NorESM3 topography update

[hgoelzer]

Update the CAM topography boundary conditions in response to ice sheet changes.

For an interactive ice sheet in NorESM2 we had an asynchronous procedure to update the CAM topography boundary conditions by modifying the restart file. --> https://github.com/NorESMhub/CISM-dynamic_atm_topo

We would like to have a similar process for NorESM3 with the new grids for both Greenland and Antarctica.

This would build on the updated process to produce CAM topography boundary conditions. --> https://github.com/NCAR/Topo

[hgoelzer]

In discussion with Mariana and Steve and with Kate at NCAR, we decided to first explore if CAM would be stable under expected topography perturbation, something Steve was concerned about. We decided to produce a set of potential topographies with changing Greenland ice sheet from standalone simulations to test CAM stability.

[hgoelzer]

I have now produced a first alternative topography file for testing with CAM.
I have added ice sheet thickness changes to the topography from a high-forcing run to year 2300. This is entering the topographic rawdata, which is then converted to cube format with bin_to_cube and subsequently to target with cube_to_target. The resulting file is here:
/cluster/projects/nn9560k/heig/topo3/dh/target/ne16pg3_gmted2010_modis_bedmachine-dh2300_nc3000_Laplace0200_noleak_greenlndantarcsgh30fac2.50_20260226.nc

I am using git@github.com:NCAR/Topo.git with input files and command taken from information in
/glade/campaign/cgd/amp/pel/topo/cesm3/ne16pg3_gmted2010_modis_bedmachine_nc3000_Laplace0200_noleak_greenlndantarcsgh30fac2.50_20250825.nc

:data_script = "./cube_to_target --grid_descriptor_file /glade/campaign/cesm/cesmdata/inputdata/atm/cam/coords/ne16pg3_scrip_170725.nc --grid_descriptor_file_gll /glade/campaign/cesm/cesmdata/inputdata/atm/cam/coords/ne16np4_esmf_c210305.nc --intermediate_cs_name /glade/campaign/cgd/amp/pel/topo/cubedata/gmted2010_modis_bedmachine-ncube3000-220518.nc --greenlndantarcsgh30_fac 2.500 --output_grid ne16pg3 --smoothing_scale 200.0000 --name_email_of_creator Peter Hjort Lauritzen, pel@ucar.edu --output_data_directory output/ --jmax_segments 100000"

[hgoelzer]

Additional input files with modified Greenland topography are now available at
/cluster/projects/nn9560k/heig/topo3/dh/target

ne16pg3_gmted2010_modis_bedmachine-dh1970_nc3000_Laplace0200_noleak_greenlndantarcsgh30fac2.50_20260228.nc
ne16pg3_gmted2010_modis_bedmachine-dh1980_nc3000_Laplace0200_noleak_greenlndantarcsgh30fac2.50_20260227.nc
ne16pg3_gmted2010_modis_bedmachine-dh1990_nc3000_Laplace0200_noleak_greenlndantarcsgh30fac2.50_20260228.nc
ne16pg3_gmted2010_modis_bedmachine-dh2000_nc3000_Laplace0200_noleak_greenlndantarcsgh30fac2.50_20260227.nc
ne16pg3_gmted2010_modis_bedmachine-dh2100_nc3000_Laplace0200_noleak_greenlndantarcsgh30fac2.50_20260227.nc
ne16pg3_gmted2010_modis_bedmachine-dh2300_nc3000_Laplace0200_noleak_greenlndantarcsgh30fac2.50_20260228.nc

The underlying ice sheet experiment is initialised to start in 1960 and follows first a historical forcing and then an extended SSP5-8.5 trajectory until 2300. The elevation changes due to the evolving ice sheet are calculated as anomalies relative to the initial state at 1960 and added to the raw topography file. ne16pg3_gmted2010_modis_bedmachine_nc3000_Laplace0200_noleak_greenlndantarcsgh30fac2.50_20250825.nc therefore represents the unperturbed observed topography (here nominally assigned to 1960), while the new input files above provide topography updates first every 10 years and then larger changes to 2100 and 2300.

The dh2300 above has been replaced by a new version ("_20260228.nc"), because it was missing the PHIS_gll variable due to premature OOM termination of the earlier run.

[hgoelzer]

The whole topography generation for the ne16pg3 grid currently takes more than 16 hours on the betzy preproc queue with one processor. The bin_to_cube script needed for preparation runs in 4 minutes, while cube_to_target uses more than 16h.

#!/bin/tcsh
#SBATCH --job-name=topo_update
#SBATCH --account=nnxxxxxk
#SBATCH --partition=preproc
#SBATCH --ntasks=1
#SBATCH --mem-per-cpu=40G
#SBATCH --time=18:00:0
./cube_to_target --grid_descriptor_file /cluster/projects/nn9560k/heig/topo3/data/ne16pg3_scrip_170725.nc --grid_descriptor_file_gll /cluster/projects/nn9560k/heig/topo3/data/ne16np4_esmf_c210305.nc --intermediate_cs_name /cluster/projects/nn9560k/heig/topo3/dh/cube/gmted2010_modis_bedmachine-ncube3000-220518-dh2300.nc --greenlndantarcsgh30_fac 2.500 --output_grid ne16pg3 --smoothing_scale 200.0000 --name_email_of_creator 'Heiko Goelzer, heig@norceresearch.no' --output_data_directory output/ --jmax_segments 100000 --source_data_identifier gmted2010_modis_bedmachine-dh2300

[hgoelzer]

In addition to the provided input files for Greenland above, I have now also processed elevation changes for the Antarctic ice sheet. The files are in the same directory /cluster/projects/nn9560k/heig/topo3/dh/target on betzy.
I have renamed the existing files to distinguish input files with modified GrIS and AIS as follows.
ne16pg3_gmted2010_modis_bedmachine-dhGrIS*
ne16pg3_gmted2010_modis_bedmachine-dhAIS*
The timing of the underlying ice sheet model experiment is a bit different for AIS, with major changes happening later. Testing should probably also start with the end member dhAIS2300.

[TomasTorsvik]

@hgoelzer - is this in any way connected with the topography change that has been introduced in CAM lately, and is now included in noresm3_0_beta12? I'm just wondering, if there will be more topography updates needed for upcoming noresm3_0_betaXX tags.