Authors: Kenneth D. Mankoff1,2, Fabien Maussion3
Affiliations:
1NASA Goddard Institute for Space Studies, New York, NY, 10025 USA
2Autonomic Integra LLC, New York, NY, 10025 USA
3School of Geographical Sciences, University of Bristol, Bristol, UK
License: CC-BY
Executive Summary
The International Association of Cryospheric Sciences (IACS) working group on the delineation of glaciers, ice sheets and ice sheet basins deliverable #1 aims to provide an overview of the state of ice sheet delineations from some of the more common and popular products. These include basic data products used by much of the glaciological community (e.g., BedMachine), observational-derived boundaries (e.g., RGI 7.0) some regional climate model masks, etc.
We reached out to the glaciological community (CRYOLIST announcement) and gathered a group of polar data experts, producers and users. We asked our members to provide information on their use of ice sheet and glacier masks and boundaries and collecte the information as GitHub Issues. The following report is based on this initial consultation and contains new analyses of the presently available datasets.
We provide an explanation for which products rely on which other products based on a community survey, and graphical statistical displays of the product union (X \union Y, or area overlap) and the product not-union (X \union Y’, or area of product X outside of product Y).
Products¶
We list all products either introducing or using a Greenland or Antarctic mask. A “mask” represents a gridded or vector files defining the spatial domain of a specific product.
This list is meant to be as exhaustive as possible. If a product is missing, please reach out.
Greenland¶
| Name | Data citation | Science citation | Details | Comments |
|---|---|---|---|---|
| MEaSUREs ITS_LIVE Greenland Monthly Ice Masks, Version 1 | Gardner (2023) | Greene et al. (2024) | IACS | |
| PISM | IACS | |||
| NORCE-CISM | IACS | |||
| MAR | IACS | |||
| NHM-SMAP | IACS | |||
| ISSM | IACS | |||
| MetUM | IACS | |||
| GlacierMIP | Hock et al. (2019) | IACS | ||
| RGI 7.0 | IACS | |||
| BedMachine | Morlighem et al. (2022) | Morlighem et al. (2017) | IACS | |
| PROMICE | Citterio & Ahlstrøm (2022) | Citterio & Ahlstrøm (2013) | ||
| ESA CCI | Harper et al. (2023) |
Antarctica¶
| Name | Data citation | Science citation | Details | Comments |
|---|---|---|---|---|
| MEaSUREs ITS_LIVE Antarctic Annual Ice Masks, Version 1 | Greene et al. (2022) | Using only 2020 | ||
| PISM | IACS | |||
| NORCE-CISM | IACS | |||
| MAR | IACS | |||
| NHM-SMAP | IACS | |||
| ISSM | IACS | |||
| MetUM | IACS | |||
| HIRHAM5 Antarctic ice mask | IACS | |||
| GlacierMIP | Hock et al. (2019) | IACS | ||
| RGI 7.0 | IACS | |||
| BedMachine | Morlighem (2022) | Morlighem et al. (2019) | IACS |
Discussion¶
The masks or boundaries themselves are rarely the primary focus of the work (with the exception of RGI, Greene et al. (2024), and a few others). For the remaining products that are not mask-producers but rather mask-users, the decision of which mask to begin with appears to be based on reasonable and justifiable decisions, but those decisions may be based on ease-of-use, familiarity, use of local or national products, etc. and not with significant consideration of overlap (or not) with peripheral products such as RGI, or ease-of-comparison with final downstream products.
Greenland¶
Several Greenlandic products use the BedMachine mask (which is based on the GIMP mask) or the GIMP mask directly, which at its release may have been the most complete, accurate, and highest resolution mask. However, the GIMP mask is now based on a 10 year old paper (Howat et al. (2014)) and seven-year-old product (Howat et al. (2017)), and the mask itself uses data spanning 15 years, which is problematic considering the large annual changes in Greenland.
A detailed examination of seven Greenlandic products (BedMachine, PROMICE, MAR, RACMO, ESA CCI, GIMP, and RGI region 05) is shown below for four regions: Qaanaaq, Sisimiut, near the Geike Plateau, and the southern tip of Greenland.
Figure 1:Overlap map of seven masks (BedMachine, PROMICE, MAR, RACMO, ESA CCI, GIMP, and RGI region 05) near Qaanaaq, Greenland. The six filled colors represent number of overlapping products when each is limited to the main (connected) ice sheet. The seventh product is RGI region 05 (peripheral Greenland, connectivity level 0 and 1) shown as a red outline. This graphic shows that some small areas that are in the RGI peripheral region are covered by two, three, or even five of the other main ice sheet masks. If the non-RGI masks include unattached glaciers, the number of overlap is even higher. Furthermore, regions of the ‘main ice sheet’ are defined differently in one mask (where overlap drops from six to five).
Figure 2:Same display as Figure 1, but here showing a large area within in the RGI peripheral region that is covered by four of the other main ice sheet masks.
Figure 3:Same display as Figure 1, but here showing only two small areas with overlap between RGI and two or three of the main ice sheet masks. The Geike Plateau (off figure to the southeast) shows one mask does not cover this area, as coverage drops from six to five.
Figure 4:Same display as Figure 1, but here showing the southern tip of Greenland. Large areas covered the RGI peripheral glacier product are also covered by three or four of the six mask products, and large areas not covered by RGI are only covered by three or five of the six mask products.
The above figure provides a qualitative view of overlapping masks in a few regions of Greenland. The figures below show a quantitative display of more masks and what areas of each mask either overlaps the other masks, or is outside of the other masks.
Figure 5:Overlapping areas between masks on X axis and on Y axis in Greenland. Scale (both number and color) is logarithmic. The diagonal is the area of a product (100 % overlap with itself). An example interpretation of this graphic follows the RGI column (first column) starting at the bottom row. The bottom left square reports 10 km2 (~2 km2) area overlap between RGI and areas flagged as “ice shelf” in BedMachine. The second row shows overlap between RGI and peripheral glaciers in the ‘harper_2023’ product is 10 km2 (~63096 km2). More interesting is overlap between RGI and the ‘main@’ rows, which are the primary ice sheet in those products. RACMO only overlaps RGI by 10 km2 (~2,500 km2) but most other products overlap by ~10x as much (seen by the difference between ~103 and ~104). The largest overlap is GIMP with 10 km2 (~25,000 km2).
Data available as a csv file.
An alternative view, rather than overlap, is area of product X outside of product Y. Because overlap is commutative (X overlap Y is the same as Y overlap X) the above figure is half empty. However, X not in Y is different than Y not in X, so the following figure contains twice as much information and occupies twice the area (same information density).
Figure 6:Non-overlapping areas between masks on X axis and on Y axis in Greenland. Scale (both number and color) is logarithmic. The diagonals are 0 because a product cannot be outside itself. An example interpretation of this graphic follows the RGI column (first column) starting at the bottom row. The bottom left square reports 10 km2 (~4,000 km2) area of BedMachine ice shelf outside of the RGI peripheral glaciers. The second row shows the ‘harper_2023’ peripheral glaciers have 10 km2 (5,000 km2) outside of the RGI peripheral glaciers. Because this graphic is log scale, a change between two values of magnitude 1 represents ~10 % of the original values, and magnitude 2 (e.g. 105 to 103) represents 1 %.
Data available as a csv file.
Antarctica¶
The state of masks in Antarctica is less homogeneous than in Greenland. The BedMachine mask uses unpublished data (see IACS
A detailed examination of six Antarctica products (BedMachine, ESA CCI, Rignot (IMBIE), Greene et al. (2022), NSIDC 0709, and RGI region 19) is shown below for two regions: Wilkins ice shelf and Abbott ice shelf.
Figure 7:Overlap map of six masks (BedMachine, ESA CCI, Rignot (IMBIE), Greene et al. (2022), NSIDC 0709, and RGI region 19) near Wilkins ice shelf, Antarctica. The five filled colors represent number of overlapping products when each is limited to the main (connected) ice sheet. The sixth product is RGI region 19 (peripheral Antarctica) shown as a thin red outline. This graphic shows one product includes the ice shelves and coastal islands.
Figure 8:Same display as Figure 7, but near Abbott, Pine Island Glacier, and Thwaites ice shelves. This graphic shows a) general outline misalignment, and b) One product includes the ice shelves and coastal islands.
The above figure provide a qualitative view of overlapping masks in two regions of Antarctica. The figures below show a quantitative display of more masks and what areas of each mask either overlaps the other masks, or is outside of the other masks.
Figure 9:Overlapping areas between masks on X axis and on Y axis in Antarctica. Scale (both number and color) is logarithmic. The diagonal is the area of a product (100 % overlap with itself). Of interest, the dark (low values) in the column over ‘peripheral@RGI’ shows overlap between RGI peripheral region and all other products. This overlap is sometimes small (~103 or 1000 km2), but sometimes 10 or 100 times as large (e.g. 105 km2).
Data available as a csv file.
An alternative view, rather than overlap, is area of product X outside of product Y. Because overlap is commutative (X overlap Y is the same as Y overlap X) the above figure is half empty. However, X not in Y is different than Y not in X, so the following figure contains twice as much information and occupies twice the area (same information density).
Figure 10:Non-overlapping areas between masks on X axis and on Y axis in Antarctica.
Data available as a csv file.
Way forward¶
The following are suggestions, and we seek feedback from working groups and team members. This is not yet a consensus document, nor a plan to be implemented. This section will be published on-line and then linked to a GitHub discussion, then a final edit based on feedback in the discussion.
Greenland¶
We can address the problem of double counting in Greenland through the following approach using new and updated products.
In mid 2025 GEUS will be releasing an update to Citterio & Ahlstrøm (2013) or a new outline of the conterminous Greenlandic ice sheet based on 2022 Sentinel imagery. We recommend using this as a broader community definition of ‘ice sheet’. RGI will update to exclude overlap, and become the definition of ‘peripheral glaciers’. We will recommend that popular products such as BedMachine and RCMs update their domains, boundaries, or masks to match these two products to avoid double counting.
This plan does not address several important but separate issues raised by this working group, specifically:
- The connectivity of ice categorized as part of the GEUS ice sheet, including ice that appears visually connected in satellite imagery but may not be connected hydrologically, dynamically, or by other definitions.
- The lack of defined interior ice sheet basins, as existing basins (e.g., Zwally, Rignot, Mouginot) may not cover the entire ice sheet based on the new boundary.
These and other secondary issues will be addressed at a later stage.
Antarctica¶
Antarctica is a more complicated environment. There is an equivalent to the GEUS 2022 Greendlandic outline found in Greene et al. (2024) which provides annual masks, but these masks include but do not distinguish ice shelves.
The following recommendations come to mind. In the following, “RGI 19” and “RGI 20” refer to the masks of region 19 and 20 in the RGI region files. Changes to these would imply altering the assignment of glacier outlines to these regions.
Use BedMachine as baseline¶
- Update BedMachine mask metadata (only metadata) to mark conterminous ice, ice shelves, and peripheral (unattached) ice.
- Update the outlines of the RGI 19 and RGI 20 glacier regions to match BedMachine conterminous and ice shelves
- Update RGI 19 region outlines to include all BedMachine peripheral and exclude the updated RGI 20 (BedMachine conterminous ice).
- Ideally, BedMachine peripheral is updated to match RGI 19 region outline, but prior discussion suggests this is not likely to happen, in which case RGI 19 region outline is likely to be a superset of BedMachine peripheral.
- Recommend to community that they use the region masks definitions provided by RGI region files. The reason to recommend RGI and not BedMachine is that users of BedMachine may include peripheral ice, which would lead to doule counting or under counting (if included in larger assessments that either include RGI 19, or assume that the Antarctica effort includes RGI 19).
Use Greene et al. (2024) as baseline¶
- Select a Greene et al. (2024) year, either 2000 (a very round number), 2020 (a recent round number) or 2022 (to match Greenland).
- Update Greene et al. (2024) metadata to delineate ice shelves.
- Update RGI 20 region files to match the updated Greene product.
- Update RGI 19 region file to avoid overlapping with RGI 20, and to include any unattached ice not in the updated Greene product (perhaps using BedMachine unattached ice).
- Recommend to community that they use RGI 20 regions or the updated Greene product for main ice sheet, and add in RGI 19 if they want peripheral ice.
Initial thoughts on the consequences of these recommendations¶
The proposed changes will likely affect the glacier and ice sheet communities differently.
The ice sheet community will have to update their masks to match the new standard, but this is a one-time effort for each sub-community (ISMIP for models, IMBIE for observations...) with very clear benefits. It must be added that despite of the benefits of having a unified definition of the ice sheet, neither ISMIP nor IMBIE have yet adopted a standard mask so far. Our hope is that this WG would help in this regard. Initial discussions seem to indicate that both IMBIE and ISMIP are interested in adopting a standard mask, but there seems to be challenges too complex to be summarized here. Regardless, both IMBIE and ISIMIP should be responsible for enforcing the use of a standard mask in their respective efforts (e.g. during post-processing if required), regardless of this WG’s recommendations.
For the glacier community, the changes as outlined above would mean that roughly ~80% of the glaciers in region 19 (Antarctic Periphery) and ~25% of the glaciers in region 05 (Greenland Periphery) would now be “part of the ice sheet” and eventually removed from RGI. This would have a significant impact on the future results of the glacier community, and would require a significant effort to update and communicate these results. This is mainly because in terms of relative area, the change is considerably larger for glaciers than it is for the ice sheets. Another reason is that the glacier community has been following the RGI standard for a long time, and results on e.g. sea-level rise contributions of glaciers in Antarctica and Greenland have been consistent over the years.
A strategy to communicate these changes to the public and community will be necessary. Fortunately, the changes could be done gradually. An ideal scenario could be:
- IMBIE and ISMIP adopt the new standard mask for both ice sheets for their next assessments (ISMIP7, next iteration of IMBIE, etc.). Hopefully this should not change their results in a way imputable only to the adoption of the new mask, so that the change goes mostly unoticed (we could be wrong with this assessment).
- Glaciers now classified as part of the ice sheet are flagged in the next RGI iteration (v7.2). These glaciers could either remain in the RGI files with a new attribute indicating their status or be removed, with a separate file provided for the removed outlines.
- The glacier community is encouraged to produce results for the entire RGI but to communicate distinctions clearly. For example, global tables like those in Hugonnet et al. (2021) or Rounce et al. (2023) could include additional rows for the “previously glaciers, now ice-sheet” contributions from regions 05 and 19.
- If the ice sheet community adopts a similar distinction (e.g., separating contiguous ice from “peripheral ice-sheet”), results from both communities could be compared, enhancing our understanding of methodological differences and allowing accross-community MIPs.
- While changes to the RGI could occur within a year, the gradual communication of updated results may take considerably longer (e.g., RGI7 is still not widely adopted one year after publication).
Given the size of the respective communities and the historically strong community adoption of the RGI, changing RGI is still the scenario which is most likely to trigger the necessary change. Potential pushbacks that could arise are:
- The argument that peripheral glaciers behave like glaciers, so why don’t the BedMachine or Greene products reflect this?
- Concerns that ice sheet models are less accurate for peripheral glaciers, potentially biasing results.
Personally, we (Ken & Fabien) believe these objections are valid but outweighed by the overarching goal of avoiding double counting in future IPCC reports and quantitative studies. We probably have missed many important aspects of the proposed change, and we are looking forward to the feedback from the community.
- Gardner, A. (2023). MEaSUREs ITS_LIVE Greenland Monthly Ice Masks, Version 1. NASA National Snow. 10.5067/579TO87M7IZB
- Greene, C. A., Gardner, A. S., Wood, M., & Cuzzone, J. K. (2024). Ubiquitous acceleration in Greenland Ice Sheet calving from 1985 to 2022. Nature, 625(7995), 523–528. 10.1038/s41586-023-06863-2
- Hock, R., Bliss, A., Marzeion, B., Giesen, R. H., Hirabayashi, Y., Huss, M., Radić, V., & Slangen, A. B. A. (2019). GlacierMIP – A model intercomparison of global-scale glacier mass-balance models and projections. Journal of Glaciology, 65(251), 453–467. 10.1017/jog.2019.22
- Morlighem, M., Williams, C., Rignot, E., An, L., Arndt, J. E., Bamber, J., Catania, G., Chauché, N., Dowdeswell, J. A., Dorschel, B., Fenty, I., Hogan, K., Howat, I., Hubbard, A., Jakobsson, M., Jordan, T. M., Kjeldsen, K. K., Millan, R., Mayer, L., … Zinglersen, K. (2022). IceBridge BedMachine Greenland, Version 5. NASA National Snow. 10.5067/GMEVBWFLWA7X
- Morlighem, M., Williams, C. N., Rignot, E., An, L., Arndt, J. E., Bamber, J. L., Catania, G., Chauché, N., Dowdeswell, J. A., Dorschel, B., Fenty, I., Hogan, K., Howat, I., Hubbard, A., Jakobsson, M., Jordan, T. M., Kjeldsen, K. K., Millan, R., Mayer, L., … Zinglersen, K. B. (2017). BedMachine v3: Complete Bed Topography and Ocean Bathymetry Mapping of Greenland From Multibeam Echo Sounding Combined With Mass Conservation. Geophysical Research Letters, 44(21). 10.1002/2017gl074954
- Citterio, M., & Ahlstrøm, A. P. (2022). Ice extent. GEUS Dataverse. 10.22008/FK2/PRWITW
- Citterio, M., & Ahlstrøm, A. P. (2013). Brief communication “The aerophotogrammetric map of Greenland ice masses.” The Cryosphere, 7(2), 445–449. 10.5194/tc-7-445-2013
- Harper, K. L., Lamarche, C., Hartley, A., Peylin, P., Ottlé, C., Bastrikov, V., San Martín, R., Bohnenstengel, S. I., Kirches, G., Boettcher, M., Shevchuk, R., Brockmann, C., & Defourny, P. (2023). A 29-year time series of annual 300 m resolution plant-functional-type maps for climate models. Earth System Science Data, 15(3), 1465–1499. 10.5194/essd-15-1465-2023
- Greene, C. A., Gardner, A. S., Schlegel, N.-J., & Fraser, A. D. (2022). Antarctic calving loss rivals ice-shelf thinning. Nature, 609(7929), 948–953. 10.1038/s41586-022-05037-w
- Morlighem, M. (2022). MEaSUREs BedMachine Antarctica, Version 3. NASA National Snow. 10.5067/FPSU0V1MWUB6
- Morlighem, M., Rignot, E., Binder, T., Blankenship, D., Drews, R., Eagles, G., Eisen, O., Ferraccioli, F., Forsberg, R., Fretwell, P., Goel, V., Greenbaum, J. S., Gudmundsson, H., Guo, J., Helm, V., Hofstede, C., Howat, I., Humbert, A., Jokat, W., … Young, D. A. (2019). Deep glacial troughs and stabilizing ridges unveiled beneath the margins of the Antarctic ice sheet. Nature Geoscience, 13(2), 132–137. 10.1038/s41561-019-0510-8
- Howat, I. M., Negrete, A., & Smith, B. E. (2014). The Greenland Ice Mapping Project (GIMP) land classification and surface elevation data sets. The Cryosphere, 8(4), 1509–1518. 10.5194/tc-8-1509-2014
- Howat, I., Ohio State University, & Byrd Polar Research Center. (2017). MEaSUREs Greenland Ice Mapping Project (GIMP) Land Ice and Ocean Classification Mask, Version 1. NASA National Snow. 10.5067/B8X58MQBFUPA