Jakobshavn Isbræ / Ilulissat Isfjord / West Greenland

[AbruptSLR]

The linked (open access) reference provides both a nice summary of Jakobshavn's recent behavior and effects to model that behavior:

Muresan, I. S., Khan, S. A., Aschwanden, A., Khroulev, C., Van Dam, T., Bamber, J., van den Broeke, M. R., Wouters, B., Kuipers Munneke, P., and Kjær, K. H.: Modelled glacier dynamics over the last quarter of a century at Jakobshavn Isbræ, The Cryosphere, 10, 597-611, doi:10.5194/tc-10-597-2016, 2016.


http://www.the-cryosphere.net/10/597/2016/

Abstract. Observations over the past 2 decades show substantial ice loss associated with the speed-up of marine-terminating glaciers in Greenland. Here we use a regional three-dimensional outlet glacier model to simulate the behaviour of Jakobshavn Isbræ (JI) located in western Greenland. Our approach is to model and understand the recent behaviour of JI with a physical process-based model. Using atmospheric forcing and an ocean parametrization we tune our model to reproduce observed frontal changes of JI during 1990–2014. In our simulations, most of the JI retreat during 1990–2014 is driven by the ocean parametrization used and the glacier's subsequent response, which is largely governed by bed geometry. In general, the study shows significant progress in modelling the temporal variability of the flow at JI. Our results suggest that the overall variability in modelled horizontal velocities is a response to variations in terminus position. The model simulates two major accelerations that are consistent with observations of changes in glacier terminus. The first event occurred in 1998 and was triggered by a retreat of the front and moderate thinning of JI prior to 1998. The second event, which started in 2003 and peaked in the summer 2004, was triggered by the final break-up of the floating tongue. This break-up reduced the buttressing at the JI terminus that resulted in further thinning. As the terminus retreated over a reverse bed slope into deeper water, sustained high velocities over the last decade have been observed at JI. Our model provides evidence that the 1998 and 2003 flow accelerations are most likely initiated by the ocean parametrization used but JI's subsequent dynamic response was governed by its own bed geometry. We are unable to reproduce the observed 2010–2012 terminus retreat in our simulations. We attribute this limitation to either inaccuracies in basal topography or to misrepresentations of the climatic forcings that were applied. Nevertheless, the model is able to simulate the previously observed increase in mass loss through 2014.

[A-Team]

he first event occurred in 1998 The second even peaked in the summer 2004.We are unable to reproduce the observed 2010–2012 terminus retreat in our simulations. inaccuracies in basal topography.

I must be in a different zone, it's 2016 here.

Landsat did get a shot of the calving front on day 78, 2016 LC80090112016078LGN00. There are some badly situated bright clouds and additionally dark shadows from them but the calving front area is visible. There are not close matching dates from 2013-15, see post #1339. Sentinel 2A so far has only gotten a region to the south.

The calving front is fairly far back with two minor regions about to calve. Probably the most interesting aspect of the image is the 8-9 different textures associated to differently sourced ice making up the ice stream. Relatively little of the calving front is comprised of fast ice from the deep interior.

[nukefix]

Hey A-Team are you also tweeting those images or how do you always get 40+ views in a heartbeat?

[A-Team]

are you also tweeting those images or how do you always get 40+ views in a heartbeat?

No tweeting at my end but a third party could be re-purposing. Espen has pursued other channels that would get us wider distribution. Forum stats show 51 guests (non-registrants) on the site now but presumably spread out over all the forums. Jakobshavn is one with a following -- climate change tourism, scientific interest, calving events, sea level issues.

Usually it takes a whole day to get 40 views and months to max out at in vicinity of a thousand. However I am very skeptical of the forum's view counting. That is, if I put up 4 images with some text, they end up with very different counts yet anyone loading the page would load all of them.

However someone with a blog or twitter following could link in directly to ones they thought were interesting. If so they're not consistently linking to the best of the four from my perspective. It is not likely that these could be tracked down with reverse image search as that is spotty even with imageRaider.

Alternatively, google image search with 'Jakobshavn glacier' will show forum image links as well as re-appropriated forum material. Which is fine, everything I post is open source. Rarely there is embedded attribution in the corner a la Wipneus or Espen. I hardly ever put in fourier transform watermarks.

As far as I am concerned, the purpose here is outreach, not barriers; if the images had any commercial value, they would be replicated from the inspiration, even the glacier animations. Scientific plagiarism is another matter and there attribution is mandatory, even if it is just for a research direction.

[Xulonn]

The Jakobshavn calving event in James Blalog's documentary film "Chasing Ice" is riveting and spectacular on my 40" LCD TV, and must have been even more incredible in a movie theater. 

I don't think anyone who has seen the video of that long, drawn-out event event will ever forget it.  The camera angles,excellent sound and the ponderous slow-motion visual effect of that event makes every other calving event video I have watched seem insignificant.  For me, Jakobshavn has become an icon that represents the significance of the melting of the world's land ice. 

Here is a link to the official YouTube clip from the film. 
 

[AbruptSLR]

The Jakobshavn calving event in James Blalog's documentary film "Chasing Ice" is riveting and spectacular on my 40" LCD TV, and must have been even more incredible in a movie theater. 

I don't think anyone who has seen the video of that long, drawn-out event event will ever forget it.  The camera angles,excellent sound and the ponderous slow-motion visual effect of that event makes every other calving event video I have watched seem insignificant.  For me, Jakobshavn has become an icon that represents the significance of the melting of the world's land ice. 

Here is a link to the official YouTube clip from the film. 
 

If climate change continues at its current rate, even more astounding videos of cliff failures and hydrofracturing in the 50-km long Thwaites Glacier gateway may be available by 2050.

[Espen]

Am I the only one having problems with the Sentinel Data Hub?

[Espen]

Jakobshavn Isbræ update, more calving activity is seen in the southern branch of Jakobshavn and another large chunk of ice is ready to take off:

Please click on image to enlarge and animate!

[A-Team]

Below is our very first Sentinel 2A of Jakobshavn. This is band 3 (green) at 10 m processed non-optimally as 8 bits jpeg2000. It is notable sharper than the 15 m Landsat LC80090112016078LGN00_B8 in #1353 or as seen below. Note Jakobshavn is about to calve as can be seen in the Sentinel 1A radar image Espen posted above for 22 Mar 16.

To make a fair comparison, it takes a cloud-free day where all three satellites took in the calving front at a similar time of day, with all three optimally processed at 16-bit to the same geometry. The Landsat can then be bumped up to Sentinel 2A resolution and subtracted to show where and how image quality varies.

Does resolution matter? It does, but the advantage varies by glacier. For Jakobshavn, we would be looking for pre-calving indicators, water in crevasses, input from side features and seasonal velocity changes. At Petermann, the interest would be in crack propagation within the ice shelf, not so much in where it meets the ocean. At Zachariae, early signs of crevassing. At Pine Island, for issues as discussed at that forum.

For most purposes, the trade-off between better resolution but exploding file and image sizes reaches a point of diminishing returns. Thus Worldview-3, which we don't seem to have access to, has 20x better resolution than Sentinel 2A. However that means 400 times the file size. To display Jakobshavn's 5 km wide calving front with 0.5 m pixels takes a 10,000 pixel wide image, requiring lots of scrolling even on a 3x3 array of largest retinal resolution monitors.

We are still wrestling over at the Zachariae forum with how to locate and download these images efficiently at the Amazon AWS mirror found by nukefix, which is somewhat funky but not nearly as funky as the ESA source. Here we want only bands 2-4 and none of ESA's auxillary files. That reduces the download from a gross 8.3 GB to 186 MB per band, a 98% reduction in aggravation.

While the 3 bands can be composited into a color RGB image, there is ordinarily very little color on an ice sheet. Looking at a limited sample, it appears that the red band alone B4 has the best contrast, notably in darker or shadowed regions. Thus for most forum users, this band alone can suffice.

The procedure is to zoom in on the Amazon map to the Jakobshavn calving front, see what coverage is available, collect the full file name and use it to construct the band 4 download link. One level of the folder hierarchy has to be guessed at -- if the '0' in orange below doesn't work, try '1', '2' etc.

S2A_OPER_PRD_MSIL1C_PDMC_20160323T234457_R068_V20160323T151927_20160323T151927_T22WEB
http://sentinel-s2-l1c.s3.amazonaws.com/tiles/22/W/EB/2016/3/23/0/B04.jp2

[nukefix]

Hmm since the resolution is 10m shouldn't you process in slightly better pixel-size (7.5m?) in order not to lose detail in resampling? Or do you even need to resample?

[A-Team]

since the resolution is 10m shouldn't you process in slightly better pixel-size (7.5m?) in order not to lose detail in resampling? Or do you even need to resample?

These particular channels are posted as 10 m pixels, there's been no resampling above. Or rather ESA did it already in the course of sensor data processing and there's no practical access to rawer data (which is just as well as it's hard to improve on the instrument designer).

They did resample up from 12- to 16-bit with bicubic which does not change the size of pixels in the image, only the number of 'decimal points' in the precision.

I think it would still look quite nice down at 5 meters. There are four common ways to get there but not clear how to measure which is best, probably sinc or bicubic. There may or may not be features of interest at Jakobshavn where this would matter.

Some of the background imagery at Google Earth is 1 m; while not snowflake, snowball, or snowman resolution, it is perhaps distracting in its detail. Note Howat's DEM is at 2 m horizontal based on Worldview 1&2. Big tiles, big files.

Ground resolution has somewhat dodgy history of definitions. In the old days, it was the minimal separation of checkerboard squares or parallel lines that could be distinguished painted on a large groundcloth spread out below. However an ice sheet has local contrast issues that override black and white.

Obviously if it's almost pure white like the interior, you can't see anything even at 0.1 m. If it's not pure white but say whites binning between 230 and 255 in 8-bit, there's some room to stretch the contrast out so features become recognizable. The histogram indicates the opportunities in some sense but it is a global statistic whereas contrast is local. ImageJ has some windowing tools that collect the local statistics that determine the information-theoretic end game.

It is imperative to do resampling, contrast enhancement, fourier bandpass filtration, rotations, PCA components, etc etc at 16-bit before dropping down to displayable 8-bit. The 12-bit sensors on Landsat-8 and Sentinel 2A don't sound like a big deal but they represent a huge technological advance over the 8-bit sensor on Landsat-7 and earlier, where round-off errors kill you.

The 5 m below was resampled at 8-bit with bicubic; if Wipneus can post the 16-bit of this band 4, we could subtract them and blow up the differences. Over at Zachariae, he just put the 3x16 bit RGB 10 m scene up on google drive for download, 800 MB.

 https://drive.google.com/file/d/0B8NUHa4P2gmlRW5RZExiX0hHeTg/view?usp=sharing

[Espen]

Jakobshavn update:

A small calving seen between March 18 and March 25 2016:

[Iceismylife]

And here is a real big one for the season:

video
of the event

[A-Team]

Here is that same Landsat from 26 Mar 15 adjusted to 10 m resolution, LC80100112016085LGN00_B8. This is an exceedingly clear image. I had wanted to compare it to the 10 m Sentinel 2A of a similar date (above) but that day had faint cirrus clouds making the comparison a bit unfair (the Landsat looks better).

Jakobshavn cannot be seen at all on the 27 Mar 16 Landsat because of thick clouds, though the radar of Sentinel 1A could penetrate this, scenes are only available for 23- 24- and 25-03-2016 according to http://ocean.dmi.dk/arctic/disko.uk.php. Note DMI does not process the contrast correctly on these previews so the calving front and fjord are completely blown out (pure white, 3rd image) even though the original Sentinel 1As are have excellent intrinsic contrast (4th image).

The Jakobshavn Landsats are mainly UTM22 mercator as are the Sentinel 2A (but not 1A). Thus it should be feasible to mix and match them in time series without one of them taking a hit in quality from re-projection. After 2B launches, this means a scene every 3-4 days. Note EarthExplorer is not keeping up at all with posting Sentinel 2As and is not reprocessing them for consistency with their Landsats.

The product options from USGS include a Full-Resolution Browse (FRB) image product generated by USGS, along with a 100Km by 100Km tile-based Level-1C top-of-atmosphere (TOA) reflectance product that is very similar (but not identical) to the currently distributed ESA Level 1C product (https://earth.esa.int/web/sentinel/user-guides/sentinel-2-msi/processing-levels/level-1).

The USGS Earth Resources Observation and Science (EROS) Center repackages each Sentinel-2 product bundle on a per tile basis, resulting in a file size of approximately 650 MB. Each Level-1C product is a 100 km x 100 km tile with a UTM/WGS84 (Universal Transverse Mercator/World Geodetic System 1984) projection and datum. The Sentinel-2 tiling grid is referenced to the U.S. Military Grid Reference System (MGRS (http://earth-info.nga.mil/GandG/coordsys/grids/mgrs_100km_dloads.html)).

The Landsat shows quite clearly the bands contributing to the calving front but no persuasive indications of the next piece to be calved. In the past, that has not seem correlated with any particular band or group of bands, even though some of these should have weaker ice than others.

[Espen]

Jakobshavn Isbræ update:

Due to missing data from March 13 2016 (suspect a human error probably due to the fact the data is released manually?), so only March 1 and March 25 is shown.

Please click on image to enlarge and start animation!
 

[AbruptSLR]

Jakobshavn Isbræ update:

Due to missing data from March 13 2016 (suspect a human error probably due to the fact the data is released manually?), so only March 1 and March 25 is shown.

Based on rate of retreat of the calving face, one might think it was summer instead of spring.

[oren]

Jakobshavn Isbræ update:

Due to missing data from March 13 2016 (suspect a human error probably due to the fact the data is released manually?), so only March 1 and March 25 is shown.

Please click on image to enlarge and start animation!
 

Superb animation. Clearly shows signs of calving on all fronts: north branch, main south branch, and north wall of south branch.

[A-Team]

Same but just the calving front at its posted 10 m resolution and a slight fix to registration. Sentinel 1A doesn't stick to its 12 repeat for GRD IW with much regularity; it was overhead but operating in some other mode on 13 Mar 16. Data cannot be not manually dropped as it is all in a pipeline over to the pipelines of certain heavy users as you can see from the kml metadata.

During the 24 days between scenes, Jakobshavn would have advanced ~30m/d x 24d = 720 m or 72 pixels as shown by the white point growing to a 72 pixel white line. It is not so easy to recognize features and follow their movement on Sentinel 1A pairs; Sentinel 2A is a lot more favorable for this.

[johnm33]

This might indicate an extreme breakout of WGSC in Disko and its associated turblence. If it's that warm at the surface the water passing over the sill will be much warmer [well maybe 5C] and should cause accelerated melt at the base of the calving front. Particularly interesting as tides are nowhere near peak.

[A-Team]

Always good to provide a source link for image data, the file name ss+(2016-03-29+at+08.56.36) above is opaque. The West Greenland Shelf (or Slope) Current does not reach Disko Bay. Did you intend West Greenland Current WGC instead of WGSC?

I find it implausible that anything unusual is happening in Illulisat Fjord or Jakobshavn Isbrae in March. The task really is to understand where the data is coming from and what could give rise to artifacts in it.

The Davis Strait provides a convenient choke point for monitoring temporal and spatial variability of Arctic outflow while also characterizing a critical upstream boundary condition for Labrador Sea convection. Fluxes through the Strait represent the net integrated Canadian Archipelago throughflow, over 50% of the Arctic's liquid freshwater discharge, modified by terrestrial inputs and oceanic processes during its southward transit through Baffin Bay.

By the time they reach Davis Strait, Arctic waters already embody most of the transformations they undergo prior to exerting their influence on the deepwater formation sites in the Labrador Sea. An ongoing program has characterized Davis Strait volume, freshwater and heat flux since September 2004. Measurements include continuous velocity, temperature and salinity time series collected by a moored array, autumn ship-based hydrographic sections and high-resolution sections occupied by autonomous gliders.

These data show large interannual variability in volume and freshwater transport, with no clear trends observed between 2004-2010. Average volume, liquid freshwater and sea ice transports are -1.6 +- 0.2 Sv, -93 +- 6 mSv and -10 +- 1 mSv, respectively (negative indicates southward transport).

Large-scale atmospheric teleconnections, such as the AO and NAO, correlate poorly with Davis Strait volume transport and are likely only an indicator of transport variability when the indices are strong. http://tinyurl.com/hujt374

[johnm33]

There's an animation here of the eddies breaking out and heading north, after day 215.
http://www.oc.nps.edu/~pips3/anim/eke_anim.html
I meant the slope current, I agree mostly that stays south constrained by the bathymetry of the 'Labrador basin'[?] but given the right wind and combination of tidal forcing instead of forming the Irminger rings it breaks out climbs the shelf and scours whatever it passes over or into. Try as I might the tidal elements remain opaque.  http://www.esr.org/AOTIM/ArcticTides_CoeffMaps.html
This paper which I think I found upthread explains it. http://ocean.dmi.dk/staff/mhri/Docs/Holland_et_al_NatureGeoscience2008.pdf
The harbour at Illulisat gets occasional intrusions of turbulent water, less frequently recently. So there may be some action there.

[AbruptSLR]

The linked abstract discusses the impacts of the basal ice meltwater discharge from beneath Jakobshavn:

http://www.igsoc.org/symposia/2015/cambridge/proceedings/procsfiles/procabstracts_68.html


73A1816
Parameterization of melting along the calving face of Jacobshavn Glacier, Greenland
Pierre Mathiot, Adrian Jenkins, David Holland
Corresponding author: Pierre Mathiot
Corresponding author e-mail: piethi@bas.ac.uk
Today ocean models at climate resolution (O(1–10 km)) are not able to compute the melt rate driven by subglacial discharge along the calving face of a tidewater glacier such as Jacobshavn Glacier. The reasons are simple: the processes along the calving face need a very high horizontal resolution O(1–10 m) to be explicitly simulated, and subglacial discharge is neglected in most ocean models. However, this deep input of fresh water has two major consequences. The first one is that the waters confined within the overdeepened basin of a fjord can be renewed in summer by a vigorous overturning circulation driven by the subglacial discharge and the associated, enhanced terminus melting. The second one is that the elevated melt rates may contribute to glacier recession in Greenland, which is focussed on its marine outlet glaciers. In order to incorporate these processes in a relatively coarse-resolution ocean model (~5 km), we implement a parameterization based on the theory of buoyant plumes in unstratified environments. Although Greenlandic fjords are far from unstratified, their structure is more accurately approximated using two or three well-mixed layers separated by sharp interfaces than by a constant linear stratification. In an unstratified environment there are two length scales that govern plume behaviour: one based on the ratio of the initial input of buoyancy to the buoyancy input by melting; the other based on the change in the freezing point with depth. The first is appropriate for times of high subglacial discharge, while the latter applies when the discharge is low or non-existant. Suitable nondimensionalization of the plume equations using these length scales allows the melt rate to be approximated reasonably well by simple polynomial functions. The melt rates computed by these parameterizations and the subglacial runoff are included in an ocean model using techniques recently developed to include the melting of floating ice shelves around Antarctica. As a proof of concept, this method is implemented in an idealized fjord configuration at coupled climate model resolution (~5 km) and compared with observations and results from high-resolution ocean models (~10 m at the calving face).

[Wipneus]

I have outlined how big the latest calving was in this Sentinel 2A animation.

[A-Team]

Nice!

Good practice to put the red boundary in a separate gif frame so that the original photo for that date is still available as download without the red. Note Wipneus intended 2016 rather than the 2015 shown (easily fixed by backing Gimp up to the floating text).

I am wondering though if the latter date is actually 26 Mar 16 rather than the 27th (which appears cloudy on the AWS site). http://sentinel-s2-l1c.s3-website.eu-central-1.amazonaws.com/#tiles/22/W/EB/2016/3/26/0/ is clear however and brackets the calving event more tightly. Note if the dates aren't 12 days apart, they won't have the same geometry and won't be alignable without passing through SNAP or similar re-projection.

Below is your 2016 band tour of Sentinel 2A plus a couple of RGBs made from various combinations of the 23 Mar 16 scene, which was cloudier than I had realized and not suitable for Landsat comparison.

ESA vs Sentinel public image browser at AWS have propagated some confusion over the two band 8's, calling them various variatons on B8A, B8B, B08. The ones we want are primarily the 10 m resolution bands, which can be identified by their larger file size.

To make the band tour, I used the AWS bands. These are exceedingly dark and only a fraction of the grayscale bins are utilized. Using the default 'Auto' at gimp 'Normalize' for consistency, these were brought into better utilization. However AWS only provides 8-bit data which is unacceptable.

http://sentinel-s2-l1c.s3-website.eu-central-1.amazonaws.com/#tiles/22/W/EB/2016/3/23/0/

[A-Team]

The area calved is not so easily calculated because the glacier is moving forward unevenly at the front over the three days that have elapsed, with the center is advancing faster than the sides. This means the actual calving front is upstream from the red line in Wipneus' post above.

However neglecting this offset's addition, the area pencils out to 1.70 km² which seems quite substantlal for March. Glaciologists have not kept records of historic calving data for Jakobshavn and indeed, see abstract #1373, don't even study it anymore when they study it.

Supposing the fastest part is moving 30 m/d but the average over the 4 km front is 20 m/d, the offset would be 0.240 km² for the 3 days or another 14% for 1.94 km². But since this is not a 12-day orbital pair, the scene times are needed as they won't differ by 3.0 days.

The second image shows a few feature displacements over the three days. It would be necessary to 'connect the dots' of those right at the calving front at the later date to add in the effect of moving. For that, it would be helpful if Wipneus could post the two band 4 (reds) as separate 16-bit tifs at his cloud site as features look more the same if all operations (enlarge to 5 m, rotate, match contrasts, lift clouds on 23rd) are done at 16-bit.

The AWS links for the two dates:

http://sentinel-s2-l1c.s3-website.eu-central-1.amazonaws.com/#tiles/22/W/EB/2016/3/23/0/
http://sentinel-s2-l1c.s3-website.eu-central-1.amazonaws.com/#tiles/22/W/EB/2016/3/26/0/

[Wipneus]

For that, it would be helpful if Wipneus could post the two band 4 (reds) as separate 16-bit tifs at his cloud site as features look more the same if all operations (enlarge to 5 m, rotate, match contrasts, lift clouds on 23rd) are done at 16-bit.

Will do so and understood. The two files are uploading now which will take a considerable amount of time.

[Wipneus]

Here are the links:

https://drive.google.com/file/d/0B8NUHa4P2gmlajk2MjA0ZHlPcUE/view?usp=sharing
https://drive.google.com/file/d/0B8NUHa4P2gmlUkJNU0RGS0xKQmM/view?usp=sharing

[A-Team]

Thanks so much! The download is very fast but for me upload would be very slow.

Some warnings but I clicked on through... ImageJ chokes on 15-bit but it imports as a 16-bit using 'Bioformats'. Size is showing up as 230 MB. The snippet shown below is unaltered, 8.3% of full size.

That size is workable for nearly exact co-registration by stack-and-crop together for 2-3 such images (at 10-day orbital repeat intervals) and so RGB false color from three such dates.

Google Drive can't scan this file for viruses.
S2A_OPER_PRD_MSIL1C_PDMC_20160323T234457_R068_V20160323T151927_20160323T151927.22WEB.B04.tiff (216M) exceeds the maximum size that Google can scan. Would you still like to download this file?

As Wipneus explains a few posts below, the correct way to get the time difference between scenes is
T153310 means 15h33m10s and T151927 15h19m27s in UTC so the two images differ 3 days and about 14 minutes.  Sentinels 2A orbital repeat is 10 days, not 12 like Sentinel 1A nor 16 like Landsat 8.

S2A_OPER_PRD_MSIL1C_PDMC_20160330T070547_R111_V20160326T153310_20160326T153310.22WEB.B04.tiff
S2A_OPER_PRD_MSIL1C_PDMC_20160323T234457_R068_V20160323T151927_20160323T151927.22WEB.B04.tiff

[A-Team]

Whoa! This Sentinel 2A imagery is just fabulous.

[oren]

WOW!

[Neven]

Wow, indeed. You can see a guy picking his nose. 

[A-Team]

Wow, indeed. You can see a guy/gal picking his/her nose.

The mystery to me is why we cannot see where it is going to calve 3 days 14 minutes [fixed] before it happens. Not seeing any fault line running across the 23rd, below the 5 m processed from the 10 m at 16-bit (enabled by Wipneus Cloud Service).

Using feature tracking on the 26th, I'll post where the interior crack 'should be' in a bit. There's no shortage of false positives -- plausible interior candidates that don't correspond to anything on the final calving front of the 26th (but could represent intermediate stages). Not that we know the event 'stopped' there -- it has been to cloudy for anything but Sentinel 1A radar.

Perhaps the not-quite-calved piece fell off, setting a rapid series of unbuttressing events in motion, leading to the impression of a large single event. In fact Jakobshavn is steadily calving small bits and pieces day and night so if the satellite snapshots are at a scale of days rather than hours, we get the impression of a massive calving, even though here there is no evidence in this instance of massive bergs.

Alternatively, recalling of the mid-August 2015 mega event, Jakobshavn may have become more susceptible to triggering of event cascades, say from enhanced internal fragmentation attributable to recent acceleration (or the usual suspects: warming, meltwater, bedrock profile).

While we got lucky with a ecotourist video of a 'complete' calving event, it would be great to have a drone overhead continuously in June - July - August. That's a stretch because an ordinary drone can only stay aloft 25 minutes. Skipping over the frequent katabatic winds, there are options under development:

Boston-based aerial robots company CyPhy Works, however, has created a drone called Parc that can stay in flight indefinitely (yes, indefinitely).

Equipped with six rotors and a HD camera capable of producing infrared footage for night vision, Parc can be launched in the sky to perform aerial surveillance and stay aloft for hours, days, or as long as you’d like. It uses a “microfilament” that transmits power and data. Although this limits the distance it can travel, it makes it perfect for home monitoring.

The microfilament tether is thinner than a headphone cable but strong enough to reel the drone in if necessary. The drone can be set to fly completely automatically at a specific altitude.

[solartim27]

I see a crack running along the circled area, which seems to match well with the recent calving.  If I am correct, then I think this could be a very active summer from similar indications running up the glacier.

[Wipneus]

...  2.4 days before ...

That reminds me about this:

T153310 time of 2016-03-26 was imaged
T151927 time of 2016-03-23 was imaged
1383 minutes? [subtracted numbers]
57.625 hours? [divided by 60]
2.40 days? [divided by 24

It is much simpler, T153310 means 15h33m10 and T151927 is 15h19m27s. That is time of day (in UTC)

So the two images differ 3 days and about 14 minutes.

Further the Sentinels 2A's orbit repeats after 10 days, not 12 as S1A or 16 for Landsat. The wider swath width enables faster repeats.

[A-Team]

that reminds me

Thx, I fixed those above to avoid confusion. Not rocket science to convert UTC to local scene center time; there was talk about 10:30 am being better than noon (fewer clouds). Sun angle and azimuth should be somewhere in the metadata.

That is an interesting point about the wider swath meaning more frequent re-visits. Swath width comparisons show 2A having 156% more width than Landsat-8. This is maybe as big an advance as the 10 m resolution (Landsat 8 also has 12-bit sensors).

Sentinel 2A: 290 km
Landsat 8: 185 km
Sentinel 1A IW GRD: ?
LANDSAT 5 TM: 185 km
LANDSAT 7 ETM: 185 km
SPOT-5: 120 km

'The Landsat 8 pushbroom sensor aligns the imaging detector arrays along the focal plane allowing it to view across the entire swath, 185 kilometers cross-track field of view, as opposed to sweeping left to right across the field of view. With  7,000 detectors per spectral band, this design results in increased sensitivity, fewer moving parts, and improved glacier surface information.'

I see a crack which seems to match well with the recent calving. this could be a very active summer from similar indications running up the glacier.

So far that lineation does not seem to mark the limiting line (which converges to the SW rocks), see preliminary blue/red feature tracking in #1376. We are in a good position to track the other circled pre-cracks if the clouds lift and stay lifted, they won't show on 1A radar.

I believe you are right about a very active summer at 2016 Jakobshavn -- this is just the end of March (though it's unclear how to compare year to year quantitatively). That doesn't necessarily imply record calving line retreat as more rapid advance could compensate for more massive calving front losses.

It will still be some years (?) before the calving front rounds the elbow curve and is freed from the buttressing forced by the rock mounds there. Unless we are witnessing a new stage in internal mechanical strength degradation.

[Tealight]

I created a RGB image from bands 2,3,4 using a standard RGB Profile and not the specific "Sentinel2 natural colour" profile. It makes the image slightly too blue, which is great to distinguish between areas with snow and bare ice.

The first attached image shows the break up of the continuous snow cover around 24km south-east of the calving front.

The second image shows the northern calving front with what I think is erosion from bedrock. I have noticed a dark line originating from the northern front before, but I wasn't exactly sure what it was.

[A-Team]

I created a RGB image from bands 2,3,4 using a standard RGB Profile and not the specific "Sentinel2 natural colour" profile.

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?

At some point, the bandwidth of the three sensors and the sensitivity across these bandwidths would come into play. These bands are only RGB initially in some sort of metaphorical sense. They certainly don't correspond to human opsin peaks and bandwidths, nor Landsat. And I wouldn't want to assume too much consistency across display devices used by ~900 registrants.

At the end of the day however, it is whatever mix of bands that resolves something that is useful to resolve. We are perhaps not so interested in replicating the view out the window of an airplane, yet people do have a lot of experience in interpreting eyeball colors.

The 'standard RGB profile' described at wiki seems geared to cathode ray tube monitors. Meanwhile Gimp provides color channel mixing under Colors --> Components but someone would have to know where they were going with it.

https://en.wikipedia.org/wiki/SRGB
http://www.kenrockwell.com/tech/adobe-rgb.htm
http://www.dpbestflow.org/color/color-space-and-color-profiles#profile

Yes, that glacier flour has been eroding from the same spot for years. Seems like it should sink so it must be embedded in calving ice (though if scraped, why not on the bottom). It has never been sampled despite the interest in the geology of Greenland and the assertion that two very old formations are fused more or less along the Jakobshavn trench (explaining its location?).

I struggle to understand what is so almighty difficult about sampling this, sampling blue ice from the big calvings, or at least measuring the temperature profile with depth of the south branch ice. They have rent-a-helicopters right there in Ilullisat. Better to stay home and write another model paper about an idealized isbrae, #1352 ?

[Martin Gisser]

I'm struggling to keep up with all the fascinating stuff here. And I regret having not enough time-energy and equipment to contribute a little to the visuals. 15y back I was a Gimp artist.

Now - if only... I would love to see some illustration of this:

It will still be some years (?) before the calving front rounds the elbow curve and is freed from the buttressing forced by the rock mounds there. Unless we are witnessing a new stage in internal mechanical strength degradation.

That is, a picture or two from above combined with a diagram of the bedrock topography. Maybe that could be a spectacular ASIF blog contribution without much words.

Will this Jakobshavn thing be sort of a phase transition? Almost as important a milestone as the first sea ice free summer?

[Wipneus]

There is an excellent image from 30 April. I have uploaded the B04 as tiff to:
https://drive.google.com/file/d/0B8NUHa4P2gmlMGc4LVBXc3E0Wnc/view?usp=sharing

[A-Team]

There is an excellent image from 30 March 16. B04 tiff below:

The clarity is incredible.

2016 04 01 T035851 2016 03 30 T150931 2016 03 30 T150931 22WEB.B04.tiff
<SENSING_TIME metadataLevel="Standard">2016-03-30T15:09:31.461Z</SENSING_TIME>

Like Landsat-8 scenes path,row 9,11 and 10,11 (but not 8,11 and 8,12) so maybe alignable by pixel shifting:

<HORIZONTAL_CS_NAME>WGS84 / UTM zone 22N</HORIZONTAL_CS_NAME><HORIZONTAL_CS_CODE>EPSG:32622</HORIZONTAL_CS_CODE>

A lot more information than I wanted about sun elevation and azimuth but I suppose it varies across the wide pushbroom swath. This very large kml file has data for each of the 12 bands and 12 detectors.

<Tile_Angles metadataLevel="Standard"><Sun_Angles_Grid>
<Zenith><COL_STEP unit="m">5000</COL_STEP><ROW_STEP unit="m">5000</ROW_STEP><Values_List>23 x 23
...
<Azimuth><COL_STEP unit="m">5000</COL_STEP><ROW_STEP unit="m">5000</ROW_STEP><Values_List>23 x 23
...
</Sun_Angles_Grid>

<Viewing_Incidence_Angles_Grids bandId="0" detectorId="1"><Zenith>
...
<Viewing_Incidence_Angles_Grids bandId="12" detectorId="12">

Hooray, mean values are supplied (but are 13 decimal points warranted?) It's taken from due south. 90º - zenith angle would give Landsat azimulthal convention, here 25.2º.

<Mean_Sun_Angle><ZENITH_ANGLE unit="deg">64.8342914108793</ZENITH_ANGLE><AZIMUTH_ANGLE unit="deg">177.428408973876</AZIMUTH_ANGLE>

[A-Team]

Here is a comparison at 5 m of the calving front on two recent clear days, 26th and 30th March. The front has advanced in the March 30th scene more than it has calved off. The first animation includes rocks to south because slightly different viewing geometry occurs when the scenes are not 10 days apart -- the alignment cannot be made perfect by simply shifting.

The second animation is in the co-moving frame of the Jakobshavn glacier, as has been used to good effect by Wip at Zachariae. Again, it is not feasible to reconcile the different imaging geometries over the 4 km vertical image at 5 m scale. As this viewing frame undoes glacier advance, the difference between the 26th (front farther left) and 30th represents minor calving.

The center of the glacier has advanced 23 pixels or 115 m in 4 days for a velocity of 29 m/day or 10.5 km/yr. This could be compared to velocity measured in late March 2013 in Joaghin 2014 though real change is problematic given different imagery and methodologies. For S2A, the greatest gain in accuracy would come from orbital scene pairs.

[Shared Humanity]

I'm struggling to keep up with all the fascinating stuff here. And I regret having not enough time-energy and equipment to contribute a little to the visuals. 15y back I was a Gimp artist.

Now - if only... I would love to see some illustration of this:

It will still be some years (?) before the calving front rounds the elbow curve and is freed from the buttressing forced by the rock mounds there. Unless we are witnessing a new stage in internal mechanical strength degradation.

That is, a picture or two from above combined with a diagram of the bedrock topography. Maybe that could be a spectacular ASIF blog contribution without much words.

Will this Jakobshavn thing be sort of a phase transition? Almost as important a milestone as the first sea ice free summer?

I believe this image just posted highlights the trigger for a major new approaching retreat of the calving face. We've all watched how rapidly the north wall of the south branch has retreated over the past 2 years. This retreat has removed a pinning effect on the main ice stream which has resulted in a fairly stable position for the calving face as it moves forward and calves.

If you look at this image, there is clear evidence that the north wall of the main stream that is southeast of the slowly emerging island is set to retreat as well. On the top left portion of this image, you will see a very stable feature, a visible slump in the ice. If you look closely, it appears that the ice has disappeared completely, replaced by a small portion of open water, with visible ice melange floating on it. There is no longer any real pinning effect here on the main ice stream. If you follow southeast along the edge of the main ice stream, it appears that a second pronounced slump is appearing as a result of the north wall retreat. As this retreat continues, the entire northern edge of the main ice stream will become unstable and a dramatic, irreversible retreat of the calving face will result.

[A-Team]

I could contribute a little to the visuals. 15y back I was a Gimp artist. illustrate: calving front rounding the elbow curve, a picture from above over a diagram of newly exposed bedrock topography.

That wouldn't be too hard, most involves matching scales and projections. We have a nice bedrock colored contour from sidd some posts back. After adjusting scale and maybe rotating, set Sentinel as top layer in gimp, add alpha transparency, use free select tool on ice stream back past elbow, delete, flatten, and post, very roughly below.

Will this Jakobshavn thing be sort of a phase transition? Almost as important a milestone as the first sea ice free summer?

Interesting question: will Jakobshavn become a runaway situation?

As the arctic sea ice diminishes, so has the albedo and thus reflectance of sunlight back into space, with the resultant warming already affecting Greenland (indeed the whole planet) as developed in the most recent J Francis paper. An entire ice free summer is just the end game; the trend is there now.

However air temperatures are not going to change the gravitational driving force on Jakobshavn much (roughly the height of the summit ridge divided by distance to calving front). Basal meltwater would make the ice/bedrock interface slippery but what comes after already slippery (buoyancy?); however more meltwater exiting at higher velocity would bring more turbulent churning of warm ocean water eroding the submerged portion of the calving front.

We are already seeing more inflow from the sides as the calving front retreats which acts as buttressing similarly to melange, slowing glacier advance. Changes in precipitation patterns -- volume, rain vs snow, surface mass balance -- are not induced solely by changes in arctic sea ice but by global weather trends. However changing deep temperatures in 2 km thick ice -- and so the flow viscosity of Jakobshavn -- by warmer air, warmer ocean, or melt-filled crevasses runs into its immense thermal inertia.

There are definitely precedents for glaciers surging. But whether that will happen here, whether it will be sooner rather than later, whether we will see troublesome indications in 2016 is a matter of wait-and-see. I wouldn't be surprised if there are surprises already this summer. No one is in a position to issue reassurances.

Has Arctic sea-ice loss contributed to increased surface melting of the Greenland ice sheet?
J Liu, Chen, JFrancis, Song2, T Mote, and Y Hu
http://dx.doi.org/10.1175/JCLI-D-15-0391.1

[A-Team]

it appears that a second pronounced slump is appearing as a result of the north wall retreat. As this retreat continues, the entire northern edge of the main ice stream will become unstable and a dramatic, irreversible retreat of the calving face will result.

We could do more here on quantitating this with the hive mind resources present on this forum. There has been previous discussion of drawing contours on a Sentinel 2A showing all the ice that will enter the fjord after 1, 2, 3... years. That can be converted to volume use the known ice thickness, or perhaps better by drawing volume contours of 5, 10, 15... km³ that will enter the fjord in succeeding years.

The image below is another variation on this. It shows concentric circles centered on the middle of the calving front. Arrows could be drawn along this circles pointing in the direction of flow and with length indicating speed. This could be quite instructive.

[Espen]

What is done at this site is beyond what is ever seen before by anyone outside!!

[oren]

What is done at this site is beyond what is ever seen before by anyone outside!!

Where are the PhD researchers and post-docs?? So many potential research issues, and lots of info and methods laid at your feet here.

[Espen]

What is done at this site is beyond what is ever seen before by anyone outside!!

Where are the PhD researchers and post-docs?? So many potential research issues, and lots of info and methods laid at your feet here.

They are all reinventing the wheel, as usual! That is the way for more funding!

[sidd]

As a sometime researcher, i want to speak in explanation and defense. These postdocs and grad peons and professors are hanging on for dear life in the face of budget cuts. They got no time, just trying to stay alive and do the work in their grant proposal and publish in time to meet deadlines, and get the next grant proposal in. It's horrible. I know of cases where people were effectively homeless, sleeping in offices and labs or living in their cars, and this in a purportedly well funded science like physics.

We in this forum are fortunate enuf to have the time, energy and ability to do what they cannot. We should not belittle them also. They can do many things we cannot. They are paid a pittance for what they (and we) do for love, and we cannot do without them. Sentinel and others were launched and operated by these unselfish women and men, and I give thanks for that. 

 

[Espen]

Sidd I agree with you to a certain degree, but I and the public sometimes see quite a few research projects that makes no sense either it is common knowledge or show the researcher is somehow living on another planet. But for sure a lot of great science is done as well!
I can think of a few Arctic related research programs where the funds could find better use, remember Arctic research is extremely expensive.

[Wipneus]

A fine image for the 2nd of April. I have uploaded the B04 (converted from jp2 to tiff) band to a folder that should be browsable from:

https://drive.google.com/folderview?id=0B8NUHa4P2gmlfnhfcUwtRHVzZmRnWnZhQ0UwSkI3ZW12R0dYQ0R4UUR5eElPeEJSYzBsRFk&usp=sharing

That woudl save me from posting the link to the file every time. If it does not work the file link is:
https://drive.google.com/file/d/0B8NUHa4P2gmlOWJOWm1UM3Y0UXM/view?usp=sharing

The file conversion was done with the command:

gdal_translate <source>.jp2 <source>.tiff

gdal is of course the Geospatial Data Abstraction Library http://www.gdal.org/index.html
gdal_translate one of the many utilities http://www.gdal.org/gdal_utilities.html