Home brew AMSR2 extent & area calculation

[epiphyte]

Looks to me as though everything in that view is less than 3km distant, and it is less than 50% open water, so my guess would be that from the POV of AMSR2 (or any other satellite bar Sentinel-1) it's all just ice...

[Wipneus]

Update 20160831.

Extent: -100.3 (-204k vs 2015, -1059k vs 2014, -1071k vs 2013, +609k vs 2012)
Area: -42.8 (-509k vs 2015, -1153k vs 2014, -1343k vs 2013, +372k vs 2012)
 
You will find the updated graphs in the top post

Regional extent declines in the CAB (-46k), Laptev (-23k) and Chukchi (-17k).

Regional area was down -25k in the CAB.

A delta map of the section once known as the Barents Bite. Plenty of little polynyas visible. Some are now at the edge of the pole hole. Most extent calculations will assume 100% extent within the pole hole, likely an overestimate at this situation.

[Wipneus]

Animation of the Canadian Archipelago, ice between the northern islands is entering the main channel of the NW passage.

[Pmt111500]

Those colors on the ice graphs may irritate some eyes so I made a recoloring of the latest one of Wipneus. It's not too accurate (fe.8% of ice is shown as no ice) but could work for color blind persons... sorry for the solid light red over the unobservable pole.

(modded soon after) Added another Wipneus chart from 2016.08.30. treated the same way

[Wipneus]

It is been a while that I posted the melting extent graphs (data from ADS/Jaxa thickness melting map). The late high melting extent ratio, exceeding 2012 is remarkable.

(in short melting extent ratio is the fraction of grid cells with ice where the ADS/Jaxa algorithm has detected melting.)

A click will get you a bigger picture.

Just to be clear - this is "top" melting, not the "bottom" melting from warm water that is often mentioned?

I guess it is measuring water over ice, and is calibrated for melt ponds during the melting season. I speculate that ice in a slushy state may show a lot of the bottom as well (9/10 of ice is below sea surface) and is reported as melt. That would explain the 15% during the freezing season as well.

[A-Team]

Wip:Plenty of little polynyas visible. Some are now at the edge of the pole hole. Most extent calculations will assume 100% extent within the pole hole, likely an overestimate at this situation.

Agree. The animation shows areas of sea ice concentration entering and leaving the pole hole, especially during the last 10 days of August (the whole month is shown below). It is feasible to pull back subsequent days into the pole hole but there has not been enough days of consistent CAA motion to fill more than a sixth (not shown).

Consistencies in blue feature shapes for several days rules out atmospheric and some other types of artifacts that microwave is prone to. It follows that the pole hole cannot have 100% extent throughout this period.

The pole hole arises from a combination of polar orbit for coverage and need for sun synchroneity in climate science (https://en.wikipedia.org/wiki/Polar_orbit) and swath width of the instrumentation (which is wide enough only for low resolution imaging at altitudes high enough to have low drag).

The north pole is mythic but is not the center of the Arctic Ocean nor the cold pole nor the center of the main ice pack. It has a special significance for maximal Coriolis effect but that is driven by the sine of latitude and so, expanding that in a power series whose next term is 3rd order, is very similar nearby.

[slow wing]

 

How big is that circle with the polar bears in it?

[Wipneus]

How big is that circle with the polar bears in it?

About 175 km diameter for AMSR2. BTW it is not really a circle but a polygon, bounded by 14 or 15 (approximately) straight lines (the satellite's swaths). You can see the shape change in A's animation.

[Bill Fothergill]

About 175 km diameter for AMSR2. BTW it is not really a circle but a polygon, bounded by 14 or 15 (approximately) straight lines (the satellite's swaths). You can see the shape change in A's animation.

Using this 175km "diameter" figure, that "pole hole" would effectively cover everything poleward of just above the 89th parallel (~ 89.2N)?

Taking it as approximately circular, this represents an area of about 24 thousand sq kms - or about 10% that of the UK. As such, whilst I think this represents yet another weakness as regards the future of Arctic sea ice, I don't think it represents any significant level of "over reporting" at present.

However, as A-Team has pointed out, Coriolis effects at high latitude could becoming interesting as the mechanical structure of the ice pack gradually degrades (or perhaps not so gradually?).

[A-Team]

The image below shows the orbital paths of some of our favorite satellites (offered as layer options in WorldView useful when swath timing of pie wedges is important). The AMRSR2 3.1 of 1 Sep 16 was rescaled to match out to 85º N.

It is not so surprising that the orbits are all about the same since several follow each other in the A-Train orbit and are constrained in any case by the physics of drag and synchronicity but the nadir track (spot on earth directly underneath the satellite) does seem rather far from the pole at 81.85º, more than instrumental swath width can make up for in the case of AMSR2.

On the WGS84 ellipsoid we are using, both km per degree of latitude and areas of the pole hole are a little tricky. The former is shown below; the latter was discussed in yesterday's post on elliptic integrals. At 111.65 km per deg of near-pole latitude, 175 km pencils out to 1.567º which puts the perimeter of the AMSR2 pole hole at 88.433ºN.

The pole hole comprises only a small fraction of the Arctic Ocean; indeed the North Pole itself -- even though it has polynyas close by today -- is a poor proxy for the state of Arctic sea ice (except in the mind of the public). 

Overall that state cannot be described by any single number; the reason we do so is that the real numbers form an ordered field, facilitating daily and yearly comparisons. It would take quite a stack of GIS layers (maps of ice distribution and thickness, salinity, temperature, etc etc) to describe the state adequately, with trends in these layers more informative but also more difficult to describe.

Edit: see #2762 and #2770 for corrections. Diameter should have been used in this post (as Bill F did originally in #2770) instead of radius.

[Wipneus]

Update 20160901.

Extent: -39.5 (-208k vs 2015, -1097k vs 2014, -1103k vs 2013, +575k vs 2012)
Area: -17.0 (-493k vs 2015, -1161k vs 2014, -1345k vs 2013, +330k vs 2012)
 
You will find the updated graphs in the top post

Regional extent declined in the CAB: -35k, elsewhere very little changes.

Regional area was down -20k in the CAB and -16k in the Laptev region. Small increases elsewhere.

Attached is a delta image with the Wrangel Arm. The upper (in the image) part of the arm is increasing in extent and increasing in concentration, not much flash melting there.

[Artful Dodger]

The reference from Wipneus is buried somewhere up the thread, but this is a different product and its calibrated against visual observation of melt ponds. Its also a very noisy calibration, even by  microwave standards.

Oh, okay. Thanks, Richard.

Cheers, mate!
Lodger 

[Bill Fothergill]

...At 111.65 km per deg of near-pole latitude, 175 km pencils out to 1.567º which puts the perimeter of the AMSR2 pole hole at 88.433ºN...

The above calculation clearly uses the 175 km figure as the pole hole radius. However, Wipneus had characterised this earlier as the diameter, and, in pegging the perimeter at approximately 89.2N, I used 175 kms as the diameter. Obviously if it is the radius which is 175 kms, this would effectively move the perimeter out to around 88.4N

Can someone please confirm which it is, radius or diameter?

OFF TOPIC ALERT: One of the diagrams provided by A-Team shows that, although it varies slightly from the equator to the poles due to the Earth's oblateness, the "size" of a degree of latitude remains pretty damn close to 111 kms. Therefore, the equator to pole distance works out at around 9,990 kms.

In the wake of the French Revolution, the metre was initially defined in terms of this pole-to-equator distance. The value selected for the metre was one ten-millionth of this distance. If memory serves (still possible, but increasingly unlikely these days) the meridian - or half meridian - to be used was that which passed through Paris. Consequently, when thinking about the circumference of the Earth, I always just use 40,000 kms as a reasonable approximation.

[Pmt111500]

Just having 'fun' with the image software.

[Wipneus]

Small decline in extent and a century increase in area. First candidate date of minimum is first of September.

Update 20160902.

Extent: -23.8 (-240k vs 2015, -1125k vs 2014, -1092k vs 2013, +603k vs 2012)
Area: +104.0 (-418k vs 2015, -1034k vs 2014, -1223k vs 2013, +483k vs 2012)
 
You will find the updated graphs in the top post

Regional extent changed most in the CAB: -21k.

Regional area change is also dominated by the CAB: +112k.

Attached the delta map of the Arctic Basin.

As told on the 2016 extent and area thread, I have to take a few days break from these daily reports. Also automatic updates of data and graphics (by my Raspberry Pi server) will stop, and of course any PIOMAS updates could be delayed.

[Wipneus]

Lastly an update of the animation of Jaxa sea ice concentration. Note that it ends on 1st September.

[TerryM]

Wip
Have an uneventful move & a great life in the new digs!
Terry

[Rob Dekker]

As told on the 2016 extent and area thread, I have to take a few days break from these daily reports. Also automatic updates of data and graphics (by my Raspberry Pi server) will stop, and of course any PIOMAS updates could be delayed.

Wipneus, your updates will be missed.
Thank you for your amazing work here.
Have a good move !

[Pmt111500]

Wipneus, your updates will be missed.
Thank you for your amazing work here.
Have a good move !

Can't but to join to the wishes. Have a good move, Wipneus!

[Lord M Vader]

From Wipneus ice concentration map today, it seems like Wrangels arm is broken now... The big question now is how much additional "melting" the broken arm will suffer from, especially if a cyclone an winds will push it to Berings strait.

[A-Team]

The AMSR2 pole hole: can someone please confirm which it is, radius or diameter?

Wipneus had that right (as always!). The image below lays a 1º latitudinal grid over the standard polar stereographic map (which has 70º as the only place where area is not distorted) of 02 Sep 16 cropped to our forum maximum of 700x700 pixels, which just barely picks up extreme northern Greenland and a bit of Ellesmere Island. The pole hole lies above the 89º latitude (as you calculated).

On the right margin of the animation below, a feature on the right center margin (circled) has moved 7.8 pixels between Sept 1st and 2nd. This is 0.139 of a degree of latitude so about 15.5 km in 24 hours.

While the differences seem inconsequential, the problem with the spherical earth is that all GPS devices reference the oblate ellipsoid WGS84. So it would matter in positioning a floating buoy from its reporting coordinates onto a berg on a 10 m resolution S1A image. WGS84 in turn is not a equipotential gravitational surface which matters for satellite correction. Landsat even has to correct for earth-sun distance at the time the scene is imaged.

And so it goes, the greater the need for precision (say thinning of the Greenland ice sheet) and the more instrumentation is deployed that can provide it, the more complex the earth model is needed.

[Bill Fothergill]

The AMSR2 pole hole: can someone please confirm which it is, radius or diameter?

Wipneus had that right (as always!).

Thank you for confirming that the 175km figure is the diameter - and for the excellent overlay montage!

[A-Team]

Thank you Bill for catching this (and a Blog typo on the forum the same day!).
 
The top animation below shows the action south of the pole for the Atlantic sector over last 33 days (01 August 1st to today Sept 2nd). Translation of two features is tracked, one inside a red dotted circle and the other with magenta.

The diameter of these circles is ~0.5º in latitude which is convenient in various ways but does not capture rotation. Better if the gimp circle tool had a cross-hairs and center display options. It had to be cropped from the AMSR2 3.1k to fit forum constraints. Dates are provided as frame names if you download and view.

Despite atmospheric artifacts sweeping across some frame subsets, quite a few features are persistent enough to be reliably followed for 4-5 weeks. The red circled feature moves ~279 km over the 33 days of which some 263 km is due south. At this rate, this feature (which is already a partial polynya) won't hit the melt zone for another 41 days.

A second version is not cropped but needs a click to start. It shows ice entering the Fram being  obliterated this year just like ice along the rest of the Barents front, surface waters being very warm. It is not loading properly and interfering with the first so it's being pushed to the next comment.

[A-Team]

Here is the wider version. It had to be rescaled down from its original width of 942 pixels.. That compression ruins the color key. Forum software differs from day to day on what it allows.

[oren]

The interesting thing is that while the Barents front looks roughly the same during the animation timeframe, the area to the north keeps losing concentration and polynyas are opening up. A naive observer might assume that the melting area is near the pole.
Early in the season it has already been noted that the Atlantic sector was the killing ground, but the effect took a long time to materialize visually since the inner CAB had lots of ice to supply when the season started. Had the ice been less mobile and fractured this year, it might have survived in much greater numbers.

[A-Team]

melting area near the pole

Right, 'partial polynyas' of lesser sea ice concentration have been moving in and through the pole over the last month. Worldview has a fairly decent view right to the pole for September 3rd, below.

To accurately mark up WV, first use the snapshot tool to draw a rectangle whose two labelled corners are useful. Here the lower left corner was set to 89º latitude and the upper right set merely to label the 90º meridian. Take a whole-window screenshot as the first gimp or imageJ layer.

The 216 pixels from pole to LL corner then allowed 1º latitude circles to be drawn as integral multiples out to 84º.  Next quit the snapshot tool and take a second whole-window screenshot for the base layer. Delete all but the snapshot rectangle and collapse the two layers.

We know where the pole hole of AMSR2 resides just outside the 89º circle (88.433ºN from Bill's post, 214.6 pixels). From features on its boundary, we can recognize it on higher WV magnifications and mark the pole hole on those. However there's very little quality area on enlargements because of a thin cloud cover. Tomorrow's another day -- the scale template remains valid.

The AMSR2 is not yet available for Sept 3rd; the second image is from the 2nd. The gray box shows the approximate boundaries of the WV image with some context; the 3rd image crops and rescales to match the WV. There can be quite a bit of ice motion over a single day; AMSR2 is imperfect as well so unsurprisingly the terra visible and AMSR2 concentrations are not a great match..

[Adam Ash]

AMSR2 is broken..Rite?  Another key sensor down just when it's needed most?

[Jim Hunt]

AMSR2 is broken..Rite?

I don't think so. Just one of the the periodic glitches somewhere in Bremen's pipeline:

http://GreatWhiteCon.info/2016/09/the-2016-arctic-sea-ice-metric-minima/#comment-215522

Hamburg's AMSR2 looks OK for example:

[Tigertown]

Polarview was also out earlier, but is working now.

[Shared Humanity]

Meanwhile, Big Block continues to hang out in the Beaufort.

[etienne]

Hello,

I just updated my average and trend graph.



The slope of the trend is now steeper than in june (used to be -170 km^2 per day, is now -184km^2 per day).

If we keep this slope, being on the trend would be enough in 2018 to break the record of 2012.

Best regards,

Etienne

[seaice.de]

Have a good move Wipneus!

In the meantime here is an update from UH. Number of extent are only slightly (0.05) different from yours.

[Bill Fothergill]

If we keep this slope, being on the trend would be enough in 2018 to break the record of 2012.
Etienne

It can be interesting to watch slope developments. Back in the early months of 2010, I vaguely recall doing something similar with the 1979-2009 NSIDC September averages.

Using the then available subset, the trend line did not hit the 2007 jaw dropper until 2024. As one might expect, statistical outliers (~ 2 Stnd Dev) such as 1996, 2007 and, more recently, 2012, can wreak havoc with seemingly relatively stable trends (or the first derivatives thereof).

When one expands the data set to cover 1979-2015, the trend line (virtually) reaches the 2007 level by 2020. Extrapolating further, that 1979-2015 trend line reaches 2012 proportions by 2028.

However, using a rolling annual average smooths out more of the wild fluctuations. Consequently, as your projection shows, any extrapolation of the trend line would not take as long to reach the record low set for the period straddling 2012 and 2013. (If anyone is uncertain about the 2013 bit, that year had lower extent than 2012 over the first few months.)

++++++

Can I also echo the sentiments of so may others in wishing Wip all the best, and thanks ever so much for all your informed commentary!

[seaicesailor]

One thing for sure, one feels pretty lost of what is going on and where in the Arctic without Wipneus updates.

[Jim Hunt]

Can I also echo the sentiments of so may others in wishing Wip all the best, and thanks ever so much for all your informed commentary!

Hear hear!

Perhaps I could also take this opportunity to congratulate Wipneus on having the results of all his hard work replicated by the University of Hamburg!

[Adam Ash]

If we keep this slope, being on the trend would be enough in 2018 to break the record of 2012.
Etienne

It can be interesting to watch slope developments....

I would imagine that there is some sense in seeking an idea of the future from trends of past data provided the processes you are modelling stay roughly the same; Solar insolation, bottom melt, export via Fram etc.

But this year we have seen another mechanism come into play which is a pretty Black Swan event - viz. the impact on increasingly fine granularity of the pack ice on the rate of its export through the CAA.  Suddenly your model bucket has a hole in the side of it,  through which the dipoles are shoving significant volumes of ice which under previous models would have accumulated there as MYI.  The CAA refuge also stablisied the wind-driven circulation of the whole CAB cap which up until now inhibited the relatively rapid movement of ice to the Barents, Atlantic and Beaufort death zones we are seeing today.

So, respectfully, (because you are far better at this than I) until your slope estimates and other history-based models reflect this new mechanism, I don't think your slope-watching will have any better than a random chance of reflecting the reality of what's going on out there with any useful fidelity.

[Iceismylife]

...

So, respectfully, (because you are far better at this than I) until your slope estimates and other history-based models reflect this new mechanism, I don't think your slope-watching will have any better than a random chance of reflecting the reality of what's going on out there with any useful fidelity.

I don't think there is any way to get prediction with any kind of fidelity for what the arctic is going to do in the short term. What a difference a GAC or three makes.

[etienne]

...

So, respectfully, (because you are far better at this than I) until your slope estimates and other history-based models reflect this new mechanism, I don't think your slope-watching will have any better than a random chance of reflecting the reality of what's going on out there with any useful fidelity.

I don't think there is any way to get prediction with any kind of fidelity for what the arctic is going to do in the short term. What a difference a GAC or three makes.

My idea here is not to predict the future, but to say that a Sea Ice Extend that was a huge event in 2012 would be just normal in 2018. This scares me a little bit. If you have a linear trend on an always smaller surface, you remove each year a higher percentage of the remaining surface, and I doubt that the trend is linear.

Adam Ash also talked about a hole in the bucket, but there is also an ice input comming from Greenland, don't know how much this "protects" the sea ice, maybe this is one of the reasons why the trend seems linear. This would not be a good news.

Best regards,

Etienne

[Bill Fothergill]

I would imagine that there is some sense in seeking an idea of the future from trends of past data provided the processes you are modelling stay roughly the same; Solar insolation, bottom melt, export via Fram etc...

... until your slope estimates and other history-based models reflect this new mechanism, I don't think your slope-watching will have any better than a random chance of reflecting the reality of what's going on out there with any useful fidelity.

Adam, I'm afraid that you've misinterpreted what Etienne and I were both saying. My apologies for not being clearer in what I wrote! Let's see if I can do better this time...

Let me begin by categorically stating the following. Not for one moment would I even dream of suggesting that this form of "chartism" acts in any way as a meaningful predictor for Arctic sea ice developments.

My intention was quite the opposite: namely, to demonstrate that some threshold predicated upon historic trends, can (and has) easily happen far sooner. That is what I was alluding to when I wrote that "... As one might expect, statistical outliers (~ 2 Stnd Dev) such as 1996, 2007 and, more recently, 2012, can wreak havoc with seemingly relatively stable trends (or the first derivatives thereof) ..."

For example, a hindcast based on the 1979-2009 subset of the NSIDC September averages shows very little skill at "predicting" the 2010 - 2016 results. The only years in which the difference between "predicted" and actual values was less than a third of a million sq kilometres were 2013 (165k) and 2014 (182k).

[Adam Ash]

Thanks for the clarifications gentlemen!  I wasn't being in any way critical of your efforts, rather I (as an utter lay-person among you experts) was trying to understand where you were all heading with your discussions.

[etienne]

On the topic "Stupid Questions about DMI Data" http://forum.arctic-sea-ice.net/index.php/topic,1616.0.html
I estimated that the Greenland Ice melt trend and the Artic Sea Ice melt trend are similar in quantities (around 300 Gt of ice lost per year). If you consider that there is not only Greenland where glacier loose ice everywhere, maybe we have an explonation why the Artic Sea Ice Extend trend seems linear in a non linear climate change.

Well, the Greenland Ice melt trend doesn't look linear : http://polarportal.dk/en/groenlands-indlandsis/nbsp/total-masseaendring/.

Best regards,

Etienne

[FishOutofWater]

The large Greenland melt events of 2010 and 2012 produced a huge amount of fresh surface water that would have reduced convection in the Labrador sea in 2013 and 2014. We should not expect stability in melting patterns because one big melt year may reduce the flow of warm water northwards for several years that follow.

[icy voyeur]

Wipneas is away because he's moving and the ice stops melting.
Coincidence?  Not so sure.  But next year he might have to take one for the team and move again.

[seaice.de]

AMSR2 sea ice animation in German TV weather forecast
https://twitter.com/CENunihh/status/776357159617658880
https://uhh.de/cen-arctic-sea-ice-16

[Wipneus]

Here is an animation of the sea ice thickness and melting maps from ADS-Jaxa comparing this week with 2012. One thing that can be noticed is the rapid refreeze of the ESS in 2012 with no sign of it currently.

The near total lack of thick ice compared with other years striking, total volume is tracking well below 2012 (and 2013 -2015).

The animation needs a click to start.

[marcel_g]

The near total lack of thick ice compared with other years striking, total volume is tracking well below 2012 (and 2013 -2015).

I know I'm not adding to the discussion here, but holy crap, that is striking.

[A-Team]

holy crap, that is striking

Just have to wonder how different things would look the season before game-over. The animation below is identical to Wip's above but adds a UHH AMSR2 close-up for the same dates of 2016 for the Barents Sea front.

A finger of ice is coming in above the St Anna Trough as it often does, presumably because the Atlantic Water has already taken its dive, leaving the surface water slightly cooler. However sea surface anomalies reaching 10.1ºC (sea water temperatures 13.1ºC or 55.8ºF) in the Barents suggest this ice has no prospects for extending.

As a technical note for making nullschool displays, for example time series or multiple temperature probes or multiple settings, set up a browser tab to the desired view, then clone that tab (4x7=28 times  is feasible on a 27" iMac screen), then take whole-screen snapshots as layers (they'll all be the same size, ie aligned already) and process accordingly.

[psymmo7]

Hi
Any chance of an update of the last graphic (dynamic comparison)? As predicted refreeze conditions in the ESS seem to be very unfavorable compared to previous years. Any predictions about how long is this state of affairs likely to last?

[etienne]

Hello,

I updated my artic sea ice average graph.



I calculated that we are now at 10.33E6 km^2, on the lowest sea ice extent yearly average since January 1990, which probably means since a much longer time. For January 2013 (average 2012), I have 10.38E6; for October 2011 (averange 2011 melting season) 10.44E6; and for November 2007 (average 2007 melting season) 10.46E6.

Best regards,

[etienne]

Hello,

Just one more information on the graph above. We went below January 2013 values (2012 metling season) on October 11th 2016, which is around the time where Sea Ice Extent 2016 went below the Sea Ice Extent 2012 curve on https://ads.nipr.ac.jp/vishop/#/extent.

Well, the minimum yearly average is still heading down.

Best regards,

Etienne