the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Feb 2026
Mapping Snow on Northern Winter Roads: A Dual-Frequency Polarimetric Radar Approach for Snow Characterization over Land, Lake and Sea ice
Abstract. Winter roads are lifelines for remote northern communities. Built over land, lakes, rivers, and sea ice, these travel routes are increasingly vulnerable to warming temperatures and variable precipitation. To ensure safety and adapt to these changes, operators require high-resolution monitoring of snow depth across these diverse surfaces, as natural snow accumulation dictates ice growth rates, route viability and road stability. This study extends our polarimetric radar method, previously demonstrated on pack ice, to landfast sea ice, tundra, and frozen lakes. Results indicate consistency with earlier sea ice analyses and promising performance over frozen ground especially at Ku-band. To address the specific challenge of lake ice which includes strong returns from the ice/water interface, we present a new interface-detection technique that simultaneously retrieves snow depth and ice thickness. While current validation focuses on undisturbed snow, this approach could provide a path forward for characterizing the cryospheric environment in a way that can directly support the optimization of winter roads.
Competing interests: At least one of the (co-)authors is a member of the editorial board of The Cryosphere.
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.- Preprint
(51061 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 01 Apr 2026)
- RC1: 'Comment on egusphere-2026-212', Anonymous Referee #1, 25 Feb 2026 reply
-
RC2: 'Comment on egusphere-2026-212', Anonymous Referee #2, 27 Feb 2026
reply
Review of
Mapping snow on northern winter roads: A dual-frequency polarimetric radar approach for snow characterization over land, lake and sea ice
by
Stroeve, J., et al.
Summary: This contribution describes measurements, analyses and interpretation of ground-based radar measurements carried out at frequencies in Ku- and Ka-Band, HH and VH polarization, over measurements sites in the Canadian sub-Arctic and Arctic during winter and early spring. The main aim of these radar measurements is to retrieve the snow depth. The radar measurements were carried out along transects over landfast sea ice, lake ice (two lakes), and two different types of tundra, and were accompanied with in-situ observations of relevant snow properties. These latter observations were used to evaluate the retrieved snow depth and to feed physical models used to simulate the expected radar return of the various surfaces to interprete and better explain the results obtained. The results are quite convincing with respect to the snow depth over sea ice and, mostly, also for lake ice, following logically previously published work in this field. The results obtained for the tundra sites are less convincing.
Overall, this contribution is very well written and provides an important addition to our current knowledge in this field. I only have a few specific comments and some suggestions of editoral nature. I also formulated two general comments - which are focusing on editing rather than content.General Comments:
GC1: In the title and the abstract you stress quite a bit that you aim to improve monitoring of winter roads. However, the main content of the paper is basically about snow depth (in general) over various surfaces. Neither do you elaborate on, e.g., roughness issues, aeolian redistribution of the snow, or other factors that might hamper preparation and/or maintenance of a winter road nor do you sufficiently well elaborate on the structural / morphological changes of the snow that forms the winter road, i.e. compaction, etc. Therefore, I have to say, I find it a bit far fetched to use this aim in title, abstract and motivation of this contribution. It could mislead readers.GC2: I find the description of the experiments a bit light. One needs to read through 90% of the paper to finally learn how long the transects actually were. Also an illustration how the various in-situ measurements are located relative to the radara transects is missing and would be very nice to better understand the experiments as a whole.
Specific Comments:
L17-26: Little is said in this paragraph about muskeg and the impact of vegetation. Also, the role permafrost desintegration has on such roads and their monitoring is not laid out sufficiently well.L59: Neither here nor later I found information about the length of the transects along the measurements were carried out. The entire scale of the experiment remains vague at this point.
L89-93: I well understand that it is not an issue to overprobe the snow depth in case the surface underneath the snow cover is solid - like pebbles or ice. But for the site in Churchill you wrote about the average height of the vegetation underneath as 20 cm to 80 cm and about the type of it, being - among others - shrubs. Hence I would assume that the surface underneath the snow is not necessarily overly solid at this site. Please reflect on this and, if needed, change the text accordingly.
L107/108: Here you write that you used the density cutter at 3 cm vertical intervals; I assume this applies to both sites, Churchill and Resolute. The profiles shown in Figure 4, top, do exhibit, however, a vertical spacing of values every 2 cm. How did you realize that? I note that in Fig. 4, bottom, the vertical spacing seems indeed to be 3 cm. Table 3 reveals that the distance between successive points was also sometimes 4 cm or even more.
L149/150: Even though more details of the system are given in the paper cited in L148 a reader would appreciate to know that the antennas are side looking, right. Also, there needs to be a beam protuding far enough from the sledge to the side to allow the mentioned "nadir viewing" configuration.
L153: It is not clear what you mean by "combined cross-track and along-track tilt". How did you combine the two tilt estimates?
L204: "each grid cell" --> This seems to be out of context here because so far the paper talks about in-situ observations along tracks and about KuKa observations also along a certain track ... so far ... no mentioning of a grid took place. From your description so far it is not obvious that the KuKa provided a spatial distribution of snow depths.
L279-285: This paragraph is quite qualitative ... I was wondering whether you could give quantitative values for the degree of agreement.
L353: But this secondary peak is indeed very weak for HH-polarization; it is much more pronounced at VH polarization. Is that secondary peak significant in relation to the measurement accuracy / the noise?
L357/358: This comment connects a bit to the previous one. You see this as a downward shift of the Ku HH peak ... I would say that the air/snow interface still has a very strong peak; the saline snow layer provides another peak, that is even higher, true, but only by 3-4 dB. I am wondering, whether given the range resolution that can be achieved, you are not at the edge of what you can credibly resolve and the differences in the (three) peaks we see over these two interfaces plus the saline layer at Ku HH are perhaps not significant?
Table 3: I would be very interested how figures like Figure 15 and 16 would look like for the thicker snow layers on sea ice, i.e. snow pit 3 on April 2 and, particularly, snow pit 2 on April 6. Is the main reflrection horizon for the two bands still the snow/air interface? How much of the signal reaches below the highly saline basal snow layer? I would find a focus on these two snow pits also more interesting in terms of their representativity for future conditions in the Arctic (and current conditions in the Antarctic) with respect to snow depth on sea ice.
Figure 16: Why is the peak Ku VH return located so deep in the ice layer? Why is the return from both the snow/air interface and the saline snow layer so much weaker for this snow pit than for snow pit 6?
L371-374: Since you use "winter roads" and their monitoring as one of the main motivations of your paper (it is even in your title) I find this paragraph a bit weak and very hypothetic. Aren't there studies published, e.g. in the journal Cold Regions Research and Technology about the effect of artificial / mechanical compaction of snow on physical snow properties?
Figure 17: Did you explain these pairs of snow depth values that are kind of aligned vertically (left of 200 m and of 400 m) in the Lake Resolute snow depth somewhere?
Typos / Editoral Remarks:
L80, L84 & Figure 2 caption for wind speed: "km/hr" should be replaced by "km/h".Figure 1 caption: What is meant by "- newer"?
Figure 5, caption. Please note in the caption that the vertical axis is reversed to the one in Figure 4.
L165: 0.77 and 0.85 cannot be correct ... the speed of light in vacuum is 3 x 10^8 m/s ... so, maybe a 10^8 is missing and a unit is missing as well.
Table 2: Why is the Ku-Band VH Ice/water interface power threshold (dB) higher than the one for HH? (VH: -20dB, HH: -30dB). This seems to be rather unusual.
L206 "maps" --> It is not sufficiently clear how the kind of measurements that were carried out can results in maps.
Figure 7, caption, "Light yellow lines indicate MagnaProbe snow depths" --> I am not sure I fully understand how you overplotted the MagnaProbe Snow depth onto these echograms ... you need to have a reference height (range) from which you subtract the snow depth? So, if for Ku VH the light yellow line dips down to a range of 2.0 m .... what is the reference height (range)? 1.6 m? Some clarification might increase the readability.
Figure 14: The font size within the panels should be increased to the same size as used in Fig. 13.
Figure 15, caption: I think it needs to read "6th" snow pit - at least according to the title of the figure and the match with "SI6" in Figure 4.
L366-368: But you don't show figures like 15/16 here so maybe add "(not shown)"?
Figure 17: The font sizes are really small. You could consider increasing it.
L429: Given the complexity of the returned signal from whatever is underneath the snow in case of a tundra that you mentioned above a few times, I suggest to not include snow depth over trundra in this sentence.
Citation: https://doi.org/10.5194/egusphere-2026-212-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 221 | 98 | 11 | 330 | 23 | 33 |
- HTML: 221
- PDF: 98
- XML: 11
- Total: 330
- BibTeX: 23
- EndNote: 33
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Review of “Mapping Snow on Northern Winter Roads: A Dual-Frequency Polarimetric Radar Approach for Snow Characterization over Land, Lake and Sea ice” by Julienne Stroeve et al.
The study is comparing different snow depth mapping methods (also described in Willatt et al., 2023…) with the on-ice KuKa radar to magnaprobe measurements over different terrain used for winter roads in Canada (lake ice, sea ice, muskeg). The results are in line with earlier KuKa results on sea ice that the Ku VH-HH peak provides the best snow depth estimates while the Ku – Ka peak or centroid method does not demonstrate much skill. This finding is obviously concerning when the upcoming CRISTAL snow depth product is based on the Ku – Ka method and it is also interesting to imagine what the studies mentioned in the introduction have derived, using the same method with ALTIKA and CRYOSAT data. There are differences between radar measurements from KuKa and satellite, as also mentioned, but one thing that they have in common is the signal attenuation in the snow layer. This KuKa to satellite discussion could be strengthened. Unfortunately, there are no committed future satellite polarimetric altimeters.
The MS is well written and structured but the analysis could be improved and I have some suggestions below. I think that some parts of the discussion could be strengthened and substantiated by including the measured values in Table 3 and computing the reflectivity and penetration depth and potentially scattering, absorption etc.
In addition, I have a number of specific comments:
The abstract should be detailed with main results.
There are some differences and similarities between on-ice and satellite altimeter measurements which should be emphasized in the introduction and strengthened in the last part of the discussion.
Sometimes the discussion is really confusing and I give examples below. I would suggest a conceptual model for what happens with snow - radar interaction to illustrate the main principles. What are the snow/ ice parameters affecting the radar backscatter? And how?
The in-situ measurements have uncertainties, the derived snow depths have uncertainties. Please include a discussion and a quantification of the uncertainties.
L114: please write in the text if Churchill is sea ice or something else. The same for the other sites.
Figure 4: all these parameters could contribute to the uncertainty, I suggest to use them in the analysis for comparing the different sites and for computing reflectivity and penetration depth.
L158: the snow density in Eq. 1 is in [g/cm3]. Please specify that. Otherwise you are using [kg/m3].
L161: The sentence is unclear, please clarify. The “liquid water” which is, I think, brine, is affecting the loss (penetration depth?), but according to Eq. 1 the “wave speed” (speed of light?) is only a function of snow density. Please use the terms consistently (e.g. “liquid water” -> “brine”) and substantiate your statements with references or by computing the “loss”, “Wave speed/ speed of light” with models.
L165: “…we assume c’ values range from 0.77…”: The c’ value is computed using Eq. 1, write that instead of “assume”. The speed of light in snow is not 0.77, it is 0.77 times the speed of light in a vacuum (as written in line 158). Please clarify.
Figures 7 and 8 it is not totally clear what “Range” means. Is it the two-way travel time times the speed of light in a vacuum? Please clarify. Also could you reference these plots to the snow and ice surface? It is very difficult to see if it is a track-point misalignment or an actual variation in the snow and ice surface height. Please also write the average snow and ice thickness in the figure caption.
Figure 7: Judging from the “range” you do not get returns from deep within sea ice. Please clarify. Is it because of sidelobes or other off-nadir returns?
L265: I would suggest to replace “volume scattering” with “rough surface scattering”.
L333: This explanation is confusing. According to Eq. 1 there is no permittivity contrast between Ku- and Ka-band and if there is difference in the air/snow surface scattering then it is the wavelength dependent roughness. Both frequencies scatter at the snow surface but there may be differences in snow penetration at Ku- and Ka-band and how much reaches the snow ice interface (and back). Please clarify.
Table 3 is nice and I would even suggest to include penetration depth and Fresnel reflection coefficient as columns. However, the permittivity is not consistent with Eq. 1. I think that you could even achieve better results if you use the Table 3 permittivity for computing the speed of light in snow (Eq. 1). In any case there should be consistency between Eq. 1 and the permittivity in Table 3.
Table 3: how is the salinity computed/ measured? Density cutter-> bagged sample-> melted->salinity measured->multiplied by the density ratio?
L361: I think that this sentence is misleading. Depolarization is also happening in dry snow, so I am skeptical if it depends on brine pocket scattering. Loss is the sum of absorption (brine) and scattering (snow grains). Please define the “Mie-regime”. What is “rapid depolarization”? Please reformulate these two sentences and substantiate the statements.
L367: Compute the Fresnel reflection coefficient in Table 3 to make a substantiated assessment of the “initial reflection”. According to Eq. 1, the permittivity is only a function of snow density.
L374: why would compacted snow be “high extinction”? Please clarify and define the term extinction here.
L417: I agree that KuKa is not sensitive to large scale surface roughness in the same way as satellite altimeters. However, KuKa is still sensitive to surface roughness. Please clarify and describe what KuKa is sensitive to.
Figure 17. Are there any measurements of ice thickness and can this be used to derive the effective permittivity of the ice… or snow?