overall.

Running the ESRL file through Panoply can change the output visual appearance in many ways, for example choice of palette as continuous or banded, whether the palette is fit to data or user-specified range, whether interpolation is on or off, how contouring is set, map projection choice and parameters, how image size is set and so forth.

'Interpolation off' is supposed to show actual grid cell resolution. It is easy to lose track of this in all the pretty flowing colors of a bicubic'ed 40-frame animation or smoothed MP4 video. (In Hycom ice motion, we've seen up-forum that averaging those long black vectors over 30 days reveals a surprisingly coarse grid base.)

Panoply does allow drilling into the netCDF numeric array to copy-paste the underlying grid cell data, here for this surprisingly persistent thumb of ice off Banks Island (bights are embayments, not so applicable). That would be the literal model output, graphics are downstream.

Earlier in the year, there was a magnetometer overflight in this region that can measure the ice thickness quite accurately (as the difference between sea salinity and lidar surface elevation). These are not flown at regular intervals as far as I know. In short, there's no current observational data against which to test the models.

The animation below shows a couple of Panoply views of the thumb from REB.2017-09-03.nc and their official version from REB_plots.2017-09-03 fig.17 (which is done in some unknown software used at Noaa). I could not accurately re-project or re-scale these to match the Hycom as its map parameters aren't specified. Note Panoply can save as kmz which could provide a common meeting ground on Google Earth (though it is very buggy and has not been updated in years).

Overall, ESRL thickness nearly stalls out by 13 Sep 17 though it is perhaps less than on the 3rd (bottom animation, Panoply zonal ice thickness forecast aggregated by latitude, which is like a Hovmöller diagram, only animated). The yellow line is the 3rd, the last green is the 13th. The areas under the curve would be volume if these were equal-area pixels. As is, the 13th has 99.4% as many 'volume' pixels as the 3rd. That difference is no doubt dwarfed by model error.

https://www.esrl.noaa.gov/psd/forecasts/seaice/

[Deeenngee]

An updated descent to the min chart - 2017's path vs other years' minima.

[A-Team]

Nice presentation of trajectory, almost looks like a hurricane forecast! @zlabe posted a beautiful extent anomaly graphic today, saying no recovery this year relative to climatology...

[Thomas Barlow]

Getting close to the end of melting, it looks like the Arctic Ocean itself - apart from thick ice crushed up to land - is likely the worst state on record for volume.
Here shows Arctic Ocean comparison. CAA ice, and thick ice crushed against Greenland/CAA, taken out, to emphasize the state of the overall Arctic Ocean itself.

[Thomas Barlow]

Other considerations:

[Pavel]

The slushy ice in the CAB will refreeze and build up to 3m thickness until next June if it will stay in CAB. If The Fram export will be slow, the CAB will have quite enough volume

[magnamentis]

The slushy ice in the CAB will refreeze and build up to 3m thickness until next June if it will stay in CAB. If The Fram export will be slow, the CAB will have quite enough volume

that's a rarely mentioned corelation, thanks for the reminder that fram export is not a give thing

[Daniel B.]

An updated descent to the min chart - 2017's path vs other years' minima.

Currently, 2017 is tracking quite close to 2008.  Looking like 6th lowest minimum.

[A-Team]

This ESRL site is putting out so many excellent new graphics daily that it is a challenge to assimilate them on the forum. The question is, can we add value re-working raw netCDF file combinations in Panoply, or has ESRL already done everything worthwhile? (Answer: not yet but they are improving the site almost every day.)

NOAA-ESRL-PSD has an institutional focus on the Chukchi and Beaufort, presumably because of their adjacency to the US. While the PI's real interests are likely in climate change, the project funding rationale presumably revolves around maritime safety, drilling platforms, coastal erosion, and futuristic naval conflicts.

Indeed, model quality greatly improves on the navy's site (Hycom inset, first animation vs ESRL's version of the Beaufort thumb; the former which the oft-maligned NavGem and has not been updated since 2016. The eight comparison graphics with GFS models imply a very big improvement there as well.

Today's 5-day net melt forecast (top + bottom + sides) remains melt everywhere, no net refreezing growth anywhere in the Arctic Ocean. In the 2nd animation, keep an eye on the palette colors used for freezing -- there is no use of them yet in the 20 forecast time frames. That doesn't mean ice isn't thickening in places due to other considerations (upper left corner), nor does it mean the 5-day direction of change is not positive in places (3rd animation).

The following little database provides the current content of ESRL's elegant web page graphics, the files found in their REBplot archives. We've only explored 7 of the 31 offerings so far (x column);
a 2x2 entry indicates 4 distinct but related displays per frame. Note 8 of these are comparisons of RASM-ESRL to lower quality GFS models. Twenty png's means an animation showing 4 frames per day x five days; elsewhere ESRL is going out to day 10 so 40 frames.

https://www.esrl.noaa.gov/psd/forecasts/seaice/ front page graphics navigator

ftp://ftp1.esrl.noaa.gov/RASM-ESRL/ModelOutput download archive

1   figAR2   AlaskaRegion2   20 png   2x2      RASM-ESRL vs GFS 2m temp and surface pressure 2   figAR4   AlaskaRegion4   20 png   2x2   -   RASM-ESRL vs GFS 850 hPa temp and precip 3   figAR5   AlaskaRegion5   20 png   1x1   x   ice and snow thickness 4   figAR12   AlaskaRegion12   20 png   2x2      RASM-ESRL vs GFS surface temp and LWP 5   fig0 fig6.gif   Arctic 0   1 png   1x1   x   RASM-ESRL ice edge evolution 6   fig0 fig7.gif   Arctic 1   1 png   1x1      snow depth 5 day change 7   fig0 fig8.gif   Arctic 2   1 png   1x1   x   ice thickness 5 day change 8   fig0 fig17.gif   Arctic 3   1 png   1x1   x   initial ice thickness 9   fig0 fig18.gif   Arctic 4   1 png   1x1   x   ice thickness day 5 10   fig1    Arctic1   20 png   2x2      ice area and snow depth RASM-ESRL vs GFS 11   fig2    Arctic2   20 png   2x2      RASM-ESRL vs GFS 2m temp and surface pressure 12   fig3    Arctic3   20 png   2x2   x   ice thickness thermo + dynamics + snow melt + ice melt 13   fig4    Arctic4   20 png   2x2      RASM-ESRL vs GFS 850 hPa temp and precip 14   fig5    Arctic5   20 png   1x1      ice and snow thickness contoured 15   fig9    Arctic9   20 png   2x2      RASM-ESRL vs GFS 500-1000 hPa thickness and precip 16   fig10    Arctic10   20 png   2x2      bottom ice growth + lateral melt + snow melt + top ice melt 17   fig11    Arctic11   20 png   2x2      longwave - solar flux + shortwave + albedo 18   fig12    Arctic12   20 png   2x2      RASM-ESRL vs GFS surface temp and LWP 19   fig13    Arctic13   20 png   2x2      RASM-ESRL vs GFS surface pressure + 850 hPa height 20   fig14    Arctic14   20 png   2x2      RASM-ESRL energy flux + LWP + surface temp + IWP 21   fig15    Arctic15   20 png   2x2      melt pond fraction + SST + heat flux + wind speed 22   fig16    Arctic16   20 png   1x1   x   RASM-ESRL ice speed 23   fig19    Arctic19   20 png   1x1      RASM-ESRL500hPa height and wind vectors 24   fig20    Arctic20   20 png   1x1   *    310K potential vorticity and Surface theta PVU 25   Meteograms   Beaufort   3 png   1x1      meteogram + cross section + text 26   Meteograms   Tiksi   3 png   1x2      meteogram + cross section + text 27   Meteograms   Oliktok   3 png   1x3      meteogram + cross section + text 28   Meteograms   Eureka   3 png   1x4      meteogram + cross section + text 29   Meteograms   Barrow   3 png   1x5      meteogram + cross section + text 30   Meteograms   Alert   3 png   1x6      meteogram + cross section + text 31   Meteograms   BeringStraits   3 png   1x7      meteogram + cross section + text

[Sterks]

no net refreezing growth anywhere

Yes, which is starting to be harder to reconcile with observational products, areas shooting up by some 100k's in a few days and things like that.
The net melt map of ESRL shows positive melt indeed, but of really small magnitude (order of milimeters/day rather than cm in some places) in growingly extense area of the Arctic. It also shows an unfortunate selection of minimum melt / minimum freeze color: white for both. Therefore we cannot know if indeed there is melt everywhere or, finally, refreeze is forecasted somewherein the white areas. I cannot process the nc myself, no way to know.

[DavidR]

NOAA-ESRL temperatures for August have been released and the rankings for various areas are shown.
https://www.esrl.noaa.gov/psd/cgi-bin/data/timeseries/timeseries1.pl

These figures don't indicate a particularly cold month however they are averages for the entire month and hence the second half of the month may have rated much lower than this in the Arctic

The comparatively low ranking for Air temperatures 80+ may be the best indicator for the small decline in extent in the second half as most of the remaining ice is in that area. 

SST    80+    Rank 5
Air    80+   Rank 18
SST    67+   Rank 1
Air   67+   Rank 5
SST   Global   Rank 2
Air    Global   Rank 2

[jdallen]

no net refreezing growth anywhere

Yes, which is starting to be harder to reconcile with observational products, areas shooting up by some 100k's in a few days and things like that.
<snippage>

... which I think in fact may be related to limitations of those products trying to measure what are threshold conditions in the Arctic.

They were designed to analyze the ice based on a multi-KM grid, and I submit have serious challenges acurately representing the atomized slush we see in the region currently.

I also wonder if they are able to deal accurately with increased snowfall, which is also a new feature of seasonal behavior.  As someone else said in the forums, the Arctic is no longer (or at not completely) a desert.

I think we've already seen a few years of this, with increasing destruction of ice quality starting in 2013 - which was a watershed for quality even as it was a rebound.

We have a fraction of 2012's MYI, which has been distinctly decreasing in thickness as well as area.  We are now moved state in the system behavior where the total effective ice coverage will match very closely the total annual flux in heat entering the region.  It's only by grace of the weather we haven't passed our 1 million KM2 "Ice free" state.

The feedbacks which are preventing that are going, some year, to be overwhelmed, and that sometime is getting very close.

I think Rob Dekker is on to something watching albedo and snowfall.  Unlike the pre-2007 years, I don't think the Arctic has the buffer any more which can soak up excess heat.  We've been steadily losing ground in the refreeze.  Warmer winters are going to end up what kills the Arctic, not summer heat, I think.

[Pavel]

SST    67+   Rank 1

It's hard to expect nothing else but the very mild winter in Arctic. October-December should be exciting

[F.Tnioli]

... Warmer winters are going to end up what kills the Arctic, not summer heat, I think.

Both. Summer heat does not have to be direct insolation (though chances are, that would also contribute much), - summer heat can and certainly will be increasingly large amounts of heat transported into the Arctic, both air and water currents. I simply fail to see how that (further) increase can not happen, long-term. It's already going on, and is one major reason we have such mild winters, no?

[A-Team]

The overly complex bottom animation experiments with one of many possible options for 'adding value' via Panoply to the ESRL set of daily 5 day forecasts. It forecasts ice movement and air velocity 10 days ahead with a faint overlay of ice thickness.

Both ESRL and Panoply have widespread ambiguity issues with the same white 255 being used in too many places in palette tiles and as open water background. Incidentally, they changed that around quite a bit over night, renaming all the files more informatively, eliminating some and adding others.

ESRL is a very complex project but is probably the best thing going in terms of both forecasts and reanalysis of the coming refreeze season. It has a whole raft of data that we have not gotten into such as melt ponds, solar energy fluxes, upwelling longwave radiation, albedo, water and ice surface temperatures and so forth.

Panoply, the open-source all-platform software for .nc files such as ESRL, is as easy to use as a web browser so hopefully more people will dip into it. The palette controllers are very similar to Worldview, giving rather differently appearances to maps according to how it squeezes and truncates.

ESRL has quite a different, more nuanced take on current ice thicknesses than Piomas and as usual there is insufficient observational evidence to distinguish between them. Although it's fair to say that Cryosat differs wildly from Piomas in percentage thickness (2nd image below, cross-post of Michael's graphic), we don't yet have an experimental comparison for ESRL as it is too new.

ESRL archives three file bundles every day. The previous post looked at REB_plots which are the products such as 5 day forecasts used on their front web page.

The second bundle, eg REB.2017-09-05.nc, contains 7 plotable Geo2D files and animations that go out 10 days at 6 hour intervals for 40 frames. These are not offered on the ESRL web page. The map resolution or number of grid cells in the rectangular data array i x j appears to be 384 x 432 = 16588.

Here are the geolocated Geo2D files that can be animated. Note the two sets of velocity vectors would normally be combined into magnitude color and direction arrows:

aice_h   ice area  (aggregate)
hi_h      ice thickness
hs_h     snow thickness
Tair_h   air temperature
Tsfc_h   snow/ice surface temperature
uvel_h   ice velocity (x)
vvel_h    ice velocity (y)
uatm_h   atm velocity (x)
vatm_h   atm velocity (y)