The reaction of the Antarctic ice sheet to atmospheric and ocean forcing happens to a large extent through weakening of ice shelves, concomitant reduction in ice-shelf buttressing, leading to grounding-line retreat, inland ice acceleration and loss of grounded ice mass. While the processes governing ice-shelf weakening are quite complex, due to specific interactions with atmosphere (surface melt, meltwater percolation, refreezing) and ocean (CDW circulation changes, ice-shelf- ocean interactions), uncertainties on the response of the grounded ice sheet in response to decreased buttressing is therefore harder to assess.
ABUMIP (Antarctic BUttressing Model Intercomparison Project) aims at comparing model responses to complete loss of buttressing by investigating the end-member of ice-shelf buttressing, i.e., the total loss of ice shelves. This enables gauging the sensitivity of different ice sheet models with respect to grounding-line retreat, as a function of basal sliding, isostasy, and other model pa- rameters. The experiments are kept simple and build on existing InitMIP-Antarctica experiments within the framework of ISMIP6. The ABUMIP experiments are led by Frank Pattyn and Nicholas Golledge. This description of ABUMIP-Antarctica can be downloaded from the ISMIP6 ABUMIP github site.
Ice-shelf removal or ‘float-kill’ (abuk)
The first standard experiment starts from an initialised present-day state of the Antarctic ice sheet, as defined in InitMIP-Antarctica and which represents the present-day Antarctic ice sheet either obtained through a spinup or by optimization of unknown fields (basal friction, rheology). The experiments run for 500 years, but should be at least 200 years for models that have difficulties to cope with multi-centennial runs. At the start of the experiment, all floating ice (shelves) surrounding the ice sheet are removed and kept removed during the run (so-called ‘float-kill’). In other words, the calving front coincides during the whole run with the grounding line position. The present-day surface mass balance (SMB) and tempertures are used as boundary condition and kept constant during the run. As in InitMIP, experimenters are free in their choice of SMB field. Isostasy and sub-shelf melting (upstream of the grounding line) are not considered. A similar experiment has been done by Golledge et al. (2017; supplementary material) and Pattyn (2017). The experiment aims at global Antarctic models, although regional experiments may be considered for high-resolution models. The same conventions as InitMIP applies to those models.
Extreme sub-shelf melt (abum)
The second experiment applies a constant melt rate of 400 m a-1 underneath the floating ice (shelves) for a period of 500 years. It is always possible that some models will have difficulties with the sudden removal of ice shelves (Experiment 1). Therefore, the second experiment should be feasible for all Antarctic models.
An optional third experiment performs a simple control run using a non-evolving present-day parameterisation, to ensure that ice shelves remain close to present-day extents for the duration of the experiment period. Setup should be as for the initMIP ‘ctrl’ run, but extended to span the same length as abuk and abum (ideally 500 years).
Repeat the standard experiments with:
- Addition of isostasy during the same period (abukiso and abumiso)
- Different sliding/friction laws (abuksx and abumsx, where x = 1, 2, ...)
Note that, according to InitMIP conventions, each additional experiment implies a different model name. Supplied documentation should give sufficient details on the model and its settings.
Similar output as for the InitMIP experiments is considered, with the experiment names as listed above (abuk, abum, abuc, abukiso, abumiso, ...). However, given the longer time series, output fields of 2d variables should be given every 10 years instead of every 5 years in order to keep output volume reduced. Ouput as time series should be given every year (as in InitMIP). The same convention applies, and time series, such as grounded ice volume, will therefore be 501 elements long whereby the value at t = 0 is the initialized value.
Experiments should be submitted by 1 May 2018 via the InitMIP FTP server. A separate directory (/ISMIP6/ABUMIP) has been created.
Pattyn, F.: Sea-level response to melting of Antarctic ice shelves on multi-centennial timescales with the fast Elementary Thermomechanical Ice Sheet model (f.ETISh v1.0), The Cryosphere, 11, 1851-1878, 2017.