Very interesting and pertinent multi-method study which intends to contribute filling knowledge gaps on various structural, thermal and kinematic aspects of rock glaciers, whose current activity state, meaning their ability to transfer sediment downslope, is ranging from relict (no movement) to transitional (few localized or widespread movement).

In its current form, the paper suffers however several weaknesses.

There is a frequent conceptual mixing between landforms (rock glaciers), processes (permafrost / rock glacier creep), ground thermal state (permafrost conditions or not), ground ice occurrence at depth and kinematics, which prevents a clear understanding of the paper content (data interpretation). This must be disentangled.

There is often, but not everywhere, some lack of precision/clarity in the text (is the purpose fully clear and understandable for the reader ?).

There are several questions arising about the applied methods and the interpretation of their results.

The illustrations must be improved.

In general, the paper could be very worth of being published in ESurf. However, in its current form, it requires major improvements.

My comments are going sometimes pretty far into details, when not too much. This must be understood as a constructive contribution from my side in order for the authors to improve the pertinence of their paper.

Detailed comments

Abstract

Not fully supported by data. For instance, (L. 29) “The slow surface movement of rock glaciers in the marginal periglacial mountains is driven by the deformation of thin, frozen layers”. First, the type of movement, which appears to occur in some parts of some of the studied rock glaciers in the Retezat mountains, has not been explained (permafrost creep ? vertical subsidence ? other cause ?). Second, geophysical investigation has only been conducted on a reduced section of one single rock glacier, which prevents any generalization. Third, the statement appears to never being discussed in the paper (or I missed it). Finally, the authors’ statement cannot be extended to all “marginal periglacial mountains” on the basis of this single study.

This is just one example and an invitation to revise/verify the entire abstract once the paper will be improved.

The last sentence about “the benefit of combining Sentinel-1 SAR data with comprehensive field investigations” is common sense and out of the scope of the paper (-> to be removed).

Main text

L.35. “In high mountains, ice-rich permafrost occurrence is usually associated with rock glaciers”. Rock glacier is a feature of the mountain periglacial landscape. It is not a question of the “highness” of the mountain, but of the occurrence of periglacial conditions, whose lowermost elevation on a mountain range is very roughly related to latitude and overall climate conditions. Ice-rich permafrost is an obscure concept that must be clearly defined by the authors (what do they mean here ? ice-oversaturation ? >30, 50 or 70% of the total volume ?). There are many ice-rich (almost whatever the definition) permafrost occurrences outside of rock glaciers (e.g. moraines, talus, landslides). Ice-rich permafrost is unlikely to occur in relict rock glaciers, but permafrost occurrence cannot be excluded. I imagine that the authors would like to say that the process conducting to the development of rock glacier (permafrost / rock glacier creep) is usually associated with the occurrence of saturated to super-saturated ground ice conditions (ice-rich) on mountain slopes.

L.35-36 “Rock glaciers are masses of debris-ice mixture common in many cold mountains on Earth”. A rock glacier is basically a mass of debris (see RGIK (2023) definition), which undergoes or underwent deformation (permafrost creep). There is no ice (or only few, unsaturated ground ice) in relict rock glaciers. Cold does not mean anything clear. The sentence must be adapted accordingly.

L.36-37. When occurring (not all rock glaciers comprise it), “the coarse debris surface of rock favors ground cooling”. Agreed. It produces a so-called thermal offset. However, why should it decisively contributes to preserving permafrost over long periods ? The thermal offset effect is not expected to change if for instance the temperature is warming. So, what would the authors precisely say? I imagine that they would mean that the rock glaciers in the Southern Carpathians should have formed under colder pre-holocene (?) climate conditions, but that their coarse surface debris has contributed to preserve cold enough ground conditions for permafrost to subsist at least locally? Harris and Pedersen (1998) is maybe not the most relevant reference.

L.38. What is the “themal inertia” of the thick coarse debris layer, which, as said earlier, is not always thick ? What does “thick” mean ?

L.39. Permafrost is not resilient to climate change. It is just thermally responding with delay due to the time necessary to diffuse the surface temperature forcing at depth and the large consumption (when warming) of heat by the phase change of ice to water.

L.40. What is defining the regional limit of permafrost ? Many people have used rock glaciers to do it… Moreover, what is the significance of such geomorphologically-based limit under current changing climate conditions ?

Aside from the surface debris, the occurrence of permafrost conditions at “low” elevation as in relict rock glaciers is possibly due to the internal (advective) ventilation of the rock glacier. Some of the authors have worked on such a process. Maybe worth of adding a few words on it.

The climate is warming. In Central Europe, current MAAT could be up to 2°C warmer than a few decades ago, meaning an approximative rise of the isotherms by 400 m in elevation. One should now be very careful when referring to any MAAT value. The current permafrost occurrences are obviously not defined by the current MAAT, but still in most cases by the climate conditions of the previous decades/centuries.

At the end of this first paragraph, I don’t always understand the link between the various sentences and I don’t see where the authors would like to go. They should first state that their studied region is bearing many rock glaciers usually exposing a coarse blocky surface. According to the stability of their (vegetated) fronts and their subdued morphology, these landforms are mostly considered as relict, meaning having developed under colder conditions than those prevailing during the last centuries/millenia. However, the coarse grained surface and its cooling effect (say how) is expected to have preserved permafrost conditions in more or less large areas of these rock glaciers, in particular in their uppermost sections.

2^nd paragraph

This paragraph is focusing on rock glacier kinematics (movement). There is however no word about the mechanism (creep) making the rock glacier to develop (have developed), as well as about further processes which could contribute to a tiny deformation of the surface to occur independently of the permafrost creep process (e.g. ice-melt induced subsidence, solifluction of the active layer).

In addition to that :

L. 43. “The surface kinematics of rock glaciers have garnered significant interest from the international community in recent years (Bearzot et al. 2022)…”. Right, but the reference is not suited. This is just one study on one single rock glacier. Prefer references like Kellerer-Pirklbauer et al. (2024) (over the entire Alps), Hu et al. (2025) (review paper on rock glacier velocity as a climate indicator – okay, it was published after the submission of the present manuscript), Pellet et al. (2024) (BAMS report, worldwide) and/or other references therein (e.g. Kääb and Røste  2024 about rock glaciers in the U.S.)

L. 45. Explain why rock glaciers are moving faster under warmer conditions. Does this statement also work for very slow moving rock glaciers (< 3 cm/y) ?

L. 47. “… occasional accelerations… (Vivero et al. 2022) ». What is the meaning of “occasional” and why “accelerations” (plural) ? Why not to speak about destabilization and refer to other various related papers (e.g. Roer et al. 2008, Delaloye et al. 2013, Eriksen et al. 2018, Marcer et al. 2021, Hartl et al. 2023). The study by Vivero et al. 2022 is first and mostly a technical paper, referring to a single rock glacier, suffering a single destabilization phase (its first one in this amplitude) since a decade or so.

L. 48. What is this sentence about permafrost warming doing here (section on rock glacier kinematics) ? What is meant by “increased sensitivity” ? Maybe worth of referring also to Noetzli et al. 2024.

L.54. The “ice content in this category of rock glaciers” (transitional) is not known. That it is “below a certain saturation threshold” is only presumed. In such a case the shear strength (internal friction) is too high to favor fast creep movement, but not the “shear stress is too weak”. Is there not something younger than Barsch (1996) as a reference ? E.g. Cicoira et al. 2020 or references therein ?

I stop with the detailed commenting here, but it gives an idea about the improvement the manuscript could need in its precision.

L. 59-61. The expected relationship between surface movement, permafrost occurrence and areas bearing ground ice must be explained. Consistent surface movement (consistent/homogeneous horizontal flow field, not dominated by subsidence) could relate to permafrost creep (the latter must be explained before; note that frozen conditions down to a minimum of about 20 m are necessary for permafrost creep to occur), which point out to the occurrence of thick “ice-rich” permafrost. Isolated or thinner ice-rich permafrost cannot produce rock glacier creep, whereas the latter is also not occurring in every permafrost area.

L. 72. Express the value of the “significant warming that the Southern Carpathians are facing”. It will be very important for the interpretation of the current ground surface temperatures.

L. 85. MAAT must be dated (because of the ongoing significant warming”. MAAT = 0°C at 2000-2100 m a.s.l. is particularly low according to the apparent elevation of the transitional rock glaciers. Provide also elevations for the latter (or refer the adequate section).

L.92. Rock glaciers have developed on granites (the subjacent bedrock is granite) or have they developed from granite (the bedrock in the source area is granite) ?

L.105. It would be worth of precising how the absence of glacier development during the Younger Dryas has been stated (model ?  lack of geomorphological evidence ?). It is important as it provides an idea about when the rock glaciers started to develop ? So, the rock glaciers located in the uppermost valley sections could have already developed during the Younger Dryas or even before and for most of them have become relict or almost relict since the beginning of the Holocene about 10’000 years ago.

Figure 1 :

- Better to provide a hillshade background and the name of the highlighted cities.

- Write in the map legend “ Modelled permafrost extent” or better “Modelled permafrost extent (Popescu et al. 2024)”; Differential GNSS survey or dGNSS survey

- It is a detail, but it would be nice to keep the same orientation on b).

What is in red ? The rock glacier outline I presume. It would be good to apply the RGIK guidelines to first provide both the extended and restricted outlines and to also remove the feeding talus slopes from the rock glacier outline.

Just write P1 and P2 and do it larger.

L. 122-125 If you refer to RGIK (2023), it must be fully respected (not changing the definition), done for the entire paragraph and not mixed with Barsch (1996).

L.132. “…millimetric accuracies, similar to GNSS”: to dGNSS ? The accuracy of the latter is nevertheless not millimetric, but centimetric

L. 135-138. What is Terrasigna’s PSInSAR ? Why is it providing much more data points than EGMS ? Why is it only using one path (descending) ? Are there other differences in the computation of the data in comparison to EGMS or other PSInSAR techniques ?

L. 139. Kinematics instead of dynamics.

L. 140-141. Better formulate. The motion is measured along the SAR LOS direction, whereas the displacement of the rock glacier surface is expected to mainly occur along the slope or in the vertical direction, without that it is possible to evidence / discriminate them.

One should also note that most rock glaciers in the area are north-south or south-north oriented, making that the satellite look angle is not well suitable for a fully consistent detection/analysis of the mass movement occurring along the slope and might conduct to underestimate of the actual displacements.

L. 147. “Due to the significant atmospheric noise in steep terrain…” or in areas with large elevation differences (whatever the steepness) ?

Which mountain plateau ? Where ?

L. 148. “… enabling PSInSAR measurements to cover the summits”. What does it mean ?

Figure 3 : is not legible for me. A zoom on the area of interest or just a part of it would be beneficial. Make also the legend legible. From there, it would also be very worth to describe the limitations and error sources of the applied technique in relation to the characteristics of the investigated terrain (slope aspect and steepness, vegetation, snow patches, …)

L. 158. “According to previous studies…” -> According to RGIK (2023) guidelines

L.160. How were the MAs identified ? On the basis of Terrasigna’s PSInSAR, I guess ?

L.162-163. Again, refer to the RGIK (2023) guidelines instead of Barboux et al. 2014.

L.163-164. I am a bit lost. To my understanding shadow and layover should prevent any (coherent) data acquisition. What are they doing here ? It would have been nice to make use of Figure 3 to illustrate the issue. Then, if you don’t get valuable data, why to map the area as a moving one ? Not sure that this is following the RGIK recommendations.

L. 165. Okay, but change the label of the velocity category to 0.3 – 1 cm/y (instead of < 1 cm/year).

L.166-167. I don’t understand the meaning of the sentence “Because the number of … was low… we have (done the job)…in ArcGIS 10.8”

L. 167-168. The ideas supporting the sentence are incorrect to me or difficult to understand. There is first an implicit association of permafrost occurrence and rock glacier creep, the latter being expressed by the surface movement (here measured in the SAR LOS direction only, without distinction between the along slope and vertical movement). It looks that one can expect that any piece of permanently frozen ground moves, whatever its spatial extent. Moreover, the lowermost size limit of mapping at 300 m2 (17 x 17 m), that the authors applied, appears to be too small for them. But rock glacier creep is a process developing at about 20 m depth over significant areas. Moving areas of less than maybe about 1000 m2 (33 x 33 m) are very probably not consecutive to permafrost creep and should be disregarded in this purpose.

L. 171. Validation of what ? Just write Differential GNSS measurements

L.172 Provide for both rock glaciers the numbers used in Figure 1 and refer to the latter.

It is extremely challenging to measure so low velocities (< 3 cm/year) with sufficient accuracy using dGNSS even on a 2-year basis. The 1 cm accuracy is not fully secured. It is usually expressing a standard deviation and concerns the horizontal (planimetric) positioning (the elevation is worse). The uncertainty of the velocity is larger, up to 2 cm/year (after one year). It is also extremely important to perform the measurement on stable (non-moving) points as well, not only on rock glaciers, in order to estimate this inaccuracy. Has this step been performed ?

Another way of checking the quality of the data is to map the displacement vectors and to look for their consistency. Such kind of information must be provided. Otherwise, it is not possible to trust the data.

L. 182. The RGIK guidelines are not describing how to conduct the interpretation of interferograms, but they refer to another document.

L.186. The BTS method does not allow to map the occurrence of permafrost with sufficient precision. The threshold defined by Haeberli >50 years ago were developed on glaciers/debris-covered glaciers/moraines (if I correctly remember) and were not suited for coarse blocky surfaces. There are many factors influencing the ground surface temperature under a well-developed snow cover, the first one being the constitution of the surface ground itself. A dry and porous bouldery surface (frequent on the investigated sites) tends to systematically provide lower BTS values. The early winter snow conditions are also extremely important. Air flow within the porous surface debris or deeper throughout the landform are likely to occur. So, the BTS method is very useful to evidence areas with colder ground surface temperature from those with warmer ones, and it must be mapped like that (not just using the -2/-3°C thresholds as on fig.12). The interpretation of the values themselves should then be done with great care.

L. 202. How was WEqT computed ?

Note also earlier guidelines about WEqT by Schoeneich during the PERMANET project around 2012, which himself inspired from earlier works (e.g. PERMOS in Switzerland)

L. 246. MAs are up to 10 cm/year according to Fig. 4. The authors would like to say the maximum PSI values are in the range of 5 cm/year (which is not around the uppermost detection capacity of the method ?) ?

What have been the rules to define the occurrence and the extent of a MA ? And then to attribute a velocity class ? This is unclear to me from Fig. 4.

L. 256. To what extent is the distribution of MAs among aspects related to the specificities of the investigated area (e.g. frequency distribution of aspects) ?

Figure 4:  very difficult to read. Better to zoom on a smaller area

Figure 5:  the vertical scale in incorrect (cm); yellow curve is not legible; location map as well. It would be nice to have the location of the geophysical profiles on the location map.

I am wondering how the purple curve / time series has been produced. The winter jumps are about 2.5 to 3 cm, namely one phase on Sentinel C-band. How is the winter gap solved ? The behavior during the summer is also almost flat, at least in 2018-2019-2020. Is the total displacement really so large ?

L. 269ff. The attribution of rock glaciers to the transitional category seems to have been overestimated. Looking at figure 10, I don’t understand how have both big rock glaciers facing north close the center of the map attributed to the transitional class. There is almost no movement on a very large part of their surface.

Fig. 8. I would suggest to have the maximal elevation (2500 m a.s.l.) at the center, forming like a summit with slopes all around. Easier to read (at least for me).

Fig. 10 a and b) are both very difficult to read. They should focus on a smaller area or smaller ones.

Why the leftmost transitional rock glacier considered as such and different from fig. 9, where it is drawn as a relict one ?

Fig. 10a : what is this PSINSAR data, which looks to be different from those presented on fig. 4 ?

Fig. 10b : “… the fringe cycle (bottom right) represents the change of colour”. Please explain to the reader how to read the color (changes) according to the data which is used. What is the LOS (line-of-sight) ?

Note that the data of Fig. 10b is almost not used/discussed in the paper except two short sentences (L. 286 and 288). But the reader is left alone to read the data and calculate the displacement by him/herself.

L. 298. Are the mean DGNSS (and not DGPS) velocities calculated over 2 years ? And how ? By computing the velocity between the initial and final positions (2-year span) – which should be the best way of doing – or by averaging the two annual velocities – not recommended ?

Are the velocity range and frequency for the rock glaciers only or do they also integrate points outside ?

L. 303-4 and fig. 11a : there are to me several issues :

- GNSS : we would like to know the flow direction (in order to evaluate the coherence of the flow field and the data itself)

- At Judele, most GNSS points with velocities between 1 and 2.8 cm/y (this is mapped as 1 – 3 cm/y ) are located in either slower (< 1 cm/y) or faster (3-10 cm/y) PSINSAR-based moving areas (MAs). So, things are not working at all. Why ? Is the GNSS data failing ? The PSINSAR ? Both ?

- The MAs mapping appears weird. Purple areas (3-10 cm/y) are surrounded by green ones (< 1 cm/y), the intermediate yellow ones (1-3 cm/y) being missing. The latter are however located outside of the green ones. Such a flow field is not consistent.

- One should also note that the rock glacier could be expected to move to the NE (maybe partly to the N), the use of InSAR data in descending mode in not adequate.

L. 309. See my comment in the related methodological section

L. 310. “highest” means “warmest”, right ?

Fig. 12. Provide actual BTS values in order to see where the colder/warmer areas are (as commented earlier). I am also sceptic about the MAs mapping (it appears to be too detailed considering the uncertainties of the method)

L. 325. “in all the years” instead of “in all the seasons” ?

Fig. 13a : better to provide MAGST under the form of running means instead of just single values (hydrological year)

Fig. 13c: Great dataset ! There is probably more to exploit from it. For instance, the zero curtain phase (snow melt phase) appears to be pretty long, confirming the installation of thick insulating winter snow cover. Despite it, in many years and on many sites, there are still temperature variations during most of the winter. As in 2013/14, they even appear to be inverted for a part of the winter between Pietrele and Judele. It would be also nice (valuable) to compare them to those of the air temperature (if possible). What are these temperature variations telling us ?

L. 339. How is it possible to get consistent information on this 3-4 m superficial layer from an ERT profile using an electrode spacing of 4 m ?

L.341. Could this second patchy resistive layer (please refer explicitly to the figure and show it) alternatively be porous debris ? Why is it talked about “remnants” of ice-rich permafrost and not just about patches ? Remnants from what ?

L. 342. This layer should correspond to the rock glacier body, not to the bedrock (what is the resistivity of the bedrock ?). Note that with 20-30 kohm.m the frozen state of the medium cannot be excluded.

L. 346. One should note that talus (as relict and some transitional rock glaciers) can be ventilated and by this way, deeply frozen (unsaturated). I don’t know if it makes sense regarding the investigated site.

Fig. 14a. It looks that the “ice-rich ?” and the corresponding arrows are shifted. They are some almost invisible dotted lines (same on fig. 14b), but we don’t know what it is.

It would be nice to make the color scale more accessible for the reader (e.g. what is the yellow actually representing). There are 3 orders of magnitude represented and this is almost impossible for the reader to relate the colors with an actual resistivity value. Maybe some contour lines could be inserted. Same for Fig. 14c and 14d.

Fig. 14d. From the color scale, it is impossible to read this data (we don’t know what values are mapped). We don’t know the unit as well.

I am also very surprised that the “bedrock” shoulder (low resistivity/high velocity) provides as much ice content as the rock glacier (namely the debris-constituted ground).

L. 357. As commented above, 20-30 kohm.m is far from excluding frozen ground to occur.

L. 359. So, no ice-rich permafrost occurrence.

Discussion

L. 370. A very patchy motion signal might either be related to noise (data uncertainty) and superficial/shallow movement, but not to rock glacier creep (deformation at a shear horizon at about 20 m depth). The a priori association of (INSAR-detected) surface movement to permafrost occurrence must also be avoided.

L. 380. Matching of DGNSS and PSInSAR cannot be supported (see comment Fig 10).

L. 383. The velocity is probably too small for small but significant interannual variations to be detected.

L. 384-6 and Fig. 15 : There is a big issue to me. There are just two “families” of points : those moving around 0.5 cm/y and those (the majority) moving around 3 cm/y. And nothing between. It does not make sense. There should be intermediate values. The range is close to the SAR phase (half a wave length, about 2.75 cm). I suspect an issue regarding the solving of this phase ambiguity. We would like to better know how these values have been computed and where the points are located on the rock glacier.

L. 400. What is precisely the rule which has been applied to categorize a rock glacier as transitional or relict ? Would it not be also adequate to separate between rock glacier units and systems are recommended by the RGIK guidelines ?

L. 404. What is the sense of this comparison with Brencher et al. ?

L. 406. The occurrence of the 5-10 m depth patches of ice-rich permafrost is not supported by the RST data.

L. 414ff. GST and BTS suggest (and not confirmed) the occurrence of permafrost (see my comments above).

In addition, the current MAGST are roughly about -0.5°C. Taking into account a presumed climate warming by +1.5°C over the last decades, they could have been around -2°C before. It means that extended permafrost occurrences in the rock glaciers can be expected and that deep permafrost (to say down to 50 m or so) cannot be excluded. The absence of ice-rich (ice supersaturated) permafrost conditions at depth (in the investigated rock glacier) does not mean the absence of (ice non-saturated) permafrost, which is highly susceptible to occur according to the geophysical results and the GST values. A more detailed analysis of the BTS mapping and the GST behavior in wintertime might provide some clues about possible convective and/or advective air flow processes within the rock glacier landform, which could contribute to explain why the rock glaciers are so cold.

L. 456. Conclusions are fine and sound. It might be that they could be slightly altered after revision of the paper. The authors must insure that the abstract is strictly based on the paper’s conclusions.

References mentioned in my review (not already cited in the manuscript):

Cicoira et al. (2020). A general theory of rock glacier creep based on in-situ and remote sensing observations. https://doi.org/10.1002/ppp.2090

Delaloye et al. (2013). Rapidly moving rock glaciers in Mattertal. In: Graf, C. (ed.) Mattertal – ein Tal in Bewegung. Publikation zur Jahrestagung der Schweizerischen Geomorphologischen Gesellschaft 29. Juni – 1. Juli 2011, St. Niklaus. Birmensdorf, Eidg. Forschungsanstalt WSL. 21-30.

Eriksen et al. (2018). Recent acceleration of a rock glacier complex, Adjet, Norway, documented by 62 years of remote sensing observations. Geophys. Res. Lett., 45, 8314–8323

Hartl et al. (2023). Multi-sensor monitoring and data integration reveal cyclical destabilization of the Äußeres Hochebenkar rock glacier. https://doi.org/10.5194/esurf-11-117-2023

Hu et al. (2025). Rock Glacier Velocity: An Essential Climate Variable Quantity for Permafrost.  https://doi.org/10.1029/2024RG000847

Kääb and Røste (2024). Rock glaciers across the United States predominantly accelerate coincident with rise in air temperatures. https://doi.org/10.1038/s41467-024-52093-z

Noetzli et al. (2024). Enhanced warming of European mountain permafrost in the early 21st century. https://doi.org/10.1038/s41467-024-54831-9

Pellet et al. (2024). Rock Glacier Velocity. In “State of the Climate in 2023“. Bull. Am. Meteor. Soc. 105, 42-44. https://doi.org/10.1175/2024BAMSStateoftheClimate.1

Roer et al. (2008). Observations and considerations on collapsing active rockglaciers in the Alps, Proceedings of the Ninth International Conference on Permafrost, July 2008, Fairbanks, Alaska, 2, 1505-1510.

Schoeneich (2011). Guidelines for monitoring GST - Ground surface temperature. PERMANET. https://www.permanet-alpinespace.eu/archive/pdf/GST.pdf