tmismip3d.rst - pism - [fork] customized build of PISM, the parallel ice sheet model (tillflux branch)
(HTM) git clone git://src.adamsgaard.dk/pism
(DIR) Log
(DIR) Files
(DIR) Refs
(DIR) LICENSE
---
tmismip3d.rst (5033B)
---
1 .. include:: ../../global.txt
2
3 .. _sec-MISMIP3d:
4
5 MISMIP3d
6 --------
7
8 The ice2sea MISMIP3d intercomparison is a two-horizontal-dimensional extension of the
9 flowline case described above. As before, in MISMIP3d the grounding line position and its
10 reversibility under changes of physical parameters is analyzed. Instead of changing the
11 ice softness, however, the spatial distribution and magnitude of basal friction is
12 adjusted between experiments. The applied basal friction perturbation of the basal
13 friction is a localized gaussian "bump" and thus a curved grounding line is obtained. In
14 contrast to the flowline experiments, no (semi-)analytical solutions are available to
15 compare to the numerical results.
16
17 A full description of the MISMIP3d experiments can be found at
18
19 |mismip3d-url|
20
21 and the results are published in :cite:`MISMIP3d2013`.
22
23 A complete set of MISMIP3d experiments consists of three runs: Firstly, a flowline
24 solution on a linearly-sloped bed, similar to the flowline MISMIP experiments of the
25 previous section, is run into a steady state ("standard experiment ``Stnd``"). Then the
26 localized sliding perturbation is applied ("perturbation experiment") causing the
27 grounding line to shift and lose symmetry. Two different amplitudes of the perturbation
28 are considered ("``P10``" and "``P75``"). Finally, beginning from the final state of the
29 perturbation experiment, the sliding perturbation is removed and the system is run again
30 into steady state ("reversibility experiment"). The resulting geometry, in particular the
31 grounding line position, is expected to be close to that of the standard experiment.
32 Expecting such reversibility assumes that a particular stationary ice geometry only
33 depends on its physical parameters and boundary conditions and not on how it is
34 dynamically reached.
35
36 For these experiments in PISM, a Python script generates a shell script which has the
37 commands and options for running a MISMIP3d experiment. The python script is
38 ``createscript.py`` in the folder ``examples/mismip/mismip3d/``. Run
39
40 .. code-block:: none
41
42 ./createscript.py -h
43
44 to see a usage message. A ``README.md`` gives a tutorial on how to use ``createscript.py``
45 and do the runs themselves.
46
47 For the flowline ``Stnd`` experiment, as in the MISMIP case, a computational domain with
48 three grid points in the direction orthogonal to the ice flow (arbitrarily chosen as
49 y-direction) is chosen by ``createscript.py``. For the perturbation and reversibility
50 experiments a domain is defined which is symmetric along the ice-divide (mirror symmetry)
51 and along the center line of the ice flow, while the side boundaries are periodic, which
52 corresponds to a free-slip condition for the flow in x-direction. Though this choice of
53 the symmetric computational domain increases computational cost, it allows us to use
54 standard PISM without fixing certain boundary conditions in the code. (That is, it avoids
55 the issues addressed in the regional mode of PISM; see section :ref:`sec-jako`.)
56
57 PISM participated in the MISMIP3d intercomparison project :cite:`MISMIP3d2013` using version
58 pism0.5, and the exact results can be reproduced using that version. PISM's results, and
59 the role of resolution and the new subgrid grounding line interpolation scheme are
60 discussed in :cite:`Feldmannetal2014`.
61
62 We observed a considerable improvement of the results with respect to the absolute
63 grounding line positions compared to other models (e.g. the FE reference model Elmer/Ice)
64 and to the reversibility when applying the subgrid grounding line interpolation method;
65 see :numref:`fig-Subgl`. Furthermore, we observed that only using SSA yields almost
66 the same results as the full hybrid SIA+SSA computation for the MISMIP3D (and also the
67 MISMIP) experiments, but, when not applying the SIA computation, after a considerably
68 shorter computation time (about 10 times shorter). We explain the small and almost
69 negligible SIA velocities for the MISMIP(3D) experiments with the comparably small ice
70 surface gradients in the MISMIP3d ice geometries. See :numref:`fig-compSIASSA` for a
71 comparison of SSA and SIA velocities in the MISMIP3D geometry. Note that both Figures
72 :numref:`fig-Subgl` and :numref:`fig-compSIASSA` were generated with resolution of `\Delta
73 x = \Delta y = 1` km.
74
75 .. figure:: figures/mismip3d-subgl.png
76 :name: fig-Subgl
77
78 Comparison between the grounding lines of the higher-amplitude ("``P75``") MISMIP3d
79 experiments performed with PISM when using the subgrid grounding line interpolation
80 method (left) or not using it (right). In both cases the SIA+SSA hybrid is used.
81
82 .. figure:: figures/comp-SIA-SSA.png
83 :name: fig-compSIASSA
84
85 The SIA velocities are negligible in the MISMIP3d standard experiment ("``Stnd``"). The
86 steady state ice geometry is plotted (black) together with the computed SSA velocity
87 (red) and SIA velocity (blue). The SIA velocity reaches its maximum value of about
88 `10` m/a at the grounding line, about two orders of magnitude less than the maximum of
89 the SSA velocity.