ISMIP7 Scalar Processing

Scripts for computing sea-level contributions (SLC) and other scalar variables from ISMIP7 ice sheet model output for the Antarctic Ice Sheet (AIS) and Greenland Ice Sheet (GrIS). All scripts expect ISMIP7-compliant model output on the diagnostic ISMIP grid following ISMIP7 conventions for file names and units.

Three parallel implementations are provided: Python (primary), NCO/bash (GrIS only), and MATLAB.

Requirements

conda create -n nc
conda activate nc
conda install -c conda-forge cdo=2.4.4 nco netCDF4 scipy

Input data

In addition to model output, the scripts require masks and area factors available on the ISMIP Globus server under Output-Processing. Model-specific density parameters (rhoi, rhow, rhof) are read from a params.nc file in the model directory.

Directory structure

By default, Data/ and Models/ are expected at the repository root. Any location on the file system can be used instead via the --datapath and --modelpath command-line options.

ismip7-scalar-processing/                        # repository root
├── slc/                                         # shared SLC computation package
├── python/scalars.py                            # AIS + GrIS (--region flag)
├── matlab/scalars.m                             # AIS + GrIS (region variable)
├── nco/scalars_GrIS.sh                          # GrIS NCO/bash implementation
├── Data/                                        # default location for masks and area factors
│   ├── AIS/
│   ├── GrIS/
│   ├── MINI0/
│   └── MINI1/
├── Models/                                      # default location for model output
│   ├── AIS/{lab}/{model}/{exp}_{res}/
│   ├── GrIS/{lab}/{model}/{exp}_{res}/
│   └── MINI/ISMIP7/{MINI0,MINI1}/{exp}/
└── manual-tests/
    ├── compare_outputs.py                       # Python vs MATLAB comparison (--region filter)
    ├── test_histout.py                          # integration test for --histout
    ├── test_refyear.py                          # integration test for --refyear
    └── MINI/                                   # lightweight MINI test suite
        ├── scalars_MINI.py
        ├── run_MINI.py
        ├── output/
        └── setup/                              # experiment generation and remapping tools
            ├── derive_exp0.py
            ├── check_masks.py
            ├── ground.sh
            ├── interpolate.sh
            ├── remapCDO.sh
            ├── gdf_ISMIP7_MINI0.txt
            └── gdf_ISMIP7_MINI1.txt

Running the scripts

All scripts accept CLI arguments; the default settings match the NORCE/CISM08-MAR312-p50 (GrIS) and VUW/PISM1 (AIS) reference configurations.

Python (primary)

cd python
conda run -n nc python3 scalars.py --region AIS
conda run -n nc python3 scalars.py --region GrIS
conda run -n nc python3 scalars.py --region AIS --lab ISMIP7 --model TEST --exp exp0 --hist historical

Key arguments: --region {AIS,GrIS} (required), --lab, --model, --exp, --hist, --refyear, --histout, --res, --datapath, --modelpath, --outpath.

Output is written to ./output/ relative to the script directory.

MATLAB

cd matlab
matlab -nodisplay -nosplash -r "region='AIS'; run('scalars.m'); exit"
matlab -nodisplay -nosplash -r "region='GrIS'; run('scalars.m'); exit"

Set workspace variables before run() to override any default (lab, model, exp, hist, refyear, res, datapath, modelpath, outpath).

NCO/bash (GrIS only)

First create the model-specific parameter file, then run the main script:

cd nco
bash set_params.sh
bash scalars_GrIS.sh

MINI test suite

The MINI suite provides lightweight test cases on a coarse 11×11 grid (600 km pixels) for rapid validation.

cd manual-tests/MINI

# Run a single case
conda run -n nc python3 scalars_MINI.py --model MINI1 --exp exp0

# Run all combinations (MINI0/MINI1 × exp0/expg) and print summary
conda run -n nc python3 run_MINI.py

Experiment generation and grid remapping tools live in manual-tests/MINI/setup/.

Two grid variants are available:

Grid	Size	Origin (xfirst, yfirst)	Notes
MINI1	11×11	−3,040,000 m	primary test grid
MINI0	12×12	−3,340,000 m	half-cell-offset variant for CDO remapping tests

MINI differs from the full AIS/GrIS scripts in three ways: the first time step serves as the SLC baseline (no separate historical reference), standard densities are used directly (no params.nc), and area weighting uses dx² directly.

SLC methods

The slc/ package implements three sea-level contribution methods:

Method	Description
slc_vaf	Volume Above Flotation — ISMIP6 method, freshwater conversion
slc_G2020	Goelzer et al. 2020 (TC) — adds potential ocean volume and density corrections
slc_A2020	Adhikari et al. 2020 (TC) — absolute reference frame with grounding-line migration

Three volume diagnostics are also computed (total, grounded, floating ice volume change).

Ocean area normalization uses oarea = 3.625 × 10¹⁴ m² (Gregory et al. 2019).

Testing

After running both Python and MATLAB versions, compare outputs:

cd manual-tests
conda run -n nc python3 compare_outputs.py --region AIS
conda run -n nc python3 compare_outputs.py --region GrIS

Expected tolerance: < 1 × 10⁻¹⁰ m.

Integration tests for --histout and --refyear (require --datapath and --modelpath):

cd manual-tests
conda run -n nc python3 test_histout.py --datapath <path> --modelpath <path>
conda run -n nc python3 test_refyear.py --datapath <path> --modelpath <path>