Running with YelmoX
YelmoX is a separate repository that is designed to provide supplementary libraries and programs that allow running ice-sheet simulations with realistic boundary (e.g., climate and ocean) forcing and interactions (e.g., isostatic rebound).
Here you can find the basic information and steps needed to get YelmoX running.
Super-quick start
A summary of commands to get started is given below. For more detailed information see subsequent sections.
# Before doing anything, make sure dependencies are installed (Lis, NetCDF, runner)
##########################
# Clone repository
git clone https://github.com/palma-ice/yelmox.git
git clone git@github.com:palma-ice/yelmox.git # or via ssh
# Clone yelmo into a sub-directory too
cd yelmox
git clone https://github.com/palma-ice/yelmo.git
git clone git@github.com:palma-ice/yelmo.git # or, via ssh
# Check out your yelmox branch of interest (optional)
git checkout my-branch # for an existing branch
git checkout -b my-new-branch # or, to create a new one
# Do the same for yelmo branch if needed, and return to yelmox directory
cd yelmo
git checkout my-branch
cd ..
# Enter yelmo directory and configure it for compiling
cd yelmo
python3 config.py config/snowball_gfortran # or the right config file for your system
# Return to yelmox directory and configure it for compiling
cd ..
python3 config.py config/snowball_gfortran # or the right config file for your system
# Now, compile the default program
make clean
make yelmox
# Link to `ice_data` repository wherever you have it saved on your system
ln -s path_to/ice_data
# Copy the runylmox config file to the main directory
cp config/runylmox.js ./
# Run a test simulation of Antarctica for 1000 yrs
./runylmox -r -e yelmox -n par/yelmo_Antarctica.nml -o output/ant-test -p ctrl.time_end=1e3
Standard YelmoX simulations
To run YelmoX, by default we use the program yelmox.f90. This program currently makes use of snapclim for the climatic forcing and smbpal for the snowpack and surface mass balance calculations.
Spin-up simulation for ISMIP6-based runs (ISMIP6, ABUMIP)
First make sure your distribution of yelmox and yelmo are up to date.
cd yelmo
git pull
cd ..
# In the main yelmox directory, change to the branch 'tfm2021' or 'abumip-2021':
git pull
git checkout tfm2021
# From main directory of yelmox, also reconfigure to adopt all changes:
python3 config.py config/snowball_gfortran
cp config/runylmox.js ./
# Link to ice_data path as needed:
ln -s /media/Data/ice_data ice_data
Now compile as normal, but with the yelmox_ismip6 program:
make clean
make yelmox_ismip6
You are now ready to run some ISMIP6 simulations. If you have a spinup simulation available you can skip the rest of this section.
The next step is to get a spin-up simulation ready. To do so, we will run a small ensemble of simulations that apply different calving coefficients (ytopo.kt) and shear-regime enhancement factors (ymat.enh_shear). Each simulation will run with these parameter values set, while optimizing the basal friction coefficient field cb_ref and the temperature anomalies imposed in different basins tf_corr. To run this ensemble, use the following commands:
# Specify run choices, to run locally in the background:
runopt='-r'
# or, to submit job to a cluster, eg:
runopt='-s -q priority -w 10'
# Define the output folder
fldr=tmp/ismip6/spinup_32km_68
# Run the Yelmo ensemble
jobrun ./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -- -o ${fldr} -p ctrl.run_step="spinup_ismip6" tf_cor.name="dT_nl" marine_shelf.gamma_quad_nl=14500 ytopo.kt=1.0e-3,1.5e-3,2.0e-3,2.5e-3 ymat.enh_shear=1,3
An alternative spinup procedure
# First run for 30kyr with topo relax on to spinup thermodynamics...
runopt='-s -q priority -w 10'
fldr=tmp/ismip6/spinup_32km_69
jobrun ./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -- -o ${fldr} -p ctrl.run_step="spinup_ismip6" tf_cor.name="dT_nl" marine_shelf.gamma_quad_nl=14500 spinup_ismip6.equil_method="relax" spinup_ismip6.time_end=30e3 spinup_ismip6.time_equil=30e3 ytopo.kt=1.0e-3,1.5e-3,2.0e-3,2.5e-3 ymat.enh_shear=1
# Testing opt spinup but with 'robin' initial temp profile
runopt='-s -q priority -w 10'
fldr=tmp/ismip6/spinup_32km_70
jobrun ./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -- -o ${fldr} -p ctrl.run_step="spinup_ismip6" tf_cor.name="dT_nl" marine_shelf.gamma_quad_nl=14500 ytopo.kt=1.0e-3 ymat.enh_shear=1 opt_L21.cf_max=1.0
# robin-cold but only by -2deg instead of -10deg
runopt='-s -q priority -w 10'
fldr=tmp/ismip6/spinup_32km_71
jobrun ./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -- -o ${fldr} -p ctrl.run_step="spinup_ismip6" tf_cor.name="dT_nl" marine_shelf.gamma_quad_nl=14500 ytopo.kt=1.0e-3 ymat.enh_shear=1 opt_L21.cf_max=0.2 opt_L21.cf_init=0.2
runopt='-s -q short -w 10'
fldr=tmp/ismip6/spinup_32km_72
jobrun ./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -- -o ${fldr} -p ctrl.run_step="spinup_ismip6" tf_cor.name="dT_nl" marine_shelf.gamma_quad_nl=14500 ytopo.kt=1.0e-3 opt_L21.cf_max=10,20,40,45 ydyn.beta_u0=100,300
ISMIP6 simulations
Make sure you already have a spinup simulation available, and that the parameters of the spinup will match those supplied here. The next step is to run different experiments of interest that restart from the spinup experiment.
Some commands for running diagnostic short runs
### Diagnostic short runs ###
# Run a 16km spinup run with relaxation
runopt='-s -q priority -w 1'
fldr=tmp/ismip6/spinup_16km_72_diag
jobrun ./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -- -o ${fldr} -p ctrl.run_step="spinup_ismip6" tf_cor.name="dT_nl" marine_shelf.gamma_quad_nl=14500 ytopo.kt=1.0e-3 spinup_ismip6.equil_method="relax" yelmo.grid_name="ANT-16KM" spinup_ismip6.time_end=20 spinup_ismip6.dt2D_out=1
# Run with 16km restarting from 32km file
# (currently crashes probably because tf_corr cannot be interpolated)
runopt='-s -q priority -w 1'
fldr=tmp/ismip6/ismip_32km_68_diag
file_restart=/p/tmp/robinson/ismip6/spinup_32km_68/0/yelmo_restart.nc
./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -o ${fldr}/ctrl -p ctrl.run_step="transient_proj" yelmo.restart=${file_restart} transient_proj.scenario="ctrl" tf_cor.name="dT_nl" marine_shelf.gamma_quad_nl=14500 ${paropt} yelmo.grid_name="ANT-16KM" transient_proj.time_end=1920 transient_proj.dt2D_out=1 ytill.is_angle=False
###
Actual ISMIP6 commands
# Define output folder as a bash variable
fldr=tmp/ismip6/ismip_32km_71
# Specify run choices, to run locally in the background:
runopt='-r'
# or, to submit job to a cluster, eg:
runopt='-s -q priority -w 10'
# Set parameter choices to match those of spinup simulation
paropt="ytopo.kt=1.0e-3"
## First run a steady-state simulation to stabilize everything
# Define restart file path as a bash variable, for example, on snowball:
file_restart=/p/tmp/robinson/ismip6/spinup_32km_71/0/yelmo_restart.nc
# ctrl-0
./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -o ${fldr}/ctrl-0 -p ctrl.run_step="transient_proj" yelmo.restart=${file_restart} transient_proj.scenario="ctrl" tf_cor.name="dT_nl" marine_shelf.gamma_quad_nl=14500 transient_proj.time_end=11900 transient_proj.dt1D_out=10 transient_proj.dt2D_out=200 ${paropt}
## Next, call the Yelmo commands for the individual cases...
# Define restart file path as a bash variable, for example, on snowball:
#file_restart=/p/tmp/robinson/ismip6/spinup_32km_68/0/yelmo_restart.nc
file_restart=/p/tmp/robinson/ismip6/ismip_32km_68/ctrl-0/yelmo_restart.nc
# ctrl
./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -o ${fldr}/ctrl -p ctrl.run_step="transient_proj" yelmo.restart=${file_restart} transient_proj.scenario="ctrl" tf_cor.name="dT_nl" marine_shelf.gamma_quad_nl=14500 ${paropt}
# exp05
./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -o ${fldr}/exp05 -p ctrl.run_step="transient_proj" yelmo.restart=${file_restart} transient_proj.scenario="rcp85" tf_cor.name="dT_nl" marine_shelf.gamma_quad_nl=14500 ${paropt}
# exp09
./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -o ${fldr}/exp09 -p ctrl.run_step="transient_proj" yelmo.restart=${file_restart} transient_proj.scenario="rcp85" tf_cor.name="dT_nl_95" marine_shelf.gamma_quad_nl=21000 ${paropt}
# exp10
./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -o ${fldr}/exp10 -p ctrl.run_step="transient_proj" yelmo.restart=${file_restart} transient_proj.scenario="rcp85" tf_cor.name="dT_nl_5" marine_shelf.gamma_quad_nl=9620 ${paropt}
# exp13
./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -o ${fldr}/exp13 -p ctrl.run_step="transient_proj" yelmo.restart=${file_restart} transient_proj.scenario="rcp85" tf_cor.name="dT_nl_pigl" marine_shelf.gamma_quad_nl=159000 ${paropt}
ABUMIP
Make sure you already have a spinup simulation available.
Use the following commands to run the three main experiments of interest. Note that abuk and abum may run much more slowly than abuc. The parameter values applied in the commands below ensure that the model parameters correspond to those used in the restart simulation, although many of them like ocean temp. anomalies in different basins or calving parameters, are no longer relevant in the ABUMIP context. It is important, however, to specify ydyn.ssa_lat_bc='marine', as it is relevant for this experiment to apply marine boundary conditions. This is generally not used currently, as it makes the model much less stable.
Note that an equilibrium spin-up simulation has already been performed, which gives good agreement with the present-day ice sheet. These results have been saved in a restart file, from which your simulations will begin (see below).
# Define restart file path as a bash variable
file_restart=/p/tmp/robinson/ismip6/spinup_32km_68/0/yelmo_restart.nc
# Define output folder as a bash variable
fldr=tmp/ismip6/abumip_32km_68
# Specify run choices, to run locally in the background:
runopt='-r'
# or, to submit job to a cluster, eg:
runopt='-s -q priority -w 5'
debugopt="abumip_proj.time_end=20 abumip_proj.dt2D_out=1"
# Set parameter choices to match those of spinup simulation
paropt="ytopo.kt=1.0e-3"
# Call the Yelmo commands...
# ABUC - control experiment
./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -o ${fldr}/abuc -p abumip.scenario="abuc" ctrl.run_step="abumip_proj" yelmo.restart=${file_restart} abumip_proj.scenario="ctrl" tf_cor.name="dT_nl" marine_shelf.gamma_quad_nl=14500 isostasy.method=0 ${paropt}
# ABUK - Ocean-kill experiment
./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -o ${fldr}/abuk -p abumip.scenario="abuk" ctrl.run_step="abumip_proj" yelmo.restart=${file_restart} abumip_proj.scenario="ctrl" tf_cor.name="dT_nl" marine_shelf.gamma_quad_nl=14500 isostasy.method=0 ${paropt}
# ABUM - High shelf melt (400 m/yr)
./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -o ${fldr}/abum -p abumip.scenario="abum" ctrl.run_step="abumip_proj" yelmo.restart=${file_restart} abumip_proj.scenario="ctrl" tf_cor.name="dT_nl" marine_shelf.gamma_quad_nl=14500 isostasy.method=0 ${paropt}
# ABUK - Ocean-kill experiment (MARINE BOUNDARY CONDITIONS)
./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -o ${fldr}/abuk-marine -p abumip.scenario="abuk" ctrl.run_step="abumip_proj" yelmo.restart=${file_restart} abumip_proj.scenario="ctrl" tf_cor.name="dT_nl" marine_shelf.gamma_quad_nl=14500 isostasy.method=0 ${paropt} ydyn.ssa_lat_bc="marine"
# ABUM - High shelf melt (400 m/yr) (MARINE BOUNDARY CONDITIONS)
./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -o ${fldr}/abum-marine -p abumip.scenario="abum" ctrl.run_step="abumip_proj" yelmo.restart=${file_restart} abumip_proj.scenario="ctrl" tf_cor.name="dT_nl" marine_shelf.gamma_quad_nl=14500 isostasy.method=0 ${paropt} ydyn.ssa_lat_bc="marine"
Simulations with hyster
# Define restart file path as a bash variable, for example, on snowball:
file_restart=/p/tmp/robinson/ismip6/spinup_32km_68/0/yelmo_restart.nc
# Define output folder as a bash variable
fldr=tmp/ismip6/ramp_32km_68
# Specify run choices, to run locally in the background:
runopt='-r'
# or, to submit job to a cluster, eg:
runopt='-s -q priority -w 5'
# Set parameter choices to match those of spinup simulation
paropt="ytopo.kt=1.0e-3"
# Now run simulation
./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -o ${fldr}/r1 -p ctrl.run_step="hysteresis_proj" yelmo.restart=${file_restart} tf_cor.name="dT_nl" marine_shelf.gamma_quad_nl=14500 hysteresis_proj.time_end=30e3 hyster.method='ramp-time' hyster.df_sign=-1 hyster.dt_init=0 hyster.dt_ramp=10e3 hyster.f_min=-10 hyster.f_max=5 ${paropt}
# Try a periodic simulation
./runylmox ${runopt} -e ismip6 -n par/yelmo_ismip6_Antarctica.nml -o ${fldr}/r2 -p ctrl.run_step="hysteresis_proj" yelmo.restart=${file_restart} tf_cor.name="dT_nl" marine_shelf.gamma_quad_nl=14500 hysteresis_proj.time_end=100e3 hyster.method='sin' hyster.df_sign=1 hyster.dt_init=0 hyster.dt_ramp=20e3 hyster.f_min=-10 hyster.f_max=5 ${paropt}
That's it!