Jakobshavn Isbræ / Ilulissat Isfjord / West Greenland

[A-Team]

Jakobshavn is arguably the most important glacier in Greenland but by far the hardest to study, leading to no end of factually ungrounded model studies. A March 2016 paper from the Rignot group looks instead at Store Glacier a hundred km to the north, where marine bathymetry and bedrock topography are much better known.

I'll post more details at the Store forum in a bit but this glacier definitely provides a counterweight to speculation about a runaway future for Jakobshavn -- the calving front of Store will not retreat except under rather extreme future scenarios.

They posted six superb movies in .avi format illustrating the incredible stability of this glacier. These are 500-frame .avi movies of 80 MB size that go 20 years out. Our forum does not support this format nor size, so I cropped the movie down, rotated to horizontal 700 pixel width, and exported to gif format to give a taste of what the high quality image holds.

I picked the most extreme scenario, in which Store does retreat down its reverse slope to a new position of stability. In most of the others, it doesn't. It took 15 minutes to crop operation in ImageJ which displays .avi as a virtual stack and another 25 to export in animated gif 8-bit palettes, with a further 15 minutes in Gimp to run the gif differencing filter (a most excellent opportunity to do laundry). However even with every other frame deleted and further cropping this still left too large a file to display on the forum. Better just to look at the real thing.

http://tinyurl.com/jsu3pg4 (Movie S6)

Store has been remarkably stable over the past century both in terms of velocity and ice front position, possibly since the little ice age. Store flows below sea level for about 60 km; its ice front  =rests on a sill that is ∼450 m below sea level [[direct sonar data conflicts with 'above sea level' in Bamber 2013 DEM]]. Seaward of the sill, the fjord is ~750 m deep; the bed upstream is retrograde and reaches a depth of 800 m. The glacier flows at 6300 m/yr at the calving front and drains a catchment basin of 30,500 km² [[numbers about a third of Jakobshavn]].

Store Glacier is stabilized by a sill at its terminus. The glacier is only dislodged from the sill when ocean-induced melt quadruples, at which point the glacier retreats irreversibly for 27 km down a reverse bed. The model suggests that ice-ocean interactions are the triggering mechanism of glacier retreat but the bedrock topography controls its magnitude.

http://onlinelibrary.wiley.com/wol1/doi/10.1002/2016GL067695/full (paywalled)

[A-Team]

to a folder that should be browsable

Yes, that is working. The file names are soooo long that they don't display. The April scene is on the far right of all the icons and downloads with a click. These are 16-bit rather than the peculiar 15-bit previously and so open directly as 240 MB files in ImageJ tiff.

Thanks for supporting the non-terminal mode public! Maybe after a time, people should probably get familiar with gdal as it has many useful such functions?

Note we are still awaiting clear Jakobshavn scenes at the 10-day orbital interval needed (by us) for accurate co-registration. (Any re-projection will slightly degrade image quality.) A pair just became available for Zachariae despite it being a thousand km to the north -- Wipneus posted the very first such Sentinel 2A animation ever made for an ice sheet at that forum.

espen/Oren: What is done at this site is beyond what is ever seen before by anyone outside!! Where are the researchers?? So many potential research issues ...  and the public sometimes see quite a few research projects that makes no sense

It is puzzling indeed, going into the 4th year of Landsat-8, why Greenland 15 m scenes have never seen serious use in a cryosphere journal. The latest 2016 paper concerns the big melt year 2012 (not joking). That was Landsat-7 with the broken mirror. Whatever ever happened with record calving event of August 2015 -- we know now exactly how to process the seismic data for timing of the calving cascade -- 2017 perhaps?

Sentinel 2A launch? Forget it. Didn't happen. Not invented here. We'll get back to you on that. Queue ticket given low priority. Computers tied up doing 1D model runs.

[Wipneus]

Yes, that is working. The file names are soooo long that they don't display. The April scene is on the far right of all the icons and downloads with a click.

Agree, it was convenient for me to cut, paste and expand on esa's file names but many file selection boxes are not designed for such lengths. I will do some pruning next time.

[Espen]

Jakobshavn update:
Another big calving sequence between March 22 and April 3 2016.

Please click on image to enlarge and animate!

[A-Team]

Same pair of Sentinel 1As zoomed to full resolution at the calving front.

With 3 satellites coming in, we really need to keep better track of dates, file names, and first comment. A month from now, someone will be wanting to process them differently and need a fresh download. Or will be looking for orbital pairs. Or different satellites, same day. Or need to sharpen the timing of a calving event. Or make an animation. Or search all the forums or whole internet to see if was discussed elsewhere else.

So be sure to include the file name somewhere as text so it doesn't have to be retyped!

61   01 03 16   Sentinel 1A   #1366   Espen   s1a-iw-grd-hh-20160301 62   02 03 16   Landsat-8   #1317   Wipneus   LC80090112016062LGNOO.B8 63   03 03 16             64   04 03 16   Radarsat2   #1323   solartim27   DMI low resolution 65   05 03 16             66   06 03 16             67   07 03 16             68   08 03 16   Sentinel 1A   #1339    A-Team   s1a-iw-grd-hh-20160308 +  2013-16  index 69   09 03 16             70   10 03 16   Sentinel 2A   #1359   Espen   s1a-iw-grd-hh-20160310 71   11 03 16   Landsat-8   #1365   A-Team   LC80090112013062LGNOO.B8 72   12 03 16             73   13 03 16             74   14 03 16             75   15 03 16   Landsat-8   #1348   A-Team   LC80080122016071LGNOO.B8 76   16 03 16             77   17 03 16             78   18 03 16   Landsat-8   #1363   Espen    LC80090112016078LGN00_B8 79   20 03 16             80   21 03 16             81   22 03 16   Sentinel 1A   #1405   Espen   s1a-iw-grd-hh-20160322 82   23 03 16   Sentinel 2A   #1360   A-Team   R068_V20160323T151927 92   02 04 16   Sentinel 2A   #1375   A-Team   bands 1-13 tour 83   24 03 16             84   25 03 16   Landsat-8   #1363   Espen   LC80100112016085LGN00_B8 84   25 03 16   Sentinel 1A   #1366   Espen   s1a-iw-grd-hh-20160325 85   26 03 16   Landsat-8   #1365   A-Team   LC80100112016085LGN00_B8 86   27 03 16   Sentinel 2A   #1374   Wipneus   R111 22WEB  bands 234 87   28 03 16             88   29 03 16             89   30 03 16   Sentinel 2A   #1390    Wipneus   band 4 links #1401 90   31 03 16             91   01 04 16             92   02 04 16             92   02 04 16   Sentinel 2A   #1401    Wipneus   band 4 links #1401 93   03 04 16   Sentinel 1A   #1405   Espen   s1a-iw-grd-hh-2010403

[Tealight]

Nice initiative. Can you describe for us the operational steps in doing this, where you are getting these profiles, and how they are applied to those three bands, in what software? Or is this something provided in a black box within the Snap toolbox?

Yes it is pretty much a black box in SNAP. I download bands 2-4 from Amazon WebServices and since these are three individual images, SNAP opens them as individual products and I haven found a way yet to combine these into a single one. For multiple Products it is not possible to use the "Sentinel 2A Natural Colour" RGB profile, instead I have to choose the Red, Green and Blue bands individually. Due to a bug the RGB tool crashes when selecting multiple products, but it is possible to save the bands as virtual bands in one product and then finally by selecting the virtual bands it creates the RGB image. I know that this isn't the most convenient way, but I reported the bug already and don't want to try out lots of other programs.

For the Sentinel image/bedrock overlay wouldn't it be better to overlay the whole bedrock data instead of just the main channel? Currently we are talking a lot about side buttressing and in summer we might reach the second buttressing point, which should get interesting.

[A-Team]

For the Sentinel image/bedrock overlay wouldn't it be better to overlay the whole bedrock data instead of just the main channel? Currently we are talking a lot about side buttressing...

Sure, but be sure to use the Morlighem bedrock mass conservation DEM, not the Bamber as above. It's 'best practice' to post as a three layer animation, quick satellite with its date, quick colored bedrock, slow overlay. Then the next person has the files handy to raise or lower the transparency or put in a later/earlier/different satellite.

Right now, the north side of the south channel, especially the northeast corner which is being swept off the elbow mound, appears more active than the main ice channel from the deep interior, possibly because of the very recent loss of buttressing from the main channel.

We have not found a good way yet to put in vertical thinning -- which is hugely important for current and future behavior of the glacier, especially with regards to the sidewalls and fixed obstructions. The rate of thinning is regularly measured but I have not seen it provided in an appropriate format.

[Tealight]

Sure, but be sure to use the Morlighem bedrock mass conservation DEM, not the Bamber as above.

I used the bedrock data from Morlighem. Who is Bamber, I never heard the name before?

Do you know if the NSIDC releases a new IceBridge BedMachine version annually to update it with data from the IceBridge mission?

Edit:

With 3 satellites coming in, we really need to keep better track of dates, file names, and first comment. A month from now, someone will be wanting to process them differently and need a fresh download.

I don't think the forum layout is a great way to use as a database for images. Couldn't we create a new thread which just shows links to a server i.e. an "ASIF Google drive". There all images for a glacier get their own folder and the files names include date and satellite.

The account access data will only be shared via private message between the contributors. Right now:  Espen, Wipneus, A-team

[A-Team]

bedrock data from Morlighem. Who is Bamber, never heard

J Bamber did the heavy lifting on Goginini's ice penetrating radar data set to produce the first bedrock DEMs of Greenland. Morlighem posted a putative refinement implied by surface ice velocity that most notably deepened west Greenland marine outlet troughs. At specific locations, there are also sled radar grids and seismic surveys. Rignot has since added high resolution bathymetry of calving fronts of  numerous fjords (not including Jakobshavn).

You could well ask, never mind posted formal errors, what happens when new tracks of ice penetrating radar data that weren't included (ie 2014-15) in Bamber/Morlighem are run against the DEM. There is no annual updating that I know of. Specifically, the radar tracks are sparse so questions have been raised about the accuracy interpolating with 'ordinary kriging' whereas glaciers grinding away on bedrock create an anisotropic landscape. This would be a good project for the forum.

Meanwhile Bamber is working on fixing inter-track interpolation and improving the accuracy of basal reflection depths themselves. On Cresis radargrams, the ordinate is radar return time. Depth is secondarily inferred by assuming constant propagation velocity and attenuation of radio waves in ice whereas it is actually an exponential function of temperature (profile). His other 265 publications (5844 cites) are listed at researchgate.

http://www.the-cryosphere-discuss.net/tc-2016-8/
https://www.researchgate.net/profile/Jonathan_Bamber

None of this really helps in the specific case of Jakobshavn -- the ice is too rough and too deep. Seismic lines are more effective than radar on basal hydrated till, which approaches 100 m in depth in certain JI over-deepenings. Your bedrock DEM might unfamiliar because people usually tweak color scheme transitions to bring out humps and troughs. Hue is easily rotated in gimp.

the forum not great database for images. just show links to a server ... all images for a glacier get their own folder and the files names include date and satellite. account access data will only be shared via private message

Over the last 36 days, there have been 18 images of Jakobshavn. That number will prove erratic over the melt season due to S1A scheduling priorities and periods of cloud cover. While every 2nd day is very decent and a big improvement over Landsat alone, it is not enough to resolve details of calving cascades.

For that we need to dig into the real-time seismic record. It's hard to believe with 900 forum members, we have yet to unravel the record calving event of 15 Aug 15. It's all laid out in the first link below. We need to get it in gear NOW in case there's more of the same this summer. In fact, why stop with posting one-off images when every calving event can be resolved into its stages.

http://onlinelibrary.wiley.com/doi/10.1002/2015GL066785/full
http://library.seg.org/doi/abs/10.1190/geo2015-0154.1

Given our total dependence on open source radar, open source satellite, open source seismic, open source weather, open source GPS, open source journals, etc, it might seem churlish for us to squirrel away datasets of glacier imagery. What goes around, comes around.

True, it's time-consuming to sort through incoming for cloud-free imagery and to download multi-GB files from dorky portals, crop out the glacier, align different dates, process in 16-bit, interpret the scene for significant developments, and write up a post.

Yet duplicative proprietary efforts make no sense -- we all have too much to do already, climate change is too important to be playing games. The trick is to string together a pipeline in which each contributor can do their favorite bit (perhaps effortless in their software environment), pass the baton, and see advantages from others running with it.

This is a very different paradigm for western science, collaborative rather than competitive. We're more accustomed to the unscrupulous data-stealing Watson & Crick where it's all about establishing priority, getting credit, and glorifying the individual. However collaborations have proven more effective in data-overload situations requiring inputs from many specialties. No one knows enough any more to go it alone.

People think, 'oh if I share my idea, someone else will run with it and my career will suffer'. The fact is, it won't. It costs them nothing to add another name to a paper. And if your good ideas are so far and few between that they have to be hoarded, there's no future in academics to begin with.

[Iceismylife]

Or need to sharpen the timing of a calving event.

I have been thinking that a hydrophone in the fjord somewhere would do a lot for tightening up calving events.


I would also like to listen to it for the entertainment value.

[oren]

Stupid question of the day: Is it so difficult to set up a webcam and record its output, such as in Barrow or in Helheim? I realize there is no easy location that covers the whole front, but a sub-optimal webcam is better than no webcam.

[A-Team]

Holland had two webcams and a weather station set up for years on rocks west of the north branch. One looked straight out at passing bergs and the other was aimed at the calving front. Not much was ever done with the data. The site was too far from the calving front (or needed a telephoto lens). I checked them at the time of the big 15 Aug 15 event but they had become inoperative in late July.

It seems feasible to install a hydrophone at Jakobshavn and even more feasible to lose it to icebergs and turbulence, or at least lose the signal amidst in all the background noise. Here is a 2015 paper for Spitsbergen that has gotten good results in conjunction with a camera:

...Here we present the potential of hydroacoustic methods to investigate different modes of ice detachments. High-frequency underwater ambient noise recordings are combined with synchronized, high-resolution, time-lapse photography of the Hans Glacier cliff in Hornsund Fjord, Spitsbergen, to identify three types of calving events: typical subaerial, sliding subaerial, and submarine.

A quantitative analysis of the data reveals a robust correlation between ice impact energy and acoustic emission at frequencies below 200 Hz for subaerial calving. We suggest that relatively inexpensive acoustic methods can be successfully used to provide quantitative descriptions of the various calving types....

There are many factors that affect the intensity of underwater noise recorded during ice detachments, including different calving styles and occurrence of “mixed” events, iceberg shapes and sizes, propagation effects, and activity of other sources, e.g., breaking waves, freshwater outflows, or ship traffic. For example, like in this study, friction between calved block and glacier wall may result in greater noise levels that could be expected from visual observations.

Similarly, a larger contact zone with the sea surface results in bigger splashes and greater sound intensity underwater. Conversely, these observations fully motivate the distinction between calving styles, which was one of the main goals of this study. As was also demonstrated, the energy of the impact and acoustic emission can be correlated for both sliding and typical subaerial events, knowing the frequency content of underwater noise generated by these types of detachments. The results presented here are limited to 12 well-constrained calving events from a single tidewater outlet glacier.

http://onlinelibrary.wiley.com/doi/10.1002/2014GL062859/full free full

Note as the calving front retreats, a new Landsat path,row comes into play. This will have the effect of increasing the frequency of coverage at the calving front from this satellite despite its lesser swath width.

[Iceismylife]

https://goo.gl/maps/GpEuN7PGFfz I was thinking that this location mint be protected from mechanical damage.  Is hydrophone depth critical?

[nukefix]

If you want to improve models for the physics of calving it would probably be prudent to try to do that in a simpler environment than Jakobshaven..

[A-Team]

This location might be protected from mechanical damage.

That bay does not have a direct 'line of sight' so the sound signal will have diffraction and reflection issues. The advantage of seismic is that the stations are already established and have extensive multi-year records. Fabulous strip of WorldView2 resolution at Google Earth though tiling job is awful.

improved models for the physics of calving ... do that in a simpler environment than Jakobshavn

Store Gletscher is the #1 choice for that between Rignot's latest bathymetry and Hubbard's drones and drilling inland. The question though is how transferable that model -- ice-meltwater-ocean interactions driving calving front retreat but reverse bed controlling extent -- will be to other west Greenland and Antarctic glaciers.

At Jakobshavn, the 23 March and 02 April 2016 Sentinel 2a pair are the magical ten days apart and so suitable for velocity transect measurement (though both are less than ideal regionally because of cloud cover). Based on 2.5 m resolution for feature tracking, I measured displacements at two transects, one near the calving front and the other 2 km up-flow line up-glacier. The blue overlay indicates the position of the calving front on the 23rd.

The numbers are in pixel displacement at 5 m so should be halved to give velocities in meters per day, with 33 m/d being typical for the faster moving center. Somewhat slower values occur upstream because of stretching/thinning of the glacier as it nears the calving front. Motion on the north side is quite complicated and not shown here.

These numbers are comparable to those of late March in Fig.2 of Joachin 2014 though the measurement procedure needs to be very consistent in order to establish acceleration or de-acceleration. Note 10 m S2A pixels are 100 m² in area which offers quite an advance over 15 m Landsat pixels of area 225 m².

The algorithm of T Scambos 2016 should work unchanged on S2A; while processing/posting of all pairs was promised, nothing has been appeared yet at NSIDC. Cloud-free triple or quadruple 10-day sequences this season would take some luck with the weather but would provide a fantastic scientific opportunity.

S2A pairs
23   2 clear/clear             S2A_R068_V2016 03 23 T151927.22WEB.B04.tiff
26   5 clear/missing          S2A_R111_V2016 03 26 T153310.22WEB.B04.tiff
27   6 missing/missing   
29   8 missing/friday?
30   9 clear/saturday?       S2A_R025_V2016 03 30 T150931.22WEB.B04.tiff
02   12 clear/later             S2A_R068_V2016 04 02 T151928.22WEB.B04.tiff

[Wipneus]

The algorithm of T Scambos 2016 should work unchanged on S2A; while processing/posting of all pairs was promised, nothing has been appeared yet at NSIDC. Cloud-free triple or quadruple 10-day sequences this season would take some luck with the weather but would provide a fantastic scientific opportunity.

S2A pairs
23   2 clear/clear             S2A_R068_V2016 03 23 T151927.22WEB.B04.tiff
26   5 clear/missing          S2A_R111_V2016 03 26 T153310.22WEB.B04.tiff
27   6 missing/missing   
29   8 missing/friday?
30   9 clear/saturday?       S2A_R025_V2016 03 30 T150931.22WEB.B04.tiff
02   12 clear/later             S2A_R068_V2016 04 02 T151928.22WEB.B04.tiff

The <Rnumber> is the relative orbit number. Same number and time means the images are exact pairs. Compare with Landsat's "path".

The 27 March (R125) is available, it has 45% coverage (55% is black) and 100% cloudcover.
The 29 March data has only 2% coverage. I did not bother to see if that particular corner has any clouds.
There is also a second image on the 26th, coverage 12% to confuse things.

With the relative numbers in mind, we wait now for 5,6,9... (April). Have a look at the 6th of April.

The Zachariae images don't exactly repeat the first decade, so there may be more (optimistic) here too.

[A-Team]

That is helpful to have more of the file names parsed. Though even when they match up like the two above, I still find the geometries slightly irreconcilable (not alignable within ISO(2): horizontal and vertical shifting or rotation). I don't believe that the orbit repeats have better than a tube confinement due to gravitational perturbations (sun/moon, ellipticity).

There may be a trade-off between higher order image warping and image degradation. It looks like even gdal does not go too far with this.

Agree it is worthwhile to bother with mostly cloudy scenes hoping for a break. Retrospectively, after a big event, then it makes some sense.

Not seeing any April 6th yet at divided/gedeeld or http://sentinel-s2-l1c.s3-website.eu-central-1.amazonaws.com/#tiles/22/W/EB/2016/4

I must say, this Ingestor is impressive if it can truly host the images up within minutes of download and processing at the ESA hub. EarthExplorer seems to have a two week lag, nothing on JI last time I looked. I'm wondering if USGS is even preceding with S2A hosting given the Amazon site -- I asked a question about S2A, got a queue ticket, but never an answer. Sounds like offerings are in flux.

There was a press release about ASTER satellite images becoming free online too, with glorious samples. EarthExplorer offers that too but again I couldn't find anything recent for Greenland. At some point, satellite coverage becomes too frequent relative to anything useful we could do with it. The same could be said for ground resolution.

http://www.huffingtonpost.com/entry/nasa-earth-images-aster-environment_us_5705226fe4b053766188458e
http://www.jpl.nasa.gov/news/news.php?feature=6253

[nukefix]

S-2 orbital tube has a 50m RMS-radius. AFAIK the images should coincide without resampling if the metadata is handled correctly (should work in SNAP).

[Iceismylife]

For the Sentinel image/bedrock overlay wouldn't it be better to overlay the whole bedrock data instead of just the main channel? Currently we are talking a lot about side buttressing...

...

We have not found a good way yet to put in vertical thinning -- which is hugely important for current and future behavior of the glacier, especially with regards to the sidewalls and fixed obstructions. The rate of thinning is regularly measured but I have not seen it provided in an appropriate format.

I have been talking with a friend of mine about the needs for drones. We are thinking two classes.  One to carry ice penetrating radar and the other as a photo ops ship. 10x difference in gross.  The photo ops ship would be about 150 lbs. and have a 24 to 48 hr endurance.  The other is looking about 3,000 lbs. gross. with much less endurance.  Comments input welcome.

[Iceismylife]

This location might be protected from mechanical damage.

That bay does not have a direct 'line of sight' so the sound signal will have diffraction and reflection issues. The advantage of seismic is that the stations are already established and have extensive multi-year records. Fabulous strip of WorldView2 resolution at Google Earth though tiling job is awful.

   ...

I just want to hear them go Boom Boom Boom.  There is probably a place where we could put a hydrophone that would be more or less protected that would have a good view of the calving face.

[A-Team]

S-2 orbital tube has a 50m RMS-radius. images should coincide without resampling if the metadata is handled correctly (should work in SNAP).

Decent! We are just taking whatever whomever mirrors whenever to AWS. It doesn't seem feasible to take the last of the wobble out of the rocks (there are illumination issues). On the other hand, the alignment is very decent indeed. It is only when trying to squeeze the last drop of water out of a potato (year-on-year newtonian differencing) that it matters.

I think they are offering S2A on a provisional basis, not representing that it has settled down completely, the Oper in S2A_OPER_MSI_L1C_TL_MTI__2016...  I saw somewhere that we can kiss the rocks goodbye and trust the georeferencing in the deep interior. No way I am going to do that with Landsat. Yet that limits Landsat analysis to frames that take in these lopsided ground control points on the coast.

That T Scambos paper talks about image pair feature correlation fit to a tenth of a pixel. It is quite interesting how they do that, well explained in the paper.They say further that they can give SAR a run for the money even with Landsat so S2A will be that much better. I'd like to see some cross-validation with differential GPS on the ground.

The dynamic range of S2A has been really great considering the lack of intrinsic contrast in what they're imaging on Greenland, maybe 11-bit. Landsat, some days it can't be more than 4-bit, everything scrunched into a narrow peak at the far left of the histogram.

[Wipneus]

I have uploaded S2A_R111_V20160405T152926.22WEB.B04.tiff to the usual folder.

The S2A_R125_V20160406T145931 is disappointing, only one cloudy corner of 4.5% available.

[A-Team]

Thanks for another glorious image. Calving front processed to 5 m below (1400 pixels, click to see). I don't see any indication of incipient calving. The right corner is the most interesting area ... the confluence of main channel, mound shoulder shrug, and recently unbuttressed north bank. That would be worth doing in color for possible open water. These Sentinel 2As will make for fantastic animations when longer time frames become available.

[A-Team]

Here is another experiment with color. For this, I am just working with bands 2,3,4 as they open as an 8-bit stack in gimp. This could no doubt be done better in the SNAP toolbox, perhaps with other combinations of three bands and color spaces. Note though that the flat gray area is not liquid water -- it has the same luminance in blue, green and red channels unlike blue water. 

Also the pockmarks and depression (which invert in the lower image) do not fit with that. This is a peculiar area that, even looking at multi-year animations, the glaciological 'need' for it remains a puzzle. There is a similar area just past the elbow on the north side, 3rd image. The final image shows a complex confluence of ice streams moving at different speeds on the south bank farther upstream.

These three spots are also available on the 05 Apr 16 Landsat LC80070122016096LGN00 for which we have two tutorials on how to get amazingly natural colors on an ice sheet. However going from 15 m to 5 m would be a stretch. My download speed is 2.5x faster from Amazon than USGS or ESA. I stopped using the ESA hub because it is slow and drops the connection.

[Tor Bejnar]

Wonderful detail - you guys are awesome!
Frozen lake(s) with a few small ice bergs riding high, maybe?

I can see one main wavy lineation that quasi-parallels the calving front.  I presume it will define a future calving fronts (edit:  nowhere near as well-defined as cracks on ZI).

[A-Team]

Wonderful detail - you guys are awesome! Frozen lake(s) with a few small ice bergs riding high, maybe?  main wavy lineation that quasi-parallels the calving front. define a future calving front? so different from Zachariae.

It would be even more awesome if we could take all the kinks out of the pipeline and just pour the main products onto the forum unattended with minimal delay.

It would be more ambitious to drop a new day into the next frame of an updating animation. Cropping does not seem to be much of an issue as it would just be a fixed set of coords for each orbital 'path' granule such as R111_V...22WEB, see 4th image for picking the four corner coordinates at maximal forum animation width. Cloudy scenes could be deleted manually later at a micro level or from a higher band thresholding step after the crop.

It sounds like the geolocation data is more than good enough to adjust different viewing angles to a common geometry. There may well be scripting capability within SNAP toolbox, haven't looked. If JI could be animated, that file path would serve for Zach, Petermann, Pine Island, or anywhere for that matter using a simple web form to input parameters.

The April 5th is T152926 or 15 29 26 whereas the time of posting an hour or so earlier than now... this being April 8th. Ouch, 72 hours but I don't recall the April 5th actually being uploaded to AWS last night but cannot locate the time their page was last refreshed in developer tools.

2016 04 05 S2A_R111_V20160405T152926.22WEB.B04.tiff

I added LC80070122016096LGN00 bands 345 RGB below at comparable resolutions (not pan-sharpened with the 15 m). Quite a difference!

[Iceismylife]

Here is another experiment with color. For this, I am just working with bands 2,3,4 as they open as an 8-bit stack in gimp. This could no doubt be done better in the SNAP toolbox, perhaps with other combinations of three bands and color spaces. Note though that the flat gray area is not liquid water -- it has the same luminance in blue, green and red channels unlike blue water. 

Also the pockmarks and depression (which invert in the lower image) do not fit with that. This is a peculiar area that, even looking at multi-year animations, the glaciological 'need' for it remains a puzzle. There is a similar area just past the elbow on the north side, 3rd image. The final image shows a complex confluence of ice streams moving at different speeds on the south bank farther upstream.

These three spots are also available on the 05 Apr 16 Landsat LC80070122016096LGN00 for which we have two tutorials on how to get amazingly natural colors on an ice sheet. However going from 15 m to 5 m would be a stretch. My download speed is 2.5x faster from Amazon than USGS or ESA. I stopped using the ESA hub because it is slow and drops the connection.

I like the top photo it make the topography really stand out. looking at it from the bottom up instead of the top down.  That spot looks to be below or at sea level.

[oren]

The resolution is astounding. And that depression mighty peculiar. A hi-res animation (when enough of the material adds up) could certainly provide more insights.

[A-Team]

I'm looking forward to updating this previous multi-year Landsat animation to S2A resolution ... bizarre how this area got eaten up as the calving front retreated to its maximal position but then made a big comeback to look like its old self.

The five days available for Sentinel, shown in a somewhat dodgy alignment below, still show quite a bit of action. The flat area has been held fixed which has the effect of making the area to the east move in the 'wrong' way. The pause in the animation is on April 5th.

[oren]

I'm looking forward to updating this previous multi-year Landsat animation to S2A resolution ... bizarre how this area got eaten up as the calving front retreated to its maximal position but then made a big comeback to look like its old self.

The five days available for Sentinel, shown in a somewhat dodgy alignment below, still show quite a bit of action. The flat area has been held fixed which has the effect of making the area to the east move in the 'wrong' way. The pause in the animation is on April 5th.

If there were seven wonders of the world of ice dynamics, that peculiar area would surely be one of them. To come back from the dead in exactly the same shape as before. Amazing.

[johnm33]

Eyeballing the latest animation my take:
There's an overturning wave in the flat strip caused by the ice falling over the cliff, beyond that there's a bigger wave sustained by the pressure of the fall, and thats overturning too, my guess is for that to happen it's falling into water. What are the odds this'll all be gone again by wednesday.
 Maybe we should open a thread for image interpretation?

[iwantatr8]

My thoughts on this is that we have a very complex non-Newtonian fluid interaction taking place.

But starting with the base interactions we have an ice stream, for want of better explanation acting along the main glacier flow with turbulent interactions at the side of the channel.

The second influence is the descending ice that is falling over the mound.

The falling ice causes pressure at the edge of the turbulent flow from the main stream this pressure will give rise to a turning moment which overturns the edge of the stream. This is then pulled along with the rest of the stream causing an elongated feature along the edge.
I imagine this like a horizontal vortex acting along the ice stream edge but happening very slowly.

Just my theory.
A

[Wipneus]

I added LC80070122016096LGN00 bands 345 RGB below at comparable resolutions (not pan-sharpened with the 15 m). Quite a difference!

To demonstrate what pan-sharpening does to the Landsat images, here is a comparison (I did not try to get the colors the same as A-team's). Same Landsat product but with bands 3,4,5, and 8.

[notjonathon]

That's remarkable, Wipneus.

[Wipneus]

I found this in the new released Aster satellite archives. It is taken during the arctic night (late evening to be more precise) and originally very dark.  Of the 8 bits Aster sensors only 5-6 are actually used. After brightening it up it becomes a pretty picture anyway, with some unusual shadows.

You must click the image to see it full size.

If you think ESA makes it a big problem for users to access the images, rethink. Image data hidden in hdf files, not as images but raw binary data. After wasting my time trying following stale links, registering on sites to find out that additional permissions are needed and trying out crashing Java programs I had to write my own Python tool to get the data into image formats.

[Espen]

Nice stuff Wipneus, it is almost like a topographic map/image?

[A-Team]

After wasting my time trying following stale links, registering on sites to find out that additional permissions are needed crashing Java programs Image data hidden in hdf files, not as images but raw binary data.

Hmm, looks like they have this one at EarthExplorer ... maybe the hdf 'image' data is better (?). hdf has more than a learning curve, it is a whole way of life, 3rd image. It seems very reluctant to cough up a simple image.

Local Granule ID:AST_L1T_00307192012153010_20150611212008_72940
Coordinates:69.6258 , -50.3503
Acquisition Date:2012/07/19

Aster ... I'm still trying to determine what that NASA's big press release was actually all about. I tried my luck on Aster's home page which is all in Japanese, just clicking on this or that until the screen changed, worked ok. Supposedly this very same Aster should be on EarthExplorer but that is loaded with caveats.

I would say 99.9% of all satellite data goes to waste, never used by anybody. It's been like that for decades. Private-public partnership failed monetization schemes, large volume of huge files, registration obstacles, incompetent agency portals, restrictions on users, seriously obscure file formats, uninformative web pages, minimal wiki entries etc but mostly it's indifference.

That's a nice photo. Note the strong sun shadows coming in from northwest, unusual sun angle. Great for topography. Did they have anything around the 15 Aug 15 big event or is 19 July 2012 the latest? Looks that way from EarthExplorer portal.

We've previously determined from Landsat that the dark lines in the second image are water-filled crevasses by late July.

[charles_oil]

When Sentinel 2B is launched (June) frequency is meant to go to every 5 days - will this help & smooth the timelapse images or will it be hard to integrate? Presumably alternate frames will be from different satellites.

[A-Team]

iwantatr8 writes this is a complex non-Newtonian fluid interaction taking place: a side icestream with turbulent interactions at is junction with main glacier flow. The second influence is descending ice falling over the mound causing pressure at the edge of the turbulent flow giving rise to a turning moment  at the edge of the stream, pulled to an elongated feature along the edge. a very slow horizontal vortex acting along the ice stream edge.

Interesting ideas there, consistent with observed feature re-establishment after cycles of calving front retreat and subsequent advance. I wonder if even 10 m resolution is adequate for these edge features. I didn't follow explanation of elongation / small bumps / how these ideas can be developed further. Starting with the ambient velocity vector field, which we could have at least in plan projection.

Other than the mound, what role does bedrock topography play here? The 2nd image shows available ice penetrating radar through the feature. It is very important here to abandon the inclined flat slab model of the Jakobshavn calving area -- the surface is quite contorted and always for a good reason.

There is a similar dynamic feature on the north branch, a hole on the side of another mound. For that one, Goog Earth was hosting a 1 m zoom showing liquid water in mid summer. These shadowy images are great for topography, attached Aster.

Note we have a larger slow motion 'horizontal vortex' at Petermann on the east side arising from the interaction of the main floating ice shelf with a couple of slower moving partly pinned side tributary mountain glaciers.

[A-Team]

When Sentinel 2B is launched (June) frequency is meant to go to every 5 days - will this help & smooth the timelapse images or will it be hard to integrate? Presumably alternate frames will be from different satellites.

Depending on the commissioning schedule, S2B might not be applicable to JI this melt season. It will bring on another round of orthorectification issues for sure. Whether all the imagery (including Landsat-8) can be brought to a happy common denominator (precision commensurate with ground resolution and retention of sharpness) remains to be seen.

My sense is that they are doing this all wrong, ie basing off all the nitty-gritty satellite geometric factoids that 'should' work but don't quite go the distance.  Here I would toss the logic, mask out to fixed rocks, see what it takes empirically to get rid of their jitter (eg treat mis-fit as pseudo motion to describe with the Scambos 2016 algorithm) and apply that capture to re-sample the inverse mask. This will only be a local fix that won't remain spot-on farther inland.

In terms of smoothing time lapse movies, cloud cover is a big issue at Jakobshavn -- nothing worse than six weeks of summer blanket (though S1A is still on the job). However, with more frequent pass-overs, the odds improve for catching a window. Note rapid-fire returns are more of an issue for rapid motion that scarcely exists outside of Jakobshavn main channel. In the flanking regions, the 10-day orbital return would be satisfactory and a 5-day repeat fantastic. Glaciers elsewhere? An annual shot would suffice for some of them.

I do think we have to agitate for better portals and pipelines. We can't go on too much longer with boutique manual processing; people need to be re-purposed as analysts.

Below is a robotic animation of our first five S2A scenes. The images are cropped to a fixed upper left coordinate, width and height. Then each frame is auto-rescaled to full use of contrast based on its pixel value usage statistics, and the frame delay set at 180 ms /endless loop. There is no manual shifting of imaages -- registration out of the box is trusted. (We know this is very sub-optimal for different Landsat-8 paths.)

The second animation, made the same way, shows the movement in fixed rocks. The default registration is a good start but certainly not usable for velocity measurements.

The 3rd animation shows some action over 13 days on the north branch. It is interesting how the eye can still detect overall motion correctly despite the flawed registration -- I suppose it is no different keeping a steady eye out while
riding a bicycle down a cobblestone street.

[Shared Humanity]

The depression along the north wall of the south branch and nearest the calving face is open water, covered with bergy bits IMHO. This persistent feature is being caused by the underlying topography the ice sheet is navigating. Just to the north of this depression the ice sheet begins to fracture and slow as it grinds across a small island. While the sheet does make it over the island it is moving slower than the ice that goes around it. This is causing open water to appear.

[Shared Humanity]

The second depression that is along the north wall and further from the calving face is the result of a similar process as the ice sheet moves over the much larger western most section of the large island that is north of the sharp bend of the main ice stream.

[Shared Humanity]

By the way, these images are amazing!!!!!! You guys are very talented.

[A-Team]

Somewhere many posts ago (#989, 1342-44), the vertically exaggerated perspective of this feature, made from the 2 m digital elevation map of Howat, shows the standing waves and depressions without the ambiguity of nadir view.

If someone has the time to butt up the relevant two SETSM tiles and optimize the view for this feature, that would be great. The image below shows the western tile in the 3D Surface Plot tool of ImageJ. We also have a very fine evening oblique photo looking north to this area.

SETSM_ArcticDEM_18_39_2_1_DEM.tif
SETSM_ArcticDEM_18_39_2_2_DEM.tif

http://forum.arctic-sea-ice.net/index.php/topic,154.msg71580.html#msg71580 large head-on view

Since some people see things reversed, I joined the 5 scenes as they come with their inverse image. (This keeps them in synch as they don't always stay that way running as separated animations.) These are 3x magnifications of the original 10 m so are really jittery due to the default alignment used. Below that, an animation with the adjusted registration is shown.

[Shared Humanity]

Sure doesn't look like water there! 

[Iceismylife]

My thinking is a below sea level standing wave surface with enough fracturing to let sea water in that freezes and forms a new ice layer.  A sea water spring in a rive of ice.

[johnm33]

top from 575 Espen after Sidd  lower 588 A-Team
The area thats flat is where it drops from 100 to-1500m [or 10 to dark blue] in what ? 500m I'm guessing the ice is 1km thick and is dropping into that trough. Then from 510

for perspective and http://forum.arctic-sea-ice.net/index.php?action=post;quote=46167;topic=154.600;last_msg=73511 600 to compare

[A-Team]

Nice to pull these up. It is not so easy however to precisely locate the feature relative to the bedrock map. I'm skeptical that the physics here was ever considered in the Morlighem refinement of the bed map. It's necessary ultimately to go back to the original sparse radar data to not get ensnared by interpolation. The other resource we had was an oblique evening photo looking to the north which is quite unambiguous as to humps and dips.

The animation below shows the 25 Mar 16 with its orbital pair 03 Apr 16 enlarged to 1 m resolution for purposes of accurately measuring displacement velocity, assuming ESA had them correctly co-registered. It was not easy to find features that could be reliably tracked but the set of 4 blebs did move as a rigid block, 94 pixels = 94 meters in 10 days or 9.4 m/d or 3.4 km/yr which is about a third of the fastest flowline in the main channel (#1416).

thinking a below sea level standing wave surface with enough fracturing to let sea water freeze and form an new ice layer.

No question, the lack of crevasses is remarkable in this landscape. Flat and smooth -- is this a viable adventure campsite? Yes, it seems safe enough over the short term, better here than Barneo 2016. However tents would have to be relocated eastward every two weeks or so and the helicopter pilot should not be paid for the return trip in advance.

[johnm33]

Looking at this I'd revise my ice depth estimate to @550m so the 'cliff' would be about 400m the flat area still being about 150m asl, so a little higher than I'd expect if it's sitting in water.

from http://congrexprojects.com/docs/12c20_docs2/1-joughinesa.pdf?sfvrsn=2

[A-Team]

Nice find. A lot of interesting material in that Joughin powerpoint. The nicely tinted hillshaded DEM is based on the SETSM file mentioned in #1445 which can be found now at NSIDC. To get the full author-posted resolution of an image, try extract-images-from-pdf at ImageJ.

https://daacdata.apps.nsidc.org/pub/DATASETS/nsidc0645_MEASURES_gimp_dem_v1/30/
https://daacdata.apps.nsidc.org/pub/DATASETS/nsidc0645_MEASURES_gimp_dem_v1/30/gimpdem1_2.tif
https://daacdata.apps.nsidc.org/pub/DATASETS/nsidc0645_MEASURES_gimp_dem_v1/30/gimpdem1_2_hillshade.tif

This is not the new Jakobshavn 'article in preparation' mentioned but similar and the DEM may not be the Howat product but rather from the project described in the reference below. We have no access to Worldview imagery unfortunately:

Commercial vendors can collect on-demand, sub-meter imagery anywhere on the planet from multiple platforms, with revisit times of less than a day at higher latitudes. Resampled (0.25 m minimum GSD as of June 2014, formerly 0.5 m) image products are now available to United States federal employees and federally-funded civilian researchers [we got any here?] through the NextView license...

WorldView-1 and WorldView-2 share a similar pushbroom linescan camera with 11-bit dynamic range... This provides an effective swath width of 35,840 pixels, corresponding to ∼17.6 km at ∼0.5 m ground for nadir acquisitions... The  spacecraft acquire images at off-nadir angles from 0° to 45°, two or more images of the same target in a single orbital pass, forming an along-track stereo pair with typical convergence angles of ∼30–60° some 60 s apart so with the same illumination.

Only sub-meter  images resolve small-scale surface features (crevasses, sastrugi)  not apparent at lower resolution (e.g. ∼15–30 m Landsat imagery). This high-frequency texture enables precise image correlation for feature tracking and/or surface reconstruction.

Shean, DE O Alexandrov, Z Moratto, BE Smith, IR Joughin, CC Porter, Morin PJ. (2016), An automated, open-source pipeline for mass production of digital elevation models (DEMs) from very high-resolution commercial stereo satellite imagery, ISPRS J. Photogramm. Remote Sens, 116, 101-117, doi: 10.1016/j.isprsjprs.2016.03.012.
http://www.sciencedirect.com/science/article/pii/S0924271616300107 free full highly recommended