Postprocessed Plots for "Assessing Future Ice Shelf Hydrofracture Vulnerability in the ISMIP6 Ensemble"

Results from Main Manuscript: Sensitivity of Shelf Hydrofracture Vulnerability to Forcings and Fracture Toughness

The .zip files contain all per-ice-sheet-model plots made by the reanalysis workflow used for the associated manuscript for all analyses studying the sensitivity of predicted shelf vulnerability to:

• RCP.zip: (Representative Concentration Pathway): climate scenario
• AOGCM.zip: (Atmosphere-Ocean General Circulation Model): climate model
• Basal_Melt.zip: the parametrization used to convert ocean temperatures into a melt rate for the base of ice shelves
• Fracture_Toughness.zip: the material property the controls ice's resistance to forming crevasses

The files are organized according to Table 2 in the manuscript where each .zip is one of the analysis categories in the first column. Inside each .zip are folders named as [experiment]_[fracture toughness value]_[shelf]. For example, exp07_03_KIC200_Amery has the postproccesed results for exp07 and exp03 analyzed with a fracture toughness of 200 KPa m^1/2 for the Amery ice shelf.

In each folder there are fivesubfolders that show postprocessed plot from each ice sheet model:

• R_xx_delta: shows change in resistive stress through time for each ice sheet model
• R_xx_mifit: shows misfit of model initial state relative to 2014-2017 MEaSUREs with temperature from Comiso (2000) and rheology from Cuffey and Paterson (2010)
• R_xx_ND_exc: shows change in exceedance of the dimensionless resistive stress threshold through time
• R2_Rxx: shows per-grid point change in resitive stress versus change in thickness with the best fit line and predicted line by Equations (11, 12) in the manuscript
• thickness: shows change in thickness through time
• (In Test 2d versu 3d only) Comp_2D_v_3D: shows agreement or lack-there-of in calculating vulnerability between the calculations.

There are also the summary plots of vulnerable and non-vulnerable areas for all participants ([shelf name]_Rxx_ND_exc_area_summary.png) and per-pixel agreement of vulnerability ([shelf name]_model_exc_heatmap _[year].png). 

Finally, the spreadsheet [shelf_name]_summary.xlsx records analysis inputs and the raw outputs from the analysis run. Pages include:

• Models: records experiment used for each model, whether temperature was reported from model, and whether model was manually removed from averaged analyses
• Settings: records physical properties used
• Shelf areas: shelf areas through time (km^2)
• Vulnerable areas: vulnerable areas through time (km^2)
• Fraction vulnerable: true fraction of shelf that's vulnerable at each year
• Normed frac. vuln.: fraction of original shelf area that's vulnerable at each year (what's used for summary plots like Fig. 2)
• Models T test: shows which models were used for T test after manual removals and removals from shelf area dropping below threshold (typ. 0.8x original shelf area).
• Paired T test: t-test results
• R2 thk Rxx: R^2 values for best fit line and theoretical relationship (Equations 11,12) between resistive stress change and thickness change
• Corr per model: spatially averaged across shelf stress and thickness change correlation in each model from multiple experiments. Relevant when workflow used on all ISMIP6 results used in paper at once to make thickness correlation plots (e.g. fig 7). See version in Thickness_correlation.zip below. 

Results for Supplement: Workflow Sensitivity Checks and Additional Thickness Change Correlation 

These .zip files are:

• Stress_Error.zip: Reruns of the exp05/exp01 analyses used in the RCP and AOGCM studies with resistive stress reduced by 25% to study sensitivity to stress error caused by recalculating stress during postprocessing
  • Contains same workflow outputs as above for the Ross, Filchner-Ronne, Amery, and Larsen C shelves, experiments 05 and 01 run with stress reduced by 25%. Compare to exp05 epx01 reults in the RCP folder to see sensitivity, as was done in the supplement S1.
• Test_2d_versus_3d.zip: analyses using Rxx = 2txx rather than Rxx=2txx+tyy to assess the significance of stress calculation selection
  • Same four shelves and exp05/exp01. Has additional folder, "comp_2d_vs_3d" which contains plots showing, for each model, which pixels are vulnerable / safe in each calculation (supplement Section S3).
• Thickness_correlation.zip:
  • Has shelf averaged thickness change and stress change across experiments for each model for the same four shelves 

Code

The .zip file is:

• ISMIP6_shelf_vuln_code.zip

Main scripts:

• Final version used for sensitivity studies: shelf_stress_comp_multi_v6a_r6c.m
• Final version used for assesing vulnerability where the ISMIP6 collapse forcing was applied: shelf_stress_comp_multi_frac_v3_R1.m
• Final version for bar charts comparing experiments: shelf_stress_postprocess_v2_r1.m

The code was run on the University at Buffalo hpc, where ISMIP6 results are stored.