Support the Arctic Sea Ice Forum and Blog

Author Topic: Petermann Gletscher / Petermann Fjord / North West Greenland  (Read 273094 times)

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #350 on: December 18, 2015, 03:57:43 PM »
Very helpful posts! Amazing story of the unending adventures that underlies the dry scientific prose of eventual publications.

Given you were teaching a signal processing class this fall, could not a lot of work have gone out the door as a homework assignment?

Did you ever post the locations of the ice radar and drill sites over a 15 m 2015 Landsat? (Since the ice shelf is moving, it may be better to locate them relative to surface features rather than initial GPS.)

At one time there was talk of locating them on a repeated radar flight line. If we can trust the cresis kml waypoints on these, the navigational error on these was less than a meter. One of them continues over the terminal extension of the main basal upheaval feature which does not quite reach the grounding zone but could have oceanographic consequences if it is a major source of meltwater(1st and 2nd images). These features are not associated with surface meltlakes and streams though they could be at Eqip (3rd image).

I'm wondering too about surface AWS and radar measurement of snowfall on the Petermann ice shelf. Rignot, who spent portions of three summers camped NE of the grounding zone, reported winters of little or no net snowfall. Yet it's often said not to be reliably measurable because of wind drift and other factors.

Is firn compactification an issue out on the ice shelf or has it already happened?  I'm not recalling a neutron source on the logging wire that might have measured density going through the ice. Ditto ultrasonic probe of ice fabric (again may not be of interest on a shelf). The radar profiles make the ice look rather reworked.
« Last Edit: December 19, 2015, 12:35:14 AM by A-Team »

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #351 on: December 21, 2015, 03:49:09 PM »
The question with using repeat flights, say for progression of underside crevasses near Petermann's grounding zone, is their reproducibility: each of the 22 years used a different radar design and some years different planes and flight elevations.

Previously, track crossings within a single day's flight (cases where the plane later doubled back and crossed an earlier part of its route) have been examined to establish error bounds (~10 m) needed for the bedrock map.

Here the animations below look at the reproducibility across different years of an upheaval zone at Eqip which has a near-perfect route overlap of a north-south track from 2012 (20120421_01_052) and one from 2014 (20140414_02_021), both flown on a P3 out of Illulisat.

If the upheaval has moved or otherwise evolved over those two years, it shouldn't be expected to give the same radar profile. However the bedrock hasn't moved (elastic rebound being negligible) and should line up exactly across the whole image.

The first issue is only in 2008, 2014 and a few others did Cresis use a vertical depth scale tied to the WGS84 ellipsoid (the basis for everything in Greenland glaciology). The vertical scale ticks in other years are correctly spaced at 500 m intervals but the absolute depth numbers are meaningless because the offsets aren't provided -- the 0 is set as the mean surface over that flight frame. Here I used the true scale from the 2014 (which establishes the Eqip upheaval is slightly above sea level).

Second, Cresis crams everything into a fixed template (rather than providing images at intrinsic resolution). Since every flight has different length and depth scales, no two images will have the same vertical or horizontal scale. To compare two images, one must be rescaled by the appropriate scale factors, sometimes radically. While there exists 'smart' rescaling algorithms for 2x, 4x etc here the factors were 87.790% hor and 100.958 ver so some potential exists for introducing artifacts and degrading the image.

Third, it has been belatedly recognized that some sort of mission creep occurred in some flights in some years (they mostly fly at 500 m above the ice surface which slopes at ~1% in west-central Greenland). You can see this in the animation as a residual rotation of 1.26º left after alignment of the matched-scale images.

This could have major ramifications for the bedrock DEM, ice thickness, and modeled movement if this systematic error wasn't removed from applicable flights. It has definitely affected the Petermann area, as first noticed by F Mundel in #342 (who report tilts of 1.35° to the north and 0.98° to the west in Fig.10) and Macgregor 2015.

Cresis (resp. Bamber 2013) at this point have not corrected their archive (resp. DEM).

Overall the correspondence between the two years is fantastic at every level, much better than i had anticipated, so three cheers for Cresis and Operation Icebridge. Some slight differences can be seen in the middle region of the upheaval at 3x (yellow and green bars).

Radar tracks take in reflections over a 100 m wide swath or so, they are not line data in the sense of laser altimeters. Here the two tracks are less than 10 m apart if the plane navigational coordinates are to be believed. It's conceivable reflective properties do indeed vary over a few meters in which case the differences here do not represent time evolution. This is best determined by intensive sled radar grids; Eqip is by far the most convenient site for this.

It can be verified that these differences are perceptible in the originals and did not arise from processing (2nd image). The excellent correspondence overall but differences within the upheaval suggests (one case doesn't prove) certain details of radar reflectivity may be evolving over time within these features. The six east-west flights over this upheaval are offset by 40-200 m so are transecting slightly different sections. I'll look at the variation in these east-west tracks when I get back online; there are quite a few steps in doing this optimally.
« Last Edit: January 01, 2016, 12:45:44 AM by A-Team »

P-maker

  • Frazil ice
  • Posts: 238
    • View Profile
  • Liked: 14
  • Likes Given: 0
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #352 on: December 21, 2015, 04:06:29 PM »
A-Team,

thanks for digging this out of the archives. My respect for your way of working keeps increasing year by year, as do the clear ice domes under the GrIS.

Cheers P

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #353 on: December 21, 2015, 05:57:47 PM »
Quote
Thx for digging thru the archive!
You're welcome! I'm always on the lookout for places where we can move beyond just reporting news and add value. That takes a mix of under-exploited but accessible data in a non-proprietary format, not too technical a topic, and the research community busying itself elsewhere.

Espen has been doing this successfully on the calving fronts, the twist there is real-time incoming data that he is the first to look at and see its significance. However there are quite a few people on quite a few forums doing other things quite impressively. We are head and shoulders above anything going on at scientist twitter sites in terms of creating a useful and comprehensive resource.

I heard the science preprint server ArXiv.org will be starting up a geo/cryo section in 2016 That would be a good place for people wanting to pull together a doi-citable article on a topic (like El Nino) that is currently scattered across many hundreds of posts. ASLR earlier pioneered the use of the comment section on discussion journals as a way of creating a permanent record. It's really a valuable exercise to properly pull together what you know about a topic.

I would recommend these initially over trying a direct submission to a journal which is really expensive and problematic even on merit without listing conventional academic co-authors. But you can sometimes pick these up off a preprint. (I'm at a respectable 5061 citations from articles in other fields but am falling ever farther behind people like Eric Rignot already at 10145 with 2-3 blockbuster articles this year yet to kick out cites.)

Holiday travel looming, I'm going offline (reluctantly).

Neven

  • Administrator
  • First-year ice
  • *****
  • Posts: 7209
    • View Profile
    • Arctic Sea Ice Blog
  • Liked: 732
  • Likes Given: 477
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #354 on: December 21, 2015, 06:49:10 PM »
If you're reluctant to go off-line, it's usually good to go off-line for a while.   ;)

Thanks for all the great work.
Il faut comparer, comparer, comparer, et cultiver notre jardin

ghoti

  • Grease ice
  • Posts: 766
    • View Profile
  • Liked: 12
  • Likes Given: 15
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #355 on: January 01, 2016, 02:02:15 AM »
New post up at http://icyseas.org/ with the latest data from under the glacier. The temperature and salinity fluctuations are intriguing!

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #356 on: January 01, 2016, 04:25:43 PM »
Quote
New post up at http://icyseas.org/ with the latest data from under the floating ice shelf ... salinity fluctuations at 95 and 115 m depths are intriguing
2016 ... only one instrumented ice shelf in the northern hemisphere?

Quote
it's usually good to go off-line for a while.
Oh, I'm usually offline for the day by 8:30 am. It works better for me to bite off a small post each morning rather than take long holidays followed by hours of picking up where I left off.

Hmmm, where did I leave off anyway … some oddity in Greenland last time it got warm, beyond just meltwater raising sea level and disrupting planetary heat re-distribution. The last ice age did not wipe the slate clean: the Eemian is not pure paleo. Its legacy, in the form of basal folds and thermal upheavals, is still a driving force today. Thus it is quite wrong to compare Eemian to Holocene; the future of the latter has to build on what the former left behind.

Ice sheets have exceedingly long memories. Because of its proximity to Icelandic volcanoes, that of Greenland has been retained in exquisite detail, not only in the annual climate layers of cores but also in deep stratigraphy visible along ice penetrating radar tracks. That record, a byproduct of bedrock topography determination, handed glaciologists a book in the mid-’90’s that they didn’t want to read.

So here we are, 20 years after the first basal upheavals were observed, forced to read that book anyway — we know southern Greenland is toast, it’s basal conditions of the northern ice that matters.

That’s what I’m writing about: denial.  We cannot afford it any more within the scientific community — it’s high time to get a radar sled over Eqip and a steam drill too.
« Last Edit: January 01, 2016, 04:59:44 PM by A-Team »

Tor Bejnar

  • Young ice
  • Posts: 3115
    • View Profile
  • Liked: 409
  • Likes Given: 200
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #357 on: January 01, 2016, 04:30:04 PM »
Thanks, ghoti, it is very interesting, in deed!
New post up at http://icyseas.org/ with the latest data from under the glacier. The temperature and salinity fluctuations are intriguing!
Arctic ice is healthy for children and other living things.

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #358 on: January 01, 2016, 05:59:15 PM »
The Petermann catchment upheavals are considered from a fresh perspective in the F Mundel paper linked to in #342. This is a German bachelor's thesis (~ US masters?) rather than a journal article, but one overseen by AWI geoscientists with extensive experience on Larsen C and NEEM core layer deformations (see http://www.the-cryosphere-discuss.net/9/5817/2015/tcd-9-5817-2015.html).

From a structural geology standpoint, ice folds and flows like a salt bed or rock near its melting point, leading here to a regional interpretation of Petermann cylindrical folds as arising from 3 effects: cross-flow compression from a funnel-shaped catchment, along-flow tensile stress from increasing terminal ice velocity, shear stress from a vertical velocity gradient giving sheath folds.

None of this was developed beyond the cartoon level but a conceptual framework is probably preferable to the overly intricate model in which the core proposition is lost. The author cites the previous upheaval papers but does not compare alternative mechanisms. Very few of the scientific desiderata outlined in #346 are met here.

The main idea here is to move the ice penetrating radar grid sectioning of Petermann into commercial geology software called MOVE 2013.1 which is capable of developing 3D models of internal isochronal surfaces and integrating them with bedrock and surface DEMs as well as surface velocity.

This didn't work very well for a variety of reasons: the academic license version was hamstrung, half the grid utilized the crummy radar design on the 2010 DC8 overflights, the grid was too coarse at 8-9 km relative to feature size, no physical basis existed for interpolating between 2011 grid lines, the horizontal was badly tilted in either the radargrams or the bedrock DEM, too much data had to be discarded as not straight or parallel enough, ill-advised co-mingling of manual traces of isochrons and upheaval boundaries, use of an outdated velocity magnitude map lacking flowlines, and the highly improbable assumption that ice velocity decreases exponentially to zero on a frozen bed as claimed in a 1966 paper.

It is a little troubling to see MacGregor 2015 cited early on but its vastly more sophisticated and comprehensive isochron tracings never utilized. We've seen in earlier posts here that the only isochrons worth tracing in the Petermann area are the 14.5 kyr Holocene beginning, the three sisters at 46.5 kyrs and the two brothers at 91.5 kyr. The author here came to a similar conclusion, tracing only the Holocene and middle-sister (called a highway).

MacGregor 2015 did all this and more, earlier. That data was interpolated by a better method, gridded to the Greenland standard, and stored as a public netCDF archive, with the Holocene and end-Eemian presented as smooth rendered surfaces in a much viewed NASA video. However that effort could not provide the same microscopic evaluation to an individual grid and custom contrast rescue of problematic radar tracks. MacGregor 2015 also sidestepped the issue of upheavals for the reasons discussed in #346 -- they’ve not been proven to be distorted or diffused regional isochrons or indeed any kind of isochron.

The author considered flight intersections but mostly had to toss the poor quality DC8 north-south sections and curved paths which MOVE bizarrely cannot accommodate. There was a serious misunderstanding about lat,lon numbers provided by Cresis, or to put it more fairly, Cresis did something very dumb that later confused everyone: not to subdivide the image frames with the actual high precision GPS waystations.

The graphics below show some worthwhile products from the Mundel paper. The first shows layer tracking, the second the ‘cardboard wine carton’ display of sectional assembly (developed here earlier as a take-apart animation),  the third a perspective view of the grid, applicable shears and final surface (rendered in so-so fashion by MOVE).

As we've seen earlier in #161, the center of this fold belt actually extends all the way up to Kap Schoubye but not quite to the grounding zone. Here the author discarded the DC8 vertical transects (upper blue in flight grid, 3rd image), nearly half the available data. It is very dangerous to interpolate with software because that cannot help but produce smooth folds in every situation.

The fourth image shows regions of high velocity appear strongly correlated with the cylindrical fold belt. However according to Radarsat interferometry in #86and later Sentinel 1A, the outdated velocity map used here is grieviously in error centrally. The distribution of Greenland basal upheavals overall has no association whatsoever with high surface ice velocity. There is no supporting data for the putative inland funnel. However the structural geology approach used here may still have merit.
« Last Edit: January 03, 2016, 01:28:06 PM by A-Team »

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #359 on: January 03, 2016, 03:58:33 PM »
Last March, Wipneus located a striking image taken by Radarsat operating in Sentinel mode (post #86) and observed in #91 that the ice in the fjord proper was moving too fast for the 24 day scene interval to resolve. Andreas T found the ESA's meagre background report and noted a full fringe cycle amounts to 5.55 cm given Radarsat's 5.4 ghz wavelength in #96, while Espen noted how far inland the high velocity region extended in #97.

It is really critical to use reverse image search of the internet (imageraider) and a pdf image extractor (ImageJ) to find the highest resolution version of the image. Almost everything you see on the Internet is badly compressed. I managed to locate the original image and have posted it below (the blog software will only show a thumbnail without a click). Separate coherence and intensity grayscales are also available.

Petermann, being at latitude 81ºN, would be frozen solid on these dates (11 Apr 13 to 05 May 13) per Landsat-8 so loss of height from melt or sublimation are not plausibly contributing to line-of-sight motion seen by the satellite. This region of Greenland has very little annual snowfall so wind ablation is not a factor either.

Fringe separation thus largely represents horizontal motion of the ice sheet --  bare rocks being fixed, marine terminal velocities are fast, the dividing ridge between Hammond and Glacier falls off into the respective catchments, the upper ice sheet is hardly moving toward the interior, in excellent agreement with separately determined speed maps of Rignot and Joughin (1st animation) and more recently Sentinel.

It's easy to independently calibrate velocities using a time series of Landsat-8s at least in the vicinity of the last ground control point (Kap Schoubye, the island resembling an arrowhead).

The 3rd image shows a 2x enlargement of the curious region Espen noted earlier. The center of this corresponds to Mundel's cylindrical fold belt (4th image of previous post). The colored fringes, when broken down to separate grayscale R,G,B channels run as frames of the 2nd animation, suggest surface ice motion.

Note the fast upstream region actually forms two fingers flanking a slow island that widens down-glacier. It is this slow island -- or perhaps its velocity contrast -- that best corresponds to the main structural geology fold proposed by Mundel. Funnel compression and tensile stretching of that model way are too simplistic and would have to be re-elaborated as tensor fields that fit the data.

One could also envision support here for the traveling slip-stick zones proposed by Wolovik 2015, though there's no real experimental information about basal thermal conditions or velocity with depth. Indeed it's hard to see how the bland slope and bedrock topography could produce such a nuanced surface velocity field satisfying mass conservation without noteworthy crevassing of the brittle ice. 

The bedrock lacks anything overtly corresponding to the slow island and fast fingers but does have a broad rise between the below sea level interior and fjord. This creates a curious convergence (noted earlier in discussing longitudinal flightlines) of ice rising basally over bedrock but plunging surficially as it nears the coast.
« Last Edit: January 03, 2016, 04:34:16 PM by A-Team »

nukefix

  • Frazil ice
  • Posts: 461
    • View Profile
  • Liked: 9
  • Likes Given: 1
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #360 on: January 03, 2016, 11:44:07 PM »
The "glitches" are actually features of the data. The jumps are at SAR burst-boundaries where the observation-angle (in azimuth) changes slightly. In other words, the phase-jump is physical.

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #361 on: January 04, 2016, 02:13:45 PM »
Quote
jumps are at SAR burst-boundaries where the observation-angle (in azimuth) changes slightly.
At the end of the day, the product needs to be a smooth velocity map over the entire scene -- there aren't any glitches on the ground.

The amazing thing to me is how little impact Radarsat has had on Greenland glaciology -- here it appears capable of resolving internal detail of the Petermann velocity field that the last 20 years of interferometry have missed. One has to wonder what other fine structure has been missed elsewhere in north-central Greenland.

Petermann only got picked up as part of a casual survey that included Mt. Etna volcanic bulging and Mexico City ground subsidence. And this was 2013, a great many orbits ago. The scene is huge -- a few dozen would have sufficed for all of coastal Greenland. The casual write-up does not provide the corner lat,lon points nor any gdal guidance on how the image should be re-projected to UTM, GE or PS. Possibly the Sentinel toolbox could do something with it.

I marked up the overtly visible burst boundaries below, For some reason they are most noticeable in the lower part of the scene. This is not strictly due to larger fringe cycle (slower ice sheet motion) because the northeast corner is also slow but doesn't have the fringe displacement.

I looked at fixing things by cutting out the offending boxes and shifting/rescaling but that did not go anywhere. The intensity map (which loses phase information) was only released at reduced size; it has pronounced vertical contrast banding that went uncorrected at ESA but may just need a simple multiplicative rescaling.

Note in post #359, the data is interpreted as an anomalous slow blob occupying in the middle of a typical western Greenland velocity field (widening catchment of uniformly slowing velocity). Someone else might see two incipient ice streams, features isolated by sharp shear boundaries, with a slow island between them, in effect an early stage of NEGIS/Zachariae. The ice stream there is marked by a distinct surface trough which has no counterpart in the ultra-high resolution DEM of Petermann.

I looked at quantitating velocities, which amounts to measuring relative fringe width. These ratios could be converted later to absolute velocities using Landsat calibration (since we aren't real sure what Radarsat is doing). We do know how fast Petermann's ice shelf is moving (~3 m per day) but that internal calibration got wiped out by too much movement over the 24 days.

The 4th image shows some of the options, such as enlarging the scale, desaturation modalities, or color-picking say blue to reduce clutter. While it looks quite feasible, giving 3-4 x higher velocities in the shear zone (pink averages over ten cycles), before going too far down this road I would want to see confirmation of the slow island from a pair of Sentinel 1A images -- the 12 day interval would give better resolution in the critical zone.
« Last Edit: January 04, 2016, 06:35:42 PM by A-Team »

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #362 on: January 05, 2016, 01:58:52 PM »
The first image below compares the April 2013 Radarsat interogram with a 2007 made with a different satellite (ALOS PALSAR ) at a different time of year (summer), as discussed last June in #205. The region available for comparison is fairly small but the patterns are amazingly similar.

The second image focuse on three neighboring glaciers (Newman, Steensby and western Ryder). Here too the fringe pattern of the Radarsat interferogram makes sense in terms of the line-of-sight motion being primarily attributable to horizontal motion of the glacier (rather than vertical melt or thinning).

ALOS-1 PALSAR Global Radar Imagery, 2006-2011, is now open access. I have not done much with it to date but see 3rd image — finally, a decent preview. Note the anomalous velocity region is likely evident in the intensity view of all Palsar imagery, raising again the question of why it hasn’t been noticed earlier. Perhaps poor palette color discrimination combined with low resolution and a logarithmic velocity scale have masked it.

The 4th image shows the Parca stake line velocities along the ~2000 m contour, along with the available radar transects and the region modelled in Mundel 2013 (pale orange box) and the extended upheaval feature tracked here earlier using NS tracks in addition to EW (thick black line). Note the Parca flow lines are unremarkable except for F120 (magenta) which shows signs of converging to the faster F121 flow line. The bedrock under F120 is 96 m above sea level whereas F121 is 7 m below.

The bearings differ by 322.5 - 299.1 = 23.4º over the 42 km separating the stakes; that might not seeem like much yet the ice is incompressible so has to accommodate convergence. the velocities (and so physical displacement) by 34.49 - 28.28 = 6.2 m per year. Even though the F120 velocity has slowed by 22% relative to F121 which would introduce significant strain in the ice, that would not show up in a color palette has to range from 1-1400 m/yr.

The Parca stakes are a bit upglacier from where we need them but measure velocity and bearing (tangent to flowline) much more accurately than could ever be obtained by satellite interferometery or speckle tracking. Over the 3 years of 15m resolution Landsat-8, F121 will move 6.9 pixels and F120 only 1.2 pixels less. That’s quite problematic given the nearest ground control points are 175 km to the north. Landsat-7 would bring more years into the mix but further raise issues in precision alignment.

https://www.asf.alaska.edu/sar-data/palsar/
https://vertex.daac.asf.alaska.edu/ nuisance registration but nice interface to many radar satellites

http://forum.arctic-sea-ice.net/index.php/topic,388.0.html Espen et al on Steensby
http://forum.arctic-sea-ice.net/index.php/topic,886.0.html Espen et al on Ryder

https://glacierchange.wordpress.com/2012/08/25/steensby-glacier-calving-event-and-retreat-northern-greenland/  M Pelto on Steensby
https://glacierchange.wordpress.com/2010/10/26/ryder-glacier-northern-greenland-transient-snowline-rise/  MPelto on Ryder


A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #363 on: January 06, 2016, 04:46:18 PM »
Upper Petermann receives very regular coverage by Landsat though it's rare to get a cloud-free day of the entire upheaval area. I'm looking here for visual surface counterparts to the massive foldings underneath, hoping to independently validate the anomalous surface velocity zone seen by Radarsat, ie slip motion of consistently identifiable features in consistently registered 2013, 2014 and 2015 scenes.

WorldView3 has 30 times the resolution of Landsat-8 and so can see pixel motion over 1/30 the time interval but access is expensive. Ice motion is slow high in the ice sheet; for shear zones (differential motion) we need to distinguish slow vs slower, not interior slow vs calving front fast.

The upheaval area at Petermann involves an area of roughly 180 km x 110 km which in terms of 15 m Landsat resolution , which works out at maximal blog size of 10 km x 10 km to nearly 200 images or 40 posts at the 4 attachments max, not happening. (This is why Eqip is a better choice: much smaller and denser transects.)

The 19 July 2015 LC80360022015200LGN00 covers it all fairly well (avoiding tiling) and downloads in UTM zone 21 projection as 18,000 x 18,000 pixels. The relevant half can be cut out after rectangularization by a 33º CW rotation.

Overlaying the upheaval flight lines grid and Mundel's fold belt (polar stereographic like bedrock DEM and ice thickness) onto this Landsat raises some issues. The Cresis kml waypoints file provide lat,lon not UTM21; there are online tools for that. Google Earth displays the tracks nicely but in its own projection while also providing UTM22 coordinates. The Landsat metadata file provides the corners in lat,lon (and UTM21) which in theory might allow the Landsat to be uploaded correctly reprojected.

Since a Landsat time series is needed to detect motion and the number of distinctive surface features are far and few between this far up the ice sheet, I did not want to further degrade Landsat image quality as it has gone through many processing steps already to arrive at the UTM21 download, all that is served at EarthExplorer.

The radar stratigraphic rasters too have had a lot of processing and need a lot more. These are specified as chords to the actual curvilinear flight lines; as line segments rather than area, they are not in a projection system but simply WGS84 lat,lon,z.

The set-up is shown in the first image below. I decided to locate a central flight segment through the anomalous surface velocity region in the fold belt on the Landsat by approximate methods, as shown. This was instructive because nothing on the Landsat had the slightest promise for correlating with Radarsat or IceBridge, second image.

Further, the Landsat was lumpy but otherwise so featureless that detection of slow differential motions just wouldn't work. That's too bad because the four vertical sides of a radar grid cell would fit nicely within a blog-sized Landsat at full resolution and we could tile up from there.

However without ground control points, positioning error after all the mickey mouse couldn't be independently assessed. The low budget route, entering the grid corners as EarthExplorer search corners, solves the coordinate conversion issues but not that accurately and only overlays tracks on the greenish low resolution preview image in Google Earth projection.

Here EarthExplorer allows upload of kml and shape files to drive image selection but these are limited to 30 points (radar tracks have 40 or more waystations). Unfortunately, after selecting an image for download, the kml file is tossed rather than being reprojected to an accompanying transparent overlay of the UTM.

Here again the Eqip upheaval is more favorable since it has numerous persistent melt lakes, streams and fixed rocks being near the coast and is so close to Jakobshavn that it resides in the very same Landsats. Access is 100x easier than to the Petermann grid area and many thoroughly studied sites are in the vicinity including Store Glacier and Swiss Camp.

The latter has a very peculiar thermal gradient between two nearly adjacent drill holes, one of which in view of Eqip, may have bored through a radar-unobservable basal deformation. Viewing these as buoyancy imbalances, there won't be radar reflectors if the process is essentially a moving thermal front (called zone refinement in metullurgy) with few impurities to bring out.
« Last Edit: January 06, 2016, 05:02:47 PM by A-Team »

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #364 on: January 07, 2016, 07:03:05 PM »
Ice penetrating radar is so critical to every aspect of Greenland glaciology that it is worth re-processing the entire Cresis archive. There's been considerable progress on some angles of that, see below.

In addition to correcting and uniformatizing the vertical and horizontal scales, fixing the unspeakable tilt error, improving contrast adaptively, I'm looking how best to add compass direction of the flight, direction and magnitude of ice flow across the flight line, replacing horizontal scale with precision GPS waystations, and utilizing curvilinear and haphazard flight segments as held-backs to assess true bedrock, ice thickness and fold model interpolation error. Note surface and bedrock slope are already baked into the images; aircraft bearing is inferable but ice surface velocity is not.

It’s tempting too to plot a track along an overall bedrock DEM 3D visualization but the fact is, no locally relevant data whatsover exists for bedrock other than the track at hand and its intersections.

Petermann is one of the most intensely surveyed regions in Greenland yet a typical 50 km track segment will only have 5 intersecting tracks. A whole lot can happen in topography over 10 km, for example a whole 3000 m mountain range in Nevada could come and go without being detected. However based on flights subsequent to Bamber 2013 it appears more likely that Greenland's bedrock is mostly unremarkable.

I'll also be looking at another mystery: where does the ice go as it is displaced from below by an upheaval? Higher layers are displaced upwards to a certain extent but that effect is soon damped out, not typically resulting in a surface bulge nor accounting for the area of the upheaval as seen in cross section.

Newer ice layers also thin above basal upheavals, possibly anisotropically with respect to prevailing speed and slope, as can be seen from a decreasing inter-sister distance over a hump or as mid-sister to Holocene cross-sectional area. This effect is related to but distinct from draping (or not) of stratigraphy over bedrock topography.

The issue here is recovering internal ice viscosity (temperature); the mystery is how 20 years x 500 glaciolgists can pass by without that being determined.

That has taken a new twist this year as NEEM researchers have started looking more closely at what produced micro-layering of ice crystals in the core. Glen's flow law, used in all modern ice sheet models, is partly derived from short-term experiments on lab ice utilizing enormous external stresses.

However the strain response of grain boundaries etc to low stresses over long glacial time frames is altogether different. This means folds in the NEEM core -- and thus the history and  future of the Greenland ice sheet -- cannot be accurately reproduced using Glen's law.

http://www.the-cryosphere-discuss.net/9/5555/2015/ T Goossens 2015
http://www.the-cryosphere-discuss.net/9/5817/2015/ D Jansen 2015
http://www.the-cryosphere.net/8/1129/2014/tc-8-1129-2014.pdf M Montagnat ‎2014
http://onlinelibrary.wiley.com/wol1/doi/10.1002/2015JD023290/full I Oyabu 2015
https://agu.confex.com/agu/fm15/meetingapp.cgi/Paper/83886 K Keegan 2015
https://agu.confex.com/agu/fm15/mediafile/Handout/Paper78793/GPT_Jose_AGU_CReSIS.pdf G Tsoflias 2015
https://epic.awi.de/33070/1/Eichler2013.pdf J Eichler 2013

I just noticed that the Radarsat velocity anomaly above is visible to a certain extent in Fig.4e of the original Bell 2014 paper, shown below with and without the overlay. The source of the velocity map seems to be Joughin 2010 which represents InSAR data from 4-5 years earlier. It is consistent with our Radarsat but lacks its fine structure resolution.
« Last Edit: January 08, 2016, 12:12:26 PM by A-Team »

Espen

  • Young ice
  • Posts: 3251
    • View Profile
  • Liked: 177
  • Likes Given: 0
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #365 on: January 07, 2016, 08:18:09 PM »
"I'll also be looking at another mystery: where does the ice go as it is displaced from below by an upheaval? Higher layers are displaced upwards to a certain extent but that effect is soon damped out, not typically resulting in a surface bulge nor accounting for the area of the upheaval as seen in cross section."

I see no mystery, because the way I look at it, the glacier ice behaves very much like water in a stream  or a river, where the shape of the bottom is not shown on the surface?
Have a ice day!

P-maker

  • Frazil ice
  • Posts: 238
    • View Profile
  • Liked: 14
  • Likes Given: 0
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #366 on: January 07, 2016, 11:26:30 PM »
Espen,

I do agree with you that overlying layers of ice simply will flow to the sides, as the upheavals rise from the bottom.

A-Team,

Thanks again for your great efforts and extensive attempts at documenting everything in 3D. At times however, I would prefer to see all your ramblings on this issue written up in a neat paper.

I did not check all your references but based on the stuff I did read, I think a number of facts emerge, which I have tried to nail her:

1)   Units of Disrupted Radiostratigrahy (UDRs) mainly occurs under the northern half of GrIS, where ice is cold and bottom freezing is most likely to occur
2)   Most UDRs appear near the heads of ice streams requiring some speed (and surface slope) to produce isolated patches of meltwater through pressure melting near the bottom (or drainage channels from the surface)
3)   A number of elongated (cigar-shaped) UDRs have now been identified and they all seem to follow flow lines near the bottom of ice streams
4)   At least two papers have now identified UDRs under the upper reaches of the Jakobshavn ice stream
5)   It is no wonder that there is a “pervasive tilt error” along these ice streams, in particular not if you consider that they are rapidly disintegrating and thinning near the front and at the same time UDRs are growing from the bottom near their valley heads.
6)   If these UDR upheavals are actually taking place more rapidly over the past 20 years, it must be tricky to figure out the isostatic recovery patterns under these circumstances
7)   It is well known that freezing water breaks even the strongest rocks, thus if you can get the freezing process going near the bottom, it should be a simple matter to lift the entire ice sheet from the bottom and let the overlying layers flow to the sides, thus disrupting the original stratigraphy.

If it is refrozen surface water, you will eventually end up with the youngest ice near the bottom.

sidd

  • Young ice
  • Posts: 4987
    • View Profile
  • Liked: 389
  • Likes Given: 0
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #367 on: January 08, 2016, 05:11:27 AM »
"If these UDR upheavals are actually taking place more rapidly over the past 20 years, ... "

how strong is the evidence for this ?

P-maker

  • Frazil ice
  • Posts: 238
    • View Profile
  • Liked: 14
  • Likes Given: 0
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #368 on: January 08, 2016, 10:22:38 AM »
Sidd,

I agree. This statement should not have been put under the headline "facts". I was simply referring to the observation made by A-Team that these features were not known before the 1990ies. It remains to be seen whether some clever guy can actually document the rise of an UDR over the last couple of decades. This may have a number of implications for moderne glaciology, if this is the case.

Cheers P

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #369 on: January 08, 2016, 02:59:16 PM »
Quote
the glacier ice behaves very much like water in a stream  or a river, where the shape of the bottom is not shown on the surface?  overlying layers of ice simply will flow to the sides, as the upheavals rise from the bottom.
Whoa there. Bottom upheavals raise the ice above them, sometimes many cubic km of ice forced thousands of meters overhead. The nearly flat isochronal surfaces in that 95,000 years of upper ice are bent smoothly upwards well into the Holocene and sometimes into firn.

Indeed the Panton paper uses slope departure from horizontal as a proxy for automatic identification of basal upheavals (which have poor contrast in most radar designs). The angle of bending tapers off smoothly in younger ice. That is what's being measured in Fig.1B.

Bottom upheaval pressure also cause the ice above to flow slightly to the sides, manifested as isochronal lines becoming slightly closer over the bump than before. This got caught to a certain extent in the MacGregor 2015 dataset which did not consider deformations (#346 #358). However the radar archive is so vast they could not consider specific cases at high resolution in their Greenland-wide study.

Ice being incompressible, any new volume injected or forced by folding into the bottom has to be accommodated between these two mechanisms, thinning and uplift. This fractional allocation, maybe 15% thin plus 85% uplift, has never actually been measured in a specific case.

For starters, no one has ever tabulated the volumes of ice displaced. That would require much more focused radar tomography than present-day surveys (as called for in the Panton paper). Many uplifted regions have only a single cross section not aligned with flow nor positioned representatively with respect to the global deformation structure. Think of a  low budget CAT scan -- the doctor wants to see if your liver is enlarged but is only given a random abdominal plane.

I would do this by working solely with the two brothers (91 kyr), three sisters (46.5 kyr) and Holocene start *14.5 kyr) -- they're ubiquitous and an adequate proxy for the hundreds of minor isochrons. It's not difficult to pick the layers with the bezier tool of gimp, measure the intervening area, and plot the separation between them.

It's probably better to precede inductively from case studies of experimentally accessible, limited-extent, isolated low-flow deformations like Eqip having many radar cross sections than to initially take on the remote, massive, inadequately gridded Zachariae and Petermann. No one has a clue yet what is going on -- that is why Panton terms them UDRs and why MacGregor sidesteps the whole subject.

A single flight line often provides internal calibration via long uneventful stretches preceding the deformation in which a constant separation defines the unperturbed normal. A few isochrons thus suffice to estimate how thinning thins out with vertical height. Panton gives three reasons to expect linearity -- can't go wrong with that as everything is linear to first order, all that we could aspire to here.

Basal deformations are not always easy to detect, delineate or differentiate from draping over basal topography. Panton's automated method discards sloping isochrons when there is significant bedrock slope but still comes up with  6-7% Greenland affected. However steep bedrock prominences may actually be conducive to deformation initiation in view of ice sheet flow.

Contrast in radargrams is a complex issue. The remedies have some similarities to those for Landsat and Sentinel in that optimal corrections must differ regionally to respect patchiness. Isochron striations bring in linear anisotropy as do crevasse fields that must be exploited by the filters. There are also very pronounced flow anisotropy effects in Petermann fold belt radar as noticed by Mundel.

Image intensity in satellite and ice radar data actually have physical meanings. Those becomes altered, perverted, or even lost in enhancement processes. Some people choke on the whole idea. However it's ultimately about having to optimize information extraction from very expensive data according to the purpose at hand. The original doesn't cut it.

Panton 2014 introduced elliptical band pass filtering empirically adapted to sloped and even curved isochrons. That had a wonderous effect on inter-band noise (and so tracing isochrons) but was ill-adapted to diffuse deformations, even making them worse. Panton 2015 sought to apply these to the voluminous Cresis archive with minimal human intervention.

Image segmentation followed by say ImageJ adaptive contrast applied to deeper ice does much better in pulling out faint upheaval regions. That can be problematic depending on the existence of reflectors and their diffusivity. Tracking isochrons through flares is yet another special situation in radargram enhancement important to isochron continuity In Photoshop or Gimp terms, this is masking the image appropriately before each effect is applied and then reassembling from the parts.

The creative part comes with making an effective mask without tedious manual tracing of boundaries on thousands of individuals radargrams that have built up over 22 years of idleness. Masks can often be defined by semi-automated methods but it's hard to envision these working for more than one year's radar design.

The alternative is optimizing the radar design so that it does a better job on the deformation layer from the get-go. That may be feasible for sled radar at selected sites (eg Eqip or the 2015 Renland ice cap).

The inconvenient truth is a very substantial part of north-central Greenland has significant basal issues, much more than we are acknowledging today. It is complete folly to keep applying the same old classical ice sheet models to the GrIS without first getting to the bottom of these deformations.

Below I relocated some text and links that were previously above. Panton's work was previously covered on these forums at the links below.

http://forum.arctic-sea-ice.net/index.php/topic,400.msg36170.html#msg36170 Zachariae
http://forum.arctic-sea-ice.net/index.php/topic,154.msg34112.html#msg34112 Jakobshavn
http://forum.arctic-sea-ice.net/index.php/topic,909.msg36465.html#msg36465 radon transform

Quote
Automated mapping of near bed radio-echo layer disruptions in the Greenland Ice Sheet
C Panton NB Karlsson
http://www.sciencedirect.com/science/article/pii/S0012821X15006597

One of the key processes for modulating ice flow is the interaction between the ice and the bed, but direct observations of the subglacial environment are sparse and difficult to obtain. In this study we use information from an extensive radio-echo sounding dataset to identify areas of the Greenland Ice Sheet where internal layers have been influenced by near-bed processes. Based on an automatic algorithm for calculating the slope of the internal radio-echo layers, we identify areas with disrupted layer stratigraphy. We find that large parts of the northern portion of the ice sheet are influenced by locally confined mechanisms that produce up-warping or folds in the layer stratigraphy inconsistent with the surface and bed topography. This is particularly evident at the onset of ice streams, although less dynamic areas close to the ice divide also contain imprints of layer disturbances. Our results show that the disturbances are found in many different flow and thermal regimes, and underscore the need to understand the mechanisms responsible for creating them.

Quote
Tracing Internal Radar Layers in the Greenland Ice Sheet
Christian Panton PhD thesis U Copenhagen 108 pages
http://www.nbi.ku.dk/english/research/phd_theses/phd_theses_2015/christian_panton/Christian_Panton.pdf

Internal layers in radio-echograms from the sounding of ice sheets have long been a valuable resource in glaciology, but their usefulness have been limited by availability of traced (digitized) layers. To speed up this process, we have developed an algorithm for semi-automatic tracing the internal layers and a fully automated algorithm for mapping the layer slope.

The layer slope is inferred by the intensity response to a slanted Gaussian filter, from which layers can be traced using an active contour model. With these techniques we show that it possible to trace internal layers over distances of hundreds kilometers with minimal operator intervention, and the methods have been successfully validated between two Greenland deep ice cores with internal match points.

In order to remove any operator assistance, we show how the layer slope can be used to detect disturbances in the deep radio-stratigraphy of the Greenland Ice Sheet. We find that the disturbances are scattered over the northern part of the ice sheet, with the highest density upstream from the Petermann glacier. The disturbances do not seem to be correlated with surface velocities and can be found on, and close, to the ice divide. These results highlight the need for high resolution mapping of the interior ice sheet to understand the dynamical nature of the basal environment.
« Last Edit: January 08, 2016, 05:46:51 PM by A-Team »

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #370 on: January 09, 2016, 01:38:14 PM »
Quote
If these UDR upheavals are actually taking place more rapidly over the past 20 years,  tricky to figure out the isostatic recovery patterns
The animation below shows how to filter the Cresis ice penetrating archive for repeated flight tracks with the greatest separation in time (3rd frame) that intersect a known upheaval (4th frame). 

Camp Century, summit ridge drill sites and certain glacier outlets were favorite waystations, so some tracks from the mid-90’s were indeed re-flown in 2011-14 (3rd frame). Early radar designs were already quite decent but center frequencies, bandwidth and power were quite different from post-2010 radar, as was processing. Power is important at depth.

We looked at a 2 year Eqip pair in #351; for an example of a 15 year pair, segments 19990514_01_004 to 008 can be compared to 20140514_01_048 to 063 (magenta line in 3d frame). This gives about 800 km of common track that has upheavals much of the way, compounded by bedrock-driven draping in the southeast. Google search has just discovered Cresis kml so today those search terms and a click can magically display those track segments in Google Earth as well as their pdf booklets.

https://data.cresis.ku.edu/data/rds/1999_Greenland_P3/pdf/19990524_01.pdf
https://data.cresis.ku.edu/data/rds/2014_Greenland_P3/pdf/20140514_01.pdf

The second animation compares the upheavals in a specific instance over a 15 year interval (19990514_01_008 vs 20140514_01_51). The flight lines were 84 m apart in this region according to onboard GPS. There are issues with vertical and horizontal rescaling, as well as radar resolving power and correctable contrast. However looking closely at this pair, the upheavals look very similar despite some wagging at the upheaval top. Certainly the 2014 upheaval is not new.

Keep in mind radargrams are terribly misleading as they stand. The vertical exaggeration is typically 10-20x but can be much higher. This tends to overly dramatize their appearance, squashing the upheaval into a very small horizontal extent. Somewhere way back, I posted the same scene from 1:1 to 50:1 exaggerations, that was an eye-opener -- these are very drawn out features, not the Grand Tetons.

Overall, there isn't any information currently about the temporal origin, initial geographic distribution, initiation mechanisms or subsequent evolution of upheaval systems. Cold ice is quite brittle but here it evidently has had ample time to deform smoothly upwards and flow slowly to the sides. The parameter range allowing that and the very size of some features provides a time constraint (slow development).

It might be worth developing 2-3 scenarios at the extremes and asking what could distinguish between them using either the radar we've got or by targeted experiment. My sense is that warmth of the end-Eemian set the stage for the upheavals as opposed say to awakening volcanism, geothermal excursions, the Holocene Thermal Maximum, or meteoric moulin meltwater.

In this scenario, they began to develop as the ice age had laid down a thick layer of ice of contrasting temperature and continue to growing to this very day, influenced of course by bedrock and slope, depth, temperature and velocity of ice overhead. The evolution is likely slow, steady and incremental rather than sudden and disruptive.
« Last Edit: January 09, 2016, 06:27:47 PM by A-Team »

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #371 on: January 12, 2016, 10:17:12 PM »
Quote
C Panton and NB Karlsson
Highlights (http://www.sciencedirect.com/science/article/pii/S0012821X15006597)

• Large parts of the ice sheet are influenced by near-bed processes.
No correlation between the size of the features, frequency and surface velocity.
• Folds and up-warping can occur in absence of basal melt.
The first point is strongly confirmed by advanced contrast enhancement methods that unambiguously bring out lesser deformations in many older tracks. It seems that Bell 2014 only pulled fairly extreme deformations and those only in selected years. It’s instructive just to look at post-Bell flight data (eg 2014) as this reveals many new deformations per km of novel flight track as well as increasing the documented size of previously known complexes.

The second point goes up against a deeply entrenched misconception that ice flow is somehow responsible for these upheavals. However many deformations occur in essentially stagnant ice. Panton 2015 performed formal correlation statistics, finding R = 0.03 in supplemental.

The third point cannot be fully evaluated until basal melt is better mapped. However it’s fair to say basal deformations are so extensive from Jakobshavn north that naive models (frozen vs unfrozen bed, deviatoric strain) have little applicability to the future of Greenland’s ice. That’s because these deformations introduce a whole new unexplored dimension to ice rheology, temperature and crystal fabric.

Panton 2015 suggests a persistent localized source near NEEM, what might be called an emitter. Over time, steadily formed upheaval would be entrained by ambient ice sheet flow and so move downstream from its fixed point of origin, giving rise eventually to an elongated cylindrical structure aligned along flow. In this situation, neither ice flow per se nor structural geological folding have anything to do with creation of deformation structures.

According to the intersection of the upheaval inventory with long-term repeated flight trackss (4th frame of animation above), examples of repeat flight tracks might exist within the fast-moving region of the Petermann grid (though not at less-flown Zachariae. We’ve only look so far at short time intervals between repeat flights in  fairly slowly moving coastal ice (Eqip) and the stagnant interior (posts #370 and #351).

For grid ice moving at 1 km/day (a third of calving front speed), with 15 years possible between upheaval snapshots, the surface ice will have moved 4 km. That’s easily detectable on radargrams as it corresponds to a 74 pixel displacement. Ice moves slower between surface and bottom but prospects are still good for detecting structural changes in deformations over time. A longitudinal track is more favorable than a cross-flow transect here. Unfortunately the older longitudinals at Petermann have either poor horizontal resolution (eg 1997) or poor radar design (2010).

Back at Eqip, the deformation is smaller and the track crossings more numerous. The first image below shows a new container (template) to uniformatize deformation data from over-flights in many different years and radar designs. The image below shows a complicated but unfortunately common situation where the flight sections are curved. We can’t just throw all the curved data away because much of Greenland is never going to be visited again by ice penetrating radar.

I added a compass that shows flight bearing at each track waystation and surface ice velocity at that point, a bounding box for the deformation and the area it occupies there, the google earth kml thumbnail of the relevant over-flight, and a 1:1 scale in addition to bringing data to a standard 10:1 vertical exaggeration.

Someone at Cresis had the very poor idea of labelling intermediate lat,lon points across the chord instead of along the track (however that can remedied via kml nitty-gritty). The track over the disturbance was further broken here into two independently rescaled segments that have to be tiled up again, with the parameters for that having been discarded. This is on top of mis-placed tick marks for ice depth markers, tick-marks obscuring experimental data, non-use of WGS84 calibration, and non-correction for flight incline (tilt).

Some of the Eqip flights probably followed what the crew perceived as flowlines. We have recent high quality products on velocity magnitude but not direction or trajectory. It’s not so easy to superimpose projections of the flight lines (kml files displaying in non-proprietary Google Earth), Sentinel 1A surface radar, Landsat and Cresis mercator, and various polar stereographic resources such as bedrock DEM.

This is critical though to determining whether the Eqip deformation is attributable (or at least correlated with) bedrock topography, age stratigraphy, flow orientation, or surface features. Note the very poor quality base photo from 25 Aug 2012 used by Google Earth shows, after substantial color tweaking, many small meltwater rivulets running directly over the deformation though no apparent moulins (2nd image).

The two-frame animation shows the relevent tracks within the deformation’s surface bounding box over the quaint 1990’s photomontage of Greenland that Cresis uses as locator map and a 10 Jan 16 Sentinel 1A. It’s quite difficult to register modern satellite imagery as the melt features and extent of ablation change so frequently. I have Landsat frame registered too and will add to this animation in a bit.

Fortunately satellite imagery for Jakobshavn almost always extends up to Eqip and Store. Since I have all the good ones for JI already, it won’t take long to make a high resolution velocity map for surface ice over the upheaval area, the expectation being that Eqip ice is under considerable tensile stress from ice piracy from these larger glaciers to both north and south.

Wipneus

  • Citizen scientist
  • Young ice
  • Posts: 4009
    • View Profile
    • Arctische Pinguin
  • Liked: 644
  • Likes Given: 0
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #372 on: February 08, 2016, 05:58:56 PM »
Images from the Sentinel 1A in EW (extra wide) mode are normally in what they call medium resolution (GRDM), bu over the Petermann they are delivered in high resolution (GRDH) as well. AFAICR that is 40 m/pix and 25 m/pix.
They can be in dual polarization as well. Here is a sequence of images from 20151220 and 20160206, 48 days apart. Color is derived from the polarization, Red=H, Green=V.

On the south bank a crack brightens and gets longer. I don't look at the Petermann images often enough to know whether it is significant.

Click for the animation to start.

Andreas Muenchow

  • New ice
  • Posts: 64
    • View Profile
    • IcySea
  • Liked: 2
  • Likes Given: 0
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #373 on: February 08, 2016, 06:53:30 PM »
When similar cracks in 2010 and 2012 reached the central channel separating the eastern from the western section of the glacier, a large break-up occurred. So yeah, I think this is important.
A Sailor in a Changing Climate
http://IcySeas.org

Espen

  • Young ice
  • Posts: 3251
    • View Profile
  • Liked: 177
  • Likes Given: 0
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #374 on: February 08, 2016, 10:30:45 PM »

On the south bank a crack brightens and gets longer. I don't look at the Petermann images often enough to know whether it is significant.

Petermann showed the same behavior during the freezing season 2014-2015, when the crack(s) expanded, and then slowed down again during the melt season in 2015. Why? I am not sure, only having some theories involving plasticity.
« Last Edit: February 08, 2016, 10:36:14 PM by Espen »
Have a ice day!

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #375 on: February 08, 2016, 11:18:24 PM »
The main crack on the southeast bank on the re-oriented image has not yet elongated though I expect it will this summer if the secondary crack below doesn't take over.

There are some changes though on the upper west bank. Basically the ice shelf is moving a little faster centrally than on this edge against the wall, with the effect of slightly straightening the cracks.
« Last Edit: February 08, 2016, 11:32:24 PM by A-Team »

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #376 on: February 09, 2016, 03:19:19 PM »
Ice sheets have long memories and those past conditions may remain partially determinative today and in the future. Looking at a Petermann region snippet in MacGregor 2016 got me wondering if the long-standing confluence of ice sheets  across the Nares could somehow explain today's anomalously fast velocity of Petermann well into the interior, as well as its extensive field of basal deformations.

There has been minimal coverage of the Innuitian Ice Sheet chronology on these forums despite very extensive research on the Ellesmere Island side. I located un-paywalled locations for the key articles:

http://forum.arctic-sea-ice.net/index.php/topic,176.msg43315.html#msg43315
https://www.researchgate.net/publication/223443510_The_Innuitian_Ice_Sheet_Configuration_dynamics_and_chronology J. England
ftp://soest.hawaii.edu/coastal/Climate%20Articles/Greenland%20thinning%20Holocene%202009.pdf B Vinther
http://www-udc.ig.utexas.edu/external/joemac/pdf/MacGregor_2016_Science.pdf MacGregor 2016

MacGregor 2016 write:
Quote
the northwestern sector of the GrIS was connected to the Innuitian Ice Sheet across Nares Strait. After the last deglaciation, the GrIS thinned rapidly at Camp Century. This thinning was attributed to the collapse of the Nares Strait ice bridge ~10 ka, and residual thinning may be ongoing.

The Holocene-averaged flow of this sector of the ice sheet was significantly faster than at present, and its subsequent dynamic deceleration is at least an order of magnitude greater than can be attributed to the LGP-Holocene viscosity contrast. This sector’s Holocene-averaged accumulation rate was significantly lower than at present, which also suggests that substantial dynamic thinning occurred there.

Together, these patterns indicate that faster ice flow in northwestern Greenland during the Holocene included a dynamic response to ice-bridge collapse.
The first image shows flow from the two opposing ice sheets meeting in a saddle across from present-day Hammond Glacier, with flow then diverging to the south and north (Petermann). The Ellesmere Ice Divide was not too far inland, meaning the bulk of Innuitian ice was to its west.

The long bay west of Judge Daly Promontory (opposite Petermann) is apparently where it reached Nares Strait. The whole thing started to collapsed around 10 kyr or some 1700 years into the Holocene, removing an immense buttressing effect from Petermann between 9.0 and 8.5 ky ago. This was followed by post-glacial rebound (3rd image).

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #377 on: February 13, 2016, 02:00:01 PM »
Petermann Glacier jumps out on every display of ice sheet velocities on the Greenland Ice Sheet though the explanation does not (unlike for NEGIS ice stream for which a geothermal hot spot is invoked). The rapid loss of buttressing from the Innuitian Ice Sheet across the Nares Strait in Canada may have played a role in creating the enhanced velocities we see today. However post-glacial uplift of 140 m at fjord mouth of Petermann, by reducing slope, would have the opposite effect.

Mostly I have been puzzling over the treatment of NW Greenland in MacGregor 2016 (see post #376), specifically why the authoritative account of the IIS in England 2006 (and its 91 subsequent citations) is relied on but seeming misquoted. This is not a trivial matter since the 9 kyr isochron plays a central role relative to the quoted 10 kyr collapse of the Nares Strait buttressing bridge versus the 8 kyr date actually provided in England 2006.

Quote
Granite erratics on northeast Ellesmere Island record the onshore flow of Greenland ice that extended p15km inland, overriding local summits (up to 840m above sea level (asl)) and contacting local glaciers that flowed east off the Hazen Plateau...

The absence of Greenland erratics farther inland indicates blockage by Innuitian ice that occupied tributary valleys west of Nares Strait. Farther south, at the north end of Kane Basin, the uppermost erratics (840 masl) are of Ellesmere Island provenance, indicating that the confluence between Greenland and Innuitian ice lay offshore.

Judge Daly Promontory glaciomarine sediments that mark Holocene marine limit (120 masl [date] the progressive entry of the sea from both the north and south ends of Nares Strait starting 9.0 ka BP. Deglaciation progressed to the centre of the strait by 7.5 ka BP, reconnecting the Arctic Ocean with Baffin Bay.

Buttressing by the Greenland Ice Sheet would have reduced the outflow of Ellesmere trunk glaciers into Nares Strait, allowing them to fill their fiords and result in the eastward migration of the Ellesmere Ice Divide.

At the south end of Nares Strait, convergent Innuitian and Greenland ice formed the Smith Sound Ice Steam that eroded large streamlined bedforms on land and possibly on the sea floor in northern Baffin Bay.

After 8 ka BP, the most persistent margin of the IIS occupied Nares Strait, where it remained in contact with the Greenland Ice Sheet until 7.5 ka BP.
Part of this may be due to respective uses of calendar years (from ice core layers or better, tree ring counts) versus C14 dates. Because C14 is not evenly produced in the upper atmosphere, it is necessary to use a calibration curve to relate the two. Because of its half-life, C14 is only useful for about 50 kyr; over that time frame, its dates are about 8% too low (inset, first image).

However it's not that simple because C14 levels are ambiguous especially for the key early Holocene -- a given proportion of C14 can equally well correspond to multiple calendar dates, up to six as shown in the second image. For the 2 kyr span shown, nearly 30% of C14 dating is ambiguous to a couple hundred years. Although that's only to a few percent, it matters greatly relative to the timing of dramatic calendar year events of the end-glacial and early Holocene.

Using the online calibration tool CalPal 2007 at http://www.calpal-online.de/  a C14 date of 8500 years prior to 1950 in England 2006 corresponds to a calendar age of 9521 BP which is in the right direction for MacGregor 2016. If the uncertainty was initially 250 years, the calendar uncertainty would be ±342.

However England 2006 does not date wood but rather marine clam shells in deglacialted deposits. These raise two additional issues, a lag in marine carbon dioxide equilibration with upper atmosphere C14 formation variously estimated as 300-800 years and whether carbonate deposit feeding (Portlandia arctica) or suspension-feeding clams (eg Mya truncata) were utilized, the former giving dates older by 300 years and more.

The interplay of issues here -- fluctuating sea levels, glacial advances, convergences and retreats, isostatic and elastic rebound from deweighting (2nd image), divergence of the jet stream by the 4800 m high Laurentide ice sheet south of the IIS -- is so complex that accurate reconstruction of events has proven elusive. Even ice cores have had problems from the brittle nature of Holocene ice, core locations remote or unrepresentative of the Petermann area, and technical issues with analysis, proxies and shifting interpretations.

Jim Hunt

  • Young ice
  • Posts: 4289
    • View Profile
    • The Arctic sea ice Great White Con
  • Liked: 274
  • Likes Given: 28
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #378 on: February 27, 2016, 05:53:25 PM »
I wondered why on Earth a CBS producer would want to follow me on Twitter. Then I discovered this:



Quote
FLYING OVER THE PETERMANN GLACIER IN GREENLAND, AUGUST 2015
Reality is merely an illusion, albeit a very persistent one - Albert Einstein

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #379 on: February 28, 2016, 12:32:52 AM »
Except that wasn't Petermann Glacier.

Wipneus

  • Citizen scientist
  • Young ice
  • Posts: 4009
    • View Profile
    • Arctische Pinguin
  • Liked: 644
  • Likes Given: 0
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #380 on: March 24, 2016, 12:49:30 PM »
No S2 image of the calving front yet. Here is a detail of the two upstream rifts discussed some months ago.

Click for the big image.

Wipneus

  • Citizen scientist
  • Young ice
  • Posts: 4009
    • View Profile
    • Arctische Pinguin
  • Liked: 644
  • Likes Given: 0
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #381 on: March 24, 2016, 05:30:54 PM »
Downloaded as a tile from the Amazone S2 site, here is an overview of Petermann at 50 m/pixel. Compared with the previous image, I let the bright peaks saturate somewhat to show more detail in the glacier

Click for the large image.

Wipneus

  • Citizen scientist
  • Young ice
  • Posts: 4009
    • View Profile
    • Arctische Pinguin
  • Liked: 644
  • Likes Given: 0
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #382 on: March 24, 2016, 05:34:44 PM »
And in high res, 10 m/pix, the calving front.

This one requires a click as well.

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #383 on: March 24, 2016, 05:45:23 PM »
Neat!  Do you have a 10 m piece handy for the critical area of the growing tip? Thx!

The ice shelf frontage looks very much like when we last saw it in Oct 2015. The image clarity is fabulous.

Wipneus

  • Citizen scientist
  • Young ice
  • Posts: 4009
    • View Profile
    • Arctische Pinguin
  • Liked: 644
  • Likes Given: 0
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #384 on: March 24, 2016, 06:05:06 PM »
Sure.


A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #385 on: March 24, 2016, 06:37:48 PM »
Any chance you could resample the tip down to 5 m with bicubic while still in 16-bit? Just one band would do, say band 4 red.

Wipneus

  • Citizen scientist
  • Young ice
  • Posts: 4009
    • View Profile
    • Arctische Pinguin
  • Liked: 644
  • Likes Given: 0
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #386 on: March 25, 2016, 09:03:42 AM »
Of course, although "bicubic" is a bit too much for me to understand atm. Reading the ImageMagick documentation (googling) gives me the impression that bicubic is not a single but a family of options.

So I just use the default, which when specifying -set option:filter:verbose 1 says:

Quote
# Resampling Filter (for graphing)
#
# filter = Cubic
# window = Box
# support = 2
# window-support = 2
# scale-blur = 1
# practical-support = 2
# B,C = 0.333333,0.333333

I attach 16 bits png files of the region with and without scaling. Hope that the 16 bits will survive the forum software.


nukefix

  • Frazil ice
  • Posts: 461
    • View Profile
  • Liked: 9
  • Likes Given: 1
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #387 on: March 25, 2016, 11:35:18 AM »
Why bi-cubic instead of the more standard bi-linear? Bi-cubic just smooths the end-result more...

https://blog.codinghorror.com/the-myth-of-infinite-detail-bilinear-vs-bicubic/

Andreas Muenchow

  • New ice
  • Posts: 64
    • View Profile
    • IcySea
  • Liked: 2
  • Likes Given: 0
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #388 on: March 25, 2016, 03:34:16 PM »
Wonderful discussions and imagery. Here is what the LandSat Pan-chromatic Band-8 looks like at 15-m resolution of the same crack that, I am certain, will form the next ice island within a year or two.

I re-gridded and re-projected the LandSat data for ease of quantitative analyses with 'old-fashioned' codes and scripts that, god forbid, even includes Fortran ;-)

EDIT: Spelling. Downloading and processing the original 1 GByte data of a single scene (all 11 Bands) takes less than 5 minutes using nothing more than a MacBook Pro laptop with open-source software only.
« Last Edit: March 25, 2016, 03:57:11 PM by Andreas Muenchow »
A Sailor in a Changing Climate
http://IcySeas.org

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #389 on: March 25, 2016, 03:43:56 PM »
Quote
So I just use the default
tl;dr:
-- use cubic
-- do early operations in 16-bit
-- share 16-bits on cloud as 1-band tifs like Landsat

Bicubic splines (called 'cubic' in command menus) are what the Landsat-8 and Sentinel 2A have already used getting us to where we are. Sinc (Lanczos3), which has a more information-theoretic basis, was likely considered but not chosen. One is maybe better for downsampling, the other for upsampling. (We've had this discussion before.)

One thing's for sure: ESA would not post at 10 m if they could have had bragging rights for 5 m. They would not have collected at 12-bit if they could have collected 16-bit within error. So we need to be careful not to push too hard on any form of further interpolation since we have no true 5 m or same-day ground measurements to test outcomes against.

However there may be a slightly better option when upsampling to exact 2x multiples. It is not commonly implemented being too restricted (eg not applicable to making 10 m Landsat from 15 m). We commonly shoot for 700 pixel forum width which involves arbitrary multiples in either direction but can attach the original resolution or point to it as necessary.

Nearest-neighbor and bilinear are noticeably sub-optimal (as 0th and 1st order fits to the immediate neighborhood of a given pixel) and rarely used enlarging satellite ground imagery, continuous artwork or photography. They were used in times of limited desktop CPUs (pre-1995) and sometimes even today for quickie monitor blowups and special situations.

You will find no end of silliness online about enlarging abrupt-contrast pixelated cartoon icons, like the fellow above who pleads for vector enlargement, unaware the whole point of bicubic splines is a vector fit capable of arbitrary enlargement, in other words fonts for photos.

It all comes down to how abruptly contrast varies on the ground, what wavelength is under discussion, how the sensor and software decide what goes into a pixel, and what all contributes to spillover and other errors there. On an ice sheet, there actually is information about your favorite 10 m square from its neighboring squares, but with rapidly diminishing returns beyond a two pixel radius.

We are not terribly interested in the literal top-of-atmosphere numerical reflectances from the ice. Some people are. In fact the cryosphere forums are seldom concerned with single-pixel attributes (which species of tree is that?). More commonly we are looking at multi-pixel correlated attributes such as lines (crevasse propagation) or blobs (velocity measurement).

Those are a whole different topic in enhancement because a line or blob can survive enlargement much better (remain perceptually recognizable) than a single pixel because of retained correlation. And lines are anisotropic (have an inherent directionality) which can and should be exploited.

For example, photo can be rotated so lineations are east-west, allowing simple convolutions (bump maps in gimp) to be run north-south. This is a huge deal on radargrams which have many horizontal isochrons that are initially indistinct. Even better is an adaptation to local slope. Etc for surface crevasse fields.

In summary, limited bicubic enlargement beyond the posted resolution has no physical rationale for single pixels (0D) but can be justified for lines (1D) and blobs (2D) which have an additional non-local statistical coherency.
« Last Edit: March 25, 2016, 04:05:53 PM by A-Team »

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #390 on: March 25, 2016, 04:12:47 PM »
Quote
cracks, I am certain, will form the next ice island within a year or two
I am of the same opinion but am betting on strain transfer shifting to the lower crack as the ice shelf slowly moves relative to a fixed feature on the eastern shoreline that marks, but does not cause per se, crack initiation. Moving up the shelf on both sides, quite a few cracks are no longer propagating.

Assuming these cracks extend to full depth and re-freeze in winter from sea water (rather than freshwater melt), that would make for a weak briny bond, implying the strain on them has diminished or they wouldn't become inactive.

Overall the geometry of Petermann is extremely rigid except for the sides (which are very different from each other) and pivoting around a few widening cracks. By displacing Landsats down the channel, precise overlays spanning years of Landsats can be made.

If only we could get someone to install a horizontal interferometer! Then we could watch cracks widen in real time, sort of like watching Nevada stretch with real-time GPS grid tectonics as North American plate motion is opposed obliquely by that of the larger Pacific plate.

One point of confusion we've had on the forum: yes, new shelf is continuously being created off the pinning zone (~1100 meters per year of it); no, that is not remotely keeping up with averaged-out recent rate of loss of ice islands: a single calving event can undo many years of extension.

Petermann glacier is not accelerating appreciably though its velocity is noteworthy well back into the interior, which I proposed elsewhere is a continuing response to Innuitian unbuttressing. (Not so long ago, the Greenland ice sheet  extended 15 km inland on the other side of Nares Strait.)

Thus ice shelf creation will remain slow but steady, being no match for the effect of new warm ocean water circulation under the floating ice shelf.
« Last Edit: March 25, 2016, 04:33:05 PM by A-Team »

Andreas Muenchow

  • New ice
  • Posts: 64
    • View Profile
    • IcySea
  • Liked: 2
  • Likes Given: 0
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #391 on: March 25, 2016, 04:28:04 PM »
Agreed on all points.

If only we could get someone to install a horizontal interferometer! Then we could watch cracks widen in real time, sort of like watching Nevada stretch with real-time GPS grid tectonics as North American plate motion is opposed obliquely by that of the larger Pacific plate.

I am working on getting up to Petermann this summer via helicopter from Qaanaaq to service the three ocean-glacier observing sites that we placed in the summer of 2015 from I/B Oden. I still got 3 UNAVCO fancy GPS receivers that allow sub-centimeter vertical and horizontal displacement. The 3 units could be placed in a triangle around the crack. If you know someone who has such an interferometer, please let me know (via e-mail, perhaps).

The attached plot shows the results from a 2 week deployment last summer at the grounding line of Petermann (black lines) as well as 26 km downstream on the freely floating ice shelf (red lines) and 15 km upstream where the glacier sits on bed rock (blue lines).

EDIT: On the horizontal displacement, I removed the 2-week mean displacement at each site (~700 m/year upstream and ~1250 m/year downstream), so only "anomalies" from this mean motion are shown. The vertical displacement is actual raw measured displacement estimated from 30 second samples (but I also got 1 second samples), because this is what UNAVCO reference stations on bed rock at Kap Morton and Kap Schoubye sample at.
A Sailor in a Changing Climate
http://IcySeas.org

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #392 on: March 25, 2016, 04:44:46 PM »
Very interesting.

I have some serious background chaos (visiting relatives) for the next few days. Briefly, we do have ultra-high precision historic worldview satellite data at google earth. Even with Sentinel, the cracks are not that many pixels wide, so one pixel of widening will not be satisfactorily detectable. However Sentinel is not too bad at crack lengthening, which can be leveraged to widening. I looked at all the distraction on Landsat blowups around the tip, the two little melt lakes make a great frame of reference but change seasonally in appearance.

CraigsIsland

  • Frazil ice
  • Posts: 197
    • View Profile
  • Liked: 0
  • Likes Given: 1
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #393 on: April 13, 2016, 08:58:21 PM »
Interesting eco-tourism happening on the glacier: http://the-blue-river.squarespace.com/

magnamentis

  • Guest
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #394 on: April 13, 2016, 10:31:48 PM »
are they aware that this water can be sucked into precipices at any time and them with it?
« Last Edit: April 14, 2016, 02:02:01 AM by magnamentis »

CraigsIsland

  • Frazil ice
  • Posts: 197
    • View Profile
  • Liked: 0
  • Likes Given: 1
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #395 on: April 14, 2016, 12:36:09 AM »
are they aware that this water can be sucked into precipice at any time and them with it?

lol. I have no idea. Was looking at this from a web design perspective. While "cool" in that you can have the unique blue waters, it's not a good sign for things to come.

A-Team

  • Nilas ice
  • Posts: 2453
    • View Profile
  • Liked: 305
  • Likes Given: 22
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #396 on: April 14, 2016, 04:26:15 PM »
I’ve noted a dozen or so abstract titles from the April 2016 EGU meeting on the What’s New forum and am expanding a bit on them on the appropriate forums. These two Petermann meltwater abstracts are sketchy but we can expect additional information shortly from co-author A Muenchow’s excellent web site https://icyseas.org/ as well as an open access version of whatever journal papers are in preparation.

Quote
Pathways of Petermann Glacier’s Meltwaters, Greenland
Céline Heuzé
23 Feb 2016 AGU New Orleans

Radar and satellite observations suggest that the floating ice shelf of Petermann glacier, north Greenland, loses up to 80% of its mass through basal melting, caused by the intrusion of warm Atlantic water into the fjord and under the ice shelf. Although Greenland meltwaters are key to sea level rise projections and can potentially disrupt the whole ocean circulation, the fate of Petermann’s glacial meltwater is still largely unknown. It is investigated here, using hydrographic observations collected during a research cruise onboard in August 2015.

Two layers are found: one at 200 m (ice shelf terminus depth) mostly on the eastern side of the fjord where a calving event occurred this summer, and one around 500 m (grounding line depth) on the western side.

At the sill at the end of the fjord, approximately 3 mSv* of freshwater leave the fjord around 150 m depth on the eastern side. On the western side, a more complex circulation occurs as waters intrude in. Outside of the fjord in Hall Basin, only one layer is found, around 300 m, but its oxygen content and temperature-salinity properties suggests it is a mixture between Petermann’s meltwater, meltwater from the neighbouring glaciers, surface run-off and sea ice. As Atlantic water warms up, it is key to monitor Greenland melting glaciers to properly assess sea level rise.

Quote
Pathways of Petermann Glacier meltwater, Greenland
Céline Heuzé, Anna Wåhlin, Helen Johnson and Andreas Münchow
EGU General Assembly 15 Apr 2016

… Two methods are used to detect the meltwater from Petermann: a mathematical one that provides the concentration of ice shelf meltwater, and a geometrical one to distinguish the meltwater from Petermann and the meltwater from other ice shelves. The meltwater from Petermann mostly circulates on the north side of the fjord. At the sill, 0.5 mSv of meltwater leave the fjord, mostly on the northeastern side between 100 and 350 m depth, but also in the central channel at lesser concentration.

Meltwater from Petermann is found in all the casts in Hall Basin, notably north of the sill by Greenland coast. The geometrical method reveals that the casts closest to the Canadian side mostly contain meltwater from other unidentified glaciers. As Atlantic Water warms up, it is key to monitor Greenland melting glaciers and track their meltwater to properly assess their impact on the ocean circulation and sea level rise.

*The sverdrup is an rogue honorific unit used strictly within oceanography. Not approvable within SI, its symbol sv had already been allocated to sievert, a measure of radioactivity. One sverdrup represents to an enormous current of water, one cubic kilometer of water per second crossing a flux gate (whose cross-sectional area is unspecified and possibly very broad). All the world’s rivers carry 1.2 sverdrups (making the Petermann flux 0.25% of this); the Gulf Stream varies from 30-150 sv; the Antarctic Circumpolar Current at 125 sverdrups is the largest ocean current. Petermann’s 3 milli-sverdrups represents 3,000 cubic m/s of meltwater exiting the fjord.

Sverdrups are not used in the US, river flow is measured in cfs (cubic feet per second). One sverdrup amounts to an inconvenient 35,314,667 cfs; Petermann works out to 105 944 cfs, about five times that of the Colorado River but only a fifth of the average discharge of the Mississippi River. https://en.wikipedia.org/wiki/List_of_U.S._rivers_by_discharge

These numbers seem impossibly high for Petermann but probably represent an August peak rather than a year-round average, and one that assumes zero mixing with ambient seawater. However since sverdrops are not normalized to cross-sectional area of the flux gate, a slow current across a wide and deep cross-section can yield a large number of sverdrups, extrapolation out from a small number of casts. However in terms of oceanograpphic considerations such bulk heat transport and freshwater density effects, sverdrops work well as defined.

It’s not entirely clear where this freshwater is coming from. Landsat scenes in 2015 showed very little flow in the central channel, little water puddling in the usual northeast side depressions and only briefly filled up-glacier meltlakes. Meltwater formed up-glacier exiting at the base of the grounding zone has been measured by sonar sweeps for a handful of west Greenland glaciers but not yet reported for Petermann.

Petermann is pushing 600 m thick ice across the grounding zone yet this thins to 200 m by the calving terminus. Since dynamic thinning is minimal on a floating ice shelf, the missing ice has presumably been melted by contact with circulating ocean water (which continues even in winter) and hydrostatically adjustment after surface melt. Petermann is at 80ºN so the seasonality of sunlight induced melt is strongly windowed.
« Last Edit: April 14, 2016, 04:32:55 PM by A-Team »

wili

  • Young ice
  • Posts: 2600
    • View Profile
  • Liked: 173
  • Likes Given: 305
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #397 on: April 14, 2016, 07:29:18 PM »
"Sverdrups are not used in the US"

Interesting, since Sverdrup himself lived and worked most of his adult life in the US.

There are a number of Sverdrup's in my area (Minneapolis). A building in the local college (Augsburg) bears the name, and the major interstate highway bridge (35W) that collapsed a few years ago was designed by a Sverdrup, iirc.

(Sorry for the OT comments. Back to science!)
"A force de chercher de bonnes raisons, on en trouve; on les dit; et après on y tient, non pas tant parce qu'elles sont bonnes que pour ne pas se démentir." Choderlos de Laclos "You struggle to come up with some valid reasons, then cling to them, not because they're good, but just to not back down."

Iceismylife

  • Frazil ice
  • Posts: 281
    • View Profile
  • Liked: 1
  • Likes Given: 1
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #398 on: April 14, 2016, 10:56:41 PM »
 in standard units one inch of water loss per day of exposed ice per degree c difference between sea water temp and the local ice melting point.

so 100,000 cfs needs 309 square miles of ice exposed to sea water 1 c above the local melting point.  Is this reasonable for the size of the ice tong?

Andreas Muenchow

  • New ice
  • Posts: 64
    • View Profile
    • IcySea
  • Liked: 2
  • Likes Given: 0
Re: Petermann Gletscher / Petermann Fjord / North West Greenland
« Reply #399 on: April 15, 2016, 11:17:25 PM »
A freshwater flux is

sum of (velocity * (1 - Ssalinity/Sref) times area_of_section

where Ssalinity is the measured salinity that varies with depth and across the fjord and Sref is some constant reference salinity that I take as the warm and salty Atlantic water that does the melting of the ice shelf from below. So the units are meters/second * meters * meters to give meters^3/s. A Sv (Sverdrup) is one million of those m^3/s. My favorite Delaware River discharges about 500 m^3/s on average (the Thames River near London is about half of that).

In both a prior publication, e.g., Johnson et al, 2011 (.pdf available at http://muenchow.cms.udel.edu/html/publications.html ) and a new one that uses 2015 data, we find that that indeed about 2000-3000 m^3/s of freshwater exits Petermann Fjord. Dr. Heuze reports this as 2-3 mSv, but this value includes ALL sources of freshwater such sea ice melt, run-off from land, as well as the portion glacier melted by the ocean and by air. In the new publication (not finished yet), we try to distinguish between the ocean's glacier meltwater fraction from all others and it comes out as perhaps 10-20 % of the total freshwater flux across two 2015 sections of independent estimates. These 300 m^3/s originate from the the part of Petermann Gletscher that is melted by and mixing with the warm Atlantic waters.

Just for comparison, the freshwater flux numbers for Nares Strait are about 50,000 m^3/s and that of the Amazon River are about 100,000 m^3/s. So the Arctic Ocean discharges about half an Amazon River of freshwater via Nares Strait into the North Atlantic. Petermann Fjord's contribution at 2,000 or 3000 m^3/s is a small addition that will get lost in the noise of an uncertainty estimate. More details in Muenchow (2016) which is on the above web-site as a .pdf as well (open access this one).

There are devils and details; for example, we do not measure velocity directly (proposal to do so was declined twice many years ago) and thus have to make assumptions that (a) the flow is the result of a specific and simple balance of forces (pressure gradient and Coriolis balance, so-called thermal wind balance from meteorology) and (b) that the total flux of total water is zero to a first approximation. The sum of velocity times salinity anomaly is really an integral across the section (depth across fjord) of the velocity normal to the section with velocity weighted by their salinity. So, velocities where salinity is low (fresher water close to surface) is counted more than velocity where the salinity is high (salty Atlantic waters close to bottom).

ADDENDUM: Velocity section across Petermann Fjord at the sill to Nares Strait. View is from glacier out towards Nares Strait. Red colors are flow out of the fjord, blue colors are flow into the fjord from Nares Strait. Left panel is ocean density from which I subtracted 1000 kg/m^3. Sloping lines of density correspond to velocity (differences in vertical). Stations where the ship stopped and took an ocean profile of temperature, salinity with depth are black triangles at top and thin white lines with depth.
« Last Edit: April 16, 2016, 06:50:00 AM by Andreas Muenchow »
A Sailor in a Changing Climate
http://IcySeas.org