Ice thickness and subglacial topography of Swedish reference glaciers revealed by radio-echo sounding

Wang, Zhuo; Ross, Neil; Frank, Thomas; Barnett, Jamie; Santin, Ilaria; Houssais, Martin; Dahlkvist, Johanna; Kirchner, Nina

doi:10.5194/essd-2025-745

Preprints

https://doi.org/10.5194/essd-2025-745

Preprints

21 Jan 2026

| 21 Jan 2026

Status: this preprint is currently under review for the journal ESSD.

Ice thickness and subglacial topography of Swedish reference glaciers revealed by radio-echo sounding

Zhuo Wang, Neil Ross, Thomas Frank, Jamie Barnett, Ilaria Santin, Martin Houssais, Johanna Dahlkvist, and Nina Kirchner

Abstract. Sweden currently hosts 270 glaciers, four of which belong to the 61 reference glaciers monitored worldwide. Eight Swedish glaciers disappeared during the warm summer of 2024, and under the global warming scenario associated with current climate policies, all four Swedish reference glaciers (Mårmaglaciären, Storglaciären, Rabots glaciär, and Riukojietna) are projected to vanish within this century. Such change will have implications for people, ecosystems, infrastructure, and local meteorological processes, highlighting the need to better constrain the resultant emerging post-glacial landscapes. During 2024–2025, we conducted radio-echo sounding (RES) surveys on the four Swedish reference glaciers and obtained a total of 40470 ice thickness point measurements. The mean and maximum measured ice thicknesses are 97 and 242 m for Mårmaglaciären, 88 and 225 m for Storglaciären, 85 and 158 m for Rabots glaciär, and 32 and 87 m for Riukojietna. The corresponding mean ice-thickness uncertainties are 12.0, 12.4, 11.4, and 7.5 m, respectively. The RES-measured ice thicknesses were used to produce high-resolution (10 m × 10 m) maps of ice thickness distribution and subglacial topography for each reference glacier, and to calculate their ice volumes as 0.32 (Mårmaglaciären), 0.25 (Storglaciären), 0.23 (Rabots glaciär), and 0.10 km³ (Riukojietna). The RES data for the four reference glaciers are available at https://doi.org/10.17043/tarfala-marma-res-survey-1, https://doi.org/10.17043/tarfala-storglaciaren-res-survey-1, https://doi.org/10.17043/tarfala-rabot-res-survey-1, and https://doi.org/10.17043/tarfala-rivgojiehkki-res-survey-1 (Wang et al., 2026b, d, a, c). The ice thickness and subglacial topography for the four reference glaciers are available at https://doi.org/10.17043/tarfala-marma-res-2, https://doi.org/10.17043/tarfala-storglaciaren-res-2, https://doi.org/10.17043/tarfala-rabot-res-2, and https://doi.org/10.17043/tarfala-rivgojiehkki-res-2 (Wang et al., 2025a, d, b, c).

Received: 04 Dec 2025 – Discussion started: 21 Jan 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Zhuo Wang, Neil Ross, Thomas Frank, Jamie Barnett, Ilaria Santin, Martin Houssais, Johanna Dahlkvist, and Nina Kirchner

Status: final response (author comments only)

RC1:
'Comment on essd-2025-745', Moritz Koch, 12 Mar 2026
Revision for:
Ice thickness and subglacial topography of Swedish reference glaciers revealed by radio-echo sounding – submitted to ESSD by Wang et al. 2026
The paper presents ground penetrating radar surveys for the four glaciers Mårmaglaciären, Storglaciären, Rabots glaciär, and Riukojietna, located in northern Sweden. These glaciers are part of the global reference glacier monitoring network and are expected to disappear within this century under current warming scenarios, making accurate knowledge of their geometry critical for modelling future glacier evolution and hydrology. The data were collected in 2024 and 2025 using a ground-based GPR system mounted on a snowmobile. The surveys cover almost 250 km of profiles and thus provide densely gridded observations of the bedrock topography. The authors then produced ice thickness distribution maps with a model well-suited to deal with such a density of observations. This is an important advancement in the knowledge of the volume and subglacial topography of these glaciers, especially as they are a) part of the global reference glaciers and b) expected to vanish within this century. Having accurate and reliable information on the ice thickness distribution of these glaciers is therefore essential. Given the importance of the measurements, I support publishing this work in ESSD, as it provides an essential boundary condition for future modelling of glacier dynamics, hydrology, and landscape evolution in a rapidly deglaciating environment. I also very much appreciate that the raw and processed GPR data are availiable.
Below are some suggestions that, from my perspective, could improve certain aspects of data presentation and the manuscript's readability. While I am not very familiar with the region, I approached the article as a glaciologist engaged with GPR applications.

General comments:
In the introduction, the authors mention previous GPR surveys of the glaciers; I understand that these measurements were carried out in the late 80s and 90s and are not necessarily as accurate as today's surveys. However, I believe it would be very valuable to compare the newly obtained measurements with the earlier studies in this paper. There is also a detailed overview of the methods previously used, and it would be very interesting to build on that.

You include uncertainty maps in the database, but they are not shown or discussed in the text. When I first read through the article, I thought they weren’t provided. Clarify that somewhere in the manuscript or, better still, include them as figures and discuss the spread of the uncertainty ranges.

Specific comments:

L 7: While measurement points are an interesting metric, how much spatial coverage do the radar transects in their entirety have? I think this should be mentioned here.
L 13: Is it necessary to provide the data in individual repositories? If not, I would recommend merging them and offering the dataset in a single repository; citing four different repositories seems cumbersome to me.
L 28: I am confused by the citation Houssais et al., in prep. Since there is no preprint available, there is no way to evaluate how and on what basis that point was made. If it will be available as a pre-print likely during the revision process, I don’t think that’s a problem. If not, I would reconsider citing it here.
L 34: Insert “the” before “present-day”
L40: I think there should be a citation at the end of that statement.
L43: … “with uncertainties ranging from a few meters to several tens of meters.” This is true for comparably shallow glaciers, but I would argue that the realistic uncertainty range can be significantly higher when ice thickness values are large. Especially in deep temperate glaciers, airborne RES can yield substantially higher uncertainties.
L75: maybe rephrase to: … merge into a comparably flat plateau before …
L80: … and a generally more negative mass balance? (I assume all glaciers have negative SMBs)
L87: The inset showing the location of the glaciers in Fig. 1a and the glacier outlines in Fig. 1b are barely visible. I suggest changing the colour scheme or increasing the figure's contrast size.
L94: Table1. While I think it is very nice to have such a detailed overview of the previous work conducted at the reference glaciers, have you done any comparison of your findings with the data from previous surveys? (This relates to my general comment 1.)
L96: During which season did you conduct the studies? I ask because this could influence the properties of the snow, which, depending on the depth of the snow cover, could affect the uncertainty of the measurement.
L115: I would rephrase this sentence to mention that the temperate areas of the glacier contain ice at its pressure melting point and thus water-filled cavities, pockets, and tables, which is what is visible in the radargram and also absorbs more energy than cold ice due to differences in dielectric permittivity.
L118f: It would be very interesting to know the parameters which you applied in the mentioned processing steps in Reflex (where it applies).
L122: “The radargrams have been correctly scaled according to the trace spacing recorded by the GPS.” I don’t fully understand that statement. Are you referring to matching the radar data (traces) to the coordinates you recorded with the GPS?
L124: While I agree with the subjective bias in RES data interpretation, we are not yet at the stage where human labelling is surpassed by automated methods (especially in polythermal glaciers). For example, if a single annotator selects the bed reflection, there should be some consistency within the dataset. I agree, however, that a combined approach is likely more time-efficient, but it would be interesting to know how many human annotators identified the bed reflections.
L135f: “The RES data after static correction only were first used for manual repicking.” Can you elaborate what that exactly means?
L139: “We calculated the point ice thickness …” (insert)
L160: How thick do you estimate the snow cover to be? Radio-wave velocities can travel considerably faster in dry snow; have you already accounted for that in your error budget?
L173: Prior you mentioned an acquisition velocity of 4-6 m/s. Here, you write 4 m/s. Is this based on the average, or did you choose the lower end for the error assessment?
L202ff: You mention that you have repeated the modelling with the lower and upper bounds of the measurement uncertainty for the ice thickness reconstruction. I think it would be very valuable to share the results of these modelling runs as uncertainty maps for distributed ice thickness fields. (This relates to my general comment.)
L219: On one hand, I think it's helpful to see the point value of each uncertainty source, but I found it quite difficult to interpret the distribution on the map. Perhaps you could create a four-panel plot for the measured ice thickness, with one plot for the total point uncertainty and move the other panels to the appendix? I also recommend fixing the data range for each variable to make the plots more straightforward to interpret.
L235: “the ice thickness…”
269: I think that is a very interesting point. Can you give the overestimation of the other approaches in %?
271: Since the surface lowering of these glaciers is very well known, are the larger reported values within a range you would expect?
278: I would remove “ground truth”.
290: I would not argue that this statement is incorrect, but perhaps note that global or regional studies often use coarser data, trading resolution for broader coverage.
306: Remove “clearly”.
312: change “great” to “good” or “very good” / “high”.
312: How well do the two datasets compare to each other? Can you provide an example here?
L313: Fig 6b: What do the black numbers mean?
L321: Maybe replace ground-truth with “newly obtained ice thickness measurements”?
L347: That is a very cool figure!
Citation: https://doi.org/10.5194/essd-2025-745-RC1
RC2:
'Comment on essd-2025-745', Anonymous Referee #2, 12 Mar 2026
Wang et al. present an extensive dataset of ice thickness measurements of the four reference glaciers included in the worldwide glacier monitoring in Sweden, collected during 2024-2025. The measurements cover the glaciers with an impressive density, resulting in all points of the measured glaciers being no further than 210 m from a measured point. Previous radio-echo sounding (RES) studies that have been conducted on all four glaciers are well documented in the manuscript, but none of them cover them in the same timely matching manner with high resolution. The RES based ice thickness point measurements are used as input to produce model-based ice thickness distribution and bed topography maps. Processing and analysis steps as well as uncertainty estimations are clearly described and documented, contributing to the transparency and reproducibility of the presented data. The manuscript is well written with a clear and logic structure, supported by appropriate figures and tables.
Overall, I was excited to read this article and want to commend the authors for obtaining such an extensive dataset of the Swedish reference glaciers and presenting both the manuscript and the dataset itself in a clear, and transparent way. All in all, I am very positive to this manuscript, and the presented data provide a valuable improvement of ice thickness estimates for the four glaciers, which can be used for future modelling studies, for predicting proglacial landscape changes and to study potential hazards connected to glacier retreat. I have only some minor comments and questions on the processing steps and presentation of the figures.
Section-wise general comments:
Study area: It would be nice if you could add information about the elevation range and equilibrium line altitude if available. For example, starting on L73

On the data processing workflow description on 3.2: The description of the data processing steps is not entirely clear and reproducible as it is now. Which 2-D filter/filters did you apply? Which gain filter (e.g. AGC gain or energy decay?) and which bandpass filter did you use? I also wonder why you applied migration only for RIV? Migration is a particularly important filter to collapse diffraction hyperbolas which are typically more prominent in steep terrain. Given that RIV is the only ice cap type glacier in your dataset, while SG, MG and RG are valley glaciers I would expect those to be more prone to dipping reflectors.

Another question about the data processing I have is whether you corrected for antenna distance when processing the data (dynamic correction in ReflexW or NMO in other processing softwares and packages)? This step is crucial when working with relatively low frequency/large wavelength antennas (15 m antenna separation and 16.8 m central wavelength in this case) as the measured TWTT considerably differs from the vertical TWTT, especially at more shallow depths.

In 3.2 RES data processing and interpretation, you describe the steps for automated preliminary glacier bed picks, however I could not find the code for the automated picking in the Zenodo code repository. Maybe it is there but hard to find? If that is the case, consider expanding the readme file to refer to the file containing the automated picking workflow. If the picking workflow is not in the code repository, consider adding it.

Conclusion: Add uncertainties in ice thickness measurement and distributed ice thickness values

Specific comments:
L7-9: Include uncertainties on mean ice thickness values.
L9: Consider rephrasing “RES-measured ice thicknesses” to “RES-derived ice thickness measurements”
L19: Meanwhile implies an opposing argument. The argument here is however supporting the previous one. Replace with similarly, similar or delete.
L23: Haualand et al. (2025) would be another recent example of studies on local meteorological processes.
L24: It is not entirely clear what you mean by “disappear entirely at equilibration”. Please clarify.
L39: Check the citation. To my knowledge it should be Johansson et al. 2022.
L41: Replace “integration of these two approaches” with “integration of the two”.
L43: Move reference to end of sentence, except if it does not mention uncertainties. If that is the case, you should add a reference stating the typical range of uncertainties.
L46-51: When listing the main previous studies on glacier ice thickness inversion and/or physical constraints, Millan et al. 2022 should be mentioned somewhere for completeness.
L70: Reference not needed. Delete in prep. References as this are grey literature and cannot be checked for validity. It should therefore not be referenced. Same for L193.
L76: Delete “downslope”. Surface elevations decrease downslope per definition.
L78: replace “plateau-glacier” with “small ice cap” to be conform with WGMS glacier type classification
L80: What do you mean by: “generally negative mass balance across the glacier”? Please clarify. More negative than the other glaciers or did the glacier lose its accumulation zone. Please add mass balance trends quantitatively for e.g. the last 30 years for all glaciers.
L100: Please add the centre frequency of the antenna you used, 10MHz.
L101: What is the reason that you used 125MHz sampling rate for all glaciers except RG where you used 250MHz?
L110: Could you add the total length of RES surveys to table 2 as well?
L114: You write, “in polythermal glaciers, the cold surface layer has few reflectors or scatters”. Please rephrase cold surface layer to cold layers as not all polythermal glaciers have a cold surface layer and temperate bottom layer, it can also be the other way around or they can have a cold layer in the middle. Also, add a reference to this statement e.g. (Pettersson et al. 2004).
L118: Delete link in software citation here and add it in the reference list
L121-122: It is not entirely clear to me what you mean with this sentence. Consider rephrasing it to make it clearer.
L149: add “, “ after “In total”
L242: Add “by” between observations and Holmlund.
L259-262: This sentence it very long. Consider splitting it into two sentences for better readability.
L274-276: This sentence is not entirely grammatically correct. Split into two sentences such that the second sentence becomes: Rendering the modelled thicknesses in such areas is unreliable.
L278: Delete ground-truth. RES derived ice thicknesses are, although measurements, still indirect.
L280: What do you mean by “inaccurate mass balance data”? Please clarify. What data are you referring to?
L290: Keep present tense here for consistency.
L303-304: Delete “other”. Consider rephrasing the sentence to improve grammar: Unlike the (three) valley glaciers, the glacier bed at RIV does not vary substantially in elevation.
L306-307: Please refer to figure containing the bed and surface elevation plots of selected profiles -> Figure 7.
L307-309: It is not entirely clear to me what you want to emphasise with this sentence. Consider rephrasing for more clarity and conciseness.
L315: What do you mean with “finer short-wavelength features”? Please clarify.
L321: Replace “The ground-truth ice thicknesses” with “Measured ice thickness”.
L327-329: Put references in either chronological or alphabetical order.
L335-L337: What do you mean by “in the future post-glacial period”? Please clarify and add references to support the statements.
L342: Delete first m in 10 m x 10 m.
L347: Delete “High”

Figures and tables:
Figure 1:
Consider adding a north arrow in panel (a).

In panel (b), consider using a different color for the glacier outlines or use a fill color. It is not well-distinguishable from the background.

In panel (c), the purple line is difficult to distinguish from the black lines. Consider using a lighter purple.

On the figure caption and general information: ©Esri (Esri, 2025) -> please add information about which background map you used, e.g. Worldmaps?

The maps have grids with projected coordinates; however the figure caption lacks the information about which coordinate system is used here. I see that this is included in the dataset publications, please add it in the figure caption here as well.

Why do you use Sentinel-2 L1C imagery as background maps instead of L2A products? I believe that using L2A products could improve the maps as L2A products are atmospherically corrected and thereby better represent the surface reflectance. Consider changing this for all maps where you use Sentinel-2 products as background maps (Figure 6, 8). In addition, it would be helpful for the reader if you would specify which year the Sentinel-2 images are from.

Delete Houssais in prep. reference

Table 3: Do the numbers given in the “number of crossovers” column represent the number of single crossovers or the number of points (less than 5 m apart) included in the crossover analysis? Please specify. It seems like large numbers if it is single crossovers.
Figure 3: Here I have three comments,
consider using a discrete color scale to improve the visibility

please use the same limits for the color-scale for all plots in one row (i.e. a-d) to improve comparability. It is very difficult to visually assess the meaning of the colors when the scale is different in each plot

personally, to me it would seem more logical/intuitive if the color-scale was inverted i.e. shallow=yellow, thick=blue, but I see that there is no consistency on this matter in other publications on ice thickness and the Viridis or Parula color palette (among others) has been used both in the order you present here as well as inverted. As such, this comment is more a general appraisal on the lack of consistency in the use of color palettes in the field than critique on the chosen approach here. Nevertheless, the use of the Parula color palette as you apply here is unfortunate as a very similar shade of yellow appears twice along the scale (lower and higher than orange).

Figure 4: As in figure 3, use the same limits for the color scale for each glacier, i.e. all plots showing MG should have the same limits etc.
Figure 5: As above, use unified limits for the color scale to improve comparability/readability of the plots. Consider using a discrete color scale instead of stretched.
Figure 6: Use color palette that is more suited for color blindness and align scales between glaciers.
Figure 7: Depending on which factors you want to emphasise in this figure, consider adjusting the y-scale of the different glaciers to better-visualise the differences in bed elevation along the profile (especially for panel (d)).
Table D1: SG is missing in this overview, why? Adjust northing and easting coordinate precision (decimals) according to GPS precision.
Figure E1: Do the same as for Figures 3-5 if you change the color palette.
Dataset:
The dataset, both raw and processed radar data including picks as well as distributed ice thickness and bed topography is nicely presented with clear explanations of what is what and how the data was derived.
References:
Haualand, K. F., Sauter, T., Abermann, J., de Villiers, S. D., Georgi, A., Goger, B., Dawson, I., Nerhus, S. D., Robson, B. A., Sjursen, K. H., Thomas, D. J., Thomaser, M., & Yde, J. C. : Meteorological Impact of Glacier Retreat and Proglacial Lake Temperature in Western Norway. Journal of Geophysical Research: Atmospheres, 130(13), e2024JD042715. https://doi.org/10.1029/2024JD042715, 2025
Millan, R., Mouginot, J., Rabatel, A., and Morlighem, M.: Ice velocity and thickness of the world’s glaciers, Nature Geoscience, 15, 124–129, https://doi.org/10.1038/s41561-021-00885-z, 2022.
Pettersson, R., Jansson, P., and Blatter, H.: Spatial variability in water content at the cold-temperate transition surface of the polythermal Storglaciären, Sweden, Journal of Geophysical Research: Earth Surface, 109, https://doi.org/10.1029/2003JF000110, 2004.
Citation: https://doi.org/10.5194/essd-2025-745-RC2

Zhuo Wang, Neil Ross, Thomas Frank, Jamie Barnett, Ilaria Santin, Martin Houssais, Johanna Dahlkvist, and Nina Kirchner

Data sets

Ice thickness and bed topography for Moarhmmáglaciären, northern Sweden Zhuo Wang, Neil Ross, Thomas Frank, Jamie Barnett, Ilaria Santin, Martin Houssais, Johanna Dahlkvist, and Nina Kirchner https://doi.org/10.17043/tarfala-marma-res-2

Ice thickness and bed topography for Storglaciären, northern Sweden Zhuo Wang, Neil Ross, Thomas Frank, Jamie Barnett, Ilaria Santin, Martin Houssais, Johanna Dahlkvist, and Nina Kirchner https://doi.org/10.17043/tarfala-storglaciaren-res-2

Ice thickness and bed topography for Rabots glaciär, northern Sweden Zhuo Wang, Neil Ross, Thomas Frank, Jamie Barnett, Ilaria Santin, Martin Houssais, Johanna Dahlkvist, and Nina Kirchner https://doi.org/10.17043/tarfala-rabot-res-2

Ice thickness and bed topography for Rivgojiehkki, northern Sweden Zhuo Wang, Neil Ross, Thomas Frank, Jamie Barnett, Ilaria Santin, Martin Houssais, Johanna Dahlkvist, and Nina Kirchner https://doi.org/10.17043/tarfala-rivgojiehkki-res-2

Raw and processed radio-echo sounding data for Moarhmmáglaciären, northern Sweden Zhuo Wang, Neil Ross, and Nina Kirchner https://doi.org/10.17043/tarfala-marma-res-survey-1

Raw and processed radio-echo sounding data for Storglaciären, northern Sweden Zhuo Wang, Neil Ross, and Nina Kirchner https://doi.org/10.17043/tarfala-storglaciaren-res-survey-1

Raw and processed radio-echo sounding data for Rabots glaciär, northern Sweden Zhuo Wang, Neil Ross, Johanna Dahlkvist, and Nina Kirchner https://doi.org/10.17043/tarfala-rabot-res-survey-1

Raw and processed radio-echo sounding data for Rivgojiehkki, northern Sweden Zhuo Wang, Neil Ross, and Nina Kirchner https://doi.org/10.17043/tarfala-rivgojiehkki-res-survey-1

Model code and software

Code associated with the manuscript "Ice thickness and subglacial topography of Swedish reference glaciers revealed by radio-echo sounding". Zhuo Wang https://zenodo.org/records/18001740

Zhuo Wang, Neil Ross, Thomas Frank, Jamie Barnett, Ilaria Santin, Martin Houssais, Johanna Dahlkvist, and Nina Kirchner

Viewed

Total article views: 496 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
266	207	23	496	38	33

HTML: 266
PDF: 207
XML: 23
Total: 496
BibTeX: 38
EndNote: 33

Views and downloads (calculated since 21 Jan 2026)

Month	HTML	PDF	XML	Total
Jan 2026	117	56	6	179
Feb 2026	65	94	8	167
Mar 2026	84	57	9	150

Cumulative views and downloads (calculated since 21 Jan 2026)

Month	HTML	PDF	XML	Total
Jan 2026	117	56	6	179
Feb 2026	65	94	8	167
Mar 2026	84	57	9	150

Viewed (geographical distribution)

Total article views: 471 (including HTML, PDF, and XML) Thereof 471 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 23 Mar 2026

Short summary

Under global warming, Sweden's remaining glaciers are shrinking rapidly, and all four Swedish reference glaciers (Mårmaglaciären, Storglaciären, Rabots glaciär, and Riukojietna) may disappear within this century. To better project their future evolution, we measured the ice thickness of the four glaciers using radio-echo sounding and mapped the bed topography beneath the ice. These maps provide essential insights into future landscapes, ecosystems, and policymaking.


Total:	0
HTML:	0
PDF:	0
XML:	0