The authors of this paper produce the first national-scale rock glacier inventory for the Peruvian Andes region with a coverage over 300000 km², and use quality control and cross-check to improve the quality of the dataset. This dataset has high quality and is important for the permafrost and mountain hydrology studies in Peruvian Andes. However, the wording of this paper is too long with many places showing redundant and repetitive information. I suggest this paper should improve the organization, readability, and concreteness before publication. See my comments below:

General comments:

1. According to the newest version of IPA guidelines, each rock glacier unit (RGU) and system (RGS) should have a primary marker, while this dataset only provides the footprint shapefile. Better to also incorporate the primary marker shapefile and suggest using Zenodo instead of PANGAEA to store the dataset.
2. In the results part, better to use figures instead of tables to show the results, and the presented figures are not high-quality enough. Suggest improving the figures.
3. I have some concerns about discussing controlling factors on the distribution of rock glaciers as the emergence and development of rock glaciers typically need hundreds or thousands of years (i.e., rock glaciers are some landforms that happened at least hundreds of years ago). Is it reasonable to use the modern climate to judge the distribution of rock glaciers?

Specific comments:

Line 1: Please explain what makes this inventory ‘high-resolution’, if it is because that this inventory was created using Bing Map and Google Earth, I don’t think the ‘high-resolution’ can be a highlight or advantage of this inventory as many previous inventories were also created using high-resolution Google Earth imagery.

Lines 43-63: These two paragraphs, first introduce rock glaciers, then permafrost, then rock glaciers, which reads wield. Suggest reorganizing the content, better to describe permafrost first, then introduce rock glaciers.

Line 68: “Splitting it up would reduce run-on complexity => What does it mean?

Line 100: In total, 2338 rock glaciers were mapped using these optical datasets (2095 from Bing and 243 from Google imagery) => These are results, should not appear in the Data section

Line 101: Please explain what makes this dataset complete and high-resolution

Line 105: We compiled several auxiliary datasets => topoclimatic datasets?

Lines 137-140: We primarily used Bing Satellite…Google Earth imagery for that grid cell => similar information has appeared in section 3.1 Data sources Lines 93-96: Bing Maps Aerial imagery was used as the primary data source…geodatabase creation. Suggest merging the information and write it in one place, otherwise, the readability of the paper can be reduced with many redundant and repetitive information in different places.

Line 149: Only longitudinal ridges and furrows? No latitudinal ridges and furrows?

Lines 190-217: Section 4.2 Geomorphological identification criteria: To me the Bullet point and the table are also redundant information, only keep one of them is enough (only the information in Table 2 is enough, no need to write repetitive words using Bullet points.

Lines 231-241: Again, Table 3 is sufficient to show everything clearly, no need to repeat the information using Bullet points.

Lines 250-266: What is the difference between this paragraph and Section 4.2? In section 4.2 you already describe the geomorphological identification for rock glaciers of different activities, why mention the repetitive information here?

Line 291: The smallest rock glacier included in the inventory has an area of 0.001 km², the minimum area threshold for inclusion, according to the IPA guidelines (RGIK, 2023) => Are you sure the minimum area threshold suggested by IPA guidelines is 0.001 km² but not 0.01 km²?

Line 130: 4 Methodology => The Methodology part should be reconstructed, reducing the redundant and repetitive information and making the literature more concrete. Suggestions on the subsections could be 4.1 Identification and mapping of rock glaciers 4.2 Classification of rock glaciers 4.3 Topoclimatic features 4.4 Inventory compilation and validation 4.6 Uncertainty assessment

Line 350: Morphological types: => I suppose this should be a subsection 5.2.1 Morphological types? Also the Line 357 5.2.1 Rock glacier activity: => delete ‘:’

Lines 350-356: Why the analysis of morphological types is not as long as rock glacier activity?

Line 378: he NWOT and NDOT => I suppose it should be ‘The NWOT and NDOT’.

Line 392: Elevation distribution: => I suppose it should be a subsection 5.3.1 Elevation distribution here.

Lines 392-402: For the unit of elevation, some places are m a.s.l. Some places are m. Please keep them consistent.

Lines 434-439: This paragraph is not about Aspect, may you use miss the subsection?

Lines 439-443: ‘This distribution reveals that elevation is the primary control on the presence of rock glaciers, with secondary hydrothermal modulation - evidenced by inverse AP-MAAT relationships in NDOT/SDOT/SWOT (where aridity and snow redistribution at elevation increase cooling) versus the direct correlation of NWOT (driven by thermal depression mediated by microclimatic processes). Such systematic variations underscore how Andean rock glaciers integrate macroscale climatic gradients with local topoclimatic processes.’ => Why elevation is the primary control and climatic conditions are secondary? I don’t understand how this conclusion was drawn from the results.

Lines 453-454: Individual rock glacier area uncertainty was found to range from as low as ~0.001 km² for small, clearly defined features to up to ~0.3 km² for very large or diffuse features, with a mean uncertainty of ~0.06 km². => A figure between uncertainty and area would be helpful.

Lines 504-514: I would expect more results about the comparison between the rock glacier distribution (active, transitional, relict) and the distribution of permafrost from Obu et al. (2018) instead of just stating the elevation and MAAT. Maybe better to show some example figures showing this distribution comparison, see whether the active ones are within the permafrost and the relict ones are out.

Lines 518-519: “The comparisons with global inventories (see Table 8) show that while Peru’s rock glaciers are extreme in elevation, other characteristics like slope and aspect are broadly similar to rock glaciers elsewhere” => But Table 1 only shows the elevation and does not show other characteristics like slope or aspect?

Line 537: see my general comments, not sure whether it is reasonable to use modern climate to discuss the distribution of rock glaciers as these landforms are something happened hundreds of years ago.

Lines 555-557: “The correlation of the inventory with modelled MAGT data (Obu et al. 2019) provides an independent check: nearly all active, rock glaciers lie in grid cells where MAGT is at or below 0°C, whereas relict rock glaciers occupy cells where MAGT is just above 0°C (indicating marginal permafrost conditions).” => See my comments above, I would expect more elaboration on this part. Maybe a statistics on the MAGT of the rock glaciers with different activities, or some examples showing the distribution of rock glaciers and the permafrost.

Line 675: ‘his opens’ => This opens?

Line 692-694: “Looking forward, several lines of future work are planned based on this inventory: Temporal monitoring: now that this baseline is set, repeat satellite observations (e.g., in 5–10 years) or the analysis of time series (like the 2017 vs 2024 imagery) could reveal if any rock glaciers are retreating at the margins or if new ones are forming” => The development of rock glaciers typically take hundreds of years (totally different from glaciers), I don’t think you would see significant changes on rock glaciers on decadal scale.

The authors present a geomorphological rock glacier inventory of the Peruvian Andes (i.e., PRoGI) compiled through the manual mapping on Bing (and Google Earth) optical imagery. My comments are mainly concerned with the manuscript. I didn’t have time to look at the actual shapefile. In my view, the paper is suitable for publication in ESSD after substantial revisions.

I suggest improving the logical flow of the introduction/results/discussion and provide stronger (basic and applied) motivations for compiling the inventory in the way it is presented/discussed in this paper. Considering that RG degree of activity in PRoGI relies on the visual assessment/interpretation of landforms, I suggest reducing substantially the length of the analysis (and accompanying text of the Results and the Discussion) concerned with the classification of rock glaciers into active, transitional, and relict. My suggestion is motivated by the high degree of uncertainty that is typically associated with the morphologically based classification of rock glacier activity. Please see my detailed comment to section 5.2.1.

Overall, I believe the paper would benefit from substantial trimming. A more concise paper structure would allow the main original points of the inventory to stand out more apparently.

Please consider all of my comments on the constructive side. Thank you for your effort on this work.

Francesco Brardinoni

TITLE:

On the “high-resolution” character of the inventory I share the evaluation made by referee #1. Nowadays, the use of Bing and GE imagery (comprised between 1 and 5 m resolution) represents the norm. A high-resolution inventory, in my view, should employ sub-metric RGB imagery, and possibly LiDAR-derived DTMs for increasing three-dimensional perspective and filtering out vegetation cover over vegetated RGs. For example, this is the case of a geomorphological inventory recently compiled across South Tyrol (Italy) by Scotti et al. (2024), who utilized 0.2-to-0.5 m gridded orthophoto mosaics and a 2.5 m LiDAR DTM. Incidentally, this inventory tallies a number of rock glaciers (n = 2798) comparable to PRoGI, and may serve as a useful term of comparison in Table 8.

The difference between an inventory compiled on GE imagery and one compiled on higher-resolution orthophoto mosaics coupled with LiDAR-derived DTMs was assessed by Brardinoni et al. (2019). Accordingly, it was found that “the number of mapped rock glaciers on GE imagery exhibited higher inter-operator heterogeneity (up to a factor of 3), and that using LiDAR and higher resolution orthophotos lowers this heterogeneity down to a factor of 2, while producing an increase in the number of mapped landforms, which become systematically smaller”.

To summarize, I believe that image resolution matters and the resolution of Bing and GE imagery would not warrant a consistent “high resolution” inventorying output across a large study area such as the Peruvian Andes.

Scotti R, Mair V, Costantini D, Brardinoni F. 2024. A high-resolution rock glacier inventory of South Tyrol: Evaluating lithologic, topographic, and climatic effects. In: Beddoe, R.A. and Karunaratne, K.C. (Eds.). 12th International Conference on Permafrost. 16-20 June 2024, Whitehorse, Canada: International Permafrost Association, pp. 382-389. https://doi.org/10.52381/ICOP2024.176.1

Considering that PRoGI and Scotti et al (2024) share comparable numbers of landforms, the geomorphological approach (i.e., no kinematic data are used), and a spurious RGIK inventorying approach, in this review I will refer to Scotti et al (2024) several times.

1. INTRODUCTION:

Line 39: “Nowhere is this more evident …”. Please consider tuning this sentence down. The pace of warming is widely documented in mountain and high latitude area around the globe, and the tropical Andes are just one of them.

Line 43: unclear in what sense RGs would stand out as “geomorphological archives”. Please expand and add references testifying to the geomorphic archival value of RGs.

Line 43-45: if adopting the RGIK (2023) technical definition of rock glaciers, please report the complete version (i.e., creep and “shearing at depth”; “optionally” ridge ad furrow topography, since these features do not always occur). Presently, the definition provided would apply just to active landforms (i.e., steep fronts).

Line 45: Boccali et al (2019) refers to the southeastern European Alps (the Julian Alps), which are not exactly an arid region, similar to the Peruvian Andes. Consider replacing this reference with a better fit or rewriting the sentence.

Lines 46-47: The thermal insulation (through ventilation) afforded by the surficial blocky layer is a property of rock glaciers that has been known/characterized for decades. Besides Brighenti et al (2021), please add reference to prior studies that have indeed characterized with empirical data the internal structure of rock glaciers e.g., Scapozza et al (2011), Geomorphology.

Lines 46-47: “This dual role as climate sentinels and hydrological buffers makes rock glaciers indispensable for understanding long-term environmental change”. Please clarify what is meant by climate sentinels. Are they considered climate sentinels across the Quaternary or at shorter contemporary time scales? In the former case, please add reference to Quaternary studies involving numerical dating of RGs (e.g., 10Be, 14C). In the latter case please add reference to Rock Glacier Velocity as an emergent variable of climate change.

Line 48: please cite a reference when stating the definition of permafrost.

Line 54: To acknowledge alternative views on rock glacier origin/formation that are still matter of international debate (i.e., permafrost vs glacier-to-rock glacier transition), please consider adding a sentence in which you state that in this paper you do not address the question of ice origin and rock glacier formation.

Line 60 “the first high resolution” and Line 63 “By combining sub-meter remote sensing imagery”. Given the imagery utilized (GE and Bing imagery ranging between 1 and 5 m cell size, as reported in Table 1, are not sub-metric), I don’t see the point of considering the inventory a high-resolution one. This comment applies to the paper title too.

Lines 69-70: for the sake of conciseness, please consider removing these two lines. I think the introduction would stand up fine without them.

1. STUDY AREA

The authors adopt a climatic classification of the Peruvian Andes proposed by Bonshoms et al. (2020), which in turn is based on previous climatic characterization of glaciers across the entire South America (Sagredo and Lowell, 2012).

Sagredo, E.A. and Lowell, T.V. (2012) Climatology of Andean glaciers: a framework to understand glacier response to climate change. Global and Planetary Change., 86-87, 101–109.

In the results, and especially in the discussion, it will be important to remind the reader that the four broad regions were defined solely based on climate, and therefore they do not consider interactions with the relevant terrain altitudinal distribution (i.e., how much area is available in each region for RG development above (current and former) critical isothermal altitudinal thresholds) and the dominant lithologies (i.e., propensity for rock walls to disintegrate in blocky debris, hence promote thermal ventilation and permafrost persistence). Having considered climatic regions only may explain the lack of explanatory power and the intra-regional heterogeneity observed in terms of RG spatial distribution. In this context, the subdivision of South Tyrol in physiographic zones (i.e., combining broad climatic, hypsometric, and lithologic characteristics) offered an obvious advantage, in explaining the relevant spatial variability in rock glacier density (Scotti et al., 2024).

A description (even a brief one would suffice) of the geological setting, including a list of the dominant lithologies in each of the four climatic regions is missing. Please consider adding one.

1. DATA

Line 118: “independent check”. Please expand on the likely reliability of the modelled MAGT. Similar models are known to work relatively well in low-lying Artic regions but subject to large uncertainty in rugged mountain terrain, due to high spatial heterogeneity in sediment texture and other local variables. Most importantly, how did you proceed when a RG was labelled relict but plotted withing MAGT < 0°C? Please document your set of decision rules.

1. METHODOLOGY

In the present inventory the authors partly follow the RGIK guidelines, partly do not. Consequently, the inventory would not be directly comparable with other RGIK-based counterparts compiled elsewhere around the globe. I am not suggesting that the inventory should adopt the entire RGIK identification, location, characterization, and delineation protocol; however, it is important that the authors: (i) adhere to the mandatory parts of the RGIK guidelines; and (ii) clearly summarize which components of the RGIK methodological workflow are adopted, which not, and whether a different nomenclature is applied for some of the attributes.

For example, RGIK considers as mandatory component the compilation of a shapefile of primary markers. That is, a shapefile made of point elements reporting lat, long and unique identifier for each rock glacier unit (RGU) and system (RGS). Following this logic, each rock glacier is hierarchically classified into units and systems. Similarly, the RGIK morphological-based activity classification does entail classifying rock glaciers units into active, active uncertain, transitional, transitional uncertain, relict, and relict uncertain. The present inventory encompasses solely active, transitional, and relict landforms. This is a potentially critical approach, considering the inherent uncertainty associated with the classification of transitional rock glaciers i.e., note the high degree of discrepancy between international experts in the activity classification of inactive/transitional landforms (Brardinoni et al, 2019). Indeed, the test presented by Brardinoni et al involved not only RG outline delineation (see section 4.6 of the present manuscript), but also the activity classification.

4.1.2 Digitization protocol

As per prior comment, please clarify whether the RGIK protocol was adopted, or an alternative one was implemented. The three bullet points differ from what reported in the RGIK guidelines.

Lines 154-156: “In some cases, rock glaciers exhibited very extended or degraded fronts (e.g., a “tongue-shaped” that had flowed out and thinned, or a collapsed snout). In such situations, we adopted a conservative mapping approach to remain consistent with IPA guidelines for degraded features (RGIK, 2023)”. I would like to remind the authors that the RGIK conservative approach for delineating the extended outline of RGs applies to the “exaggerated” front category, and not to the “truncated” front one (cf. Figure 3 in RGIK 2023). As per prior comment, it is critical that the authors use the same RGIK terminology, if they have decided to apply the RGIK protocol. In this context, the example provided in Figure 2, alone, may result misleading to the broader audience. Together with the present example, I suggest that the authors present a more classical example (with a steep, talus-like front and better-defined lateral margins) that would form Figure 2a: an easy to map RG case. Following this logic, the present Figure 2 could become panel b (i.e., Figure 2b) representing a more complicated RG to delineate.

In addition, the mapping example shown in Figure 2 presents a number of issues:

1) the front is located immediately above a bedrock ledge, which can confound the interpretation and the delineation of the front base.

2) The eastern lateral margin and the upper boundary of the RG outline cut across morphological flow lines. What was the rationale for drawing such outline? If you believe this is a “difficult” case characterized by complex morphology in which it is not easy to assuredly draw an outline, please acknowledge so. It would be perfectly fine to document a case in which mapping is not so straight forward.

3) The western lateral margin is composed of a number of adjacent taluses and debris cones departing from the rock glacier rooftop. I do not see any clear lateral margin along the entire western side of this rock glacier. As per prior comment, if you believe this is a “difficult” case to map/outline, please declare so.

4.1.3 Quality Control

In the RGIK protocol, the identification of RGs comes as a first step, which involves mapping primary markers and then classifying them as: rock glaciers, uncertain rock glaciers, and not rock glaciers. This means that an RGIK inventory ultimately retains some primary markers labelled as “uncertain rock glaciers”.

In the case of PRoGI, please clarify how uncertain landforms were dealt with. I can imagine that a subset of landforms has remained uncertain. Were these landforms excluded or retained in the inventory? The last two sentences (lines 187-189) are not clear on this point.

Line 171: please replace the full stop with colons at the end of the sentence and before the list.

Line 172: Sun et al 2024 is most likely a repetition and should probably be deleted from the first bullet point.

4.2 Geomorphological identification criteria

Based on its content, this part does not deal with the identification but with the classification of the activity status (or degree of activity). Please change the title of this sub-section accordingly.

Some of the morphological and thematic criteria utilized here hold the risk of reading too much in terms of activity, while not disposing of reliable kinematic data. For example:

1) Vegetation: I don’t find particularly useful using vegetation as a possible criterion for discriminating relict from active and transitional counterparts. Borrowing criteria and evidence drawn from wetter physiographic regions, such as the European Alps (e.g., Scotti et al, Colucci et al, Kellerer-Pirklbauer) is not particularly reliable in arid regions of the Peruvian Andes, where hardly any vegetation can grow in similar dry settings.

This is explicitly stated in page 10 of RGIK (2023): “In arid regions, vegetation may nevertheless be lacking on relict rock glaciers due to unfavorable environmental conditions”.

2) Ridge and furrows: this morphological element is not considered in the RGIK guidelines as diagnostic evidence for discriminating between active, transitional, and relict rock glaciers.

Based on the above, I wonder to what extent the morphological criteria detailed in Table 2 would lead to the compilation of an inventory consistent with existing (or forthcoming) RGIK-based inventories conducted elsewhere on Earth. Please elaborate.

4.4 Classification of rock glaciers

The first part of this section (lines 250-266) largely duplicates what described in section 4.2. Please move these 16 lines of text in section 4.2, ensuring to avoid repetitions.

The second part of this section is not entirely convincing, since it is mixing up the geometric characterization of simple/monomorphic landforms (termed units according to RGIK 2023) into lobate and tongue-shaped morphologies, with those of multilobe/polymorphic morphologies (termed systems according to RGIK 2023). Consequently, this hybrid system of classification does not deal solely with geometry, but also with the ability to distinguish (or not) different rock glacier units (in terms of front, lateral margins, and debris source) within a system, due to adjacency, coalescence, and overlapping of lobes.

The present classification scheme does not offer a clear protocol for consistently discriminating between complex and simple rock glacier configurations. In this context, the main rationale for proposing an RGIK hierarchical classification of rock glaciers into units and systems was exactly to mitigate mapping heterogeneity among operators when dealing with complex (multilobe) morphologies.

Please inform the reader about which geomorphic insights on rock glaciers may be gained by implementing the geometric characterization described in lines 273-280 and Table 5, and in particular the distinction between lobate and tongue-shaped morphologies. After reading section 5.2, which deals with the descriptive statistics on the above RG classification, I could not find any geomorphic insight that could help better understanding rock glacier occurrence (and relevant environmental controls) in the landscape.

4.6 Uncertainty assessment

Besides evaluating between-operator heterogeneity in terms of polygon delineation, how was the uncertainty assessment conducted in terms of degree of activity?

1. RESULTS

Please consider whether all of the sub-sections are needed to convey the main message of the paper. Some may be deleted; some may be merged. I feel that some descriptive information does not lead to original insights. Currently, the results are subdivided in many sub-sections, many of which consist of a single paragraph.

This paper section (the Results) contains both results and interpretations. Please move interpretations to the Discussion e.g., lines 323-325 and lines 332-333. Just to mention a couple of examples that I have noticed in subsection 5.1.

Subsection 5.1

Table 5: please consider moving this table in the supplementary file, while retaining just one line for the summary data relevant to each of the four climatic regions. Currently, the table contains basin-specific information which appear excessive, considering that the reader is not provided with any significant information on such drainage basins. For evaluating the relative spatial distribution of rock glaciers (abundance/paucity), I recommend that the authors use the number of rock glaciers per unit terrain area (rock glacier density) – as opposed to simple rock glacier count. Rock glacier density will be directly comparable across the relevant climatic regions. Presently, the reader does not know how large the different climatic regions are. In the text, the authors mention rock glacier density a few times, but no systematic analysis/evaluation is shown.

Lines 322-325: “Lithology is also likely a key factor. The highest percentage of inventoried rock glaciers coincides with volcanic rock outcrops, where the type of chemical alteration increases the albedo of the surfaces and enhances permafrost development and preservation (Yoshikawa et al., 2020)”. Please consider expanding on lithology or completely neglecting this factor in the manuscript. A sentence that relates RG abundance on chemical weathering (due to albedo) of volcanic rocks appears too much of a stretch. Lithological effects on rock glacier abundance have been difficult to isolate in the literature. I wouldn’t try to solve or dismiss such a complicated topic with a similar sentence on volcanics, which by the way encompass a range of lithological types. For a brief discussion on the complexity of isolating lithological effects (due to spurious interactions with hypsometry and climate), the authors may have a look at Section 4.2 in Scotti et al (2024).

Lines 332-333: “This suggests that even basins in close proximity, under the same broad climatic subregion, can exhibit different rock glacier densities and size distributions – possibly due to local geomorphological factors (such as basin lithology or glacial evolution)”. As per prior comment, this sentence appears vague without adding significant information to the paper. Moreover, RG densities are mentioned, but I couldn’t see any quantitative data on this variable. Please consider deleting this sentence.

Table 6: 1) what was the rationale for the selection of the size bins shown in this table? Two categories contain respectively just 3 and 1 rock glaciers only. The current binning does not seem appropriate. 2) What is the underlying hypothesis for presenting RG size categories as a function of RG mean elevation and slope? I don’t understand why slope and elevation should change systematically with RG size, neither I recall any empirical relation constrained along these lines in other inventories. Please consider deleting from Table 6 columns reporting mean elevation and mean slope.

Subsection 5.2:

As per prior comment made in the methods, please consider removing this subsection. It does not seem to add any insight on what may control the spatial distribution of rock glaciers across the Peruvian Andes.

Subsection 5.2.1:

Considering that RG degree of activity in PRoGI relies on the visual assessment/interpretation of landforms, I suggest reducing substantially the length of the analysis (and accompanying text of the Results and the Discussion) concerned with the classification of rock glaciers into active, transitional, and relict. My suggestion is motivated by the high degree of uncertainty that is typically associated with the morphologically based classification of inactive/transitional rock glacier activity (e.g., Table 3 and Figure 11 in Brardinoni et al., 2019), and the rate of reclassification rock glaciers may undergo, once InSAR kinematic data are integrated (e.g., Table 4 and Figure 11b in Bertone et al., 2024).

Bertone A, Jones N, Mair V, Scotti R, Strozzi T, Brardinoni F. 2024. A climate-driven, altitudinal transition in rock glacier dynamics detected through integration of geomorphologic mapping and InSAR-based kinematic information. The Cryosphere, 18, 2335–2356, https://doi.org/10.5194/tc-18-2335-2024

Indeed, Bertone et al found that 15% of the rock glaciers western South Tyrol were reclassified from relict to intact (or vice versa), as a result of InSAR data integration. This reclassification rate is likely to increase even more when 3 activity classes (as opposed to just the intact and relict ones) are considered.

Line 362: “inactive”. Please replace or remove the term inactive, as it would confound the reader. Traditionally, inactive and relict have been used to differentiate two distinct classes of activity.

Lines 369-372: “Indeed, we found that relict rock glaciers have the smallest mean size (mean area ~0.03 km²), compared to ~0.04 km² for transitional and ~0.06 km² for active rock glaciers. This pattern is consistent with expectations: once a rock glacier loses its ice (becoming relict), it may slump and shrink over time, whereas active ones are buttressed by internal ice and can maintain larger extents.”

Technically, the above sentences belong to the Discussion, as they contain interpretation of the results. Most importantly, please consider rethinking your possible explanation, since rock glaciers are complex landforms associated with millennial time scales of development. Indeed, RG size has to do with age of formation, length of (continuous or discontinuous) activity through millennia, rate of sediment supply, and available room within a valley/slope for growing in planimetric size. Attributing the smaller size of relict rock glaciers to slumping (or other mass wasting styles of obliteration) appears simplistic, considering that the vast majority of rock glaciers degrade (non-catastrophically) through subsidence. Incidentally, this interpretation contrasts with prior results from the Italian Central Alps (Scotti et al 2013), where relict RGs were found to be systematically larger than intact counterparts (i.e., Figure 8, and Figure 6a, cf. gray (intact) and white (relict) box whiskers).

Subsection 5.3.1

Lines 410-412: “This suggests that slopes around 15–20° are especially conducive to rock glacier formation/preservation, likely because they are steep enough for debris-ice creep but not so steep as to cause frequent avalanching or debris removal.”

Technically, the above sentence belongs to the Discussion, as it provides an interpretation of the results.

Subsection 5.4

Lines 455-457: “Notably, this level of uncertainty is lower than values reported in some other manual rock glacier inventories (e.g., those compiled over larger regions or by less experienced mappers, see Brardinoni et al., 2019), indicating the robustness of this methods.”

I understand the point you are trying to make, but the rates of inter-operator discrepancies shown by Brardinoni et al cannot be compared with the PRoGI’s ones. The scopes (and methods) of the two mapping exercises are completely different. The former was a mapping test conducted by international experts who were given no instructions/guidelines on how to map a rock glacier. By contrast, I imagine that the PRoGI mappers were given a set of mapping rules to warrant a certain degree of consistency. Please remove reference to Brardinoni et al in from the comparative context of this sentence.

Subsection 5.3.2

Figure 8: from the description of how mean RG aspect was computed in QGIS, I suspect that this variable might be biased. The main issue when calculating the average aspect of a surface is that aspect is a circular variable, meaning that the mean is incorrect due to the discontinuity that occurs around 360 degrees for northerly aspects (i.e., aspects approaching single-digit degrees are adjacent to aspects approaching 360 degrees). Typically, this results in underestimating the number of RGs that are dominantly facing north. Please double-check whether adequate transformation was conducted during aspect calculation and elaborate on this in the methods.

DISCUSSION

Based on the revised structure of the Results, please consider which parts of the Discussion may be really necessary to highlight the originality and robustness of your inventory, and which others may be just chocking the reader with unnecessary details (or debatable interpretations). I believe that the overall reading of the paper would highly benefit from some Discussion simplification.

Based on the above and the need for substantial revision, I will limit my comments on the Discussion to Table 8. Considering that this manuscript is not a review paper, please consider restricting your list to inventories that can help making a more straightforward and meaningful comparison with PRoGI. For example, wouldn’t be enough comparing PRoGI to other inventories in South America? Even when a worldwide comparison was deemed necessary, what is the point of comparing PRoGI with inventories that contain less (or a little more) than 100 RGs? To warrant a more reliable comparison, I would limit Table 8 to inventories that encompass thousands (or at least several hundreds) of RGs.