the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 27 Mar 2026
Brief communication: On the potential of dual-coil frequency-domain electromagnetic (FDEM) systems to detect frozen layers in mountain permafrost environments
Abstract. Frequency Domain Electromagnetic (FDEM) methods are still rarely applied in mountain permafrost environments, such as rock glaciers. Here, we test a separable dual-coil FDEM system at four mountain permafrost sites and compare the results with Electrical Resistivity Tomography (ERT), the most commonly geophysical method applied in these environments. The comparison shows that FDEM can reproduce key subsurface features identified by ERT and highlights the potential of separable dual-coil FDEM systems for a straightforward, preliminary, first-order assessment of subsurface structures in mountain permafrost environments.
- Preprint
(1052 KB) - Metadata XML
-
Supplement
(642 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-1264', Anonymous Referee #1, 08 May 2026
-
RC2: 'Comment on egusphere-2026-1264', Anonymous Referee #2, 11 May 2026
This manuscript presents a small field study aimed at demonstrating the value of FDEM measurements for detecting frozen layers in mountain permafrost environments, especially as an alternative to the more commonly used but more laborious ERT method. For a brief communication, I think this is a solid set of results, however a bit more detail about measurement acquisition, processing and discussion is required. Figures included also require a few modifications. Please refer to the list of comments below.
Line 40: Were there specific reasons behind site selection?
Line 74: How much of the data was retained after filtering? Are areas of lower sensitivity impacting the way you interpret the results?
Line 89: What is the expected measurement depth?
Line 91: This software allows you to correct EM data using ERT data? Has this been attempted?
Line 92: Is this the most optimal layer model? If so, why? Please explain the criteria that led to the selection of this model. If I understand correctly, you only have 3 data points per 1D inversion.
Line 101: Would more accurate positions of Tx and Rx coils improve data quality? Could the accurate GPS positions of the coils (rather than the midpoint) not be recorded?
Line 104: “towards”
Line 118: Would calibration help with this systematic discrepancy? How was the EM system calibrated?
Line 146: Perhaps is worth mentioning that compared to ERT, FDEM seems to lack resolution. Therefore, in my opinion, FDEM is not an exhaustive substitute for ERT. Could you make suggestions for case studies which would benefit more from the use of ERT, case studies which would benefit more from the use of FDEM and case studies which would benefit more from a combination of the two?
Line 149: As also pointed out by the editor, could you expand on the data acquisition procedure? You could also add some more specific quantitative differences between FDEM and ERT, such as differences in total survey time.
Figure 3: This figure does not have the same scale with the ERT figure, and it is a bit difficult to compare and contrast between them. Please make necessary adjustments.
Figure S2: Resistivity models in Figure 2 seem to have a larger 2D area that the sensitivity models with values >=0.1. Why are you including areas of low sensitivity in your results?
Figure S3: Are all these identical? Is that correct?
Citation: https://doi.org/10.5194/egusphere-2026-1264-RC2 -
RC3: 'Comment on egusphere-2026-1264', Anonymous Referee #3, 19 May 2026
General comments:
The authors present a well-structured, small-scale study exploring the potential of a separable dual-coil FDEM system in mountain permafrost regions, comparing it with standard ERT. The paper addresses a relevant scientific question within the scope of the journal, as rapid geophysical characterization of coarse-blocky permafrost terrain remains a significant logistical challenge. The statements and figures generally flow really well, making for an easy read, and the overall presentation is clear. However, while the abstract and conclusions highlight the logistical advantages of FDEM, the results and discussion currently lack a robust evaluation of the method's limitations. Specifically, there are concerns regarding the sensitivity analysis, the 10-layer model parameterization, and the practical challenges of deploying FDEM in extreme topography. Addressing these points will significantly improve the results.
Specific comments:
44-53: please explicitly define the measurement spacings and lengths of all EMI and ERT transects.
62-63: “and highly conductive conditions reduce effective penetration due to signal attenuation” Your focus is highly resistive conditions. How does high resistivity affect the signal?
86-88: you note that FDEM transects do not cover the full length of the corresponding ERT lines due to difficulties in optimally positioning the coils. How was it possible to collect ERT data (requiring galvanic contact) in these sections but not FDEM data? Please clarify the reasoning here.
92: how does the sensitivity analysis determine the use of a 10-layer model? This results in a model with ~2.4 m thick layers relying on only 3 depth points per FDEM measurement point, which seems like a significant red flag regarding model over-parameterization.
92 & Fig S3: the stated depth threshold is based on a 0.8 sensitivity value, but this valid only for the largest (40 m) coil spacing? Please clarify or address this.
106: Fig.2b does not seem to reach 100 kohm.m; it seems like only 2c-d reaches this magnitude.
151-153: to better support your conclusions about logistical efficiency, it would be nice to know roughly how much time it took to conduct the FDEM vs ERI surveys at these sites.
Fig 2: in 3/4 case studies presented, the ERT spatial coverage was actually longer than the FDEM coverage. Since the primary advantage of FDEM over ERT is typically rapid, wide-area coverage, I think this needs to be addressed in the discussion and conclusions. Does the extreme topography of certain glacial regions fundamentally limit FDEM from capitalizing on its primary advantages?
Fig 2-3: while the comparison is present, both the x- and y-axis scales vary significantly between all subplots. Please scale the axes equally across all FDEM and ERT plots to allow for a fair, direct visual comparison.
Fig S3: are the normalized sensitivity distributions across the sites completely identical? How is this physically possible when the ERT sensitivity models present far more heterogeneity? The S3 caption claims "resistivity ranges are very similar across the four sites", but according to the results in both Fig.s 2&3, this is incorrect. There are large variations spanning two orders of magnitude between the four sites, along with significant spatial variability. Please better explain or correct this.
Technical corrections:
95: L-BFGS-B is a solver algorithm, not a type of regularization.
116: This discrepancy is attended, -> This discrepancy is expected,
180: C93C23002690001)”. -> C93C23002690001).
Citation: https://doi.org/10.5194/egusphere-2026-1264-RC3
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 881 | 158 | 47 | 1,086 | 86 | 31 | 36 |
- HTML: 881
- PDF: 158
- XML: 47
- Total: 1,086
- Supplement: 86
- BibTeX: 31
- EndNote: 36
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This was a very interesting manuscript with convincing figures! Thanks for this. The presentation of a comparison on four different sites further strengthen the opinion expressed in the manuscript. I would encourage the users to further published the code and data associated with this manuscript as to further convince the cryosphere community to use FDEM for this kind of study. I have only minor comments and some questions before proceeding further.
General comments:
How long takes the FDEM survey and what is the errors range? How does it compare to ERT? Can you maybe advise the users on where to use ERT and when to use FDEM?
Is there any limitations from nearby metallic objects (cable car, power lines?) How does the horizontality of the coil influence the quality of the readings?
Detailed comments:
86: so it was possible to do an ERT transects but the terrain did not allow to do an FDEM measurement? I can't really imagine how but ofc, I wasn't on the field. But does this mean that terrain should be more flat for FDEM than for ERT?
89: so for each location, there were 10, 20 and 40 m spacing, so three data point per location. I guess that's not that much compared to ERT but probably enough to obtain a 2 layers model.
90: smoothed? you mean in the X direction?
95. L-BFGS-B is not a regularization it's the name of the solver. The regularization is probably a L2-norm.
118. '.. as expected lower' -> because of galvanic issue?
130. '... is attended' -> '... is expected'
131. this difference in magnitude between FDEM and ERT is large and still surprising. However, it seems systematic to between all surveys. Maybe if the Authors compared inverted ERT vs inverted FDEM, they could fit an offset between the two. I am also wondering given the rock glacier is quite a complex environment if some kind of "galvanic" isolation could not happen, which would artificially increase the resistivity measured by ERT (but would not affect FDEM measurement).
142: "the loss of VCP..." -> well, instead you could do some more HCP at different distances like 15, 25 and 30 to get more information for the inversion.
153. can you also give an idea of the time needed for acquisition compared to ERT?
fig1 and fig2: I think the comparison is quite nice and convincing! The magnitude of difference of values is a bit strange though.
174. In order to further increase the usability of the FDEM method (which I think is the purpose of the manuscript), I would strongly encourage the authors to put their data and code in open-source repository (data on zenodo and giving it a DOI so it can be cited) and code in gitlab/github/zenodo).