ocean/isomip_plus test group includes variants of the Ice Sheet-Ocean
Model Intercomparison Project, second phase (ISOMIP+) experiments from
Asay-Davis et al. (2016). These
experiments use idealized ice-shelf geometry
from the Marine Ice SheetModel Intercomparison Project, third phase (MISMIP+;
see Cornford et al. 2020)
performed with the BISICLES ice-sheet model.
Currently, only the Ocean0 experiment is supported but the plan is to add the Ocean1 and Ocean2 experiments in the next few months, and the Ocean3 and Ocean4 experiments at a later date, once MPAS-Ocean supports moving grounding lines.
By default, the test case is available at 2 km and 5 km horizontal resolution with a z-star Vertical coordinate. The test case has 36 vertical layers, each of 20-m thickness outside of the ice-shelf cavity.
The initial temperature for the whole domain is constant (1 degree Celsius), while salinity varies linearly with depth from 34.5 PSU at the sea surface to 34.7 PSU at the sea floor, which is at a constant at 2000 m depth. The conceptual overlying ice shelf depresses the sea surface height buy as much as 1990 m (leaving a 10-m water column) for the first 30 km in y. Over the next 30 km, it rises to 1490 m, then fairly abruptly to zero over the next 15 km, where it remains for the second half of the domain. The ice shelf occupies these first 75 km of the domain: fluxes from ice-shelf melting are only applied in this region.
isomip_plus test cases are composed of 3 steps that run by default:
initial_state, which defines the mesh, interpolates the ice geometry, and
computes the initial conditions for the model;
landIcePressure field to balance the
ssh field, see
Sea surface height adjustment; and
performance, which performs a 1-hour time
integration of the model and compares the results with a baseline if one is
Four additional steps can optionally be run:
simulation, which performs
one month of simulation, then updates the “evaporative” fluxes used in the test
case to prevent sea level from rising significantly due to meltwater inflow at
the ice-shelf base;
streamfunction, which computes the barotropic
(vertically integrated) and overturning streamfunctions;
viz, which plots
time series and movies of various variables of interest; and
interpolates the results to the MISOMIP comparison grid.
This test case is initialized with “warm” ocean conditions: 1 degree C at the sea floor, decreasing to -1.9 degrees C at the ocean surface. These conditions are approximately similar to those in the warmest waters on the Antarctic continental shelf in the Amundsen and Bellingshausen Seas. At the northern boundary, the temperature is restored to the same warm profile, leading to a vigorous circulation under the ice shelf that continually supplies heat and produces relatively high melt rates. Because of the rigorous flow, the simulation reaches a quasi-steady state in 2-3 years.
This test case is initialized with “cold” ocean conditions: -1.9 degree C throughout the water column. These conditions are similar to cold-shelf regions such as the Antarctic continental shelf in the Weddell and Ross Seas. At the northern boundary, the temperature is restored to the same warm profile as in Ocean0. The initially cold cavity has low melt rates and a weak flow, so that warm water from the northern boundary may take about a decade to reach the ice-shelf base. At this point, the melting and flow rapidly increase, eventually (in the coarse of ~20 years) leading to the same quasi-steady-state as in Ocean0. The ISOMIP+ protocol suggests running this simulation for 20 years.
This test case is initialized with “warm” ocean conditions as in Ocean0. At the northern boundary, the temperature is restored to the cold profile used for the initial condition in Ocean1: -1.9 degree C throughout the water column. Thus, where Ocean1 transitions from cold to warm cavity conditions, Ocean2 makes the opposite transition from warm to cold. The geometry is also taken from a different stage of the BISICLES MISIMP+ run than Ocean0 and Ocean1 in which the ice shelf has undergone significant thinning and retreat. The initially warm cavity has high melt rates and a strong flow, so that cold water water from the northern boundary will reach the ice-shelf base within a few years. At this point, the melting and flow exponentially decrease, approaching a new quasi-steady state. The ISOMIP+ protocol suggests running this simulation for 20 years, which is not long enough to reach quasi-steady state.
This test case is identical to
Ocean0 except that the land-ice pressure
and land-ice draft are prescribed to evolve in a very simple way in time.
By default, the these 2 fields start out at year 0001 with 10% of their normal
value (so the ice shelf is 10% of its thickness in a normal
Then, over the course of a year, both fields increase to 100% of their normal
value and stay there for another year. This test case is a simple way of
exploring changing ice thickness without the need to support a changing
grounding line (which remains fixed in time).
Users can modify the test case by adding or modifying entries in these config options before running the test case:
# config options for ISOMIP+ time-varying land-ice forcing [isomip_plus_forcing] # the forcing dates dates = 0001-01-01_00:00:00, 0002-01-01_00:00:00, 0003-01-01_00:00:00 # the amount by which the initial landIcePressure and landIceDraft are scaled # at each date scales = 0.1, 1.0, 1.0
Dates do not have to be the beginnings of years, they could be any list that is monotonic in time. Scales can be any fraction between 0.0 and 1.0.
isomip_plus test cases are configured for “performance” runs.
The initial condition is created, the the sea surface height and ice-shelf
pressure are adjusted to be in balance. Then, a simulation is performed for
only 1 simulated hour (appropriate for regression testing). Finally,
potential temperature and salinity are plotted at the top and bottom of the
ocean and along a cross section of through the middle (y = 40 km) of the
isomip_plus test cases can be manually configured for longer simulation
runs. First, do a performance run as described above (the default when you
compass run in the test case work directory).
Then, edit the config file in the work directory (e.g.
Ocean0.cfg) to set
setup_to_run = simulation streamfunction viz in the
at the very top. With this setting, one month of simulation will be performed,
then the streamfunctions will be computed based on the latest results in the
streamfunction step and time series plots and movies will be updated in
viz step. You can manually keep running
compass run in the test
case directory to run a month at a time, or you can create a job script to
compass run repeatedly (say 240 times for a 20-year simulation) inside