NEEM Ice Core data from NE Greenland

[CraigsIsland]

Neven - not sure if this belongs in Greenland or Consequences

I ran across an article regarding a core sample from the Eemian Period. During this period, the global mean temperature rose 2C while places in the Arctic reached 3-5C, due to natural variations in the earths orbit (being closer to the sun).

Most interesting part of article: "Based on the study of the paleoclimate record in ice cores as well as the former locations of beaches and coral reefs, researchers believe that the Eemian temperature increase likely pushed global sea levels as high as eight meters (26 feet) above where they are today."

http://www.popsci.com/science/article/2013-05/cold-hard-facts

..."This data yielded a significant discovery: Previous models had assumed that the Greenland ice sheet was at least half gone at the Eemian’s hottest point, but the ice core showed that the sheet’s total volume decreased just 25 percent. That meltwater would account for only about two meters of global sea rise, according to Dorthe Dahl-Jensen, a paleoclimate researcher at the University of Copenhagen who led the NEEM project. Therefore, the remaining six meters of water had to come mostly from Earth’s other major sea-level “co-conspirator,” as James White likes to call it: the Antarctic ice sheet. This new finding suggests that West Antarctic ice is capable of melting far more than previously thought."

[Neven]

Here is fine, Craigsisland, although I'm quite certain it was discussed before somewhere.

[Anne]

Seems to be something of a logic fail in Patrick Michaels's interpretation of the paper today for his Cato Institute blog describing his trip to Greenland.

Whew!  Thus does one revolutionary paper shoot pretty much the entire global warming sea-level catastrophe—the one worth being concerned about—through the heart.  Antarctica is so cold that it is projected to gain ice in the coming century, as slightly increased precipitation—which may have recently been detected—falls as more snow, which compacts into more ice.

This puts any sea-level crisis out in the hundreds-of-years realm, at least, and probably far beyond our current era of burning hydrocarbons for energy and heat. In other words, forever.

To paraphrase his conclusion:  "Dahl-Jensen says Greenland didn't melt as much as we thought therefore sea level won't rise as much. I don't believe anything about projected Antarctic ice loss. Besides, I didn't see much melting when I flew over Greenland. Nothing to see here, move along."

(While he doesn't dispute that there is warming, or that there will be other consequences than sea-level rise, he thinks those consequences are likely to be beneficial.)

He says in the blog that in preparation for his trip, "I read just about everything I could get my hands on" yet it's this one single paper, and this single strand of its conclusion that he seizes on. Here's the Science Daily account of that paper, with a link to the full paper.

Cato Institute has a reputation for downplaying the effects of climate change, which surely has nothing to do with the nature of its funding.

[Neven]

Yes, I saw his remarkable thinking on WUWT as well, besides his obvious lack of knowledge when it comes to the fjord where Jakobshavn ice bergs end up.

The gist was: it will take thousands of years for the GrIS to melt. But of course, it's all about the first metre of sea level rise.

Oh well, at least someone made a living out of spreading propaganda.

[A-Team]

it's all about the first meter of sea level rise.

Right. Academic research is tilted today towards Antarctica but Greenland may be actually provide that first-meter risk.

That's because the old narrative of a gently sloping ice sheet frozen to bedrock slowly slouching towards the sea has been replaced with warmer higher ice (basal deformations) not even contacting bedrock (~50 m of intervening hydrated deformable till), not to mention unforseen issues with warm ocean currents, rain-on-snow and moulin meltwater drawing down the calving front.

I'm resurrecting this old forum because it's aptly titled despite its disproportionately meagre coverage of what NEEM receives in the scientific literature: 'NEEM ice core Greenland' has 1230 matches at Google Scholar, of which 186 are for 2015 and 8 already in January 2016. Not 1% of these articles has been discussed at all our forums combined.

Deep ice cores are far and few between so every conceivable analytic technique is applied to them. The goals in studying this 110,000 years of climate archive are quite varied, ranging from atmospheric gas composition (especially methane), to prevailing temperature (from isotopes), snowfall, global volcanic activity, aridity of dust sources, frequency of boreal fire, behavior of ice fabric under slow strain, solar excursions (from Be¹⁰ and Cl³⁵), stratigraphy coregistered with global (Antarctic, speleotherm, marine sediment layers, tree rings, pollen) records, all at near-annual resolution.

No working scientist believes the Eemian per se can serve as a template for what will happen in the late Holocene given such-and-such a CO₂ level, though that's wrongly argued from daisy world modeling. The Eemian didn't just come and go, it left its mark on the deeper ice of Greenland whose effects continue, even dominate, basal ice today.

There is more than ice in the NEEM core if you look hard enough. The impurities are a very informative part of the record and need to be understood first because they provide the isochronal reflections that record the evolutionary history of the Greenland Ice Sheet (and so by implication set the ice's internal physcial parameters compatible with that history).

It's fair to ask whether the properties of ice pulled from great depth/pressure are retained during transport, cold storage and sample processing. For example, CFA (continuous flow analysis) melts the ice on a hot block and so could measure total Na⁺ and total Cl^- but not determine NaCl because some might be Na₂SO₄ or MgCl₂.

Similarly, as salts dissolve rapidly in meltwater, CFA cannot distinguish wet-deposited individual salt molecules intercalated into ice crystals (or their grain boundaries) from sea salt deposited as mineral particles. In the article below, the authors use a sublimation chamber (ice to vapor) to get rid of NEEM water, then catch the macroscopic particles -- whether they were inherently soluble or not -- on a filter.

They characterized some 32,000 captured particles from representative historic periods and analyzed them with micro-Raman and SEM-EDS X-ray spectroscopy. The former works because calcium carbonate CaCO₃ peaks at 1086 cm⁻¹ because of a symmetric C–O stretching mode which distinguishes it from dolomite CaMg(CO₃)₂ at 1095 cm⁻¹ and nitrocalcite Ca(NO₃)₂ x 4H₂O at 1051 cm⁻¹ (N–O vibrational mode) and gypsum (CaSO₄⋅2H₂O) at 1008 cm⁻¹ (S–O stretching mode).

Unsurprisingly, kaolinite (clay) and quarz silicates comprise the dominant insoluble components.

The authors look at ice representative of four periods in the history of the Greenland ice sheet. I am especially interested in these because the Bølling-Allerød produces the a prominent dark band in ice penetrating radargrams that is preceded by the non-reflective zone of Younger Dryas and the many distinct recent reflectors. (The article does not discuss whether the particles are themselves the reflectors or merely co-located with whatever dielectric contrast is causing it.)

Holocene 11,700 - 600 (warmest period 9,000 -5,000) more Na than Ca
Younger Dryas 12,600 – 12,000 (cold) more Ca salts than Na
Bølling-Allerød 14,600 - 12,900 (warm and moist interstadial) similar Na and Ca
Last Glacial Maximum 26,900 - 15,000 (cold) more Ca salts than Na in cloudy vs clear bands

In air masses reaching Greenland, primary land and sea aerosols need to be distinguished from secondary gas-to-particle conversion and chemical reactions (notably with sulfuric acid) in the atmosphere. During transport, both can undergo physical and chemical processes that change particle size, structure and composition.

In the Arctic, the natural soluble aerosols originate from the primary emission of sea salt (NaCl and MgCl₂)  and terrestrial materials (CaSO₄ and CaCO₃). Biomass burning plumes also represent a significant source of aerosol (KNO₃, K₂SO₄, and KCl).

Soluble aerosols include NH3 emitted by bacterial decomposition in soils and biomass burning, sulfur from dimethyl sulfide (DMS)) emitted by marine biological activity and volcanoes (SO₂), and nitrogen oxides emitted from  soil by microorganisms and biomass burning or produced within the troposphere (lightning) and stratosphere (N₂O oxidation). Secondary aerosol contains salts such as NH₄NO₃, NH₄HSO₄, and (NH₄₎₂SO₄.

Ca⁺⁺, a terrestrial proxy, decreases from cold to warm periods up to a factor of 80 between the LGM and the Holocene in the GRIP and GISP2 ice cores. The Na⁺ concentration, a proxy of sea salt, varies less than Ca⁺⁺ but is still much more concentrated in cold periods than in warm periods. The NH₄⁺ concentration, originating mainly from continental biogenic emissions, starts to increase around the Bølling-Allerød period.

Note that NH₄⁺ is a major cation together with H⁺ on a molar basis. The SO₄^-- concentrations are higher in cold periods than in warm periods. Whereas noneruptive volcanic emissions and marine biogenic emissions are the main sulfate sources during the Holocene, the large imbalance observed between cations and anions in LGM Greenland ice suggests that the strong increase of sulfate in glacial ice reflects increased terrestrial inputs (direct emissions of gypsum and CaCO₃ neutralized in the atmosphere by H₂SO₄).

Chemical compositions of solid particles present in the Greenland NEEM ice core over the last 110,000 years
I Oyabu et al
27 Sep 2015  DOI: 10.1002/2015JD023290
http://onlinelibrary.wiley.com/doi/10.1002/2015JD023290/full

[A-Team]

Here are some images derived from ice-penetrating radar flights that passed over the NEEM drill site. The folded fold at depth detected from analyzing the core is not apparent, raising the question of why not (or rather, what other deep folds have been overlooked).

[A-Team]

The high resolution version of the NASA-processed MacGregor-enhanced image (5760 x 3240 pixels) shows some prospects for showing overturned folds at NEEM. The flight line accession was not provided, only a date (02 May 2011) so more work is needed to precisely locate the NEEM drill hole on the image. 

The second image really could go with post #4; it shows the dark and light bands fitting warm and cold periods which fits with their relative particle accumulation. However these regions could have similarly high resp low levels of non-particulate salts that furnish the dielectric contrast.

The 3rd image compresses the original radargram flown over the Neem drill site to 15% of its original width. This increases the vertical exaggeration (VE) from its initial 10.6:1 (horizonal to vertical km) to 70.9:1 (inset). Note that although isochrons in the original image appear flat, they actually are not. In the horizontally compressed image, the upper Holocene stratigraphy is still flat but the lower isochrons are humped.

Although it is common for middle Ice Age layers to drape over topography, that cannot provide an explanation here as the bedrock profile stays fairly level and does not develop a parallel hump. Further, adjacent tracks both parallel and perpendicular fail to disclose significant topography that might be affecting the NEEM sight indirectly.

This implies that the NEEM drill site was actually sited over a basal deformation.

This small undramatic upheaval can be seen to have diffuse radar reflectors but not the strong features seen in Petermann and Zachariae deformations. Nonetheless, this upheaval could explain the double folding observed in the ice chronology at 2200 m and deeper.

More broadly, it would have been better had Cresis also annotated the 46.5 kyr 'middle sister' at the same time surface and bedrock lines were being drawn. Then it would have been easy to map all the height discrepancies Greenland-wide, ie places where draping can not account for isochron warping.

NASA SVS site:

https://svs.gsfc.nasa.gov/cgi-bin/details.cgi?aid=4249

See also http://forum.arctic-sea-ice.net/index.php?topic=867.0 for a different take on Greenland ice sheet sections.

[A-Team]

I'm looking in this post at potential benefits that might come from extreme 'sideways' compression of radargrams. These are posted with variable levels of vertical compression (ie ratio of pixels provided for a 1000 m of ice depth to pixels representing a km of flight track).

Typically the VE is around 10:1 but it can vary widely. Previous posts have looked at how to best uniformatize the VE across many radargrams without degrading image quality. Even a moderate VE can be very misleading, bringing drama to a feature like bedrock canyon or basal deformation that they just do not have at 1:1 scale. For example, Greenland's  'Grand Canyon' is really just a gentle swale that a person would hardly be aware of walking down it.

The post above looked at 20110329_02_028 over the NEEM drill hole at its intial posting and at 15% sideways compression achieved with bicubic interpolation in gimp (see first image). This had the surprising effect of greatly improving detection of humped isochrons for which bedrock draping could not be responsible. Error in slope measurement seems to be reduced.

More recent Holocene layers stayed flat, serving as an internal control for artifact introduction. The tilt artifact remains unresolved but not critical as it is constant 1º-2º across radargram segments. In working with basal deformations, it might be advantageous to expand the vertical scale at the same time the horizontal is compressed. The resulting too-tall image can always be cropped down later to the Bølling-Allerød line.

Gimp provides four options to figure out pixel values for a given scale change specification: nearest neighbor, bilinear, bicubic and sinc.

Which is best in this application (enhancing faint basal deformations)? While some guidance is supplied on gimp forums, as a practical matter it is easy to perform all four and compare visually and quantitatively (eg gimp subtraction can find the specific differences). ImageJ has mediocre default options for rescaling but a sophisticated plugin 'Resize' installed effortlessly for me.

The standard method for size reduction or enlargement is to fit the pixels with a continuous model and to resample the function at the desired rate. In the case of image reduction, the creation of aliasing is the major problem of this approach. Blocking or smoothing artifacts may also be introduced when low-order models are used.

These undesired effects are reduced if we approximate the continuous model by its least-squares projection onto a given space prior to resampling. This new approach is a generalized version of the anti-aliasing filtering operation. It is optimal in the case of image reduction. It can accomodate any arbitrary (non-necessarily rational) scaling factor. When the scale parameter is an integer power of 2, our approach is equivalent to wavelet processing because splines satisfy a two-scale relation. The boundary conditions are mirror-symmetric.

http://bigwww.epfl.ch/algorithms/ijplugins/resize/.
http://www.ee.cuhk.edu.hk/~tblu/monsite/pdfs/munoz0101.pdf

Rescaling is a big topic -- Wikipedia has an excellent treatment (24 different algorithms https://en.wikipedia.org/wiki/Image_scaling) whose open source implementations might be worth pursuing.

However the vast majority of applications call for resizing evenly in horizontal and vertical directions. Our issue is different because vertical is held fixed, contrast is a huge issue, and anisotropy is present (isochron horizontal striping) and potentially exploitable.

The four gimp options define how to scale the image. Each option describes an algorithm used to do this. None is inherently superior in all situations. 

None: The nearest-neighbor algorithm is used. There is no smoothing after scaling.
Linear: Touching pixels average their values.
Cubic: Touching pixels average their values so central pixels maintain the most value.
Lanczos: Pixels are passed into an algorithm that averages their color/alpha using sinc functions (similar to sine interpolation, somewhat like cubic).

None (nearest-neighbor): Use when you want absolutely no sampling (blurring) of the image.
Linear: Use when you have very small text; cubic interpolation is usually better otherwise. This produces blurred, but jagged, edges.
Cubic: Use for most images. Unless the image is very small or incredibly detailed, cubic and bicubic interpolation helps keep edges smooth. According to Wikipedia, it can sometimes increase perceived contrast or cause artifacts.
Lanczos: This interpolation method is much like cubic except that instead of blurring, it creates a "ringing" pattern. The benefit is that it can handle detailed graphics without blurring like the cubic filters. [explanation provided by Joshua Lamusga]

I'll add some comparisons of variously compressed radargram imagery in a bit.

[A-Team]

I've been calling for experimental drilling of a basal deformation (the one near Eqip) to see what exactly is going on. It now appears that might already have been done (Neem) and we just didn't realize it. Even if the case, this deformation under a spur of the summit ridge may not be representative of extensive deformation fields in much shallow non-Eemian ice at Petermann and Zachariae.

The intent at Neem was to reach undisturbed Eemian ice so the overturned folds came as a complete -- and unwelcome -- surprise. Site selection began well into the ice penetrating radar era so the question becomes, how did the radar miss these upheavals? The previous two posts suggest that the radar of the day did in fact detect the deformation but required better processing to bring it out.

Flight segment 20110329_02_028 passed over the Neem drill site early in its northwesterly path. For better context, I've tiled it with 20110329_02_027 and cropped so to center on the vertical hole representing the ice core. The ground resolution of these two radargrams is about one pixel per 45 meters so the aim is a more regional view that could establish that the drill hole is on the NE shoulder of a large but moderately sloped basal upheaval that only comes into view upon strong horizontal compression of the radargram.

That is, curvature of mid-aged isochrons brought out by horizontal compression can be used in a region of flat bedrock as a proxy for poorly resolved (or altogether invisible) basal deformations.

The image below compares various options for compression. Here I cropped to 1400 pixel width and shrunk sideways by 12.5% or 1:8  to 175 pixel width (as this works out to four abreast at forum limit of 700 pixels), while only slightly dropping original height. Resizing algorithms don't require an integral multiple like this but it simplifies explanation of how the final grayscale pixel value is figured by the various algorithms. Most will use a neighborhood information from above, below and to the sides so are not one-dimensional.

To a certain extent, the optimal method depends on human visual perception in addition to what is best from the information-theoretic perspective. And it may be that no approach is capable of sharpening the image if radar reflectors in deformations are inherently blurry at depth.

We've previously considered the view that dark bounding lines in deformations are not isochrons at all but rather wave fronts of impurities carried by upwelling temperature, pressure, buoyancy, or ice fabric contrasts, the glacial counterpart to zone refinement in metallurgy.

http://forum.arctic-sea-ice.net/index.php/topic,867.msg45581.html#msg45581
http://forum.arctic-sea-ice.net/index.php/topic,867.msg45620.html#msg45620

In zone refinement, a moving band of high temperature locally melts an initially impure crystal; as it cools and recrystallizes behind the melt, impurities are excluded from the crystal lattice and pushed forward and (after many cycles) out the end.

Diffusion of heat is quite slow, enabling persistence of massive blocks of ice heat relics from warmer eras, eg temperate paleo-firn described by Luethi in Greenland ice near Swiss Camp. At Neem, questions have been raised about both the chronology and the O¹⁸ temperature proxy. The data applicable to the upheaval zone needs to be revisited under the assumption that it represents a sample through the flank of a basal deformation.

http://www.the-cryosphere.net/9/245/2015/tc-9-245-2015.pdf
http://forum.arctic-sea-ice.net/index.php/topic,984.msg36237.html#msg36237

[A-Team]

The images below look at another flightline near Neem (which is also rife with semi-cryptic basal deformations) and compare the three main interpolation methods (37.5% horizontal squeeze to the same vertical exaggeration) by grain merge (subtracting, adding middle gray, equalizing histogram). The differences are real but very subtle.

Ground radar was also towed in a grid around Neem in 2008. The path lengths are on the order of 0.5 km per radargram full width versus ~50 km. These don't show any sign of basal deformation -- it appears compression of longer flight paths is necessary to bring up a larger context: deformation of lower isochrons regionally in response to upward ice forcing. Again, the surface ice has a slope of 1-2% but the bedrock here is basically flat, whereas the lower isochrons are forming very substantial relative slopes upon compression.

[A-Team]

I've been scratching around for graphical methods to disentangle what happens regionally with basal ice deformations. Overlying isochronal layers are pushed up but they also thin as some flows out to the sides, attenuating the deformation of younger layers.

Ice layers also drape conformally over bedrock topography, in some areas more noticeably and at finer scale than others; this effect also attenuates with height and scarcely affects the surface in areas of deeper ice. (Hence the need for ice penetrating radar, the bedrock topography cannot be deduced from the surface DEM.)

For a given radar flight segment, Cresis already provides digitized bedrock and surface elevations derived from manual tracing of those contours as simple spreadsheet cvs tables. They elected not to trace any intervening prominent layers such as Younger Dryas, the 46.5 triple, or an even older double.

The much more comprehensive layer tracing in MacGregor 2015 surely included these within their analysis of the radar archive but, as far as I can tell, released only a Greenland-wide gridded product but no cvs files that go with individual radargrams that would have supplemented the top and bottom Cresis lines with these prominent interior isochrons.

Below I re-traced the bedrock and 46.5 kyr isochron on a passing over the Neem drill hole, then filled the intervening space with a vertical grayscale gradient (first image). If top and bottom grayscale values could be read out along the bounding lines, subtracting them would give the ice thickness between them, in effect removing bedrock topographic variations, with the expectation that excursions from uniform thickness would reveal basal deformations whether overt or cryptic.

Repeating the same process with the isochron defining Holocene onset would give a flatter line because of attenuation from thinning. This would allow not only quantitation of basal deformation volume but also partitioning between forcing younger ice layers up and thinning them to the sides. However labeling heights with at best 256 grays followed by subtraction causes accuracy to suffer. Surprisingly neither ImageJ nor Gimp is set up to readily read out a list of these numerical values.

I found another method to do this graphically in Gimp. After tracing the isochrons (manually, with deformable splines, or magic-wand selection) and deleting regions outside the lines to alpha transparency, the residual image is cut into many one pixel wide full height slices using the grid-guides.scm plugin discuss a few posts back. This tool resides in Image --> Guides --> Grid. Applying the Filters --> Web --> Slice tool to these guides names and saves all these slices to disk.

The image is then reconstituted by File --> Opening as Layers followed by Filters--> Combine --> Filmstrip provided the stack of layers (one pixel slices) is in the right order. However these need Layers --> Autocrop to remove transparency What makes this all work is that Filmstrip discards transparency. This has the desired effect of dropping each slice down to the x axis accomplishing the necessary subtraction (second image). Most people working on similar issues recommend ImageMagick for better control on batch processes.

While this is largely an automated procedure already thanks to grid-guide, Slice and Filmstrip, those scripts need to be wrapped in a larger script to do this analysis in selected regions on the scale of the Cresis archive. That would not make sense if the key isochrons could be found already numerically digitized within the MacGregor 2015 project. However they may only have sampled and kriged to a grid rather than fit analytic bezier curves to individual radargram lines.

[A-Team]

The first image below, based on data of MacGregor 2015 and worked up into a video that we've seen before at NASA SVC , shows what is left after all the Holocene ice is stripped off (that is, down to the 11.7 kyr isochron or end of Younger Dryas or top of the ubiquitous bright band in ice penetrating radargrams that lies between the dark band of Bolling-Allerod and the dark band of Holocene).

The view is in (undescribed) perspective with quite a bit of (unstated) vertical exaggeration. It may be a 2D spline fit to an (unlinked) portion of the xy grid that was posted In MacGregor 2015, then rendered for highlights and shadows. It is not so easy to reproduce the quantitative underpinnings of this diagram from the direct radargram tracings -- too bad because we also need two other surfaces, the 46.5 kyr and an even older isochron but still well above the Eemian.

The Holocene bottom is still a rather bumpy surface.  It is not quite a map of Greenland-wide basal deformations (cryptic plus overt) because surface tilt and bedrock draping have not been subtracted off. While these are fairly minor corrections for this shallow depth due to attenuation of layer deformation with younger age, the corrected 46.5 kyr isochronal surface would be a whole lot bumpier and so more favorable for low error. 

Imagining now that we had these three corrected intermediate isochronal  surfaces, it would then be possible from subtracting them to measure attenuation with height. That goes beyond just mapping basal deformations to estimating how the ice above is responding, which in turn is a measure of rheology with depth and so internal temperature.

Since we have no real idea how long or forcefully the basal forces have been acting, this is going to end up only as a constraint bounded by the three extreme scenarios for basal deformations (early vs continuous vs recent).

There's no known relationship between basal till and basal deformation, so even carrying out this program would leave us in the dark in terms of till depth, hydration state, and deformability (which will greatly exceed that of ice assumed frozen to bedrock).

On the methods side, recall Isaac Newton originally worked with gridded data of increasingly finer mesh but eventually took things to the limit, the continuous functions of calculus because of their many operational conveniences. Note the fancier interpolation methods discussed above also leave the pixel world, get their work done, and return. The same is going on with the SVS graphic above, internally with the Landsat motion-processing scheme of Fahnestock 2016 and the bedrock DEM kriging of Bamber 2013, and the spherical harmonic expansions of GRACE (which I've argued elsewhere are better done with cylindrical Bessel functions because the GrIS is an elongated dome).

These posts may have left the impression of Greenland in tumult. Actually a lot of the subsurface is exceedingly boring (eg inland of Hammond glacier, or the east-central ridge) with nothing whatsoever going on in the layers. I'll post in a bit some very plain flight segments as well as some pure bedrock drapes from East Greenland to provide some balance.

However, even if half or more of Greenland has a classical ice-bedrock interaction, the problem here is that ice-till interaction now appears very widespread from seismic even in unlikely places like Neem and basal deformations though limited happen to be at their extreme in regions such as Petermann and Zachariae/Negis already presenting the worst velocity risk for providing that first meter of sea level rise.

It's hard to make a case for the rosy scenario when so many new developments, here and indeed throughout the cryosphere, point in the other direction.

[A-Team]

Neem is finished in the sense that the dome and drilling apparatus have moved on to the new EGRIP site in northeastern Greenland but work continues on multiple analyses of its frozen ice core and the environs.

While site selection did not prove optimal for undisturbed Eemian ice, that had the ironic benefit of rich ice core complexity that can still be unraveled with additional work. The capped drill hole has been filled with antifreeze and that  a 0.01 °C precision temperature profile will be re-read after things equilibrate (2-3 years) along with borehole inclination, azimuth, diameter and pressure (ie deformation: surface velocity is ~5 m/yr).

However for brittle ice in the Holocene, the new Renland core through stagnant ice in the southeast will prove more suitable. The new 2550 m core on the fast moving ice stream NEGIS has been deliberately situated on conformally draped ice that seems to lack basal deformation (images below), focusing on ice rheology, deformation, and dynamics of basal sliding and basal water processes. The radargram 1990525_01_10 shows the drill site.

EGRIP will be the first ice core through fast-flowing ice (65 m/yr) and so a drilling technical challenge as during the three years it takes to drill to bedrock the 13 cm diameter borehole wil move 180 meters to the northeast.
hole site initially at 75.6268N 35.9915.

http://recap.nbi.ku.dk/forsidebokse/fieldplan-2015-recap-and-egrip/Renland-EGRIP2015FieldPlan.pdf
http://www.the-cryosphere.net/8/1275/2014/ see Fig.3

Previous posts have discussed possible errors in the chronological dating, whether nitrogen isotopes make for a better temperature proxy, that drilling stopped in icy till 50-100 m above bedrock according to seismic survey, and whether observed layer folding was only to be expected on the shoulder of a large-scale basal deformation.

AGU2015 had an additional two sessions of interest. The first is very much on-topic, suggesting that impurity levels (which presumably dropped sharply as the Eemian ended) resulted in  a boundary layer deformation effect as the ice crystal fabric changes.

Deformation Studies of NEEM, Greenland Basal Folded Ice
K Keegan D Dahl-Jensen M Montagnat I Weikusat
(no poster at AAGU, no paper has appeared yet)

Deep Greenland ice cores and airborne radio echo sounding (RES) images have recently revealed that basal ice flow of the Greenland Ice Sheet is very unstable. In many locations, a basal layer of disturbed ice is observed. At the NEEM, Greenland site this folding occurs at the boundary between the Eemian and glacial ice regimes, indicating that differences in physical properties of the ice play a role in the disturbance.

Past work in metallurgy and ice suggests that impurity content controls grain evolution and therefore deformation. We hypothesize that the differences in ice flow seen deep in the NEEM ice core are controlled by differences in the impurity content of the ice layers. Here we present results of fabric, grain size, impurity content, and deformation studies from samples above and below this unstable boundary in the ice sheet.

In an earlier EGU2915 abstract, the same authors developed the metallurgy analogy further, citing Burke 1957, Hammer 1978, Langway 1988, and Dahl-Jensen 1997. That meagre bibliographic detail can be unraveled via the miracle of Google Scholar:

JE Burke 1957 (DOI: 10.1111/j.1151-2916.1957.tb12580.x): the role of grain boundaries in sintering (grain boundaries act either as sinks or as diffusion paths for lattice vacancies).

CU Hammer 1978: Dating of Greenland ice cores by flow models, isotopes, volcanic debris, and continental dust (J Glaciology v20 pp.3-26 offline)

CC Langway 1988: Crystal size and orientation patterns in the Wisconsin-age ice from Dye 3, Greenland. On-topic but a bit dated; free scan online at http://tinyurl.com/jyt9uhc

D Dahl-Jensen 1997 (DOI: 10.1029/97JC0016l7): a study of the Grip core at Summit where an effect is attributed to 'diffusion of water molecules along crystal boundaries in the recrystallizing ice matrix'.

I would have cited this one instead and its modern forward cites:

Why ice-age ice is sometimes “soft”
WSB Paterson 1991 (doi:10.1016/0165-232X(91)90058-O  cited 121 times):

Data on the mechanical properties, texture, fabric, and impurity content of ice deposited during the last glaciation are reviewed. The conclusions are: (1) Chloride and possibly sulphate ions, in concentrations high relative to those in Holocene ice, impede grain-boundary migration and grain growth so that the crystals remain small. (2) Such ice, in shear parallel to the ice-sheet bed, develops a strong, near-vertical, single-maximum fabric. (3) This fabric favours further deformation and this, in turn, further strengthens the fabric and keeps the crystals small. (4) This is why the strain rate in ice-age ice, in simple shear, is some 2.5 times that in Holocene ice at the same stress and temperature. (5) Ice-age ice under other stress systems, such as ice in roughly the upper 60% of the ice thickness, in bedrock hollows, at a stationary ice divide, in ice streams and in ice shelves, will not have enhanced flow. (6) An anisotropic flow relation must be used for detailed modelling of polar ice sheets.

Fabric along the NEEM ice core, Greenland, and its comparison with GRIP and NGRIP ice cores
M Montagna et al
http://www.diss.fu-berlin.de/docs/receive/FUDOCS_document_000000021203

abric (distribution of crystallographic orienta- tions) along the full NEEM ice core, Greenland was mea- sured in the field by an automatic ice texture analyzer every 10 m, from 33 m down to 2461 m depth. The fabric evolves from a slightly anisotropic fabric at the top, toward a strong single maximum at about 2300 m, which is typical of a de- formation pattern mostly driven by uniaxial compression and simple shearing. A sharp increase in the fabric strengthening rate is observed at the Holocene to Wisconsin (HW) climatic transition. From a simple model we estimate that this depth is located at a transition from a state dominated by vertical compression to a state dominated by vertical shear...

The second undiscussed AGU2015 abstract will also be worthy of detailed coverage when it appears:

Fire impacts on the cryosphere
N Kehrwald P Zennaro M Skiles Barbante
(no poster at AAGU, no paper has appeared yet)

Continental-scale smog clouds and massive boreal smoke plumes deposit dark particles on glaciers, darkening their surfaces and altering surface albedo. These atmospheric brown clouds are primarily comprised of both fossil fuel and biomass burning combustion products. Here, we examine the biomass burning contribution to aerosols trapped in the cryosphere using levoglucosan (anhydro glucopyranose) in ice cores. Levoglucosan is only produced by cellulose combustion [though this is mildly disputed https://en.wikipedia.org/wiki/Levoglucosan]

Black carbon reflects both fossil fuel and biomass burning; their trends differ over the industrial era at NEEM, EPICA, Kilimanjaro, Tanzania and Muztag Ata. Wildfire combustion products were a major component of dark aerosols deposited on the Greenland ice sheet during the 2012 melt event. Ice core records show climate remains the primary control over fire activity despite  increased modern biomass burning and brown cloud pollution.