Spatial characterization of near-surface structure and meltwater runoff conditions across Devon Ice Cap from dual-frequency radar reflectivity
- 1Institute for Geophysics, University of Texas at Austin, Austin, TX, 78758, USA
- 2Geological Survey of Denmark and Greenland, Copenhagen, Denmark
- 3Department of Electrical Engineering, Stanford University, Stanford, CA, 94305, USA
- 1Institute for Geophysics, University of Texas at Austin, Austin, TX, 78758, USA
- 2Geological Survey of Denmark and Greenland, Copenhagen, Denmark
- 3Department of Electrical Engineering, Stanford University, Stanford, CA, 94305, USA
Abstract. Melting and refreezing processes in the firn of Devon Ice Cap control meltwater infiltration and runoff across the ice cap, but their full spatial extent and effect on near-surface structure is difficult to measure with ground-based traverses or existing satellite remote sensing. Here, we derive the coherent component of the near-surface return from airborne ice-penetrating radar over Devon Ice Cap, Canadian Arctic, to characterize firn containing centimeter to meter-thick ice layers (i.e., ice slabs) formed from refrozen meltwater in firn. Comparison with reflectivities using a thin layer reflectivity model, informed by ground-based radar and firn core measurements, indicate that the coherent component is sensitive to the near-surface firn structure composed of quasi-specular ice and firn layers, limited by the bandwidth-constrained radar range resolution. By leveraging their differences in range resolution, we assess the use of dual-frequency airborne ice-penetrating radar to characterize the spatial and vertical near-surface structure of Devon Ice Cap. Our results suggest that average ice slab thickness throughout the Devon Ice Cap percolation zone ranges from 4.2 to 5.6 m. This implies conditions that can enable lateral meltwater runoff and potentially contribute to the total surface runoff routed through supraglacial rivers down glacier. Together with the incoherent component of the surface return previously studied, our dual-frequency approach provides an alternative method for characterizing bulk firn properties, particularly where ground-based and higher frequency radar data are not available.
- Preprint
(15510 KB) -
Supplement
(1963 KB) - BibTeX
- EndNote
Kristian Chan et al.
Status: final response (author comments only)
-
RC1: 'Comment on tc-2022-181', Anonymous Referee #1, 25 Sep 2022
Review of: Spatial characterization of near-surface structure and meltwater runoff conditions across Devon Ice Cap from dual-frequency radar reflectivity - by Chan et al.
General Comments
Chan et al. investigate the surface coherent return power of reflected radar waves by applying the Radar Statistical Reconnaissance method to multiple ice penetrating radar datasets with different center frequencies over Devon Ice Cap (Canada). The data is used to better characterize the composition of the firn pack (and if the firn pack contains ice layers) of the upper meters over the ice cap, which is important to better understand melting and refreezing processes as well as meltwater infiltration and runoff. The measured reflectivities are compared with modeled reflectivities using a reflectivity model informed by existing information on the firn pack (from ground-based ice penetrating radar data and firn cores). Their results suggest meter-thick ice slabs in certain parts of the ice cap, which permits surface water runoff away from the ice cap.Overall I find the study by Chan et al. to be informative and very well written. They applied a smart approach to characterize the firnpack with existing multiple airborne radar data sets and other auxiliary data sets (such as firn cores and land-based radar data). Although the methodology is not fundamentally new and many aspects have been already analyzed and built upon previous studies (such as in Rutishauser et al., 2016), I believe that this article deserves to be published in The Cryosphere.
The basis for my decision is that, in my opinion, this is a robust study that is well structured, clearly written, and represents a significant step forward in knowledge on which future studies can build on and which is very useful for the cryosphere community. What I particularly liked is that the authors use existing data sets and put the data into a new context with their method to find out more about the first meters of firn of the Devon Ice Cap. Below I have some comments and questions that I think might help to add clarity and make the article better readable and easier to follow.
Main RemarksIntroduction:
The introduction could benefit from a small introduction on the Devon ice cap and why it is a particularly good place to characterize the firn column. Either in a new paragraph (which I would prefer) or incorporated in one of the existing paragraphs.Figure 1:
(1) It would be a nice addition to have an overview map of the Canadian arctic or Canadian-Greenlandic arctic pointing out the location of the survey area. This would give the reader a much better impression of where the Devon ice cap is located.
(2) I also would suggest finding a better solution with the contour lines and their elevation labels. They appear very chaotic at the ice caps margins, which is rather confusing than helpful information. The same applies to all other figures (also in the supplement; S4) in which the contour lines are shown. Maybe only displaying contour lines only above 600 m would make the plot less overloaded.
(3) It would also be good to state what kind of satellite image you are using as a background image.Table 1:
Please explain the symbols in the table caption (e.g., that range resolution is z_0, etc.)
In addition, but very minor: a hline between the two systems would be nice to immediately see which z0 belongs to which system.Figure 2:
What about the following idea: To give the reader a better understanding of the different depth resolution of the radar systems and which parts of the firn column are affected, one idea would be to somehow draw or indicate the depths that HiCARS & MARFA and MCoRDS3 resolve in Figure 2b.Figure 3:
(1) I think the figure could be better arranged if, for example, (a) and (b) were in a row and (c) below. Then the subfigures would be bigger and the whole figure would take probably less space in the document at the same time. The same could be done with Figure 4.
(2) Shouldn't the label of the colorbar be "dB" instead of "db"?
(3) I would suggest a different color scale, preferably linear rather than divergent. This is because in the HiCARS display, for example, the transition from -10 to -15 dB is shown as a weak color change, while from -20 to -25 dB there is a strong color change (yellow to blue). Therefore, I would suggest a linear graded color scale to better interpret the changes in dB based on a color scale across the different data sets.Figure 4:
Caption: define again that interquartile ranges is IQR and P_c surface coherent power (as in Fig. 3).Discussion:
I have a question regarding the ice slab thicknesses in Zone II. In Line 336 you state that the HiCARS/MARFA system captures the entire thickness of the ice slabs. Maybe I have missed it, but why is that the case and how do you know that the ice slabs along these radar profiles are not thicker than the range resolution of the system?
My next question is very similar and refers to the average ice slab thicknesses. You calculated a mean ice slab thickness based on the range resolution of the two different radar (groups). Wouldn’t it be rather a minimum average ice slab thickness? Because since you are only analyzing the surface return within the limits of the range resolution of the radar system, you cannot estimate if the ice slab continues with depth and is thicker, right?
For me it seems that based on the surface GPR data it is assumed that the ice slabs in this region are not thicker as what is for example shown in Figure 2b. However, it might nevertheless be possible that thicker ice slabs might be present along the airborne radar profiles where no surface radar data exists. I think this should be clarified and also mentioned in the uncertainty section.Figure 5:
Here now appears a reference to the background satellite image, but the coordinates are missing. Again, I would prefer to get rid of the contour lines and labels below a certain depth.Supplement
Figure S4: Please mention once more in the caption that P_c is coherent specular and P_n incoherent/scattered. I'm sure many readers don't, but I often have the problem that I forget the abbreviations in the text while reading and then have to look for them again in the text when they appear in a figure.Line-item Comments
L 84-86: I think that Operation Ice Bridge should be mentioned here as well in addition to the University of Kansas. Moreover, I would suggest using the acronym MCoRDS3 instead of just MCoRDS throughout the document.
L 99-101: With respect to the factors affecting permittivity, I think that temperature and the anisotropy due to the orientation of the ice crystal fabric should also be mentioned (although COF may not be so important in the firn column). In that sense you could additionally cite for example Fujita et al. (2000):
“Fujita, S.,T. Matsuoka,T. Ishida,K. Matsuoka, and S. Mae (2000), A summary of the complex dielectric permittivity of ice in the megahertz range and its applications for radar sounding of polar ice sheets, in Physics of Ice Core Records, edited by T. Hondoh, pp. 185–212, Hokkaido Univ. Press, Hokkaido, Japan. ”L 128 (and L177-178): You mention that “[...] surface roughness is not the main contributor to surface scattering over DIC (Rutishauser et al., 2016).”. It would be interesting to mention in one sentence why this is not the case. Especially because this assumption is important for the interpretation of the results.
L 137-139: Here you state that: “Previous applications of the RSR method have empirically shown that an aircraft roll of 2 to 3° allows for a stable coherent radar return.” Is there a reference for this?
L 141: The airborne radar data in your study is also "ground-penetrating". From what I understood you refer to land-based or surface radar in this section. Therefore I would suggest making clear that all radar surveys are ground penetrating and some are airborne and this one is land-based/surface radar data.
L 248-252: I am not sure if I missed it, but is the difference between the old and refined Zones shown somewhere? If not, I think it should be (maybe in the Supplement). I guess the old Zones are those displayed in Rutishauser et al. (2016) in Figures 1a and 2?
L 252-254: Here you refer to the Discussion Section but I think it would be also good to refer to Figure 5.
Thank you for the exciting read.-
AC1: 'Reply on RC1', Kristian Chan, 15 Jan 2023
The comment was uploaded in the form of a supplement: https://tc.copernicus.org/preprints/tc-2022-181/tc-2022-181-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Kristian Chan, 15 Jan 2023
-
RC2: 'Comment on tc-2022-181', Anonymous Referee #2, 31 Oct 2022
This study uses four radar datasets (3 airborne, 1 ground based) to evaluate the firn characteristics of Devon Ice Cap in the Canadian Arctic. The general characteristics of the firn were already classified using the ground-based dataset, and the new element here is using all the airborne data together to look at the firn. These data are used as a way to assess the spatial distribution of firn properties in more detail, within the general framework of the ground-based survey. Conclusions about ice-slab thickness and melt channel distribution are derived largely using the variability in return power of the surveys within different “zones” of firn, relying on the ground-based survey to get the general structure (i.e. large slabs, thin lenses, etc). The implications for meltwater runoff are discussed, making a nice story. The main novel element here is inferring properties of ice lenses in firn using multiple airborne radars that do not resolve the ice lenses/slabs explicitly, but instead have some return-power sensitivity to the near-surface properties.
This study is novel, generally well written, and well-suited to The Cryosphere. I have two major comments and a variety of small points that I think are important to address before publication, but then I think it should be a nice contribution.
Major Comments:
There is insufficient analysis of whether one could conduct a similar study in the absence of some independent radar measurements that actually resolve the bottom of the ice slabs (i.e. the GPR)—perhaps this was never the goal of the study, but the title and some of the language suggest otherwise, which I think sets the reader up to be dissatisfied at what is otherwise a nice paper. The suggestion in the title, abstract, and conclusions is that the dual-frequency reflectometry can be used on its own to garner insight into firn properties (and extra-terrestrial applications cannot rely on such validation). As I read the paper, the analysis of things like the ice-slab thickness in Zone II (Section 3.2.3 and Discussion) relies on already knowing that this area has thick ice slabs, and otherwise the variations could be misinterpreted as density variations or similar. If the paper can be altered to use the GPR as validation rather than as a necessary component, that would be ideal; for example, is there some objective measure that would allow the picking of the zone boundaries from these model results? I assume that the answer is no since otherwise it would be discussed (which is worth adding to the text); I think this study will merit publication without that analysis, although in this case I think textual/title alterations are needed throughout to make clear that what is really happening is analysis of things like ice-slab thickness when the general firn structure (zonal classification in this case) already independently known, effectively requiring a third radar dataset (GPR) or other extensive in-situ measurements.I find Section 3.1 to be lacking in purpose, in part because it reads something like a failed attempt to distinguish the zonal classification based solely on reflectometry; it is doubly unconvincing due to insufficient error analysis. In lines 201-203 there are claims about which model fits better where, but there is not even an analysis of the relative RMS misfits of the two models in the two zones. At a bare minimum, such basic model-data misfit analysis is needed to make any claim about what model fits where. However, given the section title I was hoping it would essentially answer the other main point raised above. I understand that this may be beyond the scope of the work or not supported by it, but then I am left wondering what this section really adds (perhaps adding some error analysis would change my mind, and I could better understand what we could conclude out of this section). Perhaps some roadmap under the general “Results” heading could help as well.
Line Comments:
52: I would suggest removing the IPR acronym. These are all ice-penetrating radars, and the terminology is unnecessarily confusing.
57: What such methods? The low frequency ones?
58: I am skeptical of this claim—does Mars have surface melt? Could ice lenses and slabs be possible? While other dual-frequency applications matter there, the relevance of this study should be justified or the line should be deleted.
69: What is compact ice? It is not defined nor is it a common term. I think it just means glacier ice as opposed to firn. Perhaps “fully compacted” or “fully densified” would be more appropriate. While I put this as a line comment, I think it is important to change “compact” throughout the paper, since it is not quite the technical term and the word has multiple plain-language meanings.
82: What does dual phase mean?
99: There are plenty of homogeneous media for which the arguments in line 100 apply—perhaps just delete this sentence
Table 1: The layout here is confusing. I think I would have understood better if the epsilon_eff column were deleted and there were separate columns for z0 for firn and for ice. Also should specify that this is not a universal firn number—it assumes 410 kg/m3 or something similar.
115: It would be helpful to have a half sentence about why the bin size (in spatial terms) is different for the different systems.
126: RMS height of what? I guess this should be surface elevation
129: The hypothesis that the return power variation is dominated by variations in r2 is a large and critical assumption that is brushed aside too flippantly. I guess there was some work in Rutishauser et al., 2016, to justify that it is not dominant, but I think it is a bit too important to be relegated to a reference, since strong dependence on the roughness may invalidate any conclusions. Addressing this could be as simple as estimating the maximum variation resulting from a realistic range of roughnesses compared to the variation in return power.
139: Is this a typo? Why exclude rock based on aircraft elevation rather than imagery, etc.
143: At least a brief overview of the GPR system belongs here—the reader should not have to go to Rutishauser et al. just to find out the frequency
155: Layers of what, and should this be i.e.? Generally I would assume density is the only important factor in such shallow reflections—if not, what else should be included.
326: potentially insightful
391: Rephrase slightly to clarify that the ambiguity is due to tradeoffs between density and layer thickness
395: Caution against?
411-420: I would highly recommend moving this paragraph upward into discussion—I do not find it to be particularly convincing, and I don’t really think it is a conclusion as such. It is not my place to demand such a change, but take this as a stylistic suggestion of a way to make the paper more impactful.-
AC2: 'Reply on RC2', Kristian Chan, 15 Jan 2023
The comment was uploaded in the form of a supplement: https://tc.copernicus.org/preprints/tc-2022-181/tc-2022-181-AC2-supplement.pdf
-
AC2: 'Reply on RC2', Kristian Chan, 15 Jan 2023
Kristian Chan et al.
Kristian Chan et al.
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
341 | 104 | 15 | 460 | 33 | 4 | 3 |
- HTML: 341
- PDF: 104
- XML: 15
- Total: 460
- Supplement: 33
- BibTeX: 4
- EndNote: 3
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1