Antarctic Ice Shelf Calving, 1997-2021

This repo contains Antarctic coastline data and the MATLAB scripts we developed to use all available observations we can find, to build an uninterrupted, high-resolution coastline. Our findings are described in Greene et al., 2022.

Here's an animation of the final product:

Figures & Animations

The animations folder contains:

• 19 regional animations of coastal evolution from 1997 to 2021.
• 1 animation of the modeled velocity response to hypothetical ice shelf collapse scenarios.

The regional maps folder contains:

• 19 regional high-resolution static maps of coastal evolution from 1997 to 2021.

The figures folder contains:

• the submitted versions of every figure from the Greene et al., 2022 manuscript.

Data

Final Data Products

• Text files of ~annual Antarctic coastlines. The decimal year corresponding to each mapping is in each filename.
• Gridded ~annual Antarctic coastline masks at 240 resolution can be found in icemask_composite.nc or icemask_composite.mat if you prefer MATLAB's .mat format.
• greene_iceshelf_area_and_mass.xlsx contains time series of every ice shelf's area and mass.
• calving_flux_timeseries.mat contains a .mat form of the data that goes into the Excel spreadsheet.
• The issm_gl_flux_strict.mat contains modeled grounding line flux responses to various test cases. In it the glf_0 variable is the control run. glf_cp contains the response to past calving, glf_cf contains the reponses to hypothetical future calving from 0 to 100 percent ice shelf loss by area. glf_tp is the response to past thinning, and glf_tf2 is the respones to hypothetical future thinning (where the very last entry is the 1 m ice shelf case everywhere).

Intermediate Data Products

Extruded velocity and thickness fields are currently posted here, and are described below.

Note to self: These scripts work together to create a dataset extruded_antarctica_*.h5, which is currently on devon at /mnt/devon-r2/shared_data/greene.

The data in the h5 file and in antarctica_icemasks_extruded_velocity_and_thickness.nc include:

• vx, vy: Velocity components extended beyond present-day coastlines to fill the entire map. Velocity magnitude is constantly extrapolated from the perimeter of observed velocities. Velocity direction is obtained by the Matlab inpaint_nans algorithm applied to measured velocity multiplied by measured thickness.
• v_source: Sources of the velocity components are:
  1. ITS_LIVE (Gardner et al., 2018),
  2. MEaSUREs v2 (Rignot et al., 2017),
  3. Error-weighted mean of ITS_LIVE and MEaSUREs v2.
  4. Interpolation, and
  5. Extrapolation.
• thickness: Ice thickness extrapolated beyond present-day coastlines to fill the entire map.
• thickness_source: Sources of the thickness data are prioritized in the following order, which uses the newest data available, and fills in recently calved areas (such as Larsen B) with older datasets. Mean firn air content from the Glacier Energy and Mass Balance (GEMB) model (ISSM and Gardner's thesis) was subtracted from surface elevation datasets before inverting for thickness (multiply geoid-referenced surface by 9.3364). The BedMachine thickness product is the basis, but all other "thickness" observations are inverted surface elevations.
  1. BedMachine Antarctica (Morlighem et al., 2020),
  2. REMA - GEMB,
  3. Bedmap2 - GEMB,
  4. Bamber - GEMB,
  5. RAMP2 - GEMB
• iceshelf_mask: Mask of 181 ice shelves from Mouginot et al., 2017, dilated by iceshelf_mask_generator.m and extruded by flow_dem_extend.m.

Script Workflow

1. iceshelf_mask_generator.m uses Mouginot's iceshelves_2008_v2 outlines to create iceshelf_mask.mat, which contains a 240 m resolution mask on the ITS_LIVE. This script also dilates the ice shelf mask by 100 km to account for any possible ice shelf growth. This mask is not a final product! It is modified by flow_dem_extend.m to create the iceshelf_mask in the extruded_antarctica_*.h5 file.
2. flow_dem_extend.m creates reference ice thickness and velocity grids that extend 100 km beyond the ~2008 extents of the ice sheet. Velocity is obtained by combining ITS_LIVE and MEaSUREs v2 data as an error-weighted average of the two datasets, then velocities are extrapolated as constant values ~100 km beyond the calving front, in the direction of ice flow. Ice thickness is inverted from BedMachine v2 surface elevations (which are REMA - FAC). Where calving occurred before the ~2014? nominal date of BedMachine and BedMachine contains 0 thickness, surface elevations are taken from REMA, then Bedmap2, then Bamber DEM, then RAMP2, and from each of these non-BedMachine surface elevations, FAC from GEMB is subtracted.
   
3. extruded_antarctica_plotter.m creates the five maps shown above.
4. icemask_modis.m reads and interpolates Alex Fraser's annual modis coastline. Writes the resulting ice cube to a file called icemask_modis.mat.
5. icemask_s1a.m reads and interpolates Alex Fraser's annual ice mask that was generated from Sentinel 1a. Writes the resulting ice cube to a file called icemask_s1a.mat.
6. icemask_ramp2.m masks the its-live grid using the 1997 AMM-1 (after_coast_continuous.shp) and 2000 MMAM (cst2000line.shp) annual modis coastline. Writes the resulting ice cube to a file called icemask_ramp2b.mat.
7. icemask_moa.m masks the its-live grid using the 2004, 2009, and 2014 coastlines. Writes the resulting ice cube to a file called icemask_moa.mat.
8. icemask_compiler.m Combines masks from individual sensors to create a composite, fully consistent ice cube. This script creates icemask_composite.mat, which is used in the ISSM experiment on grounding line flux response to observed calving.
9. icemask_analysis.m Analysis of coastal extent change.
10. icemask_animator.m Creates regional gifs and still maps of the observed calving tha occurred from 1997-2021.
    
11. icemask_buffer.m creates an "ice cube" of masks that buffer inland in small increments, 1% of each ice shelf's area at a time. This script creates icemask_buffered.mat, which is used in the hypothetical calving ISSM experiment.
12. iceshelf_thickness_cube_generator.m creates thickness cubes on the 240 ITSLIVE velocity grid, annually for for the observed period 1992-2018, and every two years for the next 50 years, if trends from the past quarter century continue. (Trends from the pole hole only covered by CryoSat-2 are calculated for 2010-2018). This script creates `iceshelfthicknesscube1992-2018.matandiceshelfthicknesscube_future.mat`, which are for the ISSM past and "future" thickness-change experiments.
    
13. issm_calving_melt_setup.m sets up the geometry to initialize the ISSM experiment. This script creates issm_calving_melt_setup.mat.
14. calving_flux_calculation_cartoon.m creates this figure to explain how steady-state calving flux is calculated:
15. issm_thickness_response_analysis.m converts the modeled response velocities at each ISSM node into grounding line flux. This script creates issm_gl_flux_strict.mat, which is called by all the other ISSM plotting scripts.
    
16. gridded_data_resave.m saves the time-evolving ice mask and the extruded thickness and velocity fields as NetCDFs.

Functions and other files

• inpolygon_map() is a much faster version of inpolygon and is used by iceshelf_mask_generator.m to mask ice shelf boundaries.
• iceshelves_2008_v2_names.csv contains the names and numerical indices of all 181 ice shelves in the extended ice shelf mask.
• README_issm_model_setup_files.txt Notes on the ISSM experiment setup with descriptions of the variables and how they were created.
• reference_coastline_logbook.xls a reference to help me keep straight which indices of which ice cube masks correspond to which years.

Citation

If you use this data, please cite our Nature paper!

Greene, C.A., Gardner, A.S., Schlegel, NJ. et al. Antarctic calving loss rivals ice-shelf thinning. Nature (2022). doi:10.1038/s41586-022-05037-w

Acknowledgments

This work was supported by the NASA Postdoctoral Program and the NASA MEaSUREs ITS_LIVE project.