the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 13 Jul 2026
Altimetric Ku-band Radar Observations of Snow on Sea Ice Simulated with SMRT
Abstract. Radar altimetry provides sea ice thickness estimates for polar region. However, uncertainty in the scattering horizon used to retrieve sea ice thickness arises from interactions between the emitted signal and snow cover on the ice surface. Therefore, improving our knowledge on electromagnetic waves scattering with the snowpack and ice is necessary to retrieve sea ice thickness accurately. The Snow Microwave Radiative Transfer (SMRT) model was used to simulate the low-resolution altimeter waveform echo from snow-covered sea ice, using in-situ measurements as input. In-situ measurements from four field campaigns in three distinct Canadian Arctic regions includes temperature, salinity, density, specific surface area, microstructure from X-ray tomography and surface roughness measurements using structure from motion photogrammetry. Evaluation of SMRT in altimeter mode was performed against CryoSat-2 waveform data in pseudo-low-resolution mode. Simulated and observed waveforms showed good agreement, although it was necessary to optimize the snow and sea ice roughness. In addition, simulations of backscatter in low-resolution mode in preparation for the European Space Agency’s CRISTAL mission indicated that the dominant return comes from the ice surface at Ku-band and from the snow surface at Ka-band for smooth first-year ice. However, for rougher multi-year ice, the main scattering comes from the snow surface for both Ku and Ka-band. These findings depend on the parameterisation of the roughness. This work offers insight into the dominant surface return for Ku and Ka and paves the way towards a physical retracker using SMRT to retrieve snow depth and sea ice thickness for radar altimeter missions.
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
(10980 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-3615', Stefan Hendricks, 14 Sep 2026
-
RC2: 'Comment on egusphere-2026-3615', Anonymous Referee #2, 30 Sep 2026
Review of “Altimetric Ku-band Radar Observations of Snow on Sea Ice Simulated with SMRT” by Julien Meloche. The MS is describing a comparison of simulated and CryoSat-2 waveforms from a number of sites where in situ data were collected. After tuning of the model with surface roughness and “normalization” there is a match between observations and simulations and then the model is used to simulate the waveform sensitivity to the different parameters in the system. The need for roughness tuning gives low confidence in the model’s ability to simulate observations. Either there is something missing from the model or the simulations or the roughness is not measured correctly.
SMRT is a collection of different models and as I understand it SMRT is not used for computing the waveforms. The Larue approach is used. What is actually done could be more clearly described in the introduction. Anyway, this attempt is much better than earlier attempts where surface scattering was not included in the model and in my opinion the development of an altimeter module in SMRT could have been much faster (it has taken a while) if the authors had read the literature on sea ice radar altimeter modelling. It could also have helped the planning of the field sampling to focus on the important parameters and to include sampling in the right places. For example, SSA is less important (even though the description takes up a lot of space in the text) and spatial variability, especially on MY ice, is not well described or quantified. It is not totally clear if the objective is to present a sea ice altimeter module to SMRT or to do a sensitivity analysis. This could be clarified.
The introduction is messy and this should have been rewritten before submission.
There are some references missing. These could have helped in the discussion of the results and I have suggested some in the specific comments below. There are also references from the text missing from the literature list and in some places original references are not used.
There are both bigger and smaller issues, but after major revisions I think that the MS could be publishable.
Specific comments:
L1: add “the” before “polar”
L1: The sentence is unclear; the scattering horizon can be estimated accurately but it is the interaction with the surface where existing 1D models fail (ice surface scattering only) + the buoyancy correction fails. Please clarify.
L4: add “Here” after “.”
L4: what is meant with “low-resolution”? CS-2 is a SAR, see also L7. Please clarify.
L10-11: “…simulations of backscatter…” is that your simulations? If not include your simulation results here.
L11: “…dominant return comes from the ice…” The snow could affect the track-point even if the snow return is not dominant. Please clarify.
L12: “…for rougher multi-year ice…” add “and smooth snow” or is it not like that?
L18: replace “warmer climate” by “greenhouse gas forcing”, add e.g. a reference Dirk Notz, Julienne Stroeve Observed Arctic sea-ice loss directly follows anthropogenic CO2 emission.Science354,747-750(2016).DOI:10.1126/science.aag2345.
L20: add “air” before “temperature”
L23-24: Suddenly this concept of freeboard and thickness. Please briefly introduce the (Archimedes) principle and put a reference to e.g. Laxon…
L24: list all parameters affecting the buoyancy.
L27: To that list of parameters affecting the waveform you should add “spatial variability” and add e.g. Tonboe, R. T., L. T. Pedersen, and C. Haas, 2010: Simulation of the CryoSat-2 satellite radar altimeter sea ice thickness retrieval uncertainty. Can. J. Remote Sens., 36, 55–67, https://doi.org/10.5589/m10-027.
L31: Please use original references, is it a reference to “facets at nadir” or “track-point”, I think facets. E.g. Fung and H. Eom, “Coherent scattering of a spherical wave from an irregular surface,” IEEE Trans. Antennas Propag., vol. AP-31, no. 1, pp. 68–72, Jan. 1983.
L33: this sentence is unclear: “…in order to estimate the range with the track-point correctly”. The range to what?
L34: unclear: the retracker algorithm can use thresholding to estimate the range or invert a model to estimate the surface/range… please rewrite.
L35: It is not clear what is meant: “and rely solely on obtaining a robust estimate of the range”
L36: the threshold retracker can also be used on SAR waveforms, please be specific.
L38-44: this paragraph starts with rejection of Brown, then jumps into ice type, but in the next paragraph you are still discussing retrackers… please rewrite the paragraph.
L46-47: delete “Backscattering contributions from the surface”, and then I guess that the idea with physical retrackers is to compensate for some of the surface variations… please clarify.
L50: what is the “scattering horizon” please define.
L59: please define s and l and in addition to those two roughness parameters the backscatter is also affected by the dielectric constant, wavelength, angle…
L60: the 2.2cm is that a roughness criterion or a validity region for the IEM? Please explain and give a reference.
L61: please explain the detrending procedure and perhaps the following reference could be helpful here: Dierking, W. Quantitative roughness characterisation of geological surfaces and implications for radar signature analysis. IEEE Transactions on Geoscience and Remote Sensing, 37(5):1855–1867, 1999.
L65: I am not really sure how to read the sentence starting with “The facet-based…” or if it is true… the model lacks representation of the microstructure, at least it has snow grains. Please clarify.
L67-73: you could add a reference after “…challenged” e.g. R. Tonboe, S. Andersen, and L. T. Pedersen, “Simulation of the Ku-band radar altimeter sea ice effective scattering surface,” IEEE Geosci. Remote Sens. Lett., vol. 3, no. 2, pp. 237–240, Apr. 2006… but I think that you have to state up front that you are skeptical about Lawrence… please rewrite that paragraph.
L75: add to the list “spatial variability”.
L75: “state-of-the-art” is that the emission model or the backscatter model? This is mixed up here and in the following. So while the emission model could be “state-of-the-art” I think that it is a bit early with the backscatter model, before you have even introduced it.
L92: Site and data are not part of methods.
L126- onwards: you are describing the SSA and micro-CT scans, but in the end it is not very important. Could be shortend.
Page 6 or 7 please provide a table with the input parameters to the model, including the number of layers, are there more than one snow layer? and what is the roughness of the interfaces?
L175: what is the low-rate mode? Earlier you talked about the low-resolution mode.
L177: what kind of surface scattering model is used? Please specify.
L181: that implementation is described in: https://doi.org/10.1049/PBEW052E
L185: is the IBA or the DMRT used? Please be specific. After reading this it is not clear what you have done. This is not a discussion, it is a presentation of the model.
L192: “as well as the delay implied by the horizontal extent of the beam,” this is not clear, perhaps a figure?
L217: Equation 5.1 is from the dissertation and not the reference that you are using. The dissertation is missing from the refence list: De Rijke-Thomas, C. O. T. “Investigating Airborne Ku-Band Backscatter of Snow-Covered Sea Ice to Inform Satellite-Based Sea Ice Altimetry” PhD dissertation, University of Bristol, 2024.
I suggest to replace it by the following reference because that is where the equation is from: Fung and H. Eom, “Coherent scattering of a spherical wave from an irregular surface,” IEEE Trans. Antennas Propag., vol. AP-31, no. 1, pp. 68–72, Jan. 1983.
…and for snow-covered ice, the following reference could be useful: Shi, H., & Tonboe, R. (2025). Refraction Considered Radar Equation for Snow-Covered Sea Ice Surface. IEEE Geoscience and Remote Sensing Letters, 22, Article 2000705. https://doi.org/10.1109/LGRS.2025.3576648
L237: what is “Scattering surface roughness”?
L239: you are using the LRM altimetry module but CryoSat is a SAR. Please clarify.
L290-295: By tuning roughness you are risking that you get the right shape for the wrong reasons. Please discuss the pitfalls.
L304: when is GO and IEM used, please explain.
L309: you actually measured the autocorrelation length, so why is l_snow the same as l_ice, please explain.
Figure 4: on the right side (log-scale?) only the top of the waveform is visible?
Table 3: The tuned autocorrelation lengths are much higher than the measured ones. Perhaps the length of the profile was too short? See e.g. W. Dierking. Quantitative roughness characterisation of geological surfaces and implications for radar signature analysis. IEEE Transactions on Geoscience and Remote Sensing, 37(5):1855–1867, 1999.
L373: “Also, the shape of the waveform indicates less coherent return than at Ku-band.” How do you see that please explain.
L381: how do you know that differences between Ka and Ku band is not just due to differences between GO and IEM? Please explain.
Section 3.3 can be reduced.
L438: “SMRT can successfully simulate waveforms of altimeters on sea ice provided appropriate radar-scale roughness parameters are used” Is that surprising? Roughness is perhaps the most important parameter.
L439: On MYI did you measure on hummocks or on refrozen meltponds. This could be important, anyway, please clarify… and “needs” is not the right word here.
L408: …or it represents things that you have not included in your model. Please clarify.
L420: read Dierking!
L423: please give a reference to “…transfer is still lacking”
L432: please elaborate on “… strict nadir angle…” this is not clear.
L492: please delete “state-of-the-art”, I think that there are issues with the roughness measurements and the spatial variability quantification.
Citation: https://doi.org/10.5194/egusphere-2026-3615-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 204 | 136 | 27 | 367 | 26 | 20 |
- HTML: 204
- PDF: 136
- XML: 27
- Total: 367
- BibTeX: 26
- EndNote: 20
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Review of Altimetric Ku-band Radar Observations of Snow on Sea Ice Simulated with SMRT by Stefan Hendricks
Content:
In their paper the authors use a convolutional radar altimeter waveform model fed with snow parameters from in-situ observation to weight the contribution of different backscatter sources for different scenarios. The paper contains a sensitivity analysis for different parameters and a direct comparison to average CryoSat-2 waveforms, though unconstrained roughness parameters that have an impact on the interpretations of the results had to be obtained by fitting the simulated waveform to the observed ones.
My Background and Review Focus:
My background is on operational sea ice radar altimetry and I have reviewed this paper with a focus on the implications freeboard and snow depth retrieval and less on the processing of the snow samples or the scattering implementation in SMRT.
General Comments:
I could follow the scope, chosen methods and data sources of the paper and I didn’t find substantial issues with them. Some of the terminology used for the radar altimetry seemed a bit unclear and a few citations don’t quite support the statements in which context they have been placed. I have marked them in the commented manuscript file with recommendations.
For the results and their interpretation clarification are needed and an extended discussion for a wider application beyond the location of the in-situ data would improve the paper. Please see discussion points 1 and 2 below.
There are also at least two unfinished citations (one data citation and a scientific citation) that need to be completed.
The code for generating the results is accessible and I have made use of it during my review. The software toolbox SMRT is an community tool and generating and maintaining open-source code is a laudable effort. There is a parallel paper from some of the authors including SAR Mode waveform models in SMRT (https://doi.org/10.5194/egusphere-2025-6056), which would be interesting to see applied for the cases here as well. But that would be a mere recommendation from me and is not a formal part of this review.
Besides the discussion points below I attach a version of the manuscript with comments.
[Discussion Point 1] On “scattering horizons” and “track points”:
A central concept in this work is the “scattering horizon”. In L70 , the scattering horizon is defined as the “track point” at the leading edge of the waveform which represents a certain surface, e.g. the ice surface for first-year ice at Ku-Band (L467).
This is an frequently used definition in the scientific literature, which links the backscatter elevation distribution directly with the two-way travel time obtained by the retracker. E.g. snow backscatter raises the overall backscatter elevation distribution and thus the range obtained by the track point will be shorter. This effect presents a bias for freeboard retrieval, especially at Ku-Band as well as for snow depth retrieval from dual-band (Ku and Ka-Band) range differences.
The direct relationship between backscatter elevation and retracked range is fragile in reality though because of the inherent assumption that sea ice backscatter does not depend on elevation (EM bias) or location within in the footprint. The strong backscatter difference between young, potentially snow-free sea ice and older FYI or MYI does often enough result in substantial range errors that are unrelated to backscatter elevation distribution, but the presence of strong localized backscattering surface in the off-nadir The assumption of homogeneous sea ice backscatter is a weakness of the convolutional waveform models and the only method able to include this effect are the faced-based models.
The ground location in this paper (land fast sea ice in Cambridge Bay and Eureka weather station, as well near-coastal MYI north of Alert) may be as close to the ideal case as possible and the errors by the assuming homogenous sea ice backscatter in the convolutional waveform model are likely acceptable. But I don’t recommend to generalizing the results, because there are very different backscatter environments for both MYI and FYI in the Arctic.
In this sense it would be interesting to see where the SMRT models place the track points. This would give a more quantitative assessment of expected range errors for different snow scenarios than specifying which surface yields the larger power fraction in a waveform.
Discussion Point 2: Detail on the CryoSat-2 data and implication for CRISTAL retrieval:
I find the description of CryoSat-2 data selection and waveform alignment a bit too light. From inspecting the code on github I could see that the authors are using the 1Hz pLRM waveforms from CryoSat-2 ICE L1b data and are aligned the waveforms by their maximum. This information is important for reproducibility and should be included in the text.
Also, it is not clear to me why the offset of waveform maximum (?) and the nominal tracking gate is an issue if the waveforms are aligned anyway and the offset between neighbouring waveforms is included in the variable `window_del_plrm_01_ku. Or do the author use shifts in the window delay to identify land influence?
Averaging a number of already temporally coarse waveforms means that the waveforms are representing a substantially larger area (~30km along-track) and variability than the in-situ observations. Could this be one of the reasons, of why the authors have to optimize the roughness parameters?
I recommend here that the authors discuss the impact of the coarse resolution, especially in the backdrop of a potential application to the CRISTAL retrieval (L490). For CRISTAL, Ku SARin and Ka SAR mode will be used on the full temporal resolution, which is rather different setup from the CryoSat-2 data used in this manuscript.
Conclusion:
The paper should be publishable after the authors have addressed my review at the discretion of the editor.
Best Regards,
Stefan Hendricks