EGU General Assembly 2020
© Author(s) 2022. This work is distributed under
the Creative Commons Attribution 4.0 License.

The uncertainty in Antarctic sea-level rise projections due to ice dynamics

Javier Blasco1,2, Ilaria Tabone1,2, Daniel Moreno1,2, Jorge Alvarez-Solas1,2, Alexander Robinson1,2,3, and Marisa Montoya1,2
Javier Blasco et al.
  • 1Complutense University of Madrid, Madrid, Spain (
  • 2Geosciences Institute CSIC-UCM, Madrid, Spain
  • 3Potsdam Institute for Climate Impact Research, Potsdam, Germany

Projections of the Antarctic Ice Sheet (AIS) contribution to future global sea-level rise are highly uncertain, partly due to the potential threat of a collapse of the marine sectors of the AIS. However, whether the inherent instability of such sectors is already underway or is still far away from being triggered remains elusive. One reason for ambiguity in results relies on the uncertainty of basal conditions. Whereas high basal friction can potentially prevent a collapse of the marine zones of the AIS, low basal friction can promote such a process. In addition, future sea-level projections from the AIS are generally run from an equilibrated present-day (PD) state tuned to observational data. However, this procedure neglects the thermal memory of the ice sheet. Furthermore, there is no apparent reason for ruling out that the PD may be subject to a natural drift since the onset of the last deglaciation (~20 kyr BP). Here we study the uncertainty in sea-level projections by investigating the response of the AIS to different RCP scenarios for four different basal-dragging laws. For this purpose we use a three-dimensional ice-sheet-shelf model that is spun up from a deglaciation. Model parameters of all friction laws have been optimized to simulate a realistic PD. In addition, we study the response of the AIS to a sudden CO2 drop to investigate the potential irreversibility of the ice sheet depending on the RCP scenario and friction law.

How to cite: Blasco, J., Tabone, I., Moreno, D., Alvarez-Solas, J., Robinson, A., and Montoya, M.: The uncertainty in Antarctic sea-level rise projections due to ice dynamics, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-14352,, 2020.

Display materials

Display file

Comments on the display material

AC: Author Comment | CC: Community Comment | Report abuse

Display material version 3 – uploaded on 05 May 2020
Added equations in methodology
  • AC1: Comment on EGU2020-14352, Javier Blasco, 05 May 2020

    This is a response to CC3:

    Thank you for your suggestion, i find very interesting the idea of applying a high-pass filter to the bedrock topography. 

    I also tested a spatial constant friction coefficient and found that ice streams, specially those that are connected between the EAIS and the Ronne shelf, were not fully captured. Applying a depth dependent parameterisation allowed for a more accurate representation of ice streams. 

    But i also found that velocities in the WAIS were larger compared to observations. This was somehow expected as friction in the marine parts will be low, enhancing ice flow. Maybe applying a filter as you suggested could help to improve results.

  • CC1: more on discussion about basal drag, Lev Tarasov, 05 May 2020

    I've struggled for literally decades with how to captures all the dynamical uncertainties on basal drag. Further to Michael's comments, David Pollard over a decade ago posited that unloaded beds that are submarine are sedimentary and therefore soft for defining hard/soft beds for presently subglacial surfaces. This seems a reasonable starting point. As to roughness, high bedrock points resolved in a DEM will act as pinning points and therefore are best represented as power law 3 (not 1, cf original Weertman paper on basal sliding, or just consider that pinning points will act as enhanced ice deformation points dominating power law 1 regelation). I suspect this is why some past studies have found some presumable soft-bedded ice streams best represented by @ power law 3. The whole issue of  representing subgrid iceflow/sliding still needs more attention in the community.

  • CC2: Comment on EGU2020-14352, Michael Wolovick, 05 May 2020

    Both of those effects make a lot of sense.  The interior of WAIS has an extremely deep bed but it's flowing pretty slowly, so a depth-dependent parameterization would make it move too fast.  And the ice streams feeding from EAIS into the Filchner-Ronne Ice Shelf are all controlled by bedrock troughs, so adding in a depth-dependence should help represent them better.  

    I'm glad that you like the idea of using a high-pass filtered bed topography.  It's not a perfect solution though.  Choosing the right cutoff wavelength could be a whole complicated can of worms.  The argument about local highs and local lows applies to small-scale bedrock protrusions in the middle of ice streams just as well as it applies to the large-scale troughs themselves.  However, since you're running a continental-scale model you're more interested in getting the large-scale structure of ice streams correct rather than representing individual sticky spots, so that would argue for using a larger cutoff wavelength.  You would want a wavelength larger than the width of the important troughs, so that all of the troughs show up in the high-pass bed. Something like 100-150 km, probably.  But even then, it's not so simple.  Areas like the broad sedimentary basin in the upstream onset region of Recovery Glacier won't be captured by that method, for instance (although the trough in the downstream region of that glacier will be).  Still, no method is perfect, and I think that it's valuable that you're testing the influence that different sliding parameterizations have on sea level projections. 

Display material version 2 – uploaded on 04 May 2020, no comments
Added some titles
Display material version 1 – uploaded on 03 May 2020
  • CC1: Comment on EGU2020-14352, Nicolas Jourdain, 04 May 2020

    Hi Javier,
    Thank you for your presentation. How do you calculate surface mass balance and ice-shelf basal melting until year 5000 in the 4 RCP scenarios?

    • AC1: Reply to CC1, Javier Blasco, 04 May 2020

      Hello Nicolas,

      SMB is computed through the PDD scheme. Atmospheric temperatures and precipitation at PD are obtained from RACMO2.3, forced by the ERA-Interim reanalysis data (van Wessem et al., 2014). PD basal-melting rates are obtained from Rignot et al., 2013. 

      For the future scenarios i apply a homogeneous warming over the whole Antarctic domain, as in Golledge et al., 2015. In the third slide you can see the forcing of the 4 RCP scenarios for the atmospheric temperatures. Precipitations are scaled through the Clausius-Clapeyron-relationship. Oceanic temperatures evolve in phase with atmospheric temperatures, scaled by one fourth. Oceanic temperature anomalies are converted into basal-melting rates through a heat flux coefficient and added to the PD basal-melting rates. In this study i use a value of 10 m yr^-1 K^-1 for the heat-flux coefficient, meaning that per every degree that the ocean warms, basal melting increases by 10 (Rignot and Jacobs, 2002).

  • CC2: Comment on EGU2020-14352, Michael Wolovick, 04 May 2020

    Thank you for a very good presentation.  To clarify, when you are testing different drag laws, do you use spatially constant coefficients for each drag law?


    • AC2: Reply to CC2, Javier Blasco, 04 May 2020

      Hello Michael, thank you for your comment.

      I did not specify in the presentation that the friction coefficient cb is not set spatially constant, but diminishes exponentially with bedrock depth.  This parameterisation intends to capture the fact by which sliding is more likely to happen in topographic lows, specially at the marine based zones, where bed will be softer than in high rocky regions.

      • CC3: Reply to AC2, Michael Wolovick, 04 May 2020

        That's an interesting method that I haven't seen before.  I understand the reasoning that drag ought to be lower in the troughs, but the problem is that this reasoning implies that drag ought to be connected to local deviations in the bedrock topography more than it is connected to the absolute value of topography.  So, for example, if you have a trough bottom that is 100m above sea level but 300m below its surroundings, that trough bottom ought to have lower drag than a bedrock high point which is 500m below sea level yet 200m above its surroundings.  Following this logic through, it would make more sense to connect drag to a high-pass filtered version of the bedrock topography than to the absolute value of bedrock topography.  You could also make an argument that perhaps the form of the drag law should be connected to this high-pass filtered bed topography as well: local highs are more likely to be exposed bedrock, and thus closer to a linear sliding law, while local lows are more likely to be saturated till, and thus closer to a plastic sliding law.  

        Do you have any guesses about how the decision to connect drag to absolute basal topography might effect results, as opposed to relative topography (or no spatial variability, or some other parameterization)?  Are there particular areas of Antarctica where this assumption might influence the results?