Jakobshavn Isbræ / Ilulissat Isfjord / West Greenland

[Wipneus]

S2A_R111_V20160425T152932.22WEB.B04.tiff has been uploaded.  Clear in places, but some clouds over the calving front spoil it.

[Wipneus]

Lots of little melt ponds appear in the blue tinted area visible in the thumbnail above.

[A-Team]

40 minutes

Ouch.

melt ponds

Very convincing. Surprising this early. Or not, considering weather. Next thing to watch for: melt ponds draining down moulins. The ponds can be picked by hue or arithmetic layer interaction (divide shown) fairly easily, for later differencing.

[Espen]

Sentinel 1B will be online soon, then a 6 days pass will be available:

http://www.dmi.dk/nyheder/arkiv/nyheder-2016/april/forsinket-satellit-slutter-sig-til-sentinel-soester/

[Andreas T]

Together with the bare ice at Jakobshavn there is also snowfree land just south from there. what seems odd is that the land nearer the ice free sea is still white while the land nearer the inland ice isn't. Is this unusual?

[AbruptSLR]

Together with the bare ice at Jakobshavn there is also snowfree land just south from there. what seems odd is that the land nearer the ice free sea is still white while the land nearer the inland ice isn't. Is this unusual?

As a re-post from the Greenland 2016 melting thread: the first image shows the Nullschool Surface Temp & Wind Map focused on Greenland for April 26 2016; and the second shows the comparable forecast conditions for April 30 2016.  For this entire five-day period nullschool projects above freezing temperatures along large portions of Greenland's West Coast areas.

[Cate]

Together with the bare ice at Jakobshavn there is also snowfree land just south from there. what seems odd is that the land nearer the ice free sea is still white while the land nearer the inland ice isn't. Is this unusual?

I can give an example from my own experience, which may or may not answer your question. Here in Newfoundland at this time of year when we are moving from winter to spring, it is not unusual at all to have onshore winds dumping snow on coastal areas in spring squalls and storms while inland areas remain untouched by the snow. What we get is basically sea-effect snow. I'm not saying this is what is happening here as I know nothing about Greenland weather conditions and cannot tell the distances involved in this image.

[A-Team]

The 26 Apr 16 Landsat is quite clear. The fjord is half-cleared. Jakobshavn's calving front has advanced slightly, pushing previously calved pieces out ahead at a velocity of ~32 m/day. The shape of the calving front is normal, no asymmetric lobe developing on the south. The anomalous weather has not had any noticeable impact as yet.

[nukefix]

S-1a 27.4, in 10m pixel size & UTM

[A-Team]

S-1a 27.4, in 10m pixel size & UTM22

Nice enough image. The series below looks at another adaptive contrast method (integral image filtering) that maybe improves contrast in S1A images (3rd frame, with controller in upper right). This involves using local statistics around each pixel in a block of specified size to adjust its value if warranted by its value exceeding a specified number of standard deviations.

The original image had a good enough gaussian histogram but it was offset and did not make full use of the grayscale range. The second frame simply fixes that.

Does is it improve recognizability of features and thus help fix the current inability of Sentinel 1A to determine velocities in the final stretch? We would have to check with the group working with Sentinel pairs to find out.

[Wipneus]

A rather clear S2A_R025_V20160429T151111.22WEB.B04.tiff has uploaded.

[A-Team]

Thx! I just can't get over how fast these downloads from google drive are. Goes to show my connection is ok, it's the crummy servers at the other end that are usually limiting.

Those melt ponds from last time -- did they fully freeze back over?

We are building up a very nice portfolio of R025's. A lot could be done with these regionally clear ones (but I will be offline traveling most of this month).

I did not see anything out of the ordinary in this 29 Apr 16 image, Jakobshavn is moving right along rather predictably. One thing we are not tracking so well is vertical thinning. In nadir view, that cannot be seen but maybe inferred (after much hand-waving) as it is another form of reduced buttressing as far as the sides are concerned. That is, retreat of the calving front is followed shortly by a dramatic increase in ice flowing in from the north bank whereas lowering of the main channel generates a more subtle response.

That's a rather odd iceberg, with that distinctive little white knob. It is easily tracked back to posts from 19 Apr 16 in S1A, S2A and Landsat. We may be able to follow it way out to Disko Bay and along the coast.

[Wipneus]

Those melt ponds from last time -- did they fully freeze back over?

My impression is the opposite, there are at least as many and ate least as big in this image.

I may post some evidence later.

[Wipneus]

Some melt ponds disappear, some new ones, some grow. The majority just stays the same.
Two animations, first is the same detail as before a bit to the south. One melt pond at top-right goes and a new one on the center  left.  The second one is east of the Jakobshavn calving fronts.

[magnamentis]

Some melt ponds disappear, some new ones, some grow. The majority just stays the same.
Two animations, first is the same detail as before a bit to the south. One melt pond at top-right goes and a new one on the center  left.  The second one is east of the Jakobshavn calving fronts.

is it totally of the consider that the gone pond could have emptied into a crevasse, seems not that likely that while others are popping up and/or were growing, that so close by another was refreezing but i have no scientific evidence or skills to support this, just a kind of common sense that might well be wrong. just throwing in my 2cts.

[Wipneus]

The S2A_R068_V20160502T151916_22WEB has clouds rolling in, obscuring all calving fronts. Some of the upper glacier streams are visible though, so I could upload the band-4 image. Let me know.

[johnm33]

Not up to the standard of our usual contributors but heres a detail from http://www.polarview.aq/images/105_S1jpgfull/S1A_EW_GRDM_1SDH_20160503T101537_C402_N_1.final.jpg be warned it's the 59mb full res image. [often crashes my comp.] Big retreat evident let's hope for a cloud free shot.

[Tealight]

Not up to the standard of our usual contributors but heres a detail from http://www.polarview.aq/images/105_S1jpgfull/S1A_EW_GRDM_1SDH_20160503T101537_C402_N_1.final.jpg be warned it's the 59mb full res image. [often crashes my comp.] Big retreat evident let's hope for a cloud free shot.
[often crashes my comp.]

With Microsoft Edge my computer has no problems, but Google Chrome doesn't even attempt to display an image.

Good spotting btw. I only found some melt ponds on calved icebergs, which are currently in the middle of the channel. Would be a great place to take a bath and watch the calving event.

[nukefix]

Holy crap, that looks like a big calving, especially for the time of the year. I have the feeling we're going to see some records broken in calving front postion this summer..

[johnm33]

Doesn't look like a lots going on by the sill, but it is moving.
http://www.lookr.com/lookout/1310041325-Ilulissat

[nukefix]

Doesn't look like a lots going on by the sill, but it is moving.
http://www.lookr.com/lookout/1310041325-Ilulissat

Those bergs are stuck on a shallower part of the fjord and generally move very slowly, like meters a day. Tides must affect them too.

[A-Team]

could upload the band-4 image. Let me know.

Thx but looks too cloudy to bother with. It doesn't look to me like the calving front has changed position much, it has come forward if anything. Sometimes the best indicator of calving is displacement of bergs in the bay just in front of the calving front. If the recognizable bergs are moving in synchrony and no new ones are behind them, that indicates no major calving, just the advancing calving front pushing them forward.

[TenneyNaumer]

Sorry, I don't have access to high-res photos, but wanted to point to the two rather large depressions to the east of Jakobshavn (which can be seen in photo above) -- might this have been collapse of firn?

[TenneyNaumer]

Never mind.  The depressions are the remnants of melt lakes from last year.

[Wipneus]

Nice clear S2A_R111_V20160505T152915.22WEB.B04.tiff uploaded.

[Shared Humanity]

Looks like some serious melt going on.

[Wipneus]

R025, 2016-05-09, 100% clouds. No upload.

[Wipneus]

R068, 2016-05-12 is finally in, 100% clouds. No upload.

[johnm33]

" No upload." Shame it would be interesting to get a closer look, the jam on the cill halfway to the calving front has broken.

http://ocean.dmi.dk/arctic/images/MODIS/Disko/20160516s01a.ASAR.jpg

[A-Team]

The May 15th has also shown up.  While the calving front per se is covered by some peculiar clouds or fog, other parts of the image are clear enough and appear to show quite a bit of melt water in numerous locations including the main channel (4th image, on far right middle of locator map 2nd image). Mid-May seems early for all this liquid water but that would be hard to rigorously establish over multiple years.

I just looked at the B4 band so it would be better to see how it looks processed to three band color at 16-bit.

http://sentinel-s2-l1c.s3-website.eu-central-1.amazonaws.com/#tiles/22/W/EB/2016/5/15/0/

[Wipneus]

The May 15th has also shown up.  While the calving front per se is covered by some peculiar clouds or fog, other parts of the image are clear enough and appear to show quite a bit of melt water in numerous locations including the main channel. Seems early but that would be hard to rigorously establish over multiple years.

I just looked at the B4 band so it would be better to see how it looks processed to three band color at 16-bit.

http://sentinel-s2-l1c.s3-website.eu-central-1.amazonaws.com/#tiles/22/W/EB/2016/5/15/0/

Here is the band 234 thumbnail. Let me know what you would like uploaded, the full res color in 16bits is 461 MByte and includes of course the B04 band.

[A-Team]

Thanks, that would be great if you could just crop out this braided stream(?) area inland, approximate is fine. There was some talk about improving Cryosat surface elevations from line of points to line of swaths, example given for Petermann. This is another way to get at micro-topography, following water downhill. It is evidently lower to the north side at this site.

I will put together a comparison as to what is really gained in color with 16-bit vs 8-bit (have bands 2345 open as 8-bit) along the lines of what you did above with Landsat, pansharpened Landsat, and Sentinel 2A for Zachariae.

http://forum.arctic-sea-ice.net/index.php/topic,400.msg75256.html#msg75256

[Wipneus]

Does the forum software preserve 16 bits?

It seems not, I will upload to the usual place.

Done, filename is S2A_R111_V20160515T152912_22WEB_RGB_crop_1200x980+8100+1800.png

[A-Team]

Ok I see it on your sharing site ... http://tinyurl.com/jqo2lq7 ... but as 3x8-bit channels of png so this would get into a nuisance situation of having to upload three separate 16-bit tifs. Color is quite good now however. There is talk/confusion about 16-bit per channel png which would be 48-RGB but it does not seem widely supported, notably not on the forum.

The animation shows three contrast versions at 5 m resolution starting with the sharing site image. I like the 3rd frame option which inverts V in HSV without going there and re-displays as RGB (invert value command in gimp).

There is a lot of unambiguous meltwater all over this image. The larger lakes still have floating ice in the middle indicating the water is right at 0ºC there. It's been quite cloudy for quite a while so what brought on all this melt -- rain, warm air, or sun?

On Greenland's big melt day in August 2012, thin water clouds were blamed: https://www.nsf.gov/news/news_summ.jsp?cntn_id=127438

Clouds over the central Greenland Ice Sheet last July were "just right" for driving surface temperatures there above the melting point, according to a new study by scientists funded by the National Science Foundation (NSF) and at the National Oceanic and Atmospheric Administration (NOAA).

The study, published in the April 4 edition of the journal Nature, found that thin, low-lying clouds allowed the sun's energy to pass through and warm the surface of the ice, while at the same time trapping heat near the surface of the ice cap. This combination played a significant role in last summer's record-breaking melt.

"It's kind of like the story of Goldilocks," said Von P. Walden, a co-author of the paper and principal investigator for the NSF-funded Integrated Characterization of Energy, Clouds, Atmospheric State, and Precipitation at Summit (ICECAPS) project.

"If the sky had no clouds on July 11th, it would have been too clear and cold to melt the surface. But if the clouds were too thick, it also would have been too cloudy and cold. The thin, liquid-water clouds were just right for melting the surface snow," he said.

I once had a Goldilocks coffee mug. There were helpful lines for 'too hot", "just right" and "too cold". Somehow it knew.  Right now though I am at airbnb with many house rules about kitchen use.

[Wipneus]

Ok I see it on your sharing site ... http://tinyurl.com/jqo2lq7 ... but as 3x8-bit channels of png so this would get into a nuisance situation of having to upload three separate 16-bit tifs. Color is quite good now however. There is talk/confusion about 16-bit per channel png which would be 48-RGB but it does not seem widely supported, notably not on the forum.

I am quite sure these are 16 bit png's. All image manipulations done with imagemagick.
Output from identify (part of the IM-suite):

S2A_R111_V20160515T152912_22WEB_RGB_crop_1200x980+8100+1800.png PNG 1200x980 10980x10980+8100+1800 16-bit sRGB 6.187MB 0.000u 0:00.000

Normal pngs would report as "8-bit sRGB".

Minor detail is that I should have repaged it, now the png is stored with the offset from the original from which it was cut.

I am not sure if you have a working ImageMagick available. This is the command to create 3 16-bit tiffs from the original:

Code: [Select]

convert -separate S2A_R111_V20160515T152912_22WEB_RGB_crop_1200x980+8100+1800.png separate_%d.tiff
creating 3 files: separate_0.tiff, separate_1.tiff and separate_2.tiff
 

[A-Team]

not sure if you have a working ImageMagick available

I looked at the installation yesterday. Requires MacPort or byzantine list of commands. IM requires latest el capitan OS but  I've lost track of my admin password over the years, long workaround hack to bypass that, OS requires 6 GB storage, would have to delete lots of images, time-consuming maybe better just to buy new one with retinal display blah blah blah nice spring day here get back to you.

But yes, it seems like a totally minimal set of ImageMagick commands would help a lot.

I wasn't expecting Jakobshavn to do much this early and got engaged in optimizing Worldview palette squashing of Beaufort on a per-channel basis as tiling high resolution imagery is problematic over such a vast area and orbital repeat is too long vs daily satellites. Though the Mackenzie River annual blow-out might make a neat S2A color.

http://nsidc.org/greenland-today/ shows a couple of serious melt spikes ...

[solartim27]

Whats going on with the north branch?  Is that a cloud effect, or a big advance ( with another look I say clouds)?  Looks like the south branch calving broke up the ice jam.

[TenneyNaumer]

Those are some images of incredible melting for this time of year, A-Team!

Wish I knew how to do what you do!

Thanks for doing it for us!

[TenneyNaumer]

So, solartim, did that calving front at the south branch move back to a new record?

I don't think anything big happened at the north branch -- looks like clouds.

[solartim27]

So, solartim, did that calving front at the south branch move back to a new record?

I don't think anything big happened at the north branch -- looks like clouds.

Sorry, I wouldn't be able to say.  I would think not, but to be so close at the start of the melting season is a sure sign (to me) it will be passed soon.

[Wipneus]

R125, 16 May has also arrived. This does not cover the calving fronts, just the area east of Jakobshavn. That includes the melt ponds in the last post, the expansion in just one day is very clear:

 

[A-Team]

That's amazing for one day.

Have you tried "convert *.jpg -evaluate-sequence median OUT.jpg" in ImageMagick with the separate bands in a common directory? IM is supposed to be quite fast and effective with this kind of median sharpening. Basically the satellite is taking three simultaneous images of the "same thing". Although each sensors is specialized to its wavelength range, this may still work here because the scene is mostly white and the sensors are not noise-free. The median pixel value then takes out this noise (eg 125, 128, 129 --> 128).

http://blog.patdavid.net/2013/05/noise-removal-in-photos-with-median_6.html

I've been looking at another approach, namely geometric mean, which for two grayscale layers is essentially the multiply layers command in a gimp stack. It is quite effective on the Aqua Modis base layers people use so much in the Beaufort.

The animation below looks, with limited success, at isolating the new melt areas that developed between the 15th and 16th. The R111 and R125 have slightly different geometries which prevents pixel-perfect alignment which here is important since the new melt areas are small.

[Wipneus]

Have you tried "convert *.jpg -evaluate-sequence median OUT.jpg" in ImageMagick with the separate bands in a common directory? IM is supposed to be quite fast and effective with this kind of median sharpening. Basically the satellite is taking three simultaneous images of the "same thing". Although each sensors is specialized to its wavelength range, this may still work here because the scene is mostly white and the sensors are not noise-free. The median pixel value then takes out this noise (eg 125, 128, 129 --> 128).

Are you talking about taking the median from different color bands? These images maybe whitish, but there are almost always traces of color there, if the color channels differ more than the noise then there is no reduction. Averaging will always work (somewhat), with four bands the noise can be cut in half (at best).

[Tor Bejnar]

R125, 16 May has also arrived. This does not cover the calving fronts, just the area east of Jakobshavn. That includes the melt ponds in the last post, the expansion in just one day is very clear:

The melt pond on the right appears to have drained into the pond on the left - the connection is apparent. (Watch the animation - it's better!)

[A-Team]

it's connected

If this keeps up, it will be getting very soggy out there. Some of these lakes and channels form at the same spot year after year but others do not.

The usual way to compare S2A bands in gimp is called grain extract (upper layer - low layer +128 gray) followed by contrast un-squeeze which seems to be out = in1 - in2+ 50% etc in ImageMagick. This is subtraction, followed by re-centering to avoid negative pixel scores, giving a tight gaussian around neutral, followed by multiplication by the constant that undoes the tightness so the new range is [0,255].

I recollect that the blue and green bands had very very similar histograms and that hardly anything was visible in the sea of pure gray until the contrast was really forced. In astronomy and microscopy, they tend to use geometric mean rather than arithmetic. It is not safe to assume that each sensor error is dominated by the (same) gaussian noise distribution. The idea here is to decompose the RGB to HSV and replace the V with some improved product.

Usually in multispectral satellite band analysis, principal component analysis would be used for dimensional reduction, ie  the 'improved product' would take the first component. In other words, we're not doing anything with band covariance. I'm not seeing this offered in either IM or Gimp but there do seem to be plugsin for ImageJ. R is often used for this. Definitely would want to crop to essential ROI to reduce run time.

[oren]

A-Team your graphic abilities are as always astounding.

This looks more like a river than a "pond"...

[A-Team]

This looks more like a river than a "pond"

Yes and how much of it is already ending up at the bottom of the deep channel, softening the ice above and till below which predisposes it to deformation and lubricated flow? If the calving front response has been measurable acceleration, we would have to go with S1A since it has been so cloudy.

Seems like Sentinel 2A would be even better at estimating pond depth than Landsat-8 ... of course the academic lag of 2 years means S2A hasn't even been launched yet. Note the meltlake graph only starts at June 10th.

Estimating supraglacial lake depth in West Greenland using Landsat 8 and comparison with other multispectral methods
A Pope  T Scambos  M Moussavi  M Tedesco  M Willis  D Shean and S Grigsby
http://www.the-cryosphere.net/10/15/2016/

... Digital elevation models derived from WorldView stereo imagery (pre-lake fill-ing and post-drainage) are used to validate spectrally derived depths, combined with a lake edge determination from im- agery. The optimal supraglacial lake depth retrieval is a physically based single-band model applied to two bands independently (red and panchromatic) that are then averaged together. When Landsat-8 and WorldView-derived depths are differenced, they yield a mean and standard deviation of 0.0 ± 1.6 m. This method is then applied to OLI data for the Sermeq Kujalleq (Jakobshavn Isbræ) region of Greenland to study the spatial and intra-seasonal variability of supraglacial lakes during summer 2014.

[Wipneus]

2016/5/19 ( R025 ) is in, only clouds visible. No upload.

[A-Team]

Seems like a cloudy spring. Here are the clear days I have on tap for 2013-16 up to day 180. It's feasible to poke through Landsat metadata to systematically extract cloud cover percentages but that would be regional and some days with a clear calving front would fall through the cracks.

So we are at the mercy of Sentinel 1A scheduling and low resolution Radarsat from DMI, both minimal lately. DMI does not provide the file name or link to S1A images so the May 17th is a separate chase and hassle at the ESA portal just to get a less-than-thrilling EW grade image which shows the position of the calving front but little else. It takes the IW grade imagery to track velocities. The melt region Wipneus has been studying is barely recognizable in EW.

S1A_EW_GRDH_1SDH_20160517T095942_20160517T100022_011297_0111E6_B8C2 Size: 676.85 MB
Download URL: https://scihub.copernicus.eu/dhus/odata/v1/Products('762dcc08-a6e5-415a-8404-b19e2a5dbcb2')/$value

I've always been fascinated by the dead links at ESA Copernicus that could have provided a short cut to the imagery itself: what was going through their head?

measurement

    s1a-ew-grd-hh-20160517t095942-20160517t100022-011297-0111e6-001.tiff
    s1a-ew-grd-hv-20160517t095942-20160517t100022-011297-0111e6-002.tiff

2013 101 08 11 LC80090112013101LGN01
2013 140 10 11 LC80100112013140LGN01
2013 142 08 11 LC80080112013142LGN01
2013 172 10 11 LC80100112013172LGN00

2014 040 09 11 LC80090112014040LGN00
2014 056 09 11 LC80090112014056LGN01
2014 081 08 11 LC80080112014081LGN00
2014 097 08 11 LC80080112014097LGN00
2014 159 10 11 LC80100112014159LGN00
2014 161 08 11 LC80080112014161LGN00

2015 043 09 11 LC80090112015043LGN00
2015 050 10 11 LC80100112015050LGN00
2015 052 08 11 LC80080112015052LGN00
2015 059 09 11 LC80090112015059LGN00
2015 075 09 11 LC80090112015075LGN00
2015 091 09 11 LC80090112015091LGN00
2015 107 09 11 LC80090112015107LGN00
2015 114 10 11 LC80100112015114LGN00
2015 123 09 11 LC80090112015123LGN00
2015 132 08 11 LC80080112015132LGN00
2015 146 10 11 LC80100112015146LGN00
2015 148 08 11 LC80080112015148LGN00
2015 155 09 11 LC80090112015155LGN00
2015 162 10 11 LC80100112015162LGN00
2015 178 10 11 LC80100112015178LGN00
2015 180 08 11 LC80080112015180LGN00

2016 085 10 11 LC80100112016085LGN00
2016 096 07 12 LC80070122016096LGN00
2016 110 09 11 LC80090112016110LGN00
2016 119 08 11 LC80080112016119LGN00

[Espen]

Continuing calving and retreat at Jakobshavn Isbræ.
The glacier retreated in many places further back than in the end of September 2015:

https://twitter.com/Ecoverycom

Please click to enlarge and animate!