Data and scripts to reproduce figures from "Ocean warming threatens the viability of 60% of Antarctic ice shelves"

This is the formatted data related to the paper "Ocean warming threatens the viability of 60% of Antarctic ice shelves".

Main data used to produce the main figures:> bayesian_weights_davison_varying_combined_2300_withoutGISS.nc: Bayesian weights used in the computation of weighted likelihoods and averages considering all simulations going to 2300 - used in the main study> all_fluxes_br_withoutGISS.nc: all fluxes forming the ice-shelf mass balance propagated across the different dimensions of uncertainty (CMIP models, basal melt parameterisations, bed plasticities)> hydrofracturing_limits_new.nc: File containing the percentiles of reaching the hydrofracturing criterion

Files needed to compute the weights:> area_isf_greene22.nc: ice-shelf area used in Davison et al. 23 (from Greene et al. 2022)> varying_conditions_davison23.nc: Ice-shelf mass budget fluxes from 1997 on from Davison et al. 2023> steadystate_davison23.nc: Ice-shelf mass budget fluxes from the steady state from Davison et al. 2023

Geometric information:> gridarea_ISMIP6_AIS_4000m_grid.nc: File containing area of grid cells> Mask_Iceshelf_4km_IMBIE_withNisf.nc: Mask used to define the location of the different ice shelves, based on Caillet et al. 2025 > BedMachine_4km_slope_info_bedrock_draft_latlon_oneFRIS.nc: Geometric information needed for neural network parameterisation, inferred using the multimelt package> BedMachinev2_4km_isf_masks_and_info_and_distance_oneFRIS.nc: Geometric information needed for different computations, inferred using the multimelt package> ano_choice_NEMOorISMIP_withoutGISS.nc: file recording which T and S profiles should be taken, either ISMIP climatology or NEMO hindcast, depending on their difference to the reference mass balance

Data for analysis only until 2100:> bayesian_weights_davison_varying_combined_withoutGISS.nc: Bayesian weights used in the computation of weighted likelihoods and averages considering all simulations going to 2100 (see Extended Data Fig. 5)

Additional files needed for the additional analysis on other plausible geometries: > ElmerIce_4km_2100isf_masks_and_info_and_distance_oneFRIS.nc: Geometric information needed for different computations for the 2100 ice-sheet geometry> ElmerIce_4km_2100_slope_info_bedrock_draft_latlon_oneFRIS.nc: Geometric information needed for neural network parameterisation for the 2100 ice-sheet geometry> ElmerIce_4km_2150isf_masks_and_info_and_distance_oneFRIS.nc: Geometric information needed for different computations for the 2150 ice-sheet geometry> ElmerIce_4km_2150_slope_info_bedrock_draft_latlon_oneFRIS.nc: Geometric information needed for neural network parameterisation for the 2150 ice-sheet geometry> all_fluxes_br_withoutGISS_ElmerIcegeo2100.nc: all fluxes forming the ice-shelf mass balance propagated across the different dimensions of uncertainty (CMIP models, basal melt parameterisations, bed plasticities) for the 2100 ice-sheet geometry> all_fluxes_br_withoutGISS_ElmerIcegeo2150.nc: all fluxes forming the ice-shelf mass balance propagated across the different dimensions of uncertainty (CMIP models, basal melt parameterisations, bed plasticities) for the 2150 ice-sheet geometryThe scripts are explained in the associated README in scripts_paper_iceshelf_viability.zip. You can also find them here: https://github.com/ClimateClara/scripts_paper_iceshelf_viabilityHere is a summary of the README:

- Figure 2 and 3 were done with /notebooks_for_figures/timeseries_nb_viable_isf_withoutGISS_withhydrofrac_calving0.ipynb and /notebooks_for_figures/2D_subplots_viability_proba_withoutGISS_calving0.ipynb- Figure 4 was done with /notebooks_for_figures/2D_subplots_viability_proba_withoutGISS_calving0.ipynb- Figure 5 was done with /notebooks_for_figures/timeseries_nb_viable_isf_withoutGISS_withhydrofrac_calving0.ipynb and notebooks_for_figures/2D_subplots_viability_proba_onlyhydrofrac.ipynb

- Extended Data Figures 1 to 4 were done with /notebooks_for_figures/plot_mass_fluxes.ipynb- Extended Data Figure 5 was done with /notebooks_for_figures/timeseries_nb_viable_isf_withoutGISS_withhydrofrac_calving0.ipynb- Extended Data Figure 6 was done with /notebooks_for_figures/timeseries_nb_viable_isf_withoutGISS_withhydrofrac.ipynb and /notebooks_for_figures/2D_subplots_viability_proba.ipynb- Extended Data Figure 7 was done with /notebooks_for_figures/2D_subplots_viability_proba.ipynb- Extended Data Figure 8 was done with /notebooks_for_figures/timeseries_nb_viable_isf_withoutGISS_withhydrofrac_ElmerIcegeometries.ipynb and /notebooks_for_figures/2D_subplots_viability_proba_withoutGISS_ElmerIcegeometries.ipynb- Extended Data Figures 10 to 12 were done with /notebooks_for_figures/histo_weights_new.ipynb

In the folder 'notebooks_for_datapreparation', you will find a few scripts to prepare the data. These are not as detailed and not meant to be run out of the box but permit to give insight into the practical application of the methods described in the paper. For potential inspiration of similar work :)For the hydrofracturing calculations we refer to `Jourdain et al. 2025 <https://doi.org/10.5194/tc-19-1641-2025>`_ and the associated scripts: https://doi.org/10.5281/zenodo.13756240 and https://doi.org/10.5281/zenodo.15003864.