page maintained by F. Pattyn @ ULB

MISMIP+: Marine Ice Sheet Model Intercomparison Project

Preparing a new set of grounding line benchmark experiments

Coordinators (ad interim)

Frank PATTYN, Laboratoire de Glaciologie, Université Libre de Bruxelles, Brussels, Belgium (email: fpattyn@ulb.ac.be)

Gael DURAND, LGGE, Grenoble, France (email: Gael.Durand@lgge.obs.ujf-grenoble.fr)

What are MISMIPs?

MISMIPs are Marine Ice Sheet Model Intercomparison Projects. It started several years ago under the impulse of Christian Schoof, who produced a semi-analytical solution of grounding line migration in ice sheet models based on boundary layer theory. This experiment, MISMIP, had the aim to evaluate marine ice sheet models along a flowline and to reproduce the analytical solutions due to Schoof.

A follow-up to MISMIP was ice2sea MISMIP3D applied to planview models, i.e. incorporating two horizontal dimensions. This model setup was very close to MISMIP and considered ice flow in one direction. However, the perturbation experiments result in a curved grounding line, hence leading to buttressing effects, which could not be evaluated within MISMIP. We used the full Stokes model Elmer/Ice as the benchmark to which other models will be tested, both in a diagnostic and in a prognostic fashion. Results of both MISMIP experiments were published in:

A need for a follow-up (MISMIP+)?

At the EGU Assembly in April 2014, a MISMIP splinter meeting was organized to gauge the interest in the need for a follow-up of MISMIP3d. What questions are still unanswered? Are ice sheet models ready to be coupled to ocean models, or is there a need for an intermediate evaluation (benchmark)? The results of the discussion is given below.

At the splinter meeting, the current context of marine ice sheet intercomparisons was presented by F. Pattyn, recalling the main results of the MISMIP (Pattyn et al. 2012) and MISMIP3d (Pattyn et al. 2013, Pattyn and Durand 2013) exercises, and pointing towards the CLiC-WAGOM initiative (http://www.climate-cryosphere.org/activities/targeted/wagom). The latter has the objective of coordinating international efforts that aim at building an ensemble of outlet glacier - ocean modeling systems and improve projections of West-Antarctic contribution to sea-level rise. A large consensus in the audience was that improvements in the development of ice-sheet modeling in general and of a better representation of outlet glacier dynamics in particular should continue simultaneously to the CLiC-WAGOM initiative. A new intercomparison exercise of marine ice sheet models should therefore be envisaged. The main question to be addressed by this exercise should probably be the ability of ice flow models to not diverge to much in their projection of SLR contribution on a century time scale. Therefore, in comparison with previous MISMIPs a more particular focus on interactions between outlet glacier dynamics and sub ice shelf melt rate should be envisaged.

During the discussions, they were some agreements between the attendants on the following points:

  1. A synthetic geometry should be preferred, probably inspired from Pine Island Glacier. The geometry proposed in Gudmundsson et al. 2012, probably with a smaller longitudinal extension could be a solution.
  2. Perturbations should be introduced through submarine melting. Oceanographers should be involved to ensure that the chosen parametrization is meaningful.
  3. Two sets of experiments should be envisaged: (i) running the steady state before running the melting perturbations, (ii) invert the basal drag to initialize the models to a given geometry, then run the melting experiments.
  4. Participants should probably provide results using different resolution to ensure convergence of their results.

A second planning meeting was held at the IGS Conference in Chamonix (25-30 May 2014). This meeting consisted of a broader range of ice sheet modellers, albeit that the core group from the EGU meeting was also present at Chamonix. The most important point is that MISMIP should continue with a new synthetic experiment, which would be completely complementary to other initiatives on marine ice sheet modelling. The major purpose should be to imporve ice sheet models and to seek whether they can cope with processes such as melting at the grounding line and debuttressing of ice shelves. This should lead to a reduction of uncertainties in future ice sheet modelling projections.

After a discussion of about one hour, all attendees generally agreed with previous conclusions from EGU splinter meeting. Further propositions were pushed to precise previous insights.

  1. Reproducing steady geometry on a reverse bed slope as inspired by the Gudmundsson et al. (2012) set-up should be a prerequisite to further participate on perturbations experiments. It is furthermore required that the models show the reversibility as proposed in the MISMIP3d experiments (Pattyn et al., 2013).
  2. It is important to test the predictability of models but also to improve our understanding on grounding line dynamics. Two set of experiments are therefore important. One would be to produce a steady-state geometry (depending on the physics of each of the models - different problem) and to perturb this state; a second experiment would start from a fixed (given) ice sheet geometry and a given velocity field, and each model (according to their physics) should optimize the basal friction field to be in accord with the given velocity field. From here, a perturbation experiment will be done.
  3. We agreed on a sketch of experiments to be evaluated before the intercomparison exercice could be launched.
  4. The geometry should be inspired by the one proposed by Gudmundsson et al. (2012), but with a smaller longitudinal extension
  5. A Century (100 years) is the typical duration to be envisaged for perturbations.

A preliminary schedule of actions to be engaged has been proposed. Stephen Cornford, in concordance with Hilmar Gudmunsson, will design and test a set of experiments. According to the various asymptotic expansion of Stokes equations implemented within BISICLES, this should help to embrace the response of various models.

Timeframe: initial tests should be completed during autumn 2014, before launching the intercomparison exercice to the concerned community during AGU 2014 (December).

Further developments will be announced on this website, until an experimental website (probably hosted by Bristol University), will be launched.