mismipplus

The mismipplus test group (compass/landice/tests/mismipplus) implements a limited number of the experiments from the Marine Ice Sheet Model Intercomparison Project Plus (MISMIP+; Asay-Davis et al., 2016). To date, the spinup and Ice0 are the only experiments currently supported. In the future, additional test cases may be added for the missing MISMIP+ experiments (i.e. Ice1r, Ice1ra, Ice1rr, Ice2r, Ice2r, Ice2ra, Ice2rr; Table 2. Asay-Davis et al., 2016).

shared config options

Both mismipplus test cases share the following config options:

[mismipplus]

# the number of cells per core to aim for
goal_cells_per_core = 300

# the approximate maximum number of cells per core
max_cells_per_core = 5000

The config options are all aimed at dynamically calculating the optimal number of ntasks for the run_model step based on the resolution requested by the test case. For more information about how ntasks is calculated see tasks in the developers guide page.

smoke_test

landice/mismipplus/smoke_test runs the MISMIP+ Ice0 experiment for five years using a spun-up 2km mesh. The simulation is run with the adaptive timestepper on. There are no configuration options for this test case. Currently 2000m is the only resolution supported. In the future, the spin_up test case will be used to generate new versions of the mesh at 8000m, 4000m, 2000m, and 1000m resolutions.

spin_up

landice/mismipplus/spin_up generates the files needed to run a 20,000 year spin-up using the standard MISMIP+ initial conditions outline in section 2.1 of (Asay-Davis et al., 2016). The purpose of this test case is generate a semi-realistic glacier geometry that includes an ice shelf, which can be used to the subsequent MISMIP+ experiments and/or additional experiments that require 3-D complexity without site-specific complications arising from using an actual glacier geometry.

# config options for the mesh (i.e. initial condition) setup
[mesh]

# Nominal cell spacing (m). Resolution is "nominal" since the true cell spacing
# will be determined such that the cell center to cell center length of the
# entire domain in the y direction is exactly the required y domain length (80 km).
resolution = 8000.

# length (m) to extend the eastern domain boundary by. Needed for simulations
# that use a dynamic calving law, where the calving front will be irregularly
# shaped. Any value less than `2*resolution` will be ignored, as the default
# gutter length is 2 gridcells.
gutter_length = 0.

# ice density (kg m^{-3}). MISMIP+ uses 918 (Table 1. Asay-Davis et al. 2016),
# but MALI defaults to 910. The user can choose if they want to strictly follow
# MISMIP+ or use the default MALI value.
ice_density = 918.

# Initial ice thickness (m). The MISMIP+ protocol does not specify an initial
# ice thickness for the spinup experiment. Therefore the user is allowed to
# choose their desired value.
init_thickness = 100.

# Number of vertical levels
levels = 10

# How to distribute vertical layers. Options are "glimmer" or "uniform".
# "glimmer" will distribute the layer non-uniformly following
# Eqn. (15) from Rutt et al. 2009.
vetical_layer_distribution = glimmer