4. Test simulations¶
These are a number of computationally rather inexpensive test simulations, of which the run-specs header files are contained in the SICOPOLIS repository.
- Run
repo_vialov3d25
- 3D version of the 2D Vialov profile (Vialov [59]),SIA, resolution 25 km, \(t=0\ldots{}100\,\mathrm{ka}\).Similar to the EISMINT Phase 1 fixed-margin experiment (Huybrechts et al. [41]), but without thermodynamics. Instead, isothermal conditions with \(T=-10^{\circ}\mathrm{C}\) everywhere are assumed.
- Run
repo_emtp2sge25_expA
- EISMINT Phase 2 Simplified Geometry Experiment A,SIA, resolution 25 km, \(t=0\ldots{}200\,\mathrm{ka}\) (Payne et al. [51]).The thermodynamics solver for this run is the one-layer melting-CTS enthalpy scheme (ENTM), while all other runs employ the polythermal two-layer scheme (POLY) (Greve and Blatter [31]).
- Run
repo_grl16_bm5_ss25ka
- Greenland ice sheet, SIA, resolution 16 km,short steady-state run (\(t=0\ldots{}25\,\mathrm{ka}\)) for modern climate conditions (Fig. 4.1; unpublished).
- Runs
repo_grl16_bm5_{init100a, ss25ka_nudged}
- Greenland ice sheet, SIA, resolution 16 km;\(t=-100\,\mathrm{a}\ldots{}0\) for the init run without basal sliding (…_init100a),\(t=0\,\ldots{}25\,\mathrm{ka}\) for the main run (…_ss25ka_nudged),steady-state run for modern climate conditions, free evolution during the first 10 ka, after that gradual nudging towards the slightly smoothed present-day topography computed by the init run (Fig. 4.1; unpublished).
![Ice volume for steady-state simulations for Greenland](_images/grl16_ss25ka.png)
Fig. 4.1 Ice volume for the two steady-state simulations for Greenland, repo_grl16_bm5_ss25ka (unconstrained evolution) and repo_grl16_bm5_ss25ka_nudged (topography nudging with time-dependent relaxation time after t = 10 ka).¶
- Run
repo_ant64_bm3_ss25ka
![Ice thickness and surface velocity for float-kill simulation for Antarctica](_images/repo_ant64_bm3_ss25ka0003_H_vs.png)
Fig. 4.2 Ice thickness and surface velocity for the short steady-state simulation for Antarctica with instantaneous removal of ice shelves (“float-kill”), repo_ant64_bm3_ss25ka. The West Antarctic ice sheet has largely disappeared.¶
- Run
repo_grl20_b2_paleo21
- Greenland ice sheet, SIA, resolution 20 km,\(t=-140\,\mathrm{ka}\ldots{}0\), basal sliding ramped up during the first 5 ka.Modified, low-resolution version of the spin-up for ISMIP6 InitMIP (Greve et al. [33]).
- Runs
repo_grl10_b2_{paleo21, future21_ctrl, future21_asmb}
![Ice volume above flotation for future climate simulations for Greenland](_images/grl10_b2_future21.png)
Fig. 4.3 Ice volume above flotation, expressed in metres of sea-level equivalent (m SLE), for the two ISMIP6 InitMIP future-climate simulations for Greenland, repo_grl10_b2_future21_ctrl (constant-climate control run) and repo_grl10_b2_future21_asmb (schematic surface-mass-balance anomaly applied).¶
- Runs
repo_ant64_b2_{spinup09_init100a, spinup09_fixtopo, spinup09, future09_ctrl, future09_asmb, future09_abmb}
- Antarctic ice sheet with hybrid shallow-ice–shelfy-stream dynamics(Bernales et al. [5]) and ice shelves (SSA), resolution 64 km;\(t=-140.1\ldots{}-140\,\mathrm{ka}\) for the init run without basal sliding (…_init100a),\(t=-140\,\mathrm{ka}\ldots{}0\) for the run with almost fixed topography (…_fixtopo), basal sliding ramped up during the first 5 ka,\(t=-0.5\,\mathrm{ka}\ldots{}0\) for the final, freely-evolving-topography part of the spin-up (…_spinup09),\(t=0\ldots{}100\,\mathrm{a}\) for the three future runs (…_future09_{ctrl, asmb, abmb}).64-km version of the spin-up and schematic future climate runs for ISMIP6 InitMIP
![Ice volume above flotation for future climate simulations for Antarctica](_images/ant64_b2_future09.png)
Fig. 4.4 Ice volume above flotation, expressed in metres of sea-level equivalent (m SLE), for the three ISMIP6 InitMIP future-climate simulations for Antarctica, repo_ant64_b2_future09_ctrl (constant-climate control run), repo_ant64_b2_future09_asmb (schematic surface-mass-balance anomaly applied) and repo_ant64_b2_future09_abmb (schematic sub-ice-shelf-melt anomaly applied).¶
- Runs
repo_asf2_steady
,repo_asf2_surge
- Austfonna, SIA, resolution 2 km, \(t=0\ldots{}10\,\mathrm{ka}\).Similar to Dunse et al. [15]’s Exp. 2 (steady fast flow) and Exp. 5 (surging-type flow), respectively.
- Runs
repo_nmars10_steady
,repo_smars10_steady
- North-/south-polar cap of Mars, SIA, resolution 10 km, \(t=-10\,\mathrm{Ma}\ldots{}0\).Steady-state runs by Greve [29].
- Run
repo_nhem80_nt012_new
- Northern hemisphere, SIA, resolution 80 km, \(t=-250\,\mathrm{ka}\ldots{}0\).Similar to run nt012 by Greve et al. [40].
- Run
repo_heino50_st
- ISMIP HEINO standard run ST, SIA, resolution 50 km, \(t=0\ldots{}200\,\mathrm{ka}\) (Calov et al. [12]).
Model times, time steps, computing times:
Run |
Model time |
Time step† |
CPU time‡ |
---|---|---|---|
repo_vialov3d25 |
100 ka |
20 a |
1.0 min |
repo_emtp2sge25_expA |
200 ka |
20 a |
4.6 min |
repo_grl16_bm5_ss25ka |
25 ka |
5 a |
10.8 min |
repo_grl16_bm5_init100a |
100 a |
5 a |
1.6 sec |
repo_grl16_bm5_ss25ka_nudged |
25 ka |
5 a |
10.6 min |
repo_ant64_bm3_ss25ka |
25 ka |
2 / 10 a† |
8.9 min |
repo_grl20_b2_paleo21 |
140 ka |
5 a |
0.9 hrs |
repo_grl10_b2_paleo21* |
9 ka |
1 a |
1.1 hrs |
repo_grl10_b2_future21_ctrl |
100 a |
1 a |
0.9 min |
repo_grl10_b2_future21_asmb |
100 a |
1 a |
0.9 min |
repo_ant64_b2_spinup09_init100a |
100 a |
2 / 10 a† |
4.3 sec |
repo_ant64_b2_spinup09_fixtopo |
140 ka |
5 / 10 a† |
0.7 hrs |
repo_ant64_b2_spinup09 |
500 a |
2 / 10 a† |
0.5 min |
repo_ant64_b2_future09_ctrl |
100 a |
2 / 10 a† |
6.3 sec |
repo_ant64_b2_future09_asmb |
100 a |
2 / 10 a† |
6.3 sec |
repo_ant64_b2_future09_abmb |
100 a |
2 / 10 a† |
6.3 sec |
multi_sico_1.sh
, run with SICOPOLIS v23 (revision 1353323e2) and the Intel Fortran Compiler 19.1 for Linux (optimization options -xHOST -O3 -no-prec-div
) on a 12-Core Intel Xeon Gold 6256 (3.6 GHz) PC under openSUSE Leap 15.5.repo_ant64_xxx
runs can be done on multiple cores using OpenMP for the SSA solver. However, at the employed, low resolution of 64 km the solver does not scale well, and the gain in wall clock time by using multiple cores is very small.