Support the Arctic Sea Ice Forum and Blog

Author Topic: Greenland subglacial topography  (Read 71714 times)

idunno

  • Frazil ice
  • Posts: 187
  • wonders are many
    • View Profile
  • Liked: 2
  • Likes Given: 0
Greenland subglacial topography
« on: May 19, 2014, 11:52:18 AM »

idunno

  • Frazil ice
  • Posts: 187
  • wonders are many
    • View Profile
  • Liked: 2
  • Likes Given: 0

Espen

  • Young ice
  • Posts: 3383
    • View Profile
  • Liked: 254
  • Likes Given: 1
Re: Greenland subglacial topography
« Reply #2 on: June 15, 2014, 09:02:06 PM »
Idunno,

Very interesting and surprising?
Have a ice day!

sidd

  • First-year ice
  • Posts: 5895
    • View Profile
  • Liked: 823
  • Likes Given: 0
Re: Greenland subglacial topography
« Reply #3 on: June 15, 2014, 09:53:08 PM »
i am still on the road, so i have not yet read the paper carefully. One thing that jums out is the colossal amount of heat transport involved as the meltwater carries latent heat which it gives up when refreezing, which warms surrounding ice toward lower viscocity regimes.

sidd

wili

  • Young ice
  • Posts: 3342
    • View Profile
  • Liked: 612
  • Likes Given: 409
Re: Greenland subglacial topography
« Reply #4 on: September 29, 2014, 06:36:36 PM »
This seems like an important paper (thanks to todaysguestis at robertscribbler's blog for this):

Greenland Ice Sheet more vulnerable to climate change than previously thought

http://www.eurekalert.org/pub_releases/2014-09/uoc-gis092614.php

Quote
...the new model also takes into account the role that the soft, spongy ground beneath the ice sheet plays in its changing dynamics...

"When these large ice sheets melt, whether that's due to seasonal change or a warming climate, they don't melt like an ice cube," said Dr Marion Bougamont of Cambridge's Scott Polar Research Institute, who led the research.

"Instead, there are two sources of net ice loss: melting on the surface and increased flow of the ice itself, and there is a connection between these two mechanisms which we don't fully understand and isn't taken into account by standard ice sheet models."

Whereas other models of the Greenland Ice Sheet typically assume the ice slides over hard and impermeable bedrock - an assumption which is largely practical and based on lack of constraints - this study incorporates new evidence from ground-based surveys, which show soft and porous sediments at the bed of the ice sheet, more like the soft and muddy bottom of a lake than a sheet of solid rock. The new study specifically identifies the intake and temporal storage of water by weak sediment beneath the ice sheet as a crucial process in governing the ice flow.

"Not only is the ice sheet sensitive to a changing climate, but extreme meteorological events, such as heavy rainfall and heat waves, can also have a large effect on the rate of ice loss," said Dr Christoffersen. "The soft sediment gets weaker as it tries to soak up more water, making it less resistant, so that the ice above moves faster. The Greenland Ice Sheet is not nearly as stable as we think."

While complete loss of all ice in Greenland is judged to be extremely unlikely during this century, the record extent of surface melting in the past decade clearly shows that the ice sheet is responding to Earth's changing climate.
"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."

Laurent

  • Young ice
  • Posts: 2538
    • View Profile
  • Liked: 6
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #5 on: October 27, 2014, 09:04:04 AM »
Greenland Ice Sheet Failure May Be More Rapid Than Previous Estimates
http://planet3.org/2014/10/01/greenland-ice-sheet-failure-may-be-more-rapid-than-previous-estimates/

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #6 on: November 16, 2014, 01:02:45 PM »
The Bougamont paper mentioned above is the ~6th competing mechanism proposed to explain icesheet movement via properties at the bottom. Fluidized till has also been invoked to explain rapid movement of Zachariae in northeastern Greenland. The others are basal meltwater lubrication, efficient channelization of meltwater (resp. seasonally inefficient), depth and fluidity of temperate basal ice, caterpillar motion from released frozen-on regions, and local variations in the earth's geothermal gradient.

This paper brings out the fact that actual direct observation of the icesheet/bedrock interface is minimal (nine boreholes distributed over 543,000,000 acres, two cored with rock bits). In lieu of extensive factual constraints, models assume a particular basal interface and look for inconsistencies with more easily observed surface velocities and borehole hydraulics.

In terms of GIS layers, we now have a good bedrock surface topography layer but are still missing overlays for geothermal heat; depth, distribution and hydration state of glacial till; meltwater and freeze-on sites; soft temperate ice thickness, and indeed the all-important temperature isotherms.

The geothermal heat flux at the level of bedrock does not directly affect mechanical properties of till (Bougamont paper) but does influence opportunities for its saturation by helping ice at depth attain its pressure melting point.

The images below show varies attempts at providing Greenland-wide background geothermal flux. The first is just me dinking with the gradient tool in Gimp -- the 'shaped dimple' setting causes the gradient to work its way inward from a very irregular boundary (interior at sea level or below), followed by color draping over the bedrock DEM.

This results in the geothermal gradient highest under the summit ridge, accompanied by a fortuitous (?) thermal lobes corresponding to the Petermann canyon and the NEGIS icestream in the direction of Zachariae. This makes some sense as isostatic thinning of mantle is greatest under the thickest ice, which results in less of a barrier to heat flow.

The second image shows three models with experimental underpinnings: geothermal flux averaged over similar tectonic units elsewhere, seismic tomography from lithosphere boundary reflection, and satellite mapped remnant crustal magnetization that determines the (580º C) Curie isothermal surface (±15%). These are thoroughly discussed in Rogozhina 2012 (free, full: http://tinyurl.com/qjrolar).

This paper also furnishes pressure point basal melting temperature maps under these and uniform geothermal flux scenarios (3rd image) that do not agree terribly with observed (NEEM is non-temperate, NGRIP  -2.4ºC temperate, GRIP -8.56ºC, GISP2 -9.05ºC, Dye -13.22ºC) but provide a conceptual start.

The melting point of freshwater ice drops 1 degree C per 1500m of ice sheet depth per Clausius-Clapeyron (slope of water phase diagram). However dirty ice at the bottom of a glacier has other constituents that also affect the melting point at pressure.

Greenland geothermal heat flux at bedrock boreholes varies by a factor of seven from south to central. The numbers are small (20 to 140 mW/m2) compared to mid-summer sun on the (increasingly darkening) surface which can provide ~460 w/m2 (watts vs milliwatts). There is also a pronounced west to east increasing gradient of ~30 mW/m2, attributed to thinning lithosphere said to have resulted from Greenland passing over the Iceland plume 55 myr ago (3rd figure, Jakovlev 2912).

Energy fluxes cut both ways -- the onset of late Pleistocene glaciation cooled the lithosphere to 15 km depth; that imprint still affects the geothermal gradient today, requiring that climate history be coupled to thermomechanical models of the icesheet and its underlying lithosphere (Petrunin 2013). Here heat released by radioactive decay in the crust is still a significant unknown.

The bottom line (fig.4) is considerable variability of geothermal heat flux at bedrock even within central Greenland, with important consequences for the pressure melting point, thickness of temperate ice, and resistance to sliding and deformation.

Ice-penetrating radar also has potential for determining areas of basal melt along flight lines, which somewhat oddly are predominantly north-central. That experimental data (Ostwald 2012) has yet to be integrated into comprehensive models and came too late to inform summit drill site choices.
 
Effects of uncertainties in the geothermal heat flux distribution 2 on the Greenland Ice Sheet: An assessment of existing heat flow models
I Rogozhina et al
Journal Of Geophysical Research doi:10.1029/2011JF002098, 2012
free, full: http://tinyurl.com/qjrolar

Heat flux variations beneath central Greenland’s ice due to anomalously thin lithosphere
AG Petrunin et al
Nature Geoscience 2013 doi:10.1038/NGEO1898
free full figs: http://www.nature.com/ngeo/journal/v6/n9/fig_tab/ngeo1898_F4.html

Structure of the upper mantle in the Circum-Arctic region from regional seismic tomography
AV Jakovlev
Russian Geology and Geophysics 2012 | 53 | 10 | 963-971
request at ResearchGate: http://tinyurl.com/mmexoau

Mapping Basal Melt Under the Northern Greenland Ice Sheet
GKA Oswald
IEEE Transactions On Geoscience And Remote Sensing 2012 doi: 10.1109/TGRS.2011.2162072
« Last Edit: November 16, 2014, 01:10:02 PM by A-Team »

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #7 on: November 19, 2014, 01:21:04 PM »
The bedrock DEM layer is important both in itself and in many products derived from it. The example here, taken from an obsolute 2009 paper, calculates a simplified hydraulic potential surface from ice thickness (surface DEM - bedrock DEM) scaled up by ice density and gravity to the weight of ice overhead.

Meltwater conduits, moulins and fluid dynamics not being considered in hydrostatics, the basal water pressure amounts to ice overburden. The surface slope will dominate the direction of water flow unless bedrock slope is substantially greater. From this, the paper goes on to include an englacial meltwater layer and consider the implied hydrological network draining the bottom of the icesheet. (A 2013 paper by other authors supercedes the analysis here: http://www.the-cryosphere.net/7/1721/2013/)

This is one of the very few publications in Greenland glaciology that actually uses standard earth science (calculations on stacks of co-registered layers in ArcGIS -- which could also be done in ImageJ or Gimp). Today, the two layers taken from Bamber 2001 would be replaced by greatly improved layers from Bamber 2013 but that would suffice to automatically update the two product layers.

The real idea with GIS (calculations on co-registered layers) is that a research community puts experimentally determined layers (and their periodic updates) into a shared pool in a convenient universal format. Then anyone can grab the layers needed for their subsequent projects and combine them quantitatively in a fast and intuitive photoshop-type environment, which not incidentally outputs publication quality graphic.

The vast majority of glaciology layers are scalar fields of moderate precision, meaning an 8-bit grayscale graphic is a very attractive format because this is so easily resampled to bring all layers to a common ground resolution all tied to Greenland lat,lon coordinates in polar stereographic or interconvertable mercator projection.

Vector fields such as surface velocity need RGB color to represent their Vx,Vy,Vz components from which a grayscale velocity magnitude layers follows. Glaciology uses little else besides second rank tensor products of vector fields.

One of the great peculiarities of journal publications in glaciology is GIS layers implicit in the figures are almost never provided satisfactorily in graphics or supplemental. Graphics rarely provide data ground resolution or map projection, precious data is irreversibly overlaid with clumsy coastline or gratuitous coordinate grids, ill-advised color keys make the underlying grayscale unrecoverable, and incomprehensible use of jpg compression further degrades the data.

While the paper here is one of the few that 'gets it' -- depositing ArcGis shape files at the NSIDC repository -- it too does not provide its layers as convenient and usable graphics. No 21st century scientific journal discourages or restricts supplemental material, print requirements are as irrelevant as a horse-drawn carriage, so there is no excuse for not providing the real data underlying the paper graphics.

The NASA visualization page for Greenland surface elevation change shows best practices: http://svs.gsfc.nasa.gov/cgi-bin/details.cgi?aid=584

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #8 on: November 19, 2014, 02:34:22 PM »
The bedrock DEM of Greenland should actually be provided as a function of time because it varies with the extent of loading -- glacial isostatic adjustment. This load peaked at the Last Glacial Maximum and bedrock has been rebounding -- with fits and starts -- ever since, at rates that will provide significant impacts in this century on icesheet discharge.

Here it is important not to confuse the rapid elastic lithosphere response component with much slower viscoelastic asthenosphere response that always lags the load. Hudson Bay and Fenno-Scandanavia point to the future: the former is still rising rapidly 10 kyr after the fact at the expense of sinking northern US states while latter has risen 280m already with some 700m more to go at the center of depression.

Glacial isostatic rebound in Greenland is currently measured from five longterm GPS stations embedded in margin rock  exposed now that the vastly more massive icesheet of the LGM has retreated from edge of the continental shelf.

The interpretive situation in eastern Greenland is greatly complicated by the proximity of Iceland and its (much disputed) plume of rising mantle. It's still asserted that this hotspot tracked under central Greenland at 55 myr accounting for thinning lithosphere and uneven geothermal gradient, though this no longer reflects mainstream thinking.

In fact, one influential paper of 2001 (http://tinyurl.com/l5wgh3s, 117 cites) even reasoned that the NEGIS icestream begins over a contemporary cauldera, invoking the Yellowstone hotspot for comparable conductive heat flux and extrusive bedrock structure.

While that seems preposterous geophysically then and now, recent research strongly supports a moderate excursion of the geothermal gradient in this region. The current bedrock DEM -- based on vastly more but still sparce radar tracks -- shows no extrusive structures anywhere poking through Greenland's Precambrian shield, though volcanic flows and Baffin-dipping seismic reflectors attributable to a failed ancient tectonic rift occur in the Disko region of western Greenland.

Another high visibility paper compared an apparent DEM paleo-fluvial feature to the Grand Canyon, though the authors did not make clear how the drainage system of pre-glaciated Greenland could be constrained without a pre-glaciated bedrock DEM. Greenland today resembles an atoll, with much of the interior at or even below sea level, reflecting the greatest loading under the summit ridge and so the site of greatest future rebound.

The attached animation (adapted from Fig.6 SJ Livingstone 2013 doi:10.5194/tc-7-1721-2013 free full) makes a start on providing the needed time-dependent bedrock DEM. It doesn't get too far with this because the disappearing ice sheet is shown from 19-5 kyr over the fixed 0 kyr DEM and loading over time may or may not scale proportionally to ice sheet extent.

Here the loading situation is quite muddled just back to the Eemian, with a pine twig found at the bottom of NGRIP and complete lack of pre-Eemian ice anywhere (unlike Antarctica) contrasting with repeated claims that the icesheet has perhaps been continuously in place for three million years.

It might make the most sense to start DEM(t) at the LGM because there the viscoelastic response has had a chance to catch up (approach isostatic equilibrium). DEM(t) needs the surface and thickness time dependencies; these in turn need the snow accumulation and distribution. Sources and quantity change over time but their history is becoming directly measurable from radar sounding horizons, unlike bedrock whose history is only indirectly implied.
« Last Edit: November 19, 2014, 02:54:29 PM by A-Team »

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #9 on: November 19, 2014, 03:29:21 PM »
Here is a sophisticated paper from this summer on real-time rebound that isolates elastic response from the glacial isostatic adjustment proper of the asthenosphere. The latter is large enough to affect mass balance measurements from GRACE, which provides our best overall perspective on what Greenland is contributing to sea level rise.

Despite just 5 GPS stations continuously monitoring exposed coastal bedrock elevations for less than a decade, some trends can be noted. Fig.3 of the paper provides a short time series of elastic uplift. Changes are not so easy to see so I subtracted the images in all pairwise combinations and renormalized to neutral gray (grain extract in gimp).

The melodrama over at Jakobshavn Isbrae and Kangerlugssuuq is only to be expected; the Zachariae basin is on its way up as well, but Petermann is not showing striking changes.

Free full: http://orbit.dtu.dk/services/downloadRegister/80913070/nielsen2014.pdf

Greenland sometimes gets included in studies of glacial isostatic adjustment of Hudson Bay (based on a vastly more extensive array of GPS devices). Models assume an ice history (time-dependent extent and thickness) that loads a specified rheological model of mantle viscosity, for example upper mantle viscosity of 10 exp 21 Pa s graded to 4x that at depth.

These models predict the extent and rate of uplift as land under the former ice load slowly rebounds but subsidence beyond the sheet margin as the forebulge sinks. Little horizontal motion -- attributable to relaxation of bent lithosphere and asthenosphere flow occurs at the load center but elsewhere land moves outward to a hinge line (green line in second image below, from GF Sella 2007 free full doi:10.1029/2006GL027081) whereupon it reverses and points inward (as in western Greenland).

Post-glacial rebound has also been thoroughly studied in Fenno-Scandanavia. It proved important there to get real data on the geothermal heat flux rather than rely on a few boreholes: see the contrasting distributions below. Of course the icesheet is long gone there whereas access to bedrock in Greenland is very limited.

Geothermal heat-flow data for basal temperatures and meltwater production beneath Fennoscandian ice sheet
Naslund JO 2005
http://scholarworks.umt.edu/cs_pubs/5
« Last Edit: November 20, 2014, 11:30:23 AM by A-Team »

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #10 on: November 20, 2014, 01:54:02 PM »
Finally -- 21 years into the ice-penetrating radar program  -- a UTexas team may have pulled together the internal stratification data so important to understanding past and future of the Greenland ice sheet.

While no peer-reviewed article has been published and the UT-hosted data center remains empty, NASA Cryospheric has awarded its obscure 2014 MVP Award to team leader JA MacGregor for "synthesizing ice penetrating radar records to produce a 3-dimensional age map of the Greenland Ice Sheet", apparently based on four nearly identical meeting abstracts given over the last 15 months, none of which have associated ppts. See http://www.ig.utexas.edu/people/staff/gcatania/greenradar.html

I did locate two leaked UT graphics (below) that give some idea of the output. These do not seed the recovery of additional online product using either TinEye or Google  search-by-image. The upper image may be a poster or upcoming journal cover; the lower shows fairly conventional color tracking of several dozen radar reflection horizons along very long transects that do not correspond to any actual flight line or even campaign. Good.

NSF grants to authors speak about a database called LAYERMAP being "gridded in a manner similar to that for bed elevation underneath ice sheets". More good. https://www.collectiveip.com/grants/NSF:1107753

The web page mentions "an army of UT undergraduates" suggesting semi-automated methods. Despite dozens of initial papers on automated layer-picking  none have been pursued to any meaningful extent; here the web site speaks of a novel technique developed by M Fahnestock that works across a whole spectrum of radar designs (CReSIS, sled-drawn, PARCA, IceBridge, Danish). J Paden and G Catania provide interpretive expertise on the first two; I've never been able to locate online PARCA or Danish transect repositories.

It could just be a novel technique that they hope will someday work out, accounting for the long delay in populating the database. It's not clear how they're independently validating line-picks: not every peak in ice core conductivity correlates with a radar horizon -- far from it. Radar has never been physically applied to NEEM or NGRIP cores to obtain horizons; synthetic (virtual) radar hasn't either and has had only moderate success in reproducing profiles elsewhere.

Dating radar horizons accurately is not so easy even for flight lines directly over (or tied to) NEEM and NGRIP because radar reflections are quite thick (peaks undeterminable to tens of meters) relative to annual ice core layers (perhaps a cm). Although radar reflections can be given precise dates if they correspond to a short-lived volcanic deposition of dielectric impurities, only relatively recent (historic) eruptions have accurate dates. Other radar horizons may correspond to melt layers spread out over centuries of elevated temperature.

I wish them well on this and other stated objectives that include post-processing of layer picks to get at surface elevation and mass balance history over the Holocene, accumulation history, detection of regions of basal melting and sliding, delineation of englacial frozen and thawed regions, and GIS-wide internal temperatures.

This sounds like a mix of direct experimental read-out with lots of interpolation that will precede much modelling that will come before many simplifying assumptions that will precede yet more interpolation that will utilize guesswork on the geothermal gradient that will anticipate yet more modelling that drops inconvenient terms in critical differential equations that will need guesswork on the temperature profile that will require further unvalidatable interpolation that will terminate in final modelling, prior to its contributing a paragraph to the next IPCC draft that a coal company scientist will keep out of the final.

Better that we stick to our knitting: improve experimental areal coverage and quality of interpretation. Plain vanilla layer cake is really the best place to start rather than regions with serious perturbations. I think several icesheds like JI, Zachariae and Petermann really need to be carved out as standalone units rather than assimilated in an overall ice sheet history, ditto for the southeast.

An archive of pretty pictures cannot be the final digitized product as team member S Price of Los Alamos is there to coordinate with 'the needs of the ice-sheet modelling community' to guide the radiostratigraphic database format so it conveniently provides boundary conditions.

I hope this doesn't mean another human-unreadable big endian mega-gig binary accessible only to 2-3 mainframe computers worldwide outfitted with tens of thousands in proprietary software that can run millions of lines of compiled model code written in a rarely used language by people who favor bw screens in 9 pt courier with blinking insertion points for typing in unregistered terminal commands.

Another group has published a half-baked assessment of the 14.6 kyr isochron in a seemingly more accessible but ultimately nightmarish format -- 11,000,000 raw excel cells just for one isochron (that was never imaged). Want 1,000 of those on your desktop?

My idea for a public database is rather different. Since the basic object here is a surface, notably top, firn, bedrock, temperate ice, horizon isochrons or temperature isotherms, the most convenient representation is an ordinary grayscale where the value indicates elevation relative to WGS84 sea level.

The file size is very small since the lat,lon do not need to be re-stated for each isochron, they are implicit in pixel coordinates. The discipline of 8-bit bins prevents silly levels of precision from being stored. It is very important not to carry around artificial numerical precision that vastly exceeds the experimental floor.

Another great advantage of this is the actual isosurface can be instantly displayed in 3D from any perspective in a freeware bump map, even though it is moderately complex mathematically as an analytically fitted surface. Note certain isosurfaces come as pairs (eg temperate ice upper and lower, firn depth).

(In the case of Greenland, it actually would be convenient to post the data in low-order fitted polynomial approximation because the ice sheet overall has the shape of a toroidal chord, ie isochron surfaces in east-west cross section being parallel and of constant curvature are fitted by specifying a circle radius that only varies north-south by a second radius. However such a model does not come to grips with bedrock attributes that are important to deformation and slip.)

Most number crunching can be done directly in graphics (eg subtraction of layers to get a resultant graphic of post-ablation and movement effects), the rest can be done dropping the graphic into an excel numerical array (save in BMP format, compute, reopen as graphic).

In fact hundreds of surfaces (co-registered flat software layers) can be displayed simultaneously in 3D as any subset. If that's too visually complex, it's no problem to construct vertical slice profiles for any horizontal transect (or animate those).

The data underlying radar isosurfaces is rather sparce and a given datable horizon is not always feasible to track across the whole icesheet. Colors can track of observation flight track vs reliable fill-in (layer cake) vs less reliable fill-in (lost episodically) vs indeterminable (gone, over bedrock, over freeze-on disturbances) provided that data is encoded as tints that do not disturb the underlying grayscale.

Neither vertical nor horizontal resolution will be that great, meaning that file size are inherently small. Very little being done on the Greenland icesheet requires as much as 700 x 1225 pixels (600 kb) for a completely satisfactory loss-free display of information -- see the surface slope graphic posted in the imagery forum. An ordinary desktop has the RAM to open 25,000 such files at a time.

It is very convenient to up- and down-sample resolutions in graphics software. The coherence of stratigraphic layers over tens of thousands of years means that snowflake trajectories are coherent as well, which in turn means time evolution backwards or forwards can be economically generated from a parameter or two acting on a given layer grayscale. I'll post some concrete examples in a bit.
« Last Edit: November 20, 2014, 05:02:27 PM by A-Team »

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #11 on: November 21, 2014, 06:19:31 PM »
It seems implausible that LAYERMAP could unravel the age structure, crystal fabric or temperature profile of bedrock ice upheavals (initial survey Bell 2014); it follows that northern Greenland cannot be productively modeled, as proposed, with the same old ice physics of yesteryear. The problem is, nobody seems to know what the new ice physics should be.

These huge features are located near the major marine discharge sites and may inject massive meltwater heat and physical barriers where least expected. Previous forum posts show the existing radar archive can only detect a fraction of these features because of track sparseness, low sensitivity in deeper softer ice even when flight tracks go directly overhead (the NEEM folds), ambiguous blending with features arising from conformal deflection of flow over bedrock topography, uncertainty surrounding small, newly emerging, immature or decaying relic features, and northern effort allocation bias (rare small features south of Epiq are missed).

Even where radar tracks are closely spaced over upheavals (Epiq, Zachariae, Petermann), they are never a good match to the scale of the features -- 3D geometries cannot be inferred from haphazard tomography, often oblique to the 'downwind' deflection. It is imperative to re-fly on a much tighter grid before drilling a core, with Petermann and NEGIS the top priorities.

Greenland researchers are tip-toeing around, neither embracing nor rejecting, Bell's mechanistic interpretations (supercooled water coastal, pressure meltwater interior) which cite interpreted radar profiles in the uncored Gamburtsev range in East Antarctica. Here's a very curious response from two Russian glaciologists who work in that region:

Quote
New interpretations of radar profiles with so-called ‘refrozen ice’ in the Antarctic and Greenland ice sheets
http://www.igsoc.org/symposia/2013/kansas/proceedings/procsfiles/procabstracts_63.htm
A Markov, P Тalalay

Recently, abnormal build-ups were found near the base of ice sheets on the radar profiles in the region of Gamburtsev Subglacial Mountains, East Antarctica, and the northeastern part of the Greenland ice sheet. In some places, the height of these build-ups is up to half of the total ice-sheet thickness. Initially, their origin was explained as refrozen ice.  We considered three other options for the formation of these phenomena. 

Diapir. By analogy with geological structures the abnormal build-ups can be identified as ice diapirs. Such interpretation almost completely explains not only their shape but also the genesis.

Relic ice domes. The abnormal build-ups can be also interpreted as relic ice domes, the cross-sectional shape of which matches the shape of the cross section of the wing. The structure of the flow of ice through the relic dome is similar to that of the flow of air or water masses through the wing. This configuration causes non-laminar forms of ice flow, similar to flow breakdown behind the wing.

Pressure field and structure of ice. It is well known that radar profiles can display not the flow structure but the structure of the pressure field. In this case the abnormal build-ups can bound the subsurface structure of recrystallization ice, which appeared during the ice flow through the area having special pressure and temperature conditions. Thus, the radar profiles can display the total field of lithostatic and dynamic pressure and recrystallization ice (structure, anisotropy, the orientation of the crystal c-axes, crystal size, etc.) inside the ice sheet.

The abstract does not pursue how these possibilities might be experimentally distinguished. Whether Greenland upheavals 'look like' this or that is surely in the eye of the beholder; different upheavals could have different mechanistic explanations or be a blend of several. Without clear diagnostics, this issue won't be resolved by high resolution geometry or ice core properties of upheavals.

I've attached some seismic imagery of salt and mud diapirs. It's all about buoyancy, so I would toss in mantle plumes here as well. Seismic traces look very much like radar scans because acoustic impedance has the same refractive, attenuative and reflective math as dielectric impedance. Salt formations like ice are incompressible viscoelastics that readily flow under common pressure/temperature regimes. Ice diapirs have only been reported for Europa, a moon of Jupiter, not so much on experimental evidence as lack of other options.

Gravitational sliding of salt diapirs in the North Sea extends the analogy to the Greenland upheavals, the difference being petroleum geologists have distributed geophone receivers, vastly better processing software, and the ability to graphically 'undo' a given stratigraphy (palinspastic restoration) and push it into the future. You can see a very instructive palinspastic animation of a salt diapir at http://www.beg.utexas.edu/indassoc/agl/animations/AGL98-MM-005/

Ice diapirs arise from a buoyant gravitational force attributable to density differences. It's not clear what a quantitatively sufficient source of these could be in northern Greenland ice. But then the physics of Bell's mechanisms aren't clear either whereby a thin film of refreezing water can lift and displace megatons of ice above. Water does have an odd phase diagram but the Clausius-Clapyrion constant is markedly affected by contaminants, a real issue for both glacier beds and salt layers.

Relic ice domes, the remnants from glacier retreat in north Greenland, seem a stretch. The structures today, as isolated relics, would tip over, but where might the melt/growth record be located? The oldest known ice in Greenland is Eemian but the evidence there points to meagre maximal retreat in the northern upheaval regions, not enough to provide the geographic distribution of upheavals nor explain the association with glacial outlets.

The chaotic blocks of ice seen down-flow on some Greenland upheaval transects are important clues to the physics but cannot be analogized to turbulent flow over an airplane wing, compressible air being an altogether different medium. The gravitational input here is different from buoyancy as it ultimately arises from the icesheet sloping down from the central ridge where net surface accumulation is greatest.

The best opportunity for relic ice domes is during Marine Isotope Stage 11, its maximal sea level rise at 405,000 yr. This could conceivably left ice domes in southern Greenland but even there the icesheet's slow slide to the sea would have swept them away long ago:

Quote
South Greenland ice-sheet collapse during Marine Isotope Stage 11
AV Reyes 2014
doi:10.1038/nature13456 paywalled

Varying levels of boreal summer insolation and associated Earth system feedbacks led to differing climate and ice-sheet states during late-Quaternary interglaciations. In particular, Marine Isotope Stage (MIS) 11 was an exceptionally long interglaciation and potentially had a global mean sea level 6 to 13 metres above the present level around 410,000 to 400,000 years ago  implying substantial mass loss from the Greenland ice sheet (GIS).... cessation of subglacial erosion as a result of near-complete deglaciation of south Greenland...This is evidence for late-Quaternary GIS collapse after it crossed a climate/ice-sheet stability threshold that may have been no more than several degrees above pre-industrial temperatures.
The 3rd proposed mechanism says upheavals are an illusion, there's been no bulk advection of ice because radar returns indicate flow not only as volcanic isochron deformation but also as boundary reflections from ice fabric changes that arise from differential recrystallization appropriate to temperature and net pressure from lithospheric flexure and icesheet dynamics.

Ice fabric has been studied to the nth degree by NEEM core microscopy. NEEM was inadvertently drilled through a near-bedrock double fold which however does not show up on any of its many radar transects and so are not included in the upheaval 'database'.

People have argued for years that ice crystal boundaries, growth, anisotropy and c-axis orientation should affect radar returns but I'm not aware of a single Greenland radar feature that has ever been annotated as originating from ice fabric. See Montagnat 2014 free full doi:10.5194/tc-8-1129-2014

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #12 on: November 24, 2014, 01:23:35 PM »
With five mechanisms on the table but no serious exposition to date of respective underlying propagation physics, it would be timely to develop diagnostics that would distinguish scenarios in the event an upheaval is drilled to bedrock. For example, lab experiments show that only supercooling produces herringbone ice whereas frazil and needle ice are not at all specific to this process (first image below).

Thus a table today laying out distinctive predictions for temperature, ice crystallization fabric, uplifted impurities, isotopes, age of onset, internal substructure, buoyancy, and geographical distributions (specifically explaining non-occurrence) is highly preferable to overly flexible rules of interpretations that might be twisted to fit newly acquired data.

Note the Rapid Access Ice Drill (RAID) -- which can drill to 3300 m in 10 days -- is only slated for Antarctic use and that only in late 2017. For Greenland, the only short-term prospects are very tight radar grids and possibly seismic. Gravitometry has no prospects for resolving density or impurit contrasts within ice.

Plume mechanisms invoke slow upward movement of massive blocks of ice -- tens of cubic kilometers -- and sideways displacement of equal volumes since the ice overhead is incompressible, massive and almost rigid. No plume in either Greenland or Antarctica has created a sufficient dome to accommodate plume volume, much less breached the surface (as a salt diapir does in my backyard, Onion Creek east of Moab, UT).

While no one doubts liquid water formed under ice sheets will be redistributed according to the laws of hydraulics, including to sites of altered pressure and temperature where that water (or more likely slurry) can no longer persist in the liquid state, it's far from clear that this would cause thousands of meters of vertical uplift (rather than, say, lithosphere flexure or lateral displacement) as uplift per se changes hydraulics across the bed.

It is naive in the extreme to substitute simple overburden pressure (hydrostatics) for the complex and shifting physics at the bottom of moving icesheets, as some authors do, even as the putative upwelling plume itself significantly changes temperatures, pressures, and flow dynamic assumptions.

The work necessary to lift massive plumes against immense viscoelastic resistance might come from gravitational potential which in the end is solar energy (snow accumulation lifted high on the ice sheet). Two other sources are worth consideration: the increase in volume of liquid water as it cools from maximal density at 3.998ºC, or the volume increase as it experiences the phase change to ice.

The former effect is very small -- subglacial water here is cooling from 0.001º to 0.000ºC (neglecting pressure melting considerations) rather down from four degrees. The latter effect is quite large, on the order of 9% and capable of bursting plumbing pipes, but volume increase is actually not confined laterally at the bottom-up freeze of an ice sheet (water was able to get in) and may be experience geometric control (hexagonal crystals growing horizontally).

The lack of any published catalog is quite troubling 4-5 years in; most plume imagery is published without its frame id or worse, references non-Cresis radar without a link to its online repository (if there is one). Thus it is not feasible to bin plume imagery by angle of flight track relative to flow line, latitude, proximity to coast, ice thickness, overhead velocity, downwind components, or bedrock properties, much less assemble tomography from flight tracks through the same feature.

Radar can produce visually intriguing imagery. It is easy to forget reflections only represent variations in polarizability, not necessarily a proxy for density slices of upwelling mass flares. What one person sees as a downflow recumbent doughnut from a bottom freeze-on, another interprets as a recrystallization front.

Without theory and experiment behind them, plumes and diapirs are not much more than seeing animals in clouds or human faces on icebergs (Gamburtsev radar image below).
« Last Edit: November 24, 2014, 06:45:22 PM by A-Team »

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #13 on: November 27, 2014, 03:54:47 PM »
The image below illustrates issues involved in development of a working 3D layer model of the Greenland icesheet. Only five layers are shown here in cross sections: bedrock, surface, ice thickness, age and temperature profile. These layers can be done as virtual surfaces, ie as 8-bit grayscales imputing scalar isosurfaces as bumpmaps.

It is quite feasible to do sections along flow lines and often preferable for physics. Note very large areas of northeastern Greenland are seething underneath (in addition to the 'special cases' of JI, Petermann, NEGIS) so it will not do to copy glacier basics out from Cuffey & Paterson, Physics of Glaciers, 4th Edition ($58.64 @ Amazon, 704 pp) as modelers so often do. (Bottom upheavals were already vividly imaged by 1999, below.)

Bedrock is better done with some horizontal context to distinguish hills from ridges, basins from drainages. We've seen nice examples of that in Antarctic displays. Upheavals too will need this perspective view.

Collecting the available layers is quite a task as they are published at different scales in different projections with different keys via different grids at different accuracy with varying associated formal error using different file formats, some quite unacceptable by prevailing planetary science standards. However it sounds like LAYERMAP 'gets it', as it re-grids to an established layer, rather than going off in the weeds.

A central layer repository would allow rapid updating, not just of a new or improved layer but also of all the downstream layer-dependent calculations everyone is doing according to their interests. However it is more convenient to have a central web portal along the lines of EarthExplorer or WorldView, with checkboxes for layers and a slider for scale. That way everyone is playing with a full deck.

The image below mostly concerns itself with scale. First, there is the issue of actual data resolution; we don't want to throw that away when it's high but need to recognize it varies wildly between layers. Rather than lowest common denominator, some layers will be dumbed down and others interpolated up.

The Greenland icesheet is so very flat that it is hardly ever seen at 1:1 scale (as below). Greenland is 2700 km south to north, 3700 m above sea level at Gunnbjørn Fjeld and 1400 m below at Jakobshavn, with 2700 x 5.1 being 529:1 with the icesheet itself more like 2500 x 3 or 833:1.

So to provide 800 pixels of vertical depth requires 666400 pixels of  monitor width; note web browsers will provide scroll bars for too-large images if provided as <img src="greenland1:1.jpg" width="666400" height="800" alt="dsff">. That's 347 screen widths on a 1920 pixel width monitor.

Scrunching sideways to a given vertical exaggeration (hardly ever provided, never the same) can be very visually misleading and involve multiple rounds of resampling. The blog product is maximally 700 pixels, a print journal might allow a half-width graphic, ppts have their limit and so on. Here the issue is presentation and communication -- how is a work product to be published without degrading it to a cartoon. Journal supplemental is an important part of that.

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #14 on: November 30, 2014, 02:45:50 PM »
I've embarked on a project to collect all the raster GIS layers published for the Greenland Ice Sheet (itself sometimes abbreviated GIS) and re-post them as co-registered layers in open source grayscale data format at a uniform scale, bit depth and map projection, as is customary in the mainstream sciences.

First I needed a simple guide to the original sources of the layers. Making this little database proved very instructive, as I had to re-read all the articles just to fill in the blanks of my form. Perhaps 3 out of 20 followed acceptable scientific practice; the example shown below is about perfect though.

Note it uses the Pangaea data repository, a far better idea than paywalled journal supplemental, big-on-big-endian NSIDC storage, an evanescent campus web site, or (most favored) a crappy low resolution figure having text and grid flattened over poorly colored jpg'ed data of unspecified methodology, resolution and map projection.

I had expected a clear-cut distinction between experimentally observed, theoretically determined, and derived from other layers. That proved quite wrong, even seemingly simple layers like surface motion are heterogeneous mixes of satellite and ground data, ice physics modeling, unspecified software and arbitrary parameters, and layers like slope with complex inter-dependencies.

It is troubling to see papers invoke layers (eg magnitude of the gradient vector) without providing the slightest documentation. Gradient, slope, elevation contours and flow lines are really quite problematic if determined by subtracting the elevation of a point from those nearby.

That will give small numbers with big errors, for example (3000 m ± 20 m) - (2995 m ± 20 m) = 5 m ± 28 m in a rosy scenario for combining errors, see http://www.rit.edu/cos/uphysics/uncertainties/Uncertaintiespart2.html. That's 'fixed' by smoothing the data with an unspecified 'kernel' (presumably gaussian blur convolution) at a radius some large multiple of the ice thickness layer.

However the attached graphic shows error can be highly anisotropic, streaking along orbital lines of the satellite. Systematic error can have a very different distribution from random error. I removed it in the right half of the graphic by posterization contouring, though x,y weighting to angle the applied blur is a valid alternative. Both methods confirm the southern tributary of NEGIS, the northeast Greenland ice stream, which otherwise is bizarre in having no other tributaries. This implies a secondary peak in the barely determined geothermal gradient layer (if you believe in the first peak).

While a huge amount of effort has understandably gone into surface elevation and ice thickness layers (bedrock is a derived layer, their difference), quite a few other bounding layers are missing altogether or are only tentatively presented.

For example, I could not find a surface velocity layer, only surface speed. That's puzzling because speckle processing Landsat-8 nadir pairs takes a few minutes per and the software outputs little tangents to the flow lines, admittedly in plane projection but adjustable from the slope layer to the surface. Vector fields take three grayscales to display (usualy wrapped up as a color HSV layer): ones to specify surface of tangency, magnitude and azimuthal direction.

The absence of experimentally determined internal layers will be shortly remedied, we hope, by LAYERMAP internal isochrons possibly in base 10 logs (note ImageJ can reversibly exponentiate both color and grayscale at all bit-depth).

This was fully feasible back in 1999 as the radar track shows below, so there's a major mystery here because these 2D surfaces provide the only time-dependent 3D framework necessary to understand the past and future of Greenland ice. The history of surface elevation is otherwise limited to a very short instrument record.

Radar stratigraphy also has prospects for revealing Greenland-wide internal viscosity profiles, and so the all-important temperature at depth, heat flow being orthogonal to temperature isotherms.

Layer:surface elevation
Source:experimental
Author:Helm 2014
Access:free full: www.the-cryosphere.net/8/1539/2014/tc-8-1539-2014.pdf
Graphic:Fig.7 crops to 743 x 1373 png with valid color key but grid overlay
Download:doi.pangaea.de/10.1594/PANGAEA.831393?format=html
Resolution:     1x1 km horizontal, 32-bit tiff opens in ImageJ
Time:dH/dt over 2001-2014
Error:Uncertainty map provided as Fig 2A
Projection:Polar stereographic EPSG 3413 SP 70º WGS84
« Last Edit: November 30, 2014, 11:01:06 PM by A-Team »

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #15 on: December 01, 2014, 11:54:12 AM »
The Gogenini group at Kansas has made an extraordinary contribution to both Antarctic and Greenland glaciology, not just in designing ever better iterations of ice penetrating radar but also for their extraordinary efforts to deliver the data to the scientific consumer in various convenient formats including pdf, ftp, kml, jpg and matlab (see https://data.cresis.ku.edu/data/rds/).

Despite this accessibility, very little of the radar data has been utilized by Greenland glaciologists -- only ice thickness. However the chase for a long-term climate record (to match Antarctica's million years) has proven a pipe dream and for sea level rise, the canned programs and simple-minded reality preferred by modellers clearly can't work in view of the complex internal deformations of the Greenland icesheet so inconveniently revealed by radar.

Cresis has an ambitious new data portal up and (sort of) running, the OpenPolarServer at https://ops.cresis.ku.edu/. This is very similar to Google Earth (resp. EarthExplorer) except that it serves radar trackss over your choice of GIS layer, so far bedrock, surface velocity, or 45 m Landsat7 montage. The latter is completely useless in snow-covered Greenland except for the coast and ablation zone; the EarthExplorer portal would be used to find 12-bit Landsat8 pairs for speckle-tracking.

For a poster explaining what's under the hood, see https://drive.google.com/file/d/0B3XyGOddjZU0QkVBRTVBclRwWnM/edit?pli=1. The end user gets the benefits of MatLab data-picking.

The idea is to draw a polygon around your area of interest (say the upper NEGIS icestream), pull out all and only CReSIS radar tracks that cross it, and see how the radar stratigraphy correlates with bedrock topography and ice surface movement. I haven't explored the firn, snow or accumulation radar, just the depth sounder.

The portal unfortunately does not serve the radar tracks from AWI, WISE, PARIS, HICARS, or TUD also used by Bamber www.the-cryosphere.net/7/499/2013 to create the very bed elevation map served here. It would be appropriate to also server LAYERMAP isochrons here since they largely come from CReSIS radar; we don't know yet whether those are just colored lines on radargrams or grayscale surfaces.

The polygon tool didn't work for me in either firefox or chrome -- it couldn't find closure and couldn't be re-set -- so I wasn't able to look at what the four download formats provide. There is a nice choice of projections unlike the plate carrée and varying planar perspective you are stuck with in Google Earth.

The radar image is the same frame, at smaller scale, as you would get by ftp after a lot of drilling up and down in their file system and fooling with kml files. The picked bedrock line is not fitted to a Bamber bedrock 2D perspective as in the fancy Antarctic imagery. Colors on the velocity map are way too muddy; the bedrock seems to be a shaded relief; neither layer is readily convertible to grayscale (which is all they really are). It might make more sense to use contour maps as their white space can be made fully transparent, allowing overlays.

It would be a huge timesaver if the displayed image was whittled down (using its kml waypoints) to exactly that portion that crosses the user's rectangle. That gets to be a nuisance when the radar flight line is curved as Cresis doesn't put in kml waypoints on the imagery abscissa. It is also important to provide the lat,lon of radar flight intersections.

This is a very promising system in some respects and not too far off from the GIS server that's really needed. I didn't see any transparency controls for simultaneous displays or layer ordering options; checkboxes are used where radio buttons are intended and inter-layer math operations are notsupported.

However it's easy enough to collect what you need and drop it into a working GIS environment such as ImageJ or Gimp. I made the upper NEGIS perspective image there below. However the glaciology research community needs to move to higher display (final image) for both workplace and communication to work with Greenland in 3D.
« Last Edit: December 02, 2014, 10:16:49 AM by A-Team »

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #16 on: December 02, 2014, 12:27:52 PM »
Below I've posted the error maps associated with the Helm 2014 surface elevation map H and its recent rate of change dH/dt. We hardly ever see these published but it is very important to keep an eye on error in this situation because so many other Greenland GIS layers have dependencies on the surface layer.

These both utilize a logarithmic scale which draws out variation at the low end (central Greenland) at the expense of compressing variation at the high end. Planning ahead to say combining surface and thickness errors to get bedrock error within ImageJ (ie sq root of sum of squares) might lead to different choices, indicated here by grayscale insets.

Since the error map itself will have errors, the lower graphic smooths out excessive detail, producing simplified error contour maps via a weighted averaging kernel of appropriate pixel radius and posterization to lump colors into fewer bins.

Contour maps might provide a better middle ground for numeric modelling between pointwise complexity and single-value averaging. Such maps can also make more effective overlays if the white space beween contour lines is made transparent (one click with non-contiguous color wand).

I've also attached the root mean square error map from the new bedrock DEM, found as Fig.7 in Bamber 2013. It is heavily overlain with radar tracks (appropriate, as these are sparce), does not use a logarithmic scale (or indeed many of the colors in the color key). The error is concentrated in mountainous areas on the east coast of Greenland whereas the interest lies in special cases like Jakobshavn and the central depression and its paleodrainages.
« Last Edit: December 02, 2014, 01:11:18 PM by A-Team »

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #17 on: December 02, 2014, 02:09:29 PM »
As mentioned above, the CReSIS program has not only provided 21 years of comprehensive ice penetrating radar data but also hosts an innovative open source, no hassle distribution of that data to anyone with an interest -- no registration, no passwords, no software to buy, no glaciology club membership needed.

The bedrock paper Bamber 2013, in discussing the source of its data, provides a very unusual map showing tracks of *other* contributing radar programs. Some of these were very focused on small coastal regions, others were just pilot projects or done with obsolete radars, but the Alfred Wegener Institute stands out as providing a very dense grid of tracks around the key core sites NEEM and NGRIP.

NGRIP is mushy at the bottom; NEEM is frozen solid -- there's great interest in seeing whether radar can map temperate bottom ice over all of Greenland, so these AWI tracks could be very important in these fiducial areas.

Even more importantly, though the NEEM core analysis found the annual layers at depth doubled and folded over suggesting a complex and interesting strain history not expected for the mundane location on a summit ridge, those folds do not show up on CReSIS radar tracks directly over the hole, no matter what image enhancement magic is applied to them. Yet CReSIS radar easily picks up bottom ice upheavals nearby.

Undetectable folds -- on top of the detectable ones -- raises the specter of Greenland ice interior (and its melt under climate change) being vastly more complex than currently contemplated. However no two radars are the same and it seems possible that the AWI surveys in this area could pick out the cryptic NEEM folds.

So where are the AWI radargrams available online? Bamber only cites a 2001, no url. Going to the AWI site, no apparent link to any data, despite a statement that "We just contributed another several 10,000 profile kms to ice-thickness compilations for Greenland within the EU-funded programme ice2sea". So this would be 2014 data on top of 2010. No url to ice2sea, no data link evident at the ice2sea site, no response from the AWI 'contact' person. Somehow Bamber had access.

Looking around on the AWI site, I come across a very intriguing 2014 abstract entitled "Small Scale Folding in NEEM Ice Core" by Julien Westhoff, epic.awi.de/36396/1/BA-Westhoff.pdf. As you've probably guessed: AWI has blocked the link -- access only for AWI insiders and even for them embargoed until Oct 2015 (yes, a year from now). I had to take a screenshot of that.

Hmmm, I suppose we could block AWI staff from Landsat-8 and put a year delay on downloads from US journals. However I favor the open access approach of CReSIS, not this race to the bottom.

Quote
NEEM is a drilling site in north western Greenland, from which a 2500 m long ice core has been derived. The ice has been analyzed with visual stratigraphy to make layering visible. This thesis analyzes the layering from top to bottom in terms of folding events.

Small disturbances of layers start to appear around 1560 m depth and folding is visible at 1750 m depth from the surface. Below 2160 m there has been so much deformation that a qualitive description is not possible.

From 1750 m to 2160 m there is an evolution of folding, where normal folds, then Z-folds and shear zones, and in greater depths many Z-folds in one layer appear. They are a result of increasing strain rate, leading to deformation, which in this depth is mainly ductile.

Fold types with a brittle component are also visible in form of detachment folds. The dominant structures are Z-folds located at shear zones which were created by deformation, resulting in these diagonal shear zone in the core.

These shear zones have also been analyzed with the fabric analyzer to find the main c-axis orientation within these zones. The main orientation is caused by a tilting of the grains during deformation and another part due to recrystalization processes.

The orientation of these shear zones can be estimated by using the line scanner images which show the ice in different focus depths in the horizontal level of the core and reveal a general orientation to the top left of the images, caused by shear stress from the right in a small angle.

sidd

  • First-year ice
  • Posts: 5895
    • View Profile
  • Liked: 823
  • Likes Given: 0
Re: Greenland subglacial topography
« Reply #18 on: December 02, 2014, 07:43:50 PM »
Re:AWI data

Try writing to Prof. Viet Helm. He has helped me in the past. Also look on pangaea (http://www.pangaea.de/ )

sidd

Espen

  • Young ice
  • Posts: 3383
    • View Profile
  • Liked: 254
  • Likes Given: 1
Re: Greenland subglacial topography
« Reply #19 on: December 02, 2014, 09:30:42 PM »
A-Team,

You are reaching into the hinterland of science! 8)
Have a ice day!

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #20 on: December 03, 2014, 01:02:12 AM »
Quote
.. reaching into the hinterland of science!

Indeed as a schoolboy I could see that Africa and Brazil so nicely docked, it could not a coincidence be. Why a big German research institute also for me not named is?

Quote
Try writing to Prof Helm.

Good idea, I have still not heard back from the distinguished AWI helpdeskmeister, Herr Dr Dr h.c. Prof. Dip.Ing. von und zu Illulisat who is apparently just an unpaid intern.

I could maybe get the documents from NSA, they pwned AWI top to bottom even before Merkel's phone ... climate change in Greenland is a natl security issue like everything else. But the publicly funded research should be in the public domain, not require a rootkit.

Go figure, a closely related student paper on the NEEM stratigraphic scan IS available. While both are non-consumptive (ice crystal orientation, trace contaminants), any physical access to these ice cores is highly restricted, all the more reason to require prompt release of each and every study.

I just now noticed that Google translate has a talk mode, so you can work on something else while an attractive young female robot reads the document out loud, English or German, in the background.

Die Korrelation von visueller Stratigraphie mit physikalischen Eigenschaften bzw. Spurenstoffgehalten in polarem Eis am Beispiel des NEEM-Eiskerns (Grönland)
Sophie Ehrhardt http://epic.awi.de/36516/

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #21 on: December 04, 2014, 01:01:50 PM »
The internal 3D history of Greenland ice can be approached by determining the 'depositional provenance' (snowflake trajectories). That is, annual layers in a ice core such as NGRIP did not fall at the borehole lat,lon but rather as snow at sites some tens of km farther uphill.

That was modeled for EW and NS transects through the boreholes in a very instructive 2004 dissertation by N Lhomme, a revised version of which was published with GKC Clarke the following year.

Now that NEEM has been drilled, core dates improved by cross-correlation, and better resolution provided for bedrock topography, geothermal heat and surface elevations, their calculations need to be revisited (especially since something went very wrong with oldest ice determination).

Nonetheless, their dramatic conceptualizations illustrate where englacial Greenland research is headed -- the results might be wrong but the display provides a fantastic prototype for what LAYERMAP is capable of accurately producing.

For reasons that are unclear, they make no mention of radar stratigraphy, though they predict isochronal surfaces modeling out from core layer dates and in effect lay an iso-provenance trajectory coordinate system across those surfaces.

Upheaval deformations, already decisively documented in 1999 radar tracks, don't get a mention either, even though some simulation features bear an uncanny resemblance to them and indeed convergent and divergent bulk trajectories  noted provide a possible physical mechanism.

I myself would take the wealth of experimental data embodied in LAYERMAP to stand the semi-Lagrangian method of Lhomme/Clarke on its head. That is, fix its parameters and assumptions via maximal likelihood or plain monte carlo  inverse methods to produce a refinement that best fits a cost function derived from experimental isochrons.

That revised model will not just write a nice history of the Greenland icesheet (eg of non-diffusive temperature proxies such as O18) but also be an amazingly accurate predictor if it can only be given a reliable future mass balance from climatology. Thus the romantic era in Greenland research could be ending, to be followed shortly by academic, pedantic, and didactic.

Lhomme: 2004 https://tel.archives-ouvertes.fr/docs/00/04/80/44/PDF/tel-00009253.pdf
Clarke 2005: http://www.eos.ubc.ca/research/glaciology/research/Publications/ClarkeLhommeMarshall(QSR-2005).pdf
Goelles 2014: http://epic.awi.de/34727/1/Goelles-etal_GMD14.pdf
« Last Edit: December 04, 2014, 01:08:40 PM by A-Team »

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #22 on: December 08, 2014, 02:28:56 PM »
The images below illustrate some of the issues involved in developing a working map layer server for Greenland, which does its actually calculations on 2D bump map surfaces but must be ultimately communicated in 3D perspective.

That can get fairly complicated as shown in the example below, in which a mythical investigator is looking at annual net accumulation over bedrock (which makes some sense), in the context of radar stratigraphy isochrons E to
W and snowflake trajectories N to S that meet at the Dye3 drill site to utilize its core profiles (overly ambitious).
« Last Edit: December 08, 2014, 02:34:52 PM by A-Team »

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #23 on: December 09, 2014, 05:04:59 PM »
Quote
Try writing to Prof. Viet Helm. Also look on pangaea (http://www.pangaea.de/ )

Good suggestion, sidd. I wish all glaciologists would use www.pangaea.de and provide doi's for their Greenland products -- it's a far better system than stringing them out across dozens of journal supplementals. However I searched that site fairly thoroughly and could not find a single other layer posted there.

A field like molecular biology has a counterpart to this repository called GenBank. If you submit a paper without having already submitted your DNA sequence to GenBank and included the accession numbers in the article, it will be immediately rejected without even going out for review. No one puts up with this mickey maus in biomedical research.

I looked again at the Helm DEM using some very clever online tools posted by David Tschumperlé of U. Caen Basse-Normandie https://gmicol.greyc.fr/ that can also be done within Gimp or ImageMagik.

G'mac has some very useful offerings in the Contours section such custom convolution, gaussian curvature, gradient norm, isophotes, and laplacian. I was interested here in adaptive smoothing driven by error map but noticed that the Greenland surface DEM has some non-orbital artefacts (orange lines, lower half) that Prof. Helm will want to remove:

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #24 on: December 11, 2014, 01:08:57 AM »
This surface elevation map represents the culmination of many years of data taking and careful post-processing. This layer is very important because so many others (like bedrock) have dependencies on it.

It's been posted at 32-bit grayscale, a bit odd since 8-bit already provides 256 elevation bins (admittedly not enough given Summit, Greenland lies at 3,216, ie only 12.5 m bins) but 16-bit is good enough to specify 24 mm bins. Surely we are not measuring the surface elevation to the width of a snowflake (think wind). This would be quite the vertical mismatch with horizontal ground resolution, 1 km x 1 km.

Be that as it may, ImageJ can conduct a number of operations making full utilization of 32-bit files and I stuck to those in the processing steps below, first applying 'Find Edges' which uses a standard Sobel edge detector that accentuates local changes in elevation.  This applies two 3 × 3 convolution kernels (transposes) that take the discrete counterparts to NS and EW derivatives and combine as hypotenuse, the square root of the sum of the squares (ie norm slope, see http://rsbweb.nih.gov/ij/docs/guide/146-29.html)

 1   2   1
 0   0   0
−1  −2  −1

After renormalizing contrast in no-compromise 32-bit, this produces a number of odd visual effects on the surface elevation DEM (first image below). Some are minor stitching or re-gridding artifacts, others may arise from random error terms swamping out posted elevation differences between a pixel and its neighborhood, a few may represent uncharacterized systemic error, and still others (like the one below) could reveal unsuspected properties of the ice sheet.

In west-central Greenland, the ice slopes down very slowly and uniformly westward of the summit ridge. This means the Sobel edge detector largely acts on EW flow lines, with very little contribution from orthogonal NS contours. There's no ice sheet physics here, just internal properties of the DEM that are not at all apparent in the original.

Notice not only the peculiar wavy bands but also the extensive 'flat' area upstream of Jakobshavn, also seen above Russell glacier ~140 km to the south but nowhere else (second image enlarged in 32-bit but displayed as 8-bit).

However the output near Jakobshavn bears an uncanny resemblance to imagery in a recent ice physics paper by OV Sergienko 2014 (paywalled, doi: 10.1002/2014GL059976) reporting "patterns in driving and basal stresses" (third image below ~ Fig.3). That's because the gravitational driving stress on the icesheet is proportional to the slope (and the slowly varying ice thickness averaged over the scale of resolution). But are we really looking at glacier physics here or just an obscure feature of the DEM?

Here we have two very low angle Landsat-8 sun shadows -- LC80090112014040LGN00 09 Feb    14 sun 5.8º from the west and LC80812332014208LGN00 27 Jul 14  sun 5.7º from the south (fourth image). These provides very high resolution surface topography not subject to interpretational uncertainty -- there won't be a shadow unless there is a ridge or bump. Note 15 m resolution provides 4,444 pixels for each DEM pixel with 1 km resolution.

I've called attention earlier to overlapping surface waveforms in the NE at the Jakobshavn forum, ruling out cloud artifacts (such s those from standing atmospheric waves) on the basis of their consistent appearance on dozens of dates with a wide range of weather, satellite azimuths and sun angles. These don't show up in the DEM because it averages elevations over a sq km which washes out shorter wavelength features.
« Last Edit: December 11, 2014, 11:59:18 AM by A-Team »

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #25 on: December 12, 2014, 01:56:10 PM »
Despite the new surface elevation layers and their accompanying slope and contour maps, I have not seen one key layer -- talked about for years and utilized locally on ice setups -- the flow lines.

These are orthogonal to contours and so together these form a natural coordinate system for ice physics (which to an approximation are vertical sheets moving under 1D flow lines).

The velocity vector field is everywhere tangent to the flow lines, defining them uniquely. Over the Greenland Ice Sheet, velocity is an experimental satellite observable, yet one often deprecated to speed, direction (and so flow lines) being discarded in all but a handful of papers (along with divergence and curl of momentum; three really bad ideas in one).

It's very easy to make a high resolution velocity field (plot and tabular) from an image pair (example below) using the  particle image velocimetry plugin of ImageJ https://sites.google.com/site/qingzongtseng/piv#sys] [url]https://sites.google.com/site/qingzongtseng/piv#sys[/url]

It's not so easy to find software that offers flow lines at all, much less does them well:

Quote
"After hours trying to find a workaround trying various hydrological toolboxes, I found a solution using the opensource GIS program GRASS http://grass.osgeo.org/. The necessary tool is r.flow http://grass.osgeo.org/grass64/manuals/r.flow.html, with options available to set the spacing of the calculated flowlines and maximum number of segments per flowline. Below is an example from the Welsh Ice Cap made from an ice-surface elevation raster, with 50 m contours (note the flowlines are perpendicular)." H Patton, Aberystwyth
The spectacular image of below -- a colored contour map with flow lines translucently over a shaded DEM -- looks so much like Petermann and Zachariae in northern Greenland that it may fool you for a bit, but it is peak ice age Wales.

Patton also made two very effective animations of waxing and waning of the Welsh Ice Sheet through the last glacial maximum into the Holocene, as driven by two climate models derived from Greenland's GISP2 core. While these are just scenarios that fit moraine and sediment data but not Irish Sea effects, they illustrate exactly what we should be doing for Greenland -- but aren't, despite a vastly larger research community.

Very much worth viewing: http://henrypatton.org/academic-research/ice-sheet-models
« Last Edit: December 12, 2014, 05:08:45 PM by A-Team »

sidd

  • First-year ice
  • Posts: 5895
    • View Profile
  • Liked: 823
  • Likes Given: 0
Re: Greenland subglacial topography
« Reply #26 on: December 12, 2014, 08:02:53 PM »
GRASS is nice. Have you looked at QGIS also ?

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #27 on: December 12, 2014, 09:25:08 PM »
sidd,

No, I have not, thanks for the suggestion. It says here that QGIS runs GRASS as a core plugin. I suppose the future is some portal that can run any image processing command from anywhere on anything. Still, a lot of looking under the hood involved to understand what is actually being done to the data.

http://www.qgis.org/en/site/
http://docs.qgis.org/2.2/en/docs/user_manual/preamble/features.html#analyse-data

That would be cool if you could post a run of GRASS's rflow for us on the new Greenland DEM. It seems like line integrals along these flow lines would let us define multi-year discharge boundaries on the ice streams, eg where is the line on Jakobshavn showing all the ice that will be gone five years from now.

I have just moved over today to FIJI, a much-enhanced successor to ImageJ. I am looking in particular at several must-have plugins: laplacian of gaussian, particle image velocimetry, bunwarpJ deformation, and enhance local contrast (CLAHE). My sense is that a whole lot of remote sensing and glaciology products are under-analyzed and could maybe be done a whole lot better.

http://fiji.sc/Main_Page
http://rsb.info.nih.gov/ij/plugins/mexican-hat/index.html
http://homepages.inf.ed.ac.uk/rbf/HIPR2/log.htm
https://sites.google.com/site/qingzongtseng/piv#sys
http://fiji.sc/wiki/index.php/Enhance_Local_Contrast_%28CLAHE%29
http://rsbweb.nih.gov/ij/docs/guide/146-29.html

sidd

  • First-year ice
  • Posts: 5895
    • View Profile
  • Liked: 823
  • Likes Given: 0
Re: Greenland subglacial topography
« Reply #28 on: December 12, 2014, 11:49:24 PM »
there is a velocity field mapping in Rignot(2012) doi:10.1029/2012GL051634
corresponding author is Rignot, he will have a dataset for the velocity field plotted.

GRASS or QGIS will generate a set of vectors perpendicular to contours, but this is not so useful, since surface ice flow can be at arbitrary angles to contours. Rignot goes there somewhat: "To perform a zeroth-order analysis of the results, we compare the observed velocity with that calculated for an ice sheet deforming via pure internal deformation." And no sliding, so sad. Fig 3.

I would love to do some analysis, but my time is very limited. I have not been able to get back to Greenland datasets for some months. Hopefully I will have some time over the holidays, but i am not sure it will be enuf.

sidd

P.S. QGIS (but not GRASS (!)) turns Greenland upside down on some of the datasets. Annoying, but fixable.
« Last Edit: December 12, 2014, 11:56:23 PM by sidd »

sidd

  • First-year ice
  • Posts: 5895
    • View Profile
  • Liked: 823
  • Likes Given: 0
Re: Greenland subglacial topography
« Reply #29 on: December 13, 2014, 12:31:02 AM »
r.flow in GRASS is useful for surface hydrology, so it might illuminate supraglacial meltwater pattern ... but it will fail very quickly, it knows no thermo for the phase transition

snow/ice free terrain surface changes via erosion very slowly compared to timescale for ice surface at melting point, r.flow was developed for the former.

the big problem in modelling the subsurface in and around the refreeze massifs, (which are the interesrting bits for me,)  is history dependence, sorta like retrograde analysis in chess, or closer to physics, weighted sum over possible histories a la Feynman. But there may be only a few flow histories consonant with current observations. once we have some good atmosphere-precip--ice-thermo-ocean models. Getting closer every day. RACMXX and SSA+streamy shelf + "local ocean" like they are doing around antarctica ...

sidd

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #30 on: December 13, 2014, 02:03:04 PM »
Quote
there is a velocity field mapping in Rignot 2012 free full doi:10.1029/2012GL051634, he will have a dataset for the velocity field plotted.

Good spotting, excellent paper. We've made lots of previous use of Fig.2 (Greenland surface speed) but in Fig.1def are quite unusual in presenting error maps for both speed and direction (d,f) and representing velocity direction by a color wheel hue (e).

Color wheels require a lot of dithering to get radial from what is inherently rectangular; even extracting the original pdfs sent to the journal, this one ended up using 7,604 distinct colors. The map itself has 153,731 colors. However, looking at the error map, some small multiple of 360 would have sufficed for the binning.

From setting a color picker on radius 0 hue, it appears there was an intent to fix saturation around 71%. However decomposition to HSV, LAB, CMYK color spaces etc do not provide a satisfactory data channel. It might have been better to use a rectangular key of say 360 discrete hues. The image below provides a couple of those at 100% and 71% fixed saturations (with V set at 78%).

They set the distance scale at a very sensible 1 pixel to the km at 70ºN which is reduced below because of our 700 pixel maximal width. That entails a modest 2.3 million pixels for all of Greenland. The streaking cannot be helped, it is due to unsatisfactory satellite coverage.

In summary, it's not feasible to recover the original velocity directions from the article (or its supplemental). On the flanks of west-central Greenland, the substructure of the 'blue' (if there is one) cannot really be seen by eye. Yet subtle directional variations in upper NEGIS icestream are very clearly resolved and features of its overall basin are quite intriguing.

With the original data and the limited use of direction in lower NEGIS, the color key could be blown up. It's not clear though how to accommodate the error map -- perhaps tweaking the representation via extensive monte carlo sampling, rapidly animating that, and looking for stable features.

I don't favor non-publication of essential data. This is a small csv text file, a doi for it is fast and free. Emailing authors? They move on, die, retire, feud, compete, change interests, are too busy to find old files, lose laptops, a RAID fails etc. We've seen a ton of fails just in the last year (eg Landsat7, Cryosat2).

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #31 on: December 15, 2014, 12:44:14 PM »
Quote
GRASS or QGIS will generate a set of vectors perpendicular to contours, but this is not so useful, since surface ice flow can be at arbitrary angles to contours.
We have numerous types of equipotential surfaces here for which the natural (gradient driven) orthogonal coordinate systems may simplify (speed) numerical solution of the underlying differential equation, such as Stokes or heat, either by separation of variables, bringing in a symmetry or conservation law, or isolating poorly known parameters such as geothermal heat flux for rapid ensembling.

M Gunzburger has an interesting perspective on that, images below. There is a lot of model sensitivity to cutting Navier-Stokes down to full Stokes down to Shallow Shelf and Shallow Ice approximations (pg 52), from basal slip or not.

http://www.newton.ac.uk/files/seminar/20121001170018001-153378.pdf

In Greenland, I'm of the opinion that the observed direction of surface velocity will depart only slightly from the calculated tangent to flow lines (with important local exceptions). A map tinted by the value of their dot product would measure the variance. Here it might be a good idea to start a parallel track from the ice thickness elevation map, which departs significantly from the surface DEM and has more to do with basal driving stress.

Quote
r.flow in GRASS is useful for surface hydrology, so it might illuminate supraglacial meltwater pattern... need history to model subsurface around refreeze massifs .
Despite many recent studies of the melt lakes on the western flank, I've not seen that region partitioned into collection basins, which seems eminently doable. Some of the lakes reappear in the same spot like clockwork but it's not so clear whether or how basins will consolidate or if the 'lake district' will move uphill and act similarly.

A common complaint with paleo analogies is temperatures today are ramping up 100x faster than the PETM (per AbruptSLR post), raising the question of how rapidly increasing volumes of meltwater will be accommodated. A lot of latent heat is brought into the icesheet interior by englacial refreezing of water and massifs provide a barrier to getting down to bedrock conduits (which themselves can be overwhelmed).

sidd

  • First-year ice
  • Posts: 5895
    • View Profile
  • Liked: 823
  • Likes Given: 0
Re: Greenland subglacial topography
« Reply #32 on: December 15, 2014, 10:00:49 PM »
supraglacial lakes, some projections: Leeson(2014) doi:10.1038/NCLIMATE2463

"Our simulations are performed using the SGL Initiation and Growth (SLInG) model8 , a hydrologic model that routes runoff over a model of the ice-sheet surface, allowing water to form lakes in topographic depressions "

sidd

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #33 on: December 16, 2014, 06:25:37 PM »
sidd, you couldn't be more timely or on-topic!

That 15 Dec 14 article is paywalled at Nature CC but as with many journals, text can usually be viewed for free (five minutes, one article per day) along with full resolution figures, from its DeepDyve link. Co-authors Morlighem and Rignot are on ResearchGate but do not as yet provide an open source version.

U Leeds put out a moderately useful press release (picked up by wire services) with a nice enough animation of expected future change (last frame -- lakes in 2060 uphill from Russell Glacier -- captured below, surface velocity added). They are working on 20,000 sq km in the Isulluup-Russell region below JI on the western flank of Greenland but extrapolate in the article to a melt lake perimeter for 2060 for the whole of Greenland.

http://www.leeds.ac.uk/news/article/3640/migrating_supraglacial_lakes_could_trigger_future_greenland_ice_loss
http://goo.gl/Bg1bno

Quote
Supraglacial lakes (SGLs) form annually on the Greenland ice sheet; when they drain, their discharge enhances ice-sheet flow by lubricating the base and potentially by warming the ice. Today, SGLs tend to form within the ablation zone, where enhanced lubrication is offset by efficient subglacial drainage.... Our simulations show that in southwest Greenland, SGLs spread 103 and 110 km further inland by the year 2060 under moderate (RCP 4.5) and extreme (RCP 8.5) climate change scenarios, respectively, leading to an estimated 48–53% increase in the area over which they are distributed across the ice sheet as a whole. Up to half of these new lakes may be large enough to drain, potentially delivering water and heat to the ice-sheet base in regions where subglacial drainage is inefficient....

The literature here can be slightly confusing: on the margins, more meltwater early creates more efficient bottom drainage, thus removing basal lubrication and slowing glacier movement whereas at higher elevation later and less meltwater is not enough for channelization to develop and does speed basal sliding.

Infiltrating water also brings in heat (especially if it refreezes) which facilitates internal creep deformation. Faster flow in turn thins the ice sheet, steepening the slope reduces buttressing bringing more ice sheet into melt zone elevations.

The modelling question is whether this runaway positive feedback cycle will overwhelm negative feedback provided by excess early-season basal meltwater. It won't be easy because the heat is injected non-uniformly and locally but periodically.

Melt seasons will last longer and go higher with global warming. Greenland has seen episodes of that before, so the issue becomes 'is there an effect from the special rapidity of this warming onset?' It's a scale-invariance question, in raster graphics terms whether horizontal-only rescaling of a time axis fails to interconvert glaciology of slow and fast ramping eras.

I would say yes because advection of heat is much more rapid than its diffusive dissipation, whereas when warming ramps slowly there is more time for equilibration -- the ductility of ice rises quite non-linearly with temperature.

The authors do not address this head-on (having a full plate already) but note that the migration of the melt lake line has accelerated 6x over the last twenty years from rapid changes in regional temperature and increased frequency of negative NAO indices (warmer, drier conditions). Even at 2200 m above sea level in 2060, some 70% of the icesheet is still too high and cold to have melt lakes (based on Fig.3 perimeter counts).

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #34 on: December 17, 2014, 06:47:02 PM »
Below I measured the areas of the largest polygons not covered by anybody's radar data. When the bedrock topography is moderately hilly, there is reason to wonder about how blanks of 10,000 sq km can be filled in where there is no data.

Polygon area is payware in Google Earth but I found numerous free online services that will process the underlying kml files, using http://www.earthpoint.us/Shapes.aspx in this instance. It is a little vague on exactly what is being calculated on which geoid (presumably relative to ice surface) but I doubt if this corresponds to more than 1-2% error.

In retrospect, it might have been better to have flown more gridded to surface velocity flow lines and their contours. Right now, most of it is kattywumpus to ice physics.

The best radar tracks in this sense are those directional flight vectors giving maximal dot product (whitest grayscale values) with the surface velocity vectors. The former would just be unit vectors since the speed of the plane is immaterial; the latter could either be unit vectors or include magnitude for extracting the best flights over the fastest ice.

Although this would be immediate if the Greenland research community had put up a proper GIS shop, it is still feasible to do because laying out the flight tracks in googleE meant that CReSIS had to provide lots of internal way points. Any pair of successive lat,lon way points determines the plane's bearing vector. So it boils down to getting these into excel, reprocessing to put in a kml color scheme and reprojecting into the same polar stereographic grid as surface velocity.

Espen

  • Young ice
  • Posts: 3383
    • View Profile
  • Liked: 254
  • Likes Given: 1
Re: Greenland subglacial topography
« Reply #35 on: December 17, 2014, 07:23:58 PM »
"there is reason to wonder about how blanks of 10,000 sq km can be filled in where there is no data."

Could be because of the previous radar units installed by US military?

DYE Stations: http://en.wikipedia.org/wiki/DYE_Stations
Have a ice day!

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #36 on: December 18, 2014, 03:53:18 PM »
DYEs are way to the south, ice penetrating radar generally goes straight down so they get a decent return bounce. Seems like they never did figure out how to anchor a building on a moving ice sheet, history at above link has comic aspects.

Dye-2 is on the wrong side of the hill in terms of detecting a Russian bomber fleet (later found not to exist) but at least it was close to town. I don't think the core there went to bedrock.

The DEW line concept has come back to life in the form of billion dollar coastal tethered balloon radar. Cruise missile threat from yet-to-be-built navy of a yet-to-be-defined enemy. https://en.wikipedia.org/wiki/Tethered_Aerostat_Radar_System

Speaking of tinfoil hats, I wonder if it would serve any purpose to deliberately place passive radar reflectors (or even active transmitter sources) at the bottom of drill sites. Then it might be possible to look at the icesheet obliquely, even get swath reception from a satellite overhead. Or just drill in sideways from a town where mountains would not block transmission to put in a high power source.

Right now the received signal from planes is quite strongly attenuated and subject to various levels of confusion related to multiple internal reflectors contributing to the return signal. So it differs from medical tomography in which source and detector are on opposite sides of the object of interest.

The Dye-3 ice core is still around, 2037 meters 10 cm diameter drilled in 1979. It didn't go straight down but was up to ~6º off from the vertical. The surface ice velocity there is 12.5 m/yr in the direction 61.2° true; the mean surface slope is 0.48%. At 500 m above bedrock, the ice velocity is 10 m/yr, same heading. We have very little experimental data of this type even today.

Bedrock temperature was −13.22º as of 1984, so no pressure melt there. The deepest 22 m consists of silty ice with pebble content increasing towards bedrock. The ice is brittle between 800-1200m. Accumulation is 2x that of central Greenland making for favorably thick annual layers but this is offset by consequent rapid ice flow and so unwelcome layer thinning and diffusion of climatic proxies, to the extent Dye-3 annual layer counting breaks down at 8 kyr.

Espen

  • Young ice
  • Posts: 3383
    • View Profile
  • Liked: 254
  • Likes Given: 1
Re: Greenland subglacial topography
« Reply #37 on: December 18, 2014, 05:33:03 PM »
A-Team I was actually engaged in the DYE project in the early 80ties, with C-130 with skies and the the whole lot, one of my few covert operations ???
Have a ice day!

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #38 on: December 18, 2014, 10:32:23 PM »
This picture seems to show you taking a private vehicle up to the NEEM drill site 20 years later. Must have risen in the ranks. Skis on the C-130 make sense; triangular wheels on an suv could explain some posts?

Espen

  • Young ice
  • Posts: 3383
    • View Profile
  • Liked: 254
  • Likes Given: 1
Re: Greenland subglacial topography
« Reply #39 on: December 19, 2014, 04:29:45 PM »
Thanks A-Team ;)
Have a ice day!

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #40 on: January 16, 2015, 04:27:14 PM »
I'm revisiting the actual radar data underlying the "Grand Canyon" paper, Bamber 2013, first because I live near the real Grand Canyon and don't believe this Greenland feature bears the slightest resemblance, and second because only a very small fraction of the radar transect data ever got directly used in the paper, perhaps 7 of 100.

Among those not discussed was 1999 0511_01_014 which by coincidence directly follows the Canyon bottom for at least 30 km, providing an incredible opportunity to visualize it. That year provided some of the best contrast radar ever produced by Cresis, including some 19 other segments that intersected the yet-to-be discovered Canyon.

Sample radar data was never identified in terms of Cresis flight segment but I quickly figured those out after getting their mercator pj of the Canyon overlaid Google Earth. Because two figures show the Canyon relative to tracks, misalignment below can be ruled out.

No kml track, say of the low point in the Canyon bottom was ever supplied so it is not so easy to place it relative to Greenland's summit ridge which it approximately follows, crossing under midway between NEEM and NGRIP.

As noted above, it is very difficult to imagine a physical basis for interpolation of sparse tracks into a meaningful bedrock DEM because way too much topographic change happens within the scale of the holes. South of Petermann glacier has among the very highest radar track density of anywhere in Greenland, yet the flown grid is at 10 km spacing, meaning the data holes are 100 km2. I'll look at the four walls of one of these grid holes in the next post.

Greenland is exceedingly flat. It is almost never shown at true scale (1:1 vertical to horizontal) for practical reasons. Cresis images all differ in scales but this does not accompany the images and must be computed by measurement. It is equally rare for the scale distortion to be published by a journal (though this paper did).

Looking at the Canyon at true scale below, it looks more like a gentle swale than creek bottom much less a deeply eroded glacier or stream valley.

Contrast this with a profile of the real Grand Canyon (at Navajo Bridge) -- that's a step function: the side walls drop vertically 142 m, flatten, rise again over a span of just 188 m. A swale with hardly of 1º side wall slopes does not lend itself to comparison with a canyon with 90º. Sure, the Colorado River at Yuma is an old tidal flat -- but that's not part of the Grand Canyon.

I've collected all the northern radar transects and will provide a virtual up-canyon flight  in the next post. Keep in mind that radar transects are the only data that exists; interpolated elevations, no matter how sophisticated or sincere, are modelled, not experimental data.

Subsequent posts will look in greater detail at the main 2010-2011 grid, first to decide whether the Canyon is actually a lot shorter than portrayed (with Jakobshavn having the lion's share of paleo internal drainage), then to examine the possibility that the northwestern 'bottom freezeups' of Bell 2014 have no physical association with the Petermann section or Canyon meltwater, and finally examine the prospects for these features (which fail to displace their volume in stratifications overhead) simply being relic ice caps left behind after regional melting during the Eemian.
« Last Edit: January 17, 2015, 02:11:08 PM by A-Team »

Lennart van der Linde

  • Grease ice
  • Posts: 782
    • View Profile
  • Liked: 85
  • Likes Given: 6
Re: Greenland subglacial topography
« Reply #41 on: January 17, 2015, 09:13:53 AM »
Very nice, A-Team, I had no idea.

The key question remains, as I understand it: how much of the 'canyon' or 'gentle valley' lies below sea level (how deep and wide, and how far inland), and how vulnerable does that make that part of GIS compared to parts that are above sea level?

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #42 on: January 17, 2015, 02:09:33 PM »
Quote
key questions remain: how much of the 'canyon' or 'gentle valley' lies below sea level (how deep and wide, and how far inland), and how vulnerable does that make that part of GrIS compared to parts that are above sea level?
Agree. Cresis did not begin adding a sea level line to the radar imagery until 2014.  Bamber 2013 provided depth below sea level to only 3 transects of the Canyon. I've tinted two of these in the images above and below.

The way this works, the measured ice thickness is subtracted from from the surface elevation to get the bedrock DEM. The surface elevation is known very accurately but the thickness data is sparse and an interpretive call (by work-study students). Despite the care taken, I've seen quite a few places where the '2echo_picks" of bedrock are either ambiguous or definitely wrong.

In the image below (Cresis 19990518_01_021, red bar in Fig.2 of Bamber), the Canyon a few km northeast of the NEEM drill site has a greatest depth of 176 m over a width of 12.5 km. It's very flat when presented at 1:1 scale (bottom image).

For comparison, the real Grand Canyon at the South Rim is 1524 m deep over a 16 km width, which rescales to 1191 m at the Greenland Canyon width of 12.5 km. That's a 676% difference in relative depth. If placed inside the real Grand Canyon along the Bright Angel trail, this portion of the Greenland Canyon would not even extend to the first bathroom (Mile-and-a-Half).

A very curious feature of 19990518_01_021 is the poor correspondence between the wavy deformation of the stratigraphy and what the students have delineated at bedrock. Normally these two are conformal -- warmish ice deforms as it slowly flows over bedrock obstructions -- but here there is a double dip in the isochrons but only a single dip in the topography.  (There are no ice upheavals in the area.)

Possibly this could be attributed to off-transect topography -- we only see what is under the flight line, not extreme topography that might begin a couple hundred meters to the side. In this view, departures from conformal layer present an opportunity to correct otherwise data-free areas of the DEM. The northeasterly flow here is oblique to the track, implying topographic impacts from nearer the summit ridgeline or from back-compression on the downhill side..

Note that the stratigraphy itself is not disrupted in any way even directly over the Canyon; it has retained self-conformality over a period of ~90,000 years. Whatever might be going with meltwater formation or transport at the bottom of the Canyon has had little to no effect on the stratigraphy above over this time frame.

And where is the pre-Eemian ice? -- at a meter per year here, it has not had nearly enough time to reach the coast. To me, this suggests a major melt-back of even northern Greenland during the Eemian.

I have not come across a serious effort at the reconstructing a time dependent paleo bedrock DEM (which controls paleo drainage). This involves slow inelastic lagging displacement and re-emplacement of underlying asthenosphere and rapid elastic response to loading and unloading during glacial cycles.

Looking at comparable unloading centers at Hudson Bay and Sweden/Finland, the landscape appears totally flattened, much more so than interior Greenland. Thus prior to being a bowl, interior Greenland may have been quite mountainous 2-3 million years back.
« Last Edit: January 18, 2015, 09:36:41 AM by A-Team »

Lennart van der Linde

  • Grease ice
  • Posts: 782
    • View Profile
  • Liked: 85
  • Likes Given: 6
Re: Greenland subglacial topography
« Reply #43 on: January 17, 2015, 06:01:42 PM »
In the image below (Cresis 19990518_01_021, red bar in Fig.2 of Bamber), the Canyon a few km northeast of the NEEM drill site has a greatest depth of 176 m over a width of 12.5 km. It's very flat when presented at 1:1 scale (bottom image).

Very flat indeed. Is there any indication if this depth and width are about average for most of the canyon/valley, at least for the first 100-200 km or so? If the greatest depth at this specific site is 176m, then the average over the whole 12.5 km width is much less. The other site showed 450m as deepest point and maybe 7-8 km width?

Petermann itself is 15 km wide, according to Wikipedia. Does its fjord get much deeper behind the current grounding line? Or is this not such an unstable glacier, compared to Jakobshavn, not to speak of the big WAIS-glaciers, with their great potential for Marine Ice Sheet Instability?

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #44 on: January 18, 2015, 01:47:27 PM »
All good questions.

The next post will pull together all the available transects of the Canyon from where it crosses the NGRIP-NEEM summit line up to the lower part of Petermann proper. It's an open question whether the Canyon really extends to south-central Greenland (without running uphill or entirely losing canyon character) but its outlet is assuredly Petermann, which cannot be a coincidence since that is a rather small feature of this area.

The Petermann area, especially the cross-gridded areas in the map below, is very much complicated by massive bottom upheavals. These are not closely associated with the Canyon (or its putative meltwater) as sometimes suggested. Whether these are relic ice caps or freeze-ups will be the subject of yet another post.

The first image below overlays the Parca flowlines over the channel and 7 study transects of Bamber 2013. Note in upper and mid-image that the Canyon cuts almost orthogonally across the surface flow lines (which are also an excellent approximation to internal flow planes), though elsewhere it is mostly aligned with the flow. This lack of correlation with present day ice slumpage directions suggests that the Canyon is 'inoperative' today, not influencing current or near-future events.

The second image characterizes the last of the Bamber 2013 cross sections. Here the deepest part of the Canyon is 235 m below its rim and  in a region where it is 3.8 km wide, in a broad region slightly below sea level. (The Canyon itself is 380 m below sea level.)

Here the bottom trace is satisfactorily conformal to the 46.85 kyr isochronal horizon to the east and center but stays flat as the latter (and the whole foliation) continue to rise fairly steeply, which is thus not attributable to bedrock topography.

The third image completes the process of identifying the mystery transects in the original paper and its supplement. This was quite tacky of Science to let unlabelled images with quite different (and severely horizontally compressed) scales sail through.

I would also have required a grayscale bump map in upside-down view (see Jakobshavn forum) to make it clear where stretches of the Canyon goes uphill. Not to mention a thorough discussion of physical principles the authors used to determine what bedrock topography is doing in data-free areas -- radar tracks are very sparse along the Canyon even where they are the densest (~10 km spacing).

The 46.85 kyr isochron is part of an unmistakable trio of radar reflectors I call the "Three Sisters". It's seen all over central and north Greenland; its overall correlation (or not) with bedrock topography will surely receive prominent treatment in LayerMap. A forthcoming post assembles the four walls of a single rectilinear grid cell and asks how we best interpolate  smooth isochronal boundary lines into a isochronal surface.

This involves looking which of the many methods of geostatistical interpolation makes physical sense in the englacial context. The 46.85 kyr isochron must be mainly a flat surface of modest slope since samples look that way; the deformational waves too are smooth and simply modeled.

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #45 on: January 19, 2015, 03:40:41 PM »
I wondered above what the scientific basis could possibly be for infill of gaps in the sparse radar data used to determine the bedrock DEM. It seems that 'ordinary kriging' was used as the geostatistical interpolation method (even though it is inappropriate and sub-optimal here, see below). That generates an error map which I located both on M Morlighem excellent site and also at NSIDC in all its detail as a 1.4 GB netCDF download (which can be opened with Panoply freeware).

Notice the huge formal errors in interpolated bedrock elevations -- up to 600 m, which could easily amount to 25%. This is so large relative to measured bedrock elevation changes over characteristic distance scales that structural features inside the empty polygons hardly improve on guesswork. That certainly leaves paleo stream flow up in the air for gappy regions.

http://forum.arctic-sea-ice.net/index.php/topic,867.msg41632.html#msg41632
http://sites.uci.edu/morlighem/dataproducts/mass-conservation-dataset/
http://nsidc.org/forms/IDBMG4_or.html?major_version=1
http://www.giss.nasa.gov/tools/panoply/download_mac.html

I looked around for decent explanations of 'ordinary kriging' since the main wikipedia article falls into the statistical weeds already in its first sentence, never really spelling out the assumptions that need be met for kriging to be applicable, namely that the data points and their auto-covariance represent a random sample from some well-behaved  distribution.

The article goes on to put down bézier splines but the fact is these will provide a superior fit to radar isochrons because of a better fit to the physics governing ice deformation, to which kriging is totally oblivious as it is to ore body shapes. (I've always wondered if Krige's mining company went broke following his advice.)

Bézier splines were originally introduced to describe compressible hydrodynamic flow over auto bodies (at Renault and Citroën, 1960's) so they will work better for ice. Bedrock deformation under isostatic gravitational loading and glacial  erosional processes also leads to systemic effects but with different governing physics.

Kriging is sometimes described as the 'best linear unbiased prediction'. While that sounds hard to improve on, better than reality even, it's only as good its unrealistic assumptions and only as bad as the physical insights it neglects.

Kriging is another Daisy World of the modelling community, an over-simplification that can produce highly erroneous output yet somehow get billed as 'optimal', as if statistics somehow can be a workaround to (unnecessary) laws of physics. The real appeal of kriging is its quantification of error -- wrapping output in a soothing pseudo-scientific veneer.

Thus the error map below does NOT show experimental error in our knowledge of bedrock elevation beneath the Greenland ice. Instead it shows the statistical error in an abstract inapplicable statistical model world that has REPLACED the real world and taken on a 'reality' of its own. It doesn't really matter how the bedrock map is filled in -- at the end of the day,reported error should be with respect to actual physical measurement (held-back oblique radar tracks).

As a practical matter, kriging calculates the value of an unknown interior point simply as the average of nearby points, de-weighted by their distance away. Although that is traditionally taken as 1/r, it could just as well be 1/r2 etc depending on the context. See links below.

The coefficients of that power series expansion could be empirically determined for the case at hand (Greenland) by 'holding back' radar tracks that cut diagonally across grid squares, then varying the de-weighting until the parameters of best fit are determined. However I haven't noticed anyone 'holding back' tracks in the Greenland scientific literature to get a real grip on error.

The radar track portfolio does not remotely meet the criterion of random sample (stochastic process). First, track density is very strongly correlated with proximity to a few Greenland airports. For example, nearly all flights in northern Greenland originate and return to Thule, almost all passing over Camp Century.

Many flights follow the summit ridge, fly parallel to the coastline, or execute a tight grid just up from select marine terminating glaciers. Very few followed principle curvatures (elevation contours and flow lines). However each flight had distinct pre-specified scientific objectives. While all this is sensible, statistically it amounts to strongly biased sampling.

Geostatistics is a highly developed field because data is always too sparse yet interpolative infill is always wanted. However one size doesn't fit all, the optimal method must be adapted to actual sampling and expectations from the operative controlling physics. Above all, a feedback cycle from prediction validation (held-back data) must inform the interpolative method.

In Greenland radar data provices continuous vertical slices. That is already very different from point data (eg ice drill cores). The result is a partition of Greenland into vertically extruded polygons whose values are known on the walls.

Additionally the ice surface elevation itself is known everywhere to great accuracy so needs no interpolation, as with ice surface velocity. With some mild physics, these two caps to the polygons can significantly inform infill of interiors.

Slight ridging of the ice surface may have some information about what's underneath --  however it is largely unable to predict bottom upheavals or conformal displacement of isochrons. Conversely, roughness sometimes observed in the surface DEM may have no evident englacial counterpart.

Isochrons are exceedingly well-behaved on many radar tracks, easily fit to simple continuous equations, readily modified to reflect dimpling waves, thus making them accessible to reliable analytic continuation to interior surfaces of the extruded polygons. Isochrons are a more favorable situation than than bedrock topography itself which is finer-grained and so harder to interpolate.

Indeed it is the mid-depth isochrons, in conjunction with ice thickness overhead, that are best suited to refining bedrock topography by providing intermediate caps. In this view, the unusual data structure of Greenland calls for something different than 'ordinary kriging', more along the lines of incorporating polynomial trends as in 'universal kriging'.

One problem -- or is it an opportunity -- with Greenland's extruded polygons is their surfaces are generally oblique with respect to ice flow (~ surface elevation gradient) and so physically unnatural. These polygons often have 4 sides but are not generally rectilinear. (In hindsight, it might have been better to have flown more gridded flowlines and elevation contours.)

The Petermann glacier is a special case of a regular 10 km grid in which some 300 extruded polygons are both four-sided and orthogonal, with very low angles to ice flow. This area is also sliced by dozens of oblique flight lines that can be held back for validation.

Let's consider the middle isochron of the 'Three Sisters' on an opposing pair of east-west polygonal walls (flow being northward). Suppose both are fit to simple splines and the interior predicted as a parametrized deformation of one into the other that incorporates deflectional effects of the direction and magnitude of ice flow.

Repeating this process with the north-south opposing walls gives an independent prediction (anchored to the same corner boundary conditions). These two predictions then generate a difference DEM, the best least squares splinal surface fit to which is the final interior infill prediction. The hundred or so other isochronal surfaces will be a closely related one-parameter modulation of this fit.

Since dimpling of isochrons in general reflects the ice's response to passing over bedrock topography in a low pass fourier filtration sense, the interior isochron provides an inversional control on the edge interpolation of bedrock.

http://www.bisolutions.us/A-Brief-Introduction-to-Spatial-Interpolation.php
http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#//009z00000076000000.htm
« Last Edit: January 19, 2015, 07:24:07 PM by A-Team »

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #46 on: January 19, 2015, 04:30:11 PM »
We very rarely see a properly colored map coming out of Greenland glaciology. I've come to realize that this is mostly a consequence of being trapped in proprietary software packages that offer choices but very little control over the palette. There are other issues: poor taste, no familiarity with scientific palette construction, working in 'presentation' mode, and ~10% of senior authors having some form of genetic colorblindness.

I've yet to see the Greenland bedrock DEM presented in vertically unexaggerated natural 12-bit grayscale. That would be a great choice for displaying the Canyon because it runs north-south, meaning a simple horizontal detector applied to the complement would greatly enhance indistinct courses of main channel and faint tributaries.

The hue slider in Gimp is a wonderful tool for rapid palette exploration. Below, two variations on the standard spectral coloration of Greenland bedrock show that certain rotations in HSV color space bring out the Canyon more distinctly in different regions. I saw another palette that was chosen to show the Canyon somewhere where it was wanted, but as far as I can tell (see error map discussion above), isn't.
« Last Edit: January 19, 2015, 04:44:49 PM by A-Team »

Shared Humanity

  • Guest
Re: Greenland subglacial topography
« Reply #47 on: January 19, 2015, 05:01:10 PM »
A-Team....

As I plow through your posts (much of it unintelligible to me because of my lack of knowledge), I can't help but wonder what the composition of the 'bedrock' is as this'canyon' winds its way through Greenland's interior and makes its way to the coast. When I look at the faint tributaries feeding into the main canyon, an image of a large river system seems to emerge and I would not be at all surprised if the 'bedrock' is, in fact, composed largely of loose till formed by the grinding action of the massive sliding ice sheet. Is there any way that this 'bedrock' composition be discerned?

wili

  • Young ice
  • Posts: 3342
    • View Profile
  • Liked: 612
  • Likes Given: 409
Re: Greenland subglacial topography
« Reply #48 on: January 19, 2015, 07:37:36 PM »
I don't have anything at hand about the nature of the land under the GIS. But I just came across this on the Antarctic:

http://www.scientificamerican.com/article/scientists-drill-through-2-400-feet-of-antarctic-ice-for-climate-clues/

Scientists Drill through 2,400 Feet of Antarctic Ice for Climate Clues

Quote
...a team of researchers using ice-penetrating radar reported finding a wedge of sediment 30 meters thick at the grounding zone of the Whillans Ice Stream. This sandy heap actually causes the oozing ice to slow...

...When the drill’s metal nozzle was raised from the hole, workers found globs of rough, sticky mud clinging to it—evidence that the bottom really had been reached...

...the camera came to rest on the bottom beneath it, revealing it to be muddy and strewn with pebbles...

"sediment" "sandy heap" "sticky mud" " muddy" "strewn with pebbles" sure don't have the same ring as 'bedrock,' do they?
"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."

A-Team

  • Young ice
  • Posts: 2775
    • View Profile
  • Liked: 721
  • Likes Given: 35
Re: Greenland subglacial topography
« Reply #49 on: January 20, 2015, 11:56:56 AM »
Quote
...a team of researchers using ice-penetrating radar ... The broad rift zone of West Antarctica is thought to resemble the basin and range region of the American West ... the Whillans Ice Stream slides over a hidden ridge, a sticky spot. The Whillans lurches forward 45 centimeters twice a day, from ocean tides lifting the ice so that it can slide over the sticky spot, relieving mechanical strain that has built up on the glacier since the last high tide
.
Nice bit of research and a well-written SciAm account. Just to clarify, the 740 m ice-penetrating radar study reporting the till delta (sedimentary wedge) was conducted in 2007; free full at ftp://ftp.bartol.udel.edu/anita/amir/Corsika_files/Attenuation_reflectivity/1835.pdf  I consolidated the radar scan and graphics below and colored them for clarity.

The authors use phased radar (observing a Ricker wavelet/mexican hat and polarity reversal) to get at permitivity, which rules out sea water, meltwater, basal accretion ice, temperate ice, bedrock and sideswipe leaving unfrozen till layer with nearly fresh pore water as the remaining option

The SciAm article does not provide the lat,lon of the 2015 drill hole relative to the 2007 grounding line/accretionary wedge so, shorting of emailing the reporter (did that) or waiting 8 months for an article, we can't overlay the hole onto the graphic. However we can be sure that the drillers were very aware of the 2007 study.

It appears that the radar study, though it got a lot of things right, was not able to predict "pebbles that were seen on the bottom occur only in a thin surface layer of mud. Below that the mud contains only sand." This is a huge step forward to actually have mud samples to work with -- glaciology is cursed with too much modelling.

I find it very peculiar that a thick volcanic layer did not give rise to a strong radar reflector. The original scan is given in the 2007 supplementary material -- it has a few weak and wavy isochrons, nothing corresponding to what they are reporting from drilling. The depth below the surface is only described as 'part way down' -- if we knew that depth, the eruption could be dated. It may or may not be from a local volcano.
« Last Edit: January 20, 2015, 12:04:01 PM by A-Team »