In view of yet another conflicting interpretation of basal ice upheavals in northern Greenland (see first ref in #342), especially in the Petermann and Zachariae areas, I'm looking to revisit the whole issue during the slack season. A separate forum for this and related phenomenon would require moving many dozens of earlier posts and graphics.
While fascinating and unexpected, these basal structures tell us something important about melt conditions at the end Eemian. The response of the ice sheet at that time to warming is our most important paleo clue to Greenland's response to present-day anthropogenic warming and the contribution of melting ice to global sea level rise and Atlantic meridional current disruption.
The merits of explanatory proposals can be compared using the checklist below:
What's the theory predict?If competing theories can't be distinguished by experimental outcomes, then the relative validity of proposed mechanisms can't be assessed. Conversely if proponents cannot agree to observational outcomes that would invalidate their hypotheses, they have departed planet science for la-la land.
Here we don't get to nuances -- none of the competing upheaval papers explicitly commit to concrete predictions. It's not clear what these theories expect from drilling down to bedrock even through an extreme upheaval. Some will modify their proposal after the fact to fit whatever temperature, ice fabric, impurity, meltwater, conductivity, meteoric vs accreted ice, isotopic profiles emerge. A theory that flexible isn't in the science realm, it's merely curve-fitting.
Experimental ice core data from an upheaval is not in view. The only ongoing drilling to bedrock is at NEGIS. While site coordinates have only been disclosed in vague terms, we can be sure that extensive sled and air radar have shown regular stratification to bedrock -- there won't be tilted layers much less an upheaval at the drill site. That was certainly the choice made at the Renland ice cap.
Steam cores are quite feasible at accessible locations at the relevant depths. While these don't preserve the full ice archive, wire logging devices can capture many properties of the ice. Logged boreholes of Luethi at Swiss Camp and Hubbard at Store Glacier illustrate modern opportunities. It is really on these measurables that theories need to take a stand.
Sled or drone radar surveys are easier still. Radar to date has prioritized bedrock elevation and ice thickness; grid cells appropriate to that are too coarse relative to the intrinsic scale of upheavals and coherent systems of them. It is very difficult to reconstruct a 3D object from a couple of intersecting transects randomly oriented relative to key structural attributes and skew to the local ice velocity field. What's needed are radial grids in polar coordinates centered on individual features and grids of flowlines and isotachs.
Only one paper to date has attempted to reconstruct the overall regional 3D structure of an integrated upheaval system and that only for Petermann. Some papers envision recumbent and overturned folds where others see sheath folds; better radar grids can distinguish these (as can held-back or post-paper radar transects not used in model building).
Eqip is the most convenient site logistically, though it is not clear whether it (or any other upheaval) is 'representative' of all. The extensive upheaval there is 94 km from the Illulisat airport with a ship dock at the calving front hotel. Swiss Camp and Store research sites are close by. Crevasse and snow bridge risk is minimal inland within the nearly flat and slow moving Eqip ablation zone. Sled radars have used nearby, as have steam drills.
Occam's Razor: does it still cut it?This 14th century principle asserts a simplest most boring explanation is most often right. This translates here to every upheaval in all of Greenland and Antarctica having the same underlying physical mechanism. Such parsimony is perhaps unattractive given very different bedrock, geothermal gradient, basal stress, temperature, viscosity, ice thickness, ice sheet histories, and surface velocity fields where upheavals are observed. So perhaps two physical mechanisms are operative but certainly no more than three. If so, a classification scheme is needed and predictions subordinated to it. Multiple contributing mechanisms at individual features is incompatible with Occam's principle and should be considered only as a last resort.
Absence of upheavals is just as informative as presence.What accounts for the distribution of upheavals on the Greenland ice sheet? They are overwhelmingly concentrated today north of Eqip with many but not all in regions of moderate thickness and higher surface velocity. Yet for every upheaval, the same radar flight line can show many 'unoccupied' but seemingly indistinguishable sites. Any mechanistic explanation must give equal weight to predicting absence.
Are upheaval radar reflectors distorted isochrons?In Greenland, upheavals are always bounded from above by the ubiquitous 'three sister" isochrons, accurately dated at NEEM to the middle of the last ice age. The nearly as ubiquitous 'older brothers' dating to 91 kyr sometimes bound the upheaval but more often are caught up in its contortions. Below that there are few consistent markers. Numerous mechanisms attenuating radar returns arise from something other than these presumptive massive volcanic depositions.
Upheaval areas in radargrams can be delimited from above by dark lines but these are often blurrier than any isochron ever observed in calmer regions, There may also be large dark blocks that have no counterpart in isochrons. These dark areas can have elaborate substructure that also require explanation. In these regions, processing radargrams
at native resolution to optimally enhanced contrast is imperative, lest important clues be discarded. It's very poor practice to theorize without considering experimentally revealed complexity. The first image below compares a figure from a peer-reviewed geophysics at author-submitted resolution to actual upheaval data.
In their comprehensive mapping study of Greenland stratifications, MacGregor 2015 take a cautious perspective:
In numerous locations within Greenland’s interior, the deepest radiostratigraphy does not drape smoothly over the observed along-track bed topography. Instead, it is deflected upward across horizontal scales of several kilometers or more, distorting the depth-age relationship of the ice column by up to several hundred meters. These features are most prominent in the onset region of Petermann Glacier in northwestern Greenland (Animation S2).
Bell 2014 mapped the larger instances of these features. The precise cause of this disruption/deflection is unknown, but it is unlikely to be due to flow over undetected highs in the bed topography as previously hypothesized in Legarsky 1998, because extensive radar surveying of the GrIS fails to reveal these postulated highs.
This disruption may be due to buckling associated with spatially or temporally varying basal friction [Hindmarsh 2006 Wolovick 2014], basal freeze-on as inferred for East Antarctica [Bell 2011], or rheological variability within the basal shear layer [NEEM 2013].
Given the variability in magnitude of these radiostratigraphic disruptions, a combination of the above phenomena is also possible, or another as-of-yet unknown process. At least for Petermann Glacier, a plausible scenario is that recently described by Bell 2014 and modeled by Wolovick 2014: basal freeze-on uplifts and overturns the overlying strata, generating complex units that include both accreted and distorted meteoric ice.
These features can contain unusually bright, indistinct, or overturned reflections that do not resemble the monotonous radiostratigraphy of the overlying ice sheet. Unusually bright reflections may be due to a large fabric contrast between meteoric and accreted ice, as hypothesized to occur over Lake Vostok in MacGregor 2009a.
It is possible that some of these coherent reflections and diffuse reflectivity are not isochronal, although they likely remain useful indicators of past flow [NEEM 2013].
Are upheavals oriented along flow lines?That's not so easily determined because flight lines rarely followed surface flow lines. The few orthogonal grids take a skew angle with respect to flow and, at Petermann, the rare flight lines that did follow flow were unfortunately from years with unsatisfactory radars. The grids are not tight enough to pick up intervening upheavals; at best 2-3 cross sections are available for a body of unknown shape and dimensions, eg there's no reason to suppose the peak height happened to be transected.
Many of those most dramatic upheavals occur in regions of deep and essentially stagnant ice. It simply isn't credible to draw a causal association with flow (or associated frictional basal heat) at meter-per-year sites as there's hardly been any movement since the Eemian, not to mention that a few km over with slightly faster flow there are no upheavals. This leads to the equally unpalatable choices of either disregarding Occam's Razor or the extended upheavals that seem coherently oriented with flow for hundreds of km.
Is the inventory of upheavals comprehensive?No thresholding criteria were provided in the original mapping article [Bell 2014], no database of relevant Cresis accessions was begun, no kml path file exists. It appears that only the larger dramatic features were mapped based on subjective criteria. And not all of these were found because radar tracks are fairly sparse.
All the work on the Cresis image portfolio thus has to be repeated from scratch. This means the actual distribution of upheavals, perhaps stratified by upward magnitude and horizontal extent, has not been determined. This seriously undermines all manner of correlations with other basic attributes of the Greenland ice sheet.
Most troubling is the NEEM drill hole itself: overturned stratigraphy was carefully documented in the core but nothing resembling an upheaval can be seen in radar despite very intense surveys over the borehole and vicinity. This suggests the ability of radar to detect upheavals at depth is not adequate to locate them, not just at NEEM but by implication over much of north-central Greenland.
Is this an exercise in structural geology?Yes and no. Stress-strain physics of stratigraphic deformation and flow of sedimentary deposits have been studied for centuries and have clear counterparts in ice sheets. Salt beds and lava flows have temperature/viscosity relationships reminiscent of ice. There are counterparts as well to till hydration state, melt phase change, and crystallization in plate tectonics but with diminishing relevance to glaciers, melt lakes, moulins and basal drainage channels.
“When I use a word,” Humpty Dumpty said in rather a scornful tone, “it means just what I choose it to mean — neither more nor less.”
“The question is,” said Alice, “whether you can make words mean so many different things.” (Through the Looking Glass, Chapter 6)
Anticline and syncline are terms for compressional folds borrowed from structural geology often inappropriately applied to Greenland ice upheavals. It's impossible to recognize these from a single radar flight lines because the requirement for a longitudinal fold axial plane is inherently two dimensional. If this axis has not been observed, the fold type cannot be distinguished from a dome or basin (technical terms for folds with non-axial symmetries).
An upward deformation of stratigraphic isochrons might be an anticline but such an inference is premature without supporting evidence from multiple flight lines -- these are so sparse that it can't be determined when one feature stops and another has started. It's really a stretch to apply these terms to basal freeze-up deformations: mechanistically more related to salt diapirs, this concept is only dimly related to compressional waves of alternating synclines and anticlines.
In structural geology, overturned and recumbent folds are distinguished from sheath folds. That can sometimes be done from a single transect should it happen to pass through the signature concentric 'eye' of a sheath fold. The second image shows that the folds labelled 'overturns' in the figure above are part of a sheath fold. The subject of folds is very complex overall and highly developed within structural geology.
Are these fossil features or still active processes?In one scenario, upheaval processes began shortly after the Eemian ended, continued for a few tens of thousands of years, then ceased as an active process, no longer initiating new features nor building out older ones, which however persisted to the present day.
An alternative scenario posits that forces responsible for upheavals are fundamental basal processes not specifically tied to one-off historic ice warming during the Eemian that continue to the present. Processes mentioned here include freeze-on of meltwater at the ice/bedrock interface, slip-stick motion of the glacier and associated heat from friction, hotspots in the geothermal gradient, basal strain from ice sheet motion, temperate ice at the pressure melting point and associated lower viscosities, and so on.
North-central Greenland did not fully melt out during the Eemian (as mapped by MacGregor 2015 via a stratigraphic age-depth function) but what didn't melt came close. In particular, the upheaval areas of Petermann and Zachariae are not underlain with Eemian ice today even a hundred km and more inland from the grounding line.
In the first scenario, this created a persistent body of temperate ice, perhaps with pockets of meltwater or hydrated till underneath, that was subsequently buried under thousands of meters of subsequent ice age snowfall. Slow equilibration with cold from above meant the temperature and buoyancy anomalies played out slowly, perhaps along the lines of 'thermal-viscous collapse' considered in Colgan 2015 (
http://onlinelibrary.wiley.com/doi/10.1002/2015EF000301/full).
The second scenario predicts ongoing development of upheaval features and perhaps observable effects from the Holocene, its Thermal Optimum, or even contemporary anthropogenic warming. However the time scale needed for an observable effect isn't clear:
Even if the earliest radar flight line of 1993 had been fortuitously re-flown in the latest 2015, it isn't plausible than any upheaval change could be seen over 12 years, first because glacier processes generally precede at glacial paces, second the radars would be of very different design and third because a 100 m mismatch in flight lines would invalidate the comparison.
Eqip has favorable repeat radar coverage in different years but the east-west flight lines are displaced to 61 m and north-south to 29 m (3rd image). Note the bedrock drainage map resolution is delusional because kriging guesswork has filled in the blanks left by sparse radar coverage in active topography -- there is no other data other than what is seen along radar tracks (airGrav has been deployed in fjords, ice shelves and ice streams).
Is ice predictably deformed over an upheaval?It's fair to say that stratigraphic layers above basal upheavals are conformally uplifted with that effect tapering off in Holocene ice, sometimes to the point of imperceptibility. This requires reconciliation with the near-incompressibility of ice, its viscosity, depth and status of firn, and conservation of mass because the ice surface is not notably deformed.
There is no known method by which the presence of upheavals can be systemically detected by any combination of measurable surface properties (such as ultra high resolution DEMs, surface velocity, and surface slope). We only wish there were because then the upheavals could be comprehensively mapped from continuous 2D data, with sparse radar tracks only employed as a cross-check and for details.
At Petermann, the case can be made that the upheavals themselves are anisotropically distributed with respect to surface velocity flowlines (though it's not so clear how to make that point objectively). This has implications for anisotropy of younger layer deformation which so far have not been pursued.
At the Eqip upheaval on a south-to-north track of 2008, isochrons do not notably conform with either bedrock profile nor the upheaval (4th image). This suggests (if not a unique situation) a basis for upheaval categorization.