the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 22 Sep 2026
Ground ice distribution and degradation rates in the near-shore coastal zone
Abstract. Ice-rich permafrost coasts are among the most rapidly changing shorelines on Earth, and their retreat is episodic and highly variable along-shore, which makes planning and management difficult. Massive ground ice within and beneath coastal bluffs is a primary control on that retreat, yet its three-dimensional distribution cannot be assessed from surface observations and is rarely characterised at the resolution coastal protection decisions require. We present a dense three-dimensional Ground Penetrating Radar (GPR) survey, combined with passive seismic Horizontal-to-Vertical Spectral Ratio (HVSR) measurements, photogrammetry, and active layer thickness probing with validation pits, acquired across the beach at Qikiqtarjuak Tuktoyaktuk Island), Beaufort Sea, Northwest Territories, Canada, over three summer campaigns. The GPR resolves the upper surface of a massive ice body dipping seaward at approximately 1:5, from 0.4 to 2 m below the beach, with along-shore variability resolvable only through dense three-dimensional survey. Repeat survey over a five-year interval yields the first direct measurements of coastal massive ice degradation at this site: a mean of 6 ± 2 cm yr−1 along a mid-beach transect, rising to 14 ± 2 cm yr−1 in the shallowest ice sampled. The rate decays exponentially with overburden thickness, with a decay length of 40 cm. Read through the thermal damping-depth relation, that length scale implies a surface forcing period of 7-27 days, excluding the diurnal tidal constituents and the annual cycle and pointing instead to fortnightly-scale wetting of the beach by spring tides and storm surge. Rising thawing indices account for the magnitude of the mean rate; the fortnightly signal governs how it varies with depth. The depth-dependence of ground ice loss therefore appears to be set by the frequency with which the surface is wetted, rather than by overburden insulation alone. That control is not fixed: it will shift as open-water seasons lengthen and storm frequency changes, and armouring a beach necessarily alters it. The baseline established here, immediately prior to rock armour installation in March 2025, provides a direct test of that expectation.
- Preprint
(3168 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 09 Nov 2026)
-
RC1: 'Comment on egusphere-2026-5431', Anonymous Referee #1, 20 Oct 2026
reply
-
AC1: 'Reply on RC1', Craig Warren, 23 Oct 2026
reply
We thank the referee for a careful and constructive review. Our full point-by-point response is attached as a PDF.
In summary, we accept the referee’s recommendations on the processing description, the gradient quoted in the abstract, the orientation indicators, and the title, and will make those changes at revision. On the 2018 velocity model we provide the requested comparison with literature values, showing that the three campaign velocities correspond to volumetric water contents ordered consistently with the thawing indices for those years, and we set out how the associated systematic uncertainty is already propagated through the reported rates. On the HVSR acquisition duration we confirm that no longer records were taken, and justify the 600 s window against both the design of the instrumentation and the SESAME guidelines. On comparability between campaigns we note that an identical processing workflow was applied to all three, that the quantity differenced is depth below the local beach surface rather than elevation, and that beach surface change was measured rather than assumed.
-
AC1: 'Reply on RC1', Craig Warren, 23 Oct 2026
reply
-
RC2: 'Comment on egusphere-2026-5431', Anonymous Referee #2, 26 Oct 2026
reply
General Comments
The manuscript investigates the distribution and degradation of ground ice in a coastal permafrost environment using ground-penetrating radar (GPR), horizontal-to-vertical spectral ratio (HVSR) measurements, and complementary observations acquired during three field campaigns. The study addresses a relevant topic within the scope of TC, particularly in the context of ongoing permafrost degradation and coastal change in Arctic environments.
I approach this manuscript from a background in observational geophysics in cryospheric environments, with particular experience in Alpine environments, rather than from the perspective of permafrost thermal processes.
I find the study interesting and appreciate the considerable effort involved in collecting these datasets. In particular, the combination of dense, repeated GPR measurements with passive seismic observations provides an innovative approach to investigating ground ice distribution. As the authors point out, the datasets also provide an interesting opportunity for future comparisons, particularly following the installation of coastal rock armour.
Overall, the manuscript is well structured, and the authors provide a detailed uncertainty analysis, which I appreciate. However, I have several concerns regarding the reproducibility of the geophysical processing, the reliability of the 2018 survey (specifically the EM velocity) as a baseline for the temporal comparison, and the resulting interpretation of the depths of subsurface reflectors. In particular, some processing choices require further details and justification, as they directly influence the estimated reflector depths, and consequently the inferred changes in ground ice thickness.
Furthermore, I find that the broader implications of the results are not sufficiently developed in the discussion. While the manuscript presents a valuable dataset and provides insights into coastal permafrost processes, its contribution would be strengthened by a more explicit discussion of the wider relevance of these findings. I suggest that the authors revisit the context that they effectively introduce in the introduction of the manuscript and refer back to it in their discussion, i.e., how do their findings relate to potential consequences for local communities and coastal protection decisions.
Specific Comments
1. Title and scope of the study: The current title, “Ground ice distribution and degradation rates in the near-shore coastal zone”, is rather broad and suggests a more general investigation of coastal permafrost environments. The study focuses on a specific survey area. I suggest revising the title to better reflect the site-specific nature of the investigation.
2. Abstract: The current wording gives the impression that the dense 3D GPR survey was repeated over the five-year interval: L4-10 “We present a dense three-dimensional Ground Penetrating Radar (GPR) survey, combined with passive seismic Horizontal-to-Vertical Spectral Ratio (HVSR) measurements, photogrammetry, and active layer thickness probing with validation pits, acquired across the beach at Qikiqtarjuak (Tuktoyaktuk Island), Beaufort Sea, Northwest Territories, Canada, over three summer campaigns. […] Repeat survey over a five-year interval yields the first direct measurements of coastal massive ice degradation at this site: …” However, the degradation rates were derived from the comparison of a single 2018 GPR transect with the 2023 3D survey. I suggest clarifying this distinction in the abstract.
3. Introduction and broader implications: The introduction highlights the potential consequences of coastal permafrost degradation for local communities. However, this aspect is not sufficiently revisited in the discussion. It is briefly mentioned again in the last paragraph of the conclusions (L445-450), but I think it deserves more attention in the discussion itself. I suggest expanding the discussion to connect the observed subsurface changes and interpreted mechanisms to their potential practical implications for local communities and coastal protection decisions. How transferable do the authors consider their findings to other Arctic coastal settings?
4. Overview of geophysical methods in the introduction: In the section introducing geophysical methods for permafrost investigations, I suggest also mentioning active seismic techniques (e.g. SRT), which have been used extensively in permafrost environments.
5. Reliability of the 2018 GPR velocity estimate and implications for the temporal comparison: One of my main concerns relates to the electromagnetic velocity used for the depth conversion of the 2018 GPR dataset. Unlike the later surveys, for which velocity estimates were obtained from common-midpoint (CMP) measurements, the 2018 velocity was determined from diffraction hyperbola fitting. The resulting velocity differs significantly from those adopted for 2023 and 2024, and for the 2018 data there are no supporting complementary measurements.
Given that the 2018 dataset serves as the baseline for the temporal comparison, the reliability of this velocity estimate is particularly important, as it directly affects the estimated reflector depths and influences inferred degradation rates.
I would appreciate a more detailed description of the hyperbola fitting. It would also be useful to show an example of the 2018 transect and the fitted hyperbolae, potentially in the supplementary material.
Furthermore, there is a minor inconsistency in the description of the 2018 dataset. L119-120 refers to “an initial set of sparse profiles”, whereas L247 describes “a single profile”. I suggest clarifying this.
6. GPR processing and reproducibility: The description of the GPR processing workflow currently lacks sufficient detail to allow the processing to be reproduced. I recommend providing a more complete account of the parameters used, particularly for filtering and migration. For example, which migration algorithm was applied, and was a constant-velocity model assumed? Was topography incorporated into the migration? Were the same processing parameters applied to all survey years, or were they adapted to the individual datasets?
7. Uncertainty estimates: I appreciate the effort devoted to quantifying uncertainty throughout the manuscript. However, the origin of some uncertainty estimates is not entirely clear to me. For example, the caption of Figure 3 refers to error bars of “±0.05 m vertical and ±0.03 m horizontal”. How were these values determined?
Technical Comments
The following comments mainly concern the figures.
Figure 1: Please add geographical coordinates and a scale bar to the overview map to improve the geographical context and orientation.
Figure 2b: I suggest to consider annotating the 0.5 m spacing between adjacent survey lines directly in the figure. This would help readers appreciate the spatial density of the GPR acquisition.
Figure 4: I suggest displaying the conceptual models using a common depth scale to facilitate direct comparison and avoid potentially misleading visual differences. In the example showing the double peak, the “Amplitude” label overlaps with the conceptual model. The annotations indicating the two peak values could also be repositioned to make their association with the respective peaks more intuitive.
Figure 5: Why does the color scale extend to 3 m when the maximum depth discussed in the text appears to be approximately 2 m? L239 “The model reveals a consistently dipping ice surface from approximately 0.4 m at the cliff base to approximately 2 m at the seaward extent, …” I suggest to consider adjusting the color scale or explaining the choice of range. In the figure caption, the sentence “Inset locates the survey area on the island” could be made more precise by explicitly referring to the orange rectangle.
Figure 6: Two data points appear to deviate noticeably from the fitted curve. Could the authors comment on these outliers?
Figure 7b: I suggest adding a scale to the 3D visualization.
Citation: https://doi.org/10.5194/egusphere-2026-5431-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 334 | 114 | 63 | 511 | 73 | 59 |
- HTML: 334
- PDF: 114
- XML: 63
- Total: 511
- BibTeX: 73
- EndNote: 59
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Dear Editors, Authors and Readers,
I have reviewed this manuscript with great interest, as it partialy aligns with my expertise in geophysical imaging of polar regions. I find the research both interesting and valuable. The authors tackle the challenge of conducting geophysical research in the highly variable and demanding environment of coastal permafrost, specifically at Qikiqtarjuaq (Tuktoyaktuk Island, Beaufort Sea, Northwest Territories, Canada), across three summer campaigns.In general, the authors clearly present the scope and aims of their study, utilizing geophysical, in-situ, and supporting data. They provide a detailed interpretation and discussion aimed at revealing permafrost degradation patterns and their effects, highlighting key limitations. From a geophysical perspective, the choice of methods is non-standard. Specifically, the use of GPR with a passive seismic approach (HVSR in this research) is less common than the typical combination of additional Active seismic (e.g. SRT) or geoelectrical methods (e.g ERT). However, the results appear to cross-correlate well, providing conclusive data.
A key issue the authors face is the differing survey geometries between the 2018 and the 2023–2024 campaigns, which significantly limits direct comparability between the datasets. Furthermore, precise vertical localization (z-value) was unavailable for the 2018 data due to the lack of an RTK system. While the authors attempted to mitigate this by using DEM and excluding certain parts of the data from the discussion, this limitation remains a significant constraint on the robustness of time-lapse/repeated survey comparisons. A critical point at this stage is the detailed description of the processing procedures for both the GPR and the HVSR techniques. The text mentions the procedures generally but lacks specific parameter details. This information is crucial, as processing steps, especially signal filtering, have a significant impact on data comparability. Particularly concerning given the high variability of measured physical parameters reported by the authors is velocity model building. For instance, they state: “Electromagnetic velocities in the overlying wet sand between the beach surface and the top of the ground ice were derived from the CMP gathers and validated against the ALT probing measurements for 2023 (0.073 m/ns, εr = 17) and 2024 (0.075 m/ns, εr = 16). For 2018, where neither CMP gathers nor ALT probing are available, the velocity of 0.084 m/ns (εr = 12.7) was used.” The difference in velocity between 2018 and the later years is significant, and the velocity models were built using different approaches. In 2018, the value was obtained from diffraction hyperbolas: “The velocity for that year was instead obtained from eight diffraction hyperbolae fitted along a 275 m transect, which returned values between 0.070 and 0.120 m/ns with a mean of 0.089 m/ns and a standard error of 0.007 m/ns; a value of 0.084 m/ns was adopted, within one standard error of that mean.” It is difficult to verify if the 2018 value is robustly comparable to the 2023–2024 values, especially given the lack of direct validation methods in that year. I recommend the authors provide a more rigorous justification for the 2018 velocity model, including a comparison with theoretical values or literature benchmarks.
Regarding the HVSR technique, I noted that the data acquisition time was relatively short (only 10 minutes). In many similar studies, I have observed acquisition times exceeding an hour. Could the authors clarify if they compared these 10-minute datasets with longer observations at the same points? If so, did they find any significant differences, or was this duration a fixed constraint for other reasons? A brief justification for the 10-minute window would strengthen the reliability of the HVSR results.
I also found minor inconsistency. The current sentence, “The GPR resolves the upper surface of a massive ice body dipping seaward at approximately 1:5,” appears in the Abstract. I suggest replacing this with a more precise description: “dipping seaward with gradients ranging from 1:4 to 1:7.” This would accurately reflect the site and year-dependent variability reported in Section 3 (where slopes of 1:4, 1:5, and 1:7 are cited for different figures and years) and align better with the Conclusion, which correctly summarizes the range as "gradients of approximately 1:4 –1:7."
The figures 1,2,5,7 lack orientation arrows (e.g., North indicators). Adding these would help readers better orient themselves within the survey area.
The title lacks precision regarding the survey area. I recommend revising the title to clearly indicate the specific location (e.g by adding., "Coastal Permafrost at Qikiqtarjuaq, Beaufort Sea").
While I find the manuscript to be of high scientific value, I have serious concerns regarding the precision of the comparison between 2018 and later years. These concerns stem from the different survey patterns, the lack of precise vertical positioning in 2018, and the differing geophysical acquisition across the years 2018-2023 especially.