the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 20 Mar 2025
Active-passive microwave scattering in the Antarctica wind-glazed region: an analog for icy moons of Saturn
Abstract. Microwave radiometry and scatterometry, two complimentary modes of sensing the composition and structure of the top meters to hundreds of meters of the subsurface, are often difficult to reconcile, both on the Earth cryosphere and on icy moons of Saturn. To help interpret and model microwave scattering in porous, high-purity ices, we examine jointly 6.9 to 89 GHz AMSR2 radiometry in vertical (V) and horizontal (H) polarizations as well as 5.2 GHz ASCAT, 13.4 GHz QuikSCAT, and 13.5 GHz OSCAT scatterometry in the wind-glazed region of the East Antarctic ice sheet. The data are simulated using the Snow Microwave Radiative Transfer (SMRT) with a simplified snowpack with constant temperature and continuously increasing grain size and density with depth. For the first time, we show that scatterometry and 6.9 to 37 GHz radiometry at V polarization can be successfully simulated with a unique simple snowpack model, indicating that incoherent volume scattering on subsurface heterogeneities dominates both the active and passive signals. To also simulate H-polarized radiometry, a thin surface ice layer as observed in the wind-glazed regions is necessary. Additional complexity, such as seasonal temperature variations, surface roughness, or non-continuous density variations, is necessary to explain the 89 GHz data and HH-polarized backscatter. Meanwhile, applying the same approach to simulate simultaneously passive and active Ku-band observations of icy moons improves on previous attempts but remains unable to reproduce the very high backscatter observed, highlighting the importance of coherent scattering and possibly unknown large (at least millimetric) icy structures in the subsurface.
- Preprint
(3249 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-3972', Anonymous Referee #1, 26 May 2025
Review “Active-passive microwave scattering in the Antarctica wind-glazed
region: an analog for icy moons of Saturn”
Summary:
This paper presents simulations of both active and passive microwave observations over the Antarctic wind-glazed region using the SMRT (Snow Microwave Radiative Transfer) model across frequencies from 5 to 89 GHz. The study aims to evaluate the Antarctic as an Earth analog for Jupiter and Saturn’s icy moons, and applies the same modeling approach to explore the radiative properties of icy moon surfaces. The results show promising potential in using Antarctic conditions to interpret extraterrestrial scattering environments, and highlight SMRT’s flexibility in handling both passive and active configurations.
However, in my opinion several issues require clarification before publication:
The description of the scatterer configuration requires clarification and refinement. In particular, terminology such as “microwave grain size,” “optical grain size,” and “stickiness” should be more clearly defined. These terms are often used interchangeably or without context, which creates confusion. It would greatly benefit the reader if the authors could explain how these terms are derived, how they relate to measurable physical properties (e.g., grain radius, correlation length, microstructural anisotropy). A more precise treatment of these definitions would strengthen both physical interpretation and modeling transparency.
The treatment of the Coherent Backscatter Opposition Effect (CBOE) correction and the relationship between grain size and wavelength require more thorough justification, particularly in light of the frequency-dependent scattering regimes involved.
There appears to be inconsistency in the identification and justification of the best-fit model. If the simulation results are intended to represent best fits, the retrieval methodology and whether a formal optimization was performed should be explicitly described.
Despite these issues, the paper offers valuable insight into SMRT’s performance in cold, dry, high-frequency regimes, and contributes to understanding terrestrial–planetary analogs. Addressing the points above will significantly strengthen the scientific clarity and rigor of the study.
Thank you!
Specific comments:
Line 1-5:
The phrase “complimentary modes” should be corrected to “complementary modes”.
Additionally, the expression “top meters to hundreds of meters” could be refined to “upper meters to several hundred meters” for improved readability.
Line 5-10:
The repetition of “with” in “with a simplified snowpack with constant temperature” may be improved by rephrasing.
Line 15-20:
The use of “diffuse scattering” in this sentence is slightly ambiguous and breaks the otherwise quantitative list structure. Consider rephrasing or elaborating to clarify whether the diffuse behavior refers to surface roughness, volume scattering, or both, and how it contributes to the anomalous radar response.
Line 20-25:
The phrase “simple radiative transfer models” is somewhat vague. It would strengthen the introduction to specify what assumptions or limitations define these models (e.g., homogeneous layering, absence of coherent effects, neglect of subsurface structure). This would help the reader understand the nature of the modeling challenge and the physical processes potentially being overlooked.
“Indeed, the Cassini Radar instrument, which operated in the Ku-band (13.78 GHz frequency, 2.2 cm wavelength) measured high backscatter but also higher than expected microwave emissivities.”
The sentence as written is ambiguous. It is unclear whether the high backscatter and elevated emissivity are each individually unexpected, or whether it is their coexistence at the same frequency that is anomalous. Clarifying this would improve the reader’s understanding of the observational paradox being described.
Line 25-30:
“The surfaces of Saturn’s icy moons are constituted primarily of high-purity water ice…”
It would be helpful for the authors to clarify whether this refers solely to the immediate optical surface (as seen in reflectance spectroscopy), or if it includes the shallow subsurface layers that contribute to radar and microwave responses.
Line 45-50:
“…which is the least emissive and most scattering region…”
The rationale for selecting a region described as “least emissive” warrants further clarification. Given that the icy moons of Jupiter and Saturn are characterized by both high microwave emissivity and strong radar backscatter, it is not immediately clear how a low-emissivity terrestrial region serves as an appropriate analog.
The phrase “transports and sublimates snow downslope” is ambiguous. It is unclear whether the authors intend to describe snow sublimation as occurring simultaneously with transport, or as a separate surface process enhanced by katabatic wind.
Line 50-55:
“These large crystals, very scattering at microwave frequencies, are responsible for the observed low emissivities and high radar backscatter”. The expression “very scattering” is informal and potentially ambiguous. Consider rephrasing to describe the crystals as strong or efficient microwave scatterers.
Line 55-60:
The phrase “cause important scattering” is imprecise. Consider replacing it with more specific terminology, such as “enhance subsurface volume scattering” or “increase microwave scattering efficiency.” Additionally, you may wish to briefly clarify how sublimation–deposition cycles contribute to the formation of polygonal grains and grain chains.
Line 60-65:
The use of “shorter frequencies” is not standard terminology. Consider revising to “lower frequencies”
The explanation of how increased penetration depth and enhanced scattering by larger grains "compensate each other" to yield a flat brightness temperature spectrum between 19 and 37 GHz lacks clarity. It may be useful to first clarify that lower frequencies (e.g., 19 GHz) penetrate deeper and may therefore sample warmer subsurface layers, potentially increasing brightness temperatures. However, the statement that "longer wavelengths are also sensitive to larger grain sizes" is not always valid, especially considering that the wavelength difference between 19 GHz and 37 GHz is only a factor of two. It would be helpful to clarify the intent behind introducing this flat spectral behavior: is it meant to serve as evidence of grain size increasing with depth, or simply a result arising from known structural gradients? If it is intended as evidence of depth-dependent grain evolution, the current explanation is not sufficiently convincing. If it is instead presented as an outcome, then rephrasing to reflect that distinction would improve clarity and interpretation.
Line 86
The role of multi-frequency observations in constraining vertical variations in grain size and density is important, and could be more clearly articulated. It is not immediately clear what the authors mean by the “frequency dependence of grain size and density.” Grain size and density are physical properties of the snowpack, not functions of frequency. Rather, it is the instrument's sensitivity to those properties that varies with frequency, due to differences in penetration depth and scattering behavior. Clarifying this distinction would strengthen the physical interpretation and aid in understanding how multi-frequency data constrain subsurface structure.
It would be helpful for the authors to include the spatial resolution of each observational dataset in Table 1.
Line 95-100
The explanation of the Kohonen clustering could benefit from additional clarification. While it is mentioned that the algorithm identifies clusters with neighborhood constraints, it would be useful to briefly describe what the clusters represent in the context of the dataset—i.e., distinct combinations of backscatter and emissivity properties or distinct physical parameters. Clarifying how the number of clusters (10) was chosen and what physical insights each group provides would help readers interpret Figures 2 and 3 more effectively.
Line 107
The use of ERA5 skin temperature to compute observed emissivity should be more carefully justified in light of the modeling assumptions. In the simulations, a uniform temperature profile is applied, whereas the skin temperature used for observed emissivity inherently contains seasonal and diurnal variability. This mismatch is especially important at lower microwave frequencies, which probe deeper into the snowpack where temperatures remain relatively stable and are decoupled from short-term surface fluctuations. The authors should clarify how the observationally derived emissivity, based on variable skin temperatures, is meaningfully comparable to the model-derived emissivity under the assumption of isothermal conditions. A brief sensitivity analysis or discussion of potential biases would improve the credibility of the comparison.
It would be valuable to assess how the derived emissivities vary with time of year and with seasonal temperature fluctuations.
Line 109
“To remove seasonal temperature variations, AMSR2 data…”
Consider change “data” to “emissivity” to avoid confusion with brightness temperature.
Line 130
It would benefit the reader if the authors could explicitly explain the physical reasons why Ku-band is more sensitive to scattering than C-band in this context.
Line 146
Remove one “the”
Line 149
For clarity, it would be helpful to briefly explain what polydispersity is.
Line 160
There appears to be a contradiction in this paragraph. The authors initially state that random structural variations are especially important in the top few meters, but then conclude that these do not significantly affect the microwave observations presented in Section 2. This claim seems unsupported. Rather than assuming these fluctuations are negligible, the authors could instead justify this by noting that the observations are averaged over a full year, which likely smooths out the influence of transient or small-scale random variability. Rephrasing to reflect this reasoning would improve logical consistency and credibility.
Line 161-164
While the manuscript appropriately notes the limitations of assuming continuous vertical variations in grain size and density—especially in the context of H-polarized radiometry and high-frequency channels—it would significantly strengthen the analysis to include a quantitative estimate or sensitivity test illustrating how this assumption affects model performance. Even a simplified comparison could help assess whether this limitation introduces minor biases or fundamentally alters the interpretation of the signal.
Line 165
The manuscript describes a 10-layer snowpack model as a “continuous” profile. However, with discrete layers and potentially discontinuous density between adjacent layers, the profile is in fact piecewise continuous, and the SMRT model will simulate internal reflections at these interfaces. The authors should clarify whether such reflection effects are significant in their simulations, particularly for H-polarized channels.
Line 167
The authors should explicitly state what material was used as the substrate in the SMRT simulations.
Line 169
The phrase “the same model has been used on Jupiter’s icy moons by Brown et al. (2023)” is ambiguous. It would be helpful to specify whether “same model” refers to the use of the SMRT radiative transfer framework, the same physical configuration (e.g., snowpack structure, grain sizes, density profiles), or identical boundary conditions such as substrate properties. Clarifying this would help the reader understand the degree of comparability between the Earth-based and planetary modeling efforts.
Line 183
For solid ice, is grain size still meaningful?
Line 197-217
This section conflates multiple distinct scattering parameterizations available in SMRT, resulting in conceptual ambiguity. Specifically, it is unclear whether the authors are using the sticky hard sphere model—which introduces inter-particle interactions via a "stickiness" parameter—or a correlation length-based model, where scattering is governed by permittivity fluctuations described by quantities such as the Porod length and correlation length. These two approaches are physically distinct and are implemented independently in SMRT. If the authors are using a correlation-length-based scattering model, then references to “stickiness” are misleading and should be removed or rephrased appropriately.
Additionally, the paragraph references the method of Picard et al. (2022a, 2022b), but the implementation remains unclear. For instance, the sentence: “To use the correct value as correlation length, we use the microwave grain size, defined by Picard et al. (2022a)...” raises several questions. Is the microwave grain size being used directly as the correlation length lc? If not, what is the precise relationship between lc and lMW? Furthermore, the value K=0.62 from Picard et al. is described as “best for Antarctic snow,” but was derived in which region of Antarctica? Can the authors justify applying it to the current study site?
To improve clarity, the authors should explicitly (1) state which scattering model in SMRT was selected (2) define how input parameters like optical grain size, microwave grain size, and correlation length are defined, derived and linked, and (3) avoid blending terminology from physically incompatible models. Without this clarity, it is difficult to evaluate the validity or physical interpretability of the simulation results.
Line 237
The manuscript should clarify whether the results presented in Figure 5 correspond to best-fit simulations under the stated assumptions. If so, please specify the method used to identify the best-fit model: was the parameter space (e.g., grain size, density, temperature) systematically explored using an optimization or sampling approach? Or were the parameter values selected based on physical reasoning or prior field measurements?
Line 246
The author should explain why a crusted surface is necessary to explain the 89GHz data.
Line 247-254
The manuscript compares the grain sizes retrieved in this study with those reported by Brucker et al. (2010), but the basis for this comparison is unclear. Are the two studies analyzing the same geographic region? Besides grain size, are other parameters (e.g., snow density, temperature, layering assumptions) held constant? Given that the model used in this work (Picard, 2022) incorporates more advanced microstructure parameterizations than models available in 2010, a direct comparison of grain size may not be meaningful. A more physically grounded comparison would involve correlation lengths or scattering coefficients. Furthermore, it would be informative to discuss how well the Brucker et al. model fits the current observations — do those earlier results underperform compared to the current model, or do they provide complementary insights? Clarifying these points would strengthen the validity of the comparison and help readers interpret the model advancement more clearly.
Line 259
The phrase “several different datasets” is ambiguous. It would be helpful if the authors clarified what distinguishes these datasets — are they from different satellite instruments (e.g., active vs. passive), cover different frequency channels, represent distinct geographic regions, or span different time periods? Explicitly defining what is meant by “different” in this context would be helpful.
The author should also explain briefly how this range of parameter values are determined. What retrieval process is used?
Line 279
Q3 model
R^3(15m) = (0.1[mm])^3 + (50*1e-9 [mm3/m]) * 15[m] R(15m) ~ 0.1mm
Q2 model
R^2(15m) = (0.3[mm])^2 + 6e4*1e-6 [mm2/m]) * 15[m] R(15m) ~ 1 mm
“The grain radii we find in the subsurface are very large, with values around 1 mm at 15 m depths.”
Please specify this is for the Q2 model and also note the corresponding value for the Q3 model.
Line 274
The statement that “ASCAT data, which is at lower frequency than 10–89 GHz AMSR2 and therefore probes deeper…” oversimplifies the depth-sensitivity comparison between active and passive sensors. Although ASCAT operates at lower frequencies, it is an active instrument and thus involves a two-way signal path, which significantly alters penetration behavior. Please consider rephrase this argument.
“can never be reproduced for the same configuration as AMSR2 data”
But in Fig6 panel C, the region between the two yellow lines does overlap with the observation range, why does the author make this statement?
Line 288
The manuscript suggests that the H-pol radiometric signal (AMSR2) cannot be reproduced due to polarization effects related to surface or subsurface layering. However, it is unclear why similar polarization effects would not also affect the active radar backscatter at HH polarization. Since both H-pol radiometry and HH-pol radar are sensitive to horizontal interfaces and anisotropy, some discussion of why the active signal can still be matched — while the passive signal cannot — would strengthen the interpretation.
Line 298
“Indeed, it does not account for the observed ice crust in the wind-glazed regions, the variations of temperature with depth and season, and the random variations of density and grain size at these depths.”
The explanation that the spectral slope cannot be reproduced due to “variations of temperature with depth and season” seems questionable, as the observations have already been averaged over an annual cycle. This averaging should significantly reduce the impact of seasonal thermal variability. Additionally, if random fluctuations in grain size and density were responsible, it is unclear why the slope would appear consistently across the entire region of interest — random features should not produce a coherent spatial signature. Moreover, in Section 4.3, the inclusion of an ice crust still does not recover the slope, which suggests that other structural or radiative mechanisms may be at play. A more detailed investigation or alternative hypothesis may be warranted to explain this persistent model–data mismatch.
Line 380
Please define NRCS.
Line 398
The application of a factor of 2 to the total SMRT backscatter to account for the Coherent Backscatter Opposition Effect (CBOE) appears problematic. The CBOE only enhances the multiple scattering component — not the total signal, which includes both single and multiple scattering. Since the SMRT output includes both contributions, applying a multiplicative factor to the full signal likely overestimates the CBOE enhancement. A more accurate approach would be to isolate the multiple scattering term (if possible within SMRT) and apply the enhancement selectively. I recommend revisiting this correction or clarifying its justification with reference to the physical assumptions and model limitations.
Figure.8
The observed correlation between ASL and emissivity is clearly demonstrated, but the scatter in the relationship suggests that other physical factors are also influencing the outcome. It would be helpful if the authors briefly discussed the main contributors to this spread. Identifying which parameters have the strongest and weakest influence in this context would provide valuable insight, and help readers understand the limitations and sensitivities of the model.
As noted earlier, the SMRT simulated model over-estimates the emissivity in H pol. How is this effect going to affect your simulated correlation between ASL and e?
Please double check the units for Q1, Q2 and Q3 both in figures and text. There are quite a few places that the units are not correct.
Line 391
Based on the comparison presented in Fig6, the model using a Q3 grain size increase appears to fit the observations more closely than the Q2 model, with model parameters rtop=0.1mm, Q3=50um3/m. However, the text states that Q2 represents the best-fit model and with rtop=0.3mm, Q2=1e5um2/m (unit is wrong in the text, and in Fig6 Q2=6e4um2/m). This discrepancy should be addressed explicitly. If Q2 is preferred for reasons beyond data matching, this justification should be clearly stated. Otherwise, the statement about the best-fit model should be revised to align with the presented results.
Line 412
The statement that the signal is driven by “incoherent scattering on large subsurface grains” is potentially misleading. In the Q2 model, the grain size ranges from 1 mm at 15 m depth to 3 mm at 100 m. These sizes are relatively small compared to the microwave wavelengths under consideration (1–6 cm), and may not qualify as “large” scatterers in scattering theory.
Moreover, coherent scattering effects are generally most pronounced when scatterer sizes are comparable to the wavelength. The current phrasing is misleading, and I suggest reorganizing this argument to clarify the scale-dependence and the physical regime being invoked.
Line 413
The statement that “an ice crust over the snowpack is necessary to reproduce H-polarized emissivities” appears too strong based on the evidence presented. While the inclusion of an ice crust lowers the modeled TB at H pol, it doesn’t really improve agreement with observations. The paper does not demonstrate that this is the only viable configuration capable of doing so. Alternative explanations are not ruled out. I suggest rephrasing to indicate that the crust is a plausible or effective solution, rather than a proven requirement.
Citation: https://doi.org/10.5194/egusphere-2024-3972-RC1 -
AC1: 'Reply on RC1', Catherine Prigent, 05 Sep 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2024-3972/egusphere-2024-3972-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Catherine Prigent, 05 Sep 2025
-
RC2: 'Comment on egusphere-2024-3972', Anonymous Referee #2, 19 Jun 2025
-
AC2: 'Reply on RC2', Catherine Prigent, 05 Sep 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2024-3972/egusphere-2024-3972-AC2-supplement.pdf
-
AC2: 'Reply on RC2', Catherine Prigent, 05 Sep 2025
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 390 | 61 | 16 | 467 | 20 | 47 |
- HTML: 390
- PDF: 61
- XML: 16
- Total: 467
- BibTeX: 20
- EndNote: 47
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1