Updated August 2019.
This document gives an overview of the DMC:ANTARCTICA data-model comparison project. It also describes the online DMC:ANTARCTICA interface and and provides some details about calculating cosmogenic-nuclide concentrations from ice-sheet models.
1. The DMC:ANTARCTICA project
DMC:ANTARCTICA is a NSF-funded data-model comparison project whose objective is to systematically compare all known cosmogenic-nuclide data from Antarctica with predictions from numerical ice-sheet models that simulate the evolution of the ice sheet over long time periods.
Such models are important for establishing the ice-sheet's climate sensitivity and for forecasting its future behavior and contribution to sea level, however, they have limited meaning unless it can be shown that they agree with observations of past ice-sheet change. Although some simulations have, in fact, been calibrated with geologic data, these primarily cover the period between the last glacial maximum and the present. Longer simulations, spanning important time periods when the climate was as warm or warmer than present, have not been thoroughly tested.
As of August 2019, the set of Antarctic cosmogenic nuclide measurements comprised 4361 measurements on 2551 samples, which have been compiled in the ICE-D:ANTARCTICA database. While, in most cases, these data cannot be simply inverted to generate unique ice-thickness chronologies, they can be used to distinguish incorrect chronologies predicted from ice-sheet models from plausible ones. The basic idea is that, for a given rock in Antarctica, the history of exposure and ice-cover simulated by a model predicts certain concentrations of cosmogenic nuclides (e.g. He-3, C-14, Al-26, Be-10, Ne-21) at the present-day that can be directly compared to concentrations that have actually been measured in the rock. Although such comparisons have been done previously at individual sites and for small number of samples (e.g. Mukhopadhyay et al., 2012), the objective of this project is to evaluate ice-sheet models using the entire Antarctic dataset.
This is currently implemented in the DMC:ANTARCTICA web interface, which serves data-model comparisons for individual site-model pairs. Ultimately, the goal is to perform synoptic evaluation of model performance using all available Antarctic cosmogenic-nuclide measurements. That is coming.
2. Data and models
2.1. Cosmogenic-nuclide data
DMC:ANTARCTICA leverages the ICE-D:ANTARCTICA database, which contains all known published (and many unpublished) cosmogenic-nuclide data from Antarctica. Around 80% of samples in the database that have associated analyses are glacial erratics and around 20% are samples of bedrock surfaces. The maps below show that most regions where outcrops exist are well represented by each population.
Minimum and maximum circle sizes represent 1 and 88 samples, respectively.
The next set of maps show the geographic distribution by nuclide. Unsurprisingly, Be-10 is the most prevalent.
While most samples only have a single analysis of a single nuclide, multiple nuclides (e.g. Be-10 and Al-26) have been measured on some samples. These samples are important because paired nuclide data can provide information about past exposure, ice-cover, and erosion.
The temporal distribution of the database is shown in the histograms below (note logarithmic axis). The main point here is that different nuclides provide information about exposure and ice-cover on a variety of timescales extending back to the Miocene. The implication, therefore, is that these data can be used to evaluate numerical ice-sheet models that run over those different timescales.
2.2. Ice-sheet models
The numerical ice-sheet models available for comparison to cosmogenic-nuclide data are listed at the bottom of this page. Most cover the period between the last ice age and the present, but some span large portions of the Pleistocene or Pliocene. These models were initially run for different reasons; at different spatial resolutions; and use different physics, boundary conditions, and parameterizations. Nevertheless, they all predict ice-thickness changes over Antarctica, which, in turn, predict certain present-day cosmogenic-nuclide concentrations that can be compared to observations.
3. Calculating nuclide concentrations from ice-sheet models
3.1. Overview of calculations
For a given site and ice-sheet model of interest, the modeled ice-thickness history is extracted from the model at the site via bi-linear interpolation of the four surrounding model grid points. Here is an example from the model of Pollard & DeConto (2009) for Mt. Hope in the central Transantarctic Mtns.
The next step is to generate synthetic samples of bedrock and glacial erratics spanning a range of altitudes at the site. The history of exposure and ice cover experienced by each synthetic sample is determined and used to calculate the present-day concentrations of various cosmogenic nuclides.
3.4. Cosmogenic-nuclide production
Concentrations are determined for the following nuclides: He-3 (pyroxene or olivine), He-3 (quartz), Be-10 (quartz), C-14 (quartz), Ne-21 (quartz), Al-26 (quartz). Concentrations, as well as exposure ages, are calculated using the Matlab/Octave code behind the version 3 online exposure age calculator, originally described by Balco et al. (2008) and subsequently updated. Documentation for that code is located here.
Scaling methods and calibration data: Two methods are available in DMC for estimating cosmogenic-nuclide production rates at sites of interest: (i) the method of Stone et al. (2000), which cast the equations of Lal (1991) in terms of atmospheric pressure rather than elevation, and (ii) the recent and much more complicated "LSDn" method of Lifton et al. (2014; 2016), which is based on simulations of cosmic-ray fluxes in the atmosphere. The LSDn method appears to do a better job in Antarctica (further discussion can be found here and here) and is the default in DMC:ANTARCTICA. The sources of production-rate calibration data (and the method of calibration) vary for different nuclide-mineral pairs. Refer to documentation of the version 3 online exposure age calculator here for details.
Production by muons: Production by cosmic-ray muons is simulated for times when a sample is exposed at the surface, but not if the sample is covered by any thickness of ice. Although this is obviously an over simplification, for the majority of samples, it doesn't make a big difference. Future versions of DMC will deal with subglacial production by muons properly.
Changes in atmospheric depth: Over glacial-interglacial cycles, samples in Antarctica experience changes in atmospheric depth and, in turn, changes in production rates due to (i) glacial isostatic changes in the elevation of the Earth surface and (ii) atmospheric pressure changes due to changes in atmospheric temperature, sea level, and the distribution atmospheric mass. At the moment, neither effect is simulated. Realistically, simulating (ii) will probably be a long time coming because ice-sheet models are generally not coupled to atmospheric general circulation models and therefore this information is not part of standard ice-sheet model output. However, isostatic effects on production rates can and should be accounted for because ice-sheet models do typically simulate elevation changes of the solid earth surface. This is coming.
Topographic shielding: Shielding of samples from a portion of the cosmic-ray flux by surrounding topography could be accounted for using a high-resolution DEM of Antarctica. This is not done at present.
3.2. Predictions for bedrock and glacial erratics
Nuclide concentrations are simulated in synthetic samples of both bedrock surfaces and glacial erratics. All samples are assumed to have zero cosmogenic nuclides at the beginning of the model simulation. As a result, running DMC with, for example, a site where high concentrations of stable or long-live nuclides have been measured, along with a relatively short-duration model (e.g. LGM-present) will typically produce large data-model discrepancies in terms of concentrations and exposure ages.
Simulation of glacial erratics: DMC simulates periodic deposition of erratics whenever the ice sheet at a site is thinning. This is because most glacial erratics in Antarctica are ablation deposits, meaning that they were transported englacially from upstream erosion areas to ablation areas on the ice surface (where they were initially exposed to the cosmic-ray flux), and subsequently deposited on mountain flanks by ice thinning. Periodic deposition of erratics is simulated for all sites, regardless of whether erratics are actually known to be present. This is akin to asking, if erratics are present at the site, what does the model predict their nuclide concentrations to be?
Antarctic exposure dating studies commonly measure glacial deposits with a wide range of apparent exposure ages at a given site. This is usually attributed to some combination of (i) recycling of rock from older deposits, (ii) burial of older deposits by frozen-based ice and subsequent re-exposure, or (iii) deposition of rock previously exposed as bedrock. DMC assumes scenario (ii) for all simulated erratics. Once deposited, erratics do not change elevation, roll over, etc. They just stay put.
Models that simulate many glacial-interglacial cycles can therefore produce erratics at different times but at similar elevations. The effect of this is shown below for the 5 Myr simulation of Pollard & DeConto (2009) at Mt. Hope. In contrast to bedrock samples, in which nuclide concentrations always increase monotonically with altitude, glacial erratics at similar elevations can have very different present-day nuclide concentrations. An exception to this is for the nuclide C-14 (shown in orange), which decays rapidly with a 5730 year half life and therefore has no memory of the exposure and ice-cover history prior to around 30 kyr ago.
Different colors represent different nuclide-mineral pairs as follows: orange = C-14 (quartz); red = Be-10 (quartz); blue = Al-26 (quartz); neon = Ne-21 (quartz); gray = He-3 (pyroxene/olivine).
Erosion, snow cover, till cover: These processes, which have the effect of reducing nuclide concentrations relative to what they would otherwise be, are not simulated.
3.3. Elevation relative to what datum?
Whether a sample is predicted to be exposed or ice-covered by a model depends on the elevation of the sample relative to the elevation of the ice-sheet surface. Although this sounds straightforward, there are actually two vertical datums to pick from for doing this comparison: mean sea level, also known as the geoid, and the present-day ice sheet surface.
Consider an erratic located 200 m above the present ice surface, which in turn, is located 800 m above present sea level. Let's say the erratic was deposited during the LGM when the ice-sheet locally was at a highstand. The erratic therefore records 200 m of total thinning during the last deglaciation. Now also consider an ice-sheet model which correctly simulates 200 m of thinning at this site since the LGM, but, for whatever reason, thinks the present-day ice surface is not 800 m above sea level but 1000 m. If we evaluate the model using sea level as our datum, the model does terribly despite the fact that it correctly simulated the amount of ice-thickness change. On the other hand, if we use the present-day ice surface as our datum, the model does well.
The best choice of a vertical datum therefore depends on the question being asked. If we're most interested in whether the model accurately simulates the present-day ice sheet, then we should use sea level as our datum. If we're more interested in whether the model accurately simulates past ice-thickness change, then we should use the present ice surface.
DMC:ANTARCTICA uses the present-day ice surface as the datum to predict nuclide concentrations from ice-sheet model output. However, a problem arises when comparing these predicted concentrations to observed concentrations because, for most samples in the ICE-D:ANTARCTICA database, their height above the present-day ice surface is not known. One reason for this is that samples commonly are collected near sloping ice and thus determining what to use as the modern ice surface is not always straightforward. To deal with this, DMC:ANTARCTICA is currently providing data-model comparisons in terms of elevation relative to sea level, which is known for all samples. A better solution may be to retroactively estimate sample heights above the modern ice sheet using the recently released high-resolution Reference Elevation Map of Antarctica (Howat et al., 2019), but this has not been done yet.