Review Reports - Snow Cover Trends in the Chilean Andes Derived from 39 Years of Landsat Data and a Projection for the Year 2050

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

General comments

The author proposed a fragment and generated monthly Snow Line Elevation (SLE) from Landsat images. The description of methods needs to clarify, especially for the projection. The verification of SLE is not strong. How the Landsat-derived SLE support the conclusion is not clear. The potential of increasing SLE on climate and water resources is not specifically pointed out. The author should check the styles of reference carefully to meet the requirements of Remote Sensing. I suggest Reconsider after major revisions.

Specific comments

The abstract need to improve and focus on their own work and findings.

Line 18: such

Line 79-86: the objectives need to adjust to meet the requirements of Remote Sensing.

It takes more 2pages of 18 pages to describe the study area. It is too long and need to brief.

Line 93 asl ( a.s.l ) appear repeatedly in this paragraph, what it is refer to?

Line 32-34：add the reference.

Line 292-293: (1) and (2) is misleading, please correct it.

Figure 4 y axis, elevation should be capitalized. Suggest to check other figures.

The sub-figures should be coded, and explained separately in the figure caption.

Line 419 (Error! Reference source not found.).

Line 482 [28] Reference should not appear in the conclusion.

The conclusion is too simple. The first paragraph needs to remove. What is the influence of SLE projection on climate?

Comments on the Quality of English Language

The quality and writing styles of English writing needs to improve.

Author Response

OUR ANSWER: Before going into the specific comments, we want to thank the reviewer for their time and effort to provide us with valuable feedback. We know that it always takes considerable time to do these manuscript reviews. It is only thanks to the reviewers that we are able to publish our research, and we greatly appreciate it!

Specific comments

The abstract need to improve and focus on their own work and findings.

OUR ANSWER: we have added more details about our findings to the abstract, introducing the concrete trends and future projections that were the results of the study.

Line 18: such

OUR ANSWER: Changed to “these”

Line 79-86: the objectives need to adjust to meet the requirements of Remote Sensing.

OUR ANSWER: we realized that the objectives might have been formulated a bit too general, not making clear that the core objective is the use and analysis of remote sensing-based data and methods. We have added another half-sentence to the objectives to emphasize this part.

It takes more 2pages of 18 pages to describe the study area. It is too long and need to brief.

OUR ANSWER: We shortened this paragraph and moved some of the content to different sections. But we also believe the information about the study is crucial because vegetation is an important aspect when detecting snow cover, climate is essential for the dynamics and the developments we see in the Andes, and therefore we want to included as much background as possible.

Line 93 asl ( a.s.l ) appear repeatedly in this paragraph, what it is refer to?

OUR ANSWER: this refers to “above sea level” and has been adjusted to be concise and always abbreviated as “m a.s.l.”

Line 32-34：add the reference.

OUR ANSWER: The data comes from the latest census of Chile, The reference has been added.

Line 292-293: (1) and (2) is misleading, please correct it.

OUR ANSWER: We changed this sentence and made sure the (1) and (2), referring to the western and eastern part of the Maipo catchment, are not misleading anymore

Figure 4 y axis, elevation should be capitalized. Suggest to check other figures.

The sub-figures should be coded, and explained separately in the figure caption.

OUR ANSWER: We modified the figures, included codes for the subfigures.

Line 419 (Error! Reference source not found.).

OUR ANSWER: The error has been fixed

Line 482 [28] Reference should not appear in the conclusion.

OUR ANSWER: Reference has been removed

The conclusion is too simple. The first paragraph needs to remove. What is the influence of SLE projection on climate?

OUR ANSWER: We adjusted the conclusion and extended it with some more details from our results. About the influence of the SLE projection on the climate, we can’t state any effects for sure as this was not a part of the study, and certainly such effects would be very complex to model. There will most likely be effects not only for the water availability (which we discussed and addressed in great detail), but other elements such as regional albedo, biodiversity, freeze/thaw cycles of the soil and ground, or permafrost. Such effects would need to be subject of an additional study.

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript entitled “Snow Cover trends in the Chilean Andes derived from 39 years 2 of Landsat data and a projection for the year 2050” extracted monthly SLE time series from the entire achive of available Landsat data between 1985 and 2024 and analyzed the changes in SLE during the 39 years in two catchments in the Chilean Andes.

While this study is generally interesting and meaningful, there are some issues needs to be well addressed to ensure the reliability of the results and the understandability to the authors.

In introduction section, the gap of existing studies on detecting SLE changes from satellite data, particularly Landsat data, should be stengthened, as it is the major area of interest for the readers of the journal Remote Sensing.

As the paper states, “No study has yet approached the question of how the snow cover dynamics around the Metropolitan region of Santiago de Chile have changed on a catchment basis while relying on a Landsat-based snow classification together with a DEM to retrieve snow line elevation time series”. It is not clear where the gap is. Has any one conducted similar studies using Landsat data to detect snow cover changes in other regions? If yes, what is th innovation of the proposed study? It needs to be stated clearly.

In Data section, the authors mentioned that all available Landsata data, without considering the cloud coverage, were used in the study. How do you deal with the image quality flags ? Some Landsat images might be acquired with low qualities with respects to radiometric accuracies and so on. They will further influence the classification of snow cover. Consequently, how to ensure the consistency of the classification between high-quality images and low-quality images?

Additionally, for ETM SLC-off data, were the data gaps in the images filled? As the authors have processed a large achive of Landsat data, details of data preprocessing need to be introduced.

In section 3.2, lines 231-234, the author claimed that the efficacy of the snow cover classification has been assessed in the Alps. However, the study area, Chilean Andes, in this paper, is very different with the European Alps. And the snow cover classification accuracy is really important to ensure the realiabilty of the following analysis. It is necessary to conduct a in-depth analysis of the snow cover classification accuracy. The selection of test samples, their distribution in space and time, the quantitative accuracy indicator, are all very important issues.
In section 3.3, is it possible to evaluate the accuracy of extracted SLE using in-situ observations? Although the paper mentioned a series of quantitative indicators, they are all based on classification results. It is important to use other reliable data from independent sources to evaluate accuracy objectively.
In section 3.3, the indicator RMSE, is quite different from the RMSE traditionally used in accuracy assessment. This is not common. The authors also made a detailed discussion in section 5.1, stating that the RMSE does not represent error but represent variability. However, this is quite misleading. I suggest not to use the terminology RMSE, as its traditional definition is error, not variability. Consequently, the discussion in 5.1 is not necessary, just change the terminology.
In section 3. Materials and methods, I suggest a flow chart to illustrate the methods, for better readability.

Author Response

While this study is generally interesting and meaningful, there are some issues needs to be well addressed to ensure the reliability of the results and the understandability to the authors.

In introduction section, the gap of existing studies on detecting SLE changes from satellite data, particularly Landsat data, should be stengthened, as it is the major area of interest for the readers of the journal Remote Sensing.

OUR ANSWER: It was to our surprise that there are only very few Landsat-based studies in this direction, even though the unique 40+ year time series is an ideal data source. We identified only very few studies, partially done by former colleagues or our co-authors themselves. The innovation of the presented study is not so much the method, but the application to the Southern Hemisphere and the Andes Mountains, which are challenging due to considerable cloud cover during winter. The fact that SLEs would be receding so fast here was expected from other studies relying on medium resolution data, but to see this in such high resolution and so precisely for the steep and complex mountains was totally new. Our intention is also to use this study as a starting point to investigate much further the future of snow cover and water availability of the region around Santiago de Chile. As we stated in the manuscript, water scarcity will become a severe problem, and because this is known to the government, there are already some strategies being developed. We hope that our work and data can contribute to these efforts.

In Data section, the authors mentioned that all available Landsata data, without considering the cloud coverage, were used in the study. How do you deal with the image quality flags ? Some Landsat images might be acquired with low qualities with respects to radiometric accuracies and so on. They will further influence the classification of snow cover. Consequently, how to ensure the consistency of the classification between high-quality images and low-quality images?

Additionally, for ETM SLC-off data, were the data gaps in the images filled? As the authors have processed a large achive of Landsat data, details of data preprocessing need to be introduced.

OUR ANSWER: Image quality flags have not been considered for this study, but because we relied only on Landsat collection 2 level 2 data, the data already has to meet certain standards in terms of radiometric calibration and terrain correction. We assume that the quality of this data is sufficient for our analysis, and because we’re deriving a level-3 product (snow cover) which is the basis for our statistical analyses and calculation of the SLEs, slight differences or instabilities in radiometric accuracy would not be problematic.

About the SLC problem in landsat 7: we treat these strips like data gaps/clouds. They are not filtered out or interpolated. Because the SLE calculation can be performed even under partially cloud-covered conditions, the SLC-gaps are not particularly problematic as long as they don’t come together with severe cloud cover. They might – however – lead to more scenes being filtered out due to too much missing data.

In section 3.2, lines 231-234, the author claimed that the efficacy of the snow cover classification has been assessed in the Alps. However, the study area, Chilean Andes, in this paper, is very different with the European Alps. And the snow cover classification accuracy is really important to ensure the realiabilty of the following analysis. It is necessary to conduct a in-depth analysis of the snow cover classification accuracy. The selection of test samples, their distribution in space and time, the quantitative accuracy indicator, are all very important issues.

OUR ANSWER: The reviewer mentioned a very important aspect here, which we were considering thoroughly. Unfortunately, no in-situ measurements of snow cover are available for the study region. We agree that ideally, we would want to validate the landsat classification accuracy at different times throughout the year, all available Landsat sensors, and making sure that all different land cover classes are equally represented. We are very confident that the quality of our Landsat snow classifications is sufficient for several reasons though: The land cover in the Andes mountains is more permissive and favorable to snow cover classification as there is no tree cover or dense vegetation present on the ground. That makes the detection of snow on the ground easier for the algorithm. Secondly, the underlying Snowpex intercomparison of snow cover detection algorithms did a thorough accuarcy assessment and comparison of 459 different Landsat tiles distributed all over the Northern Hemisphere. Again – it is not ideal that the Andes Mountains are not included in this sample, but the validation of 4 different snow cover algorithms confirmed an overall good accuracy without considerable differences between the approaches. Thirdly, the classification scheme we applied was based on the SnowPex outcomes and the studies of Hu et al. (2019) and Koehler et al. (2022). In these studies, the Landsat snow classifications reached an overall accuracy of 96% (Hu et al. 2019) and between 87.5% (Landsat 5) and 95.5% (Landsat 7) in (Koehler et al. 2022), which was assessed comparing the results with Sentinel-2.

We have then decided to perform an additional validation of the Landsat classifications with Sentinel-2 specifically for our study region. We found 23 scenes since 2015, covering the different Landsat tiles, originating from the same day as Landsat, and featuring enough cloud-free pixels for a comparison. Based on this newly introduced accuracy assessment, the overall accuracy of the Landsat snow classifications reaches 96.8%. More details can be found in the revised manuscript and the appendix.

In section 3.3, is it possible to evaluate the accuracy of extracted SLE using in-situ observations? Although the paper mentioned a series of quantitative indicators, they are all based on classification results. It is important to use other reliable data from independent sources to evaluate accuracy objectively.

OUR ANSWER: Similar to the prior comment, we absolutely agree that an in-situ validation would be ideal. Unfortunately, no in-situ data is available for SLEs. We are very confident that the accuracy metrics we have calculated, Representative Index (RI), Error Index (EI), and Root Mean Square Error (RMSE), give a comprehensive overview of the general accuracy of the SLEs, which is in the end solely depending on the accuracy of the snow cover classification if considering the whole catchment. To validate the SLE with independent reference data, we’d need several stationary webcams which allow continuous monitoring of the SLE, and which are distributed equally over the catchment to cover all slopes and aspects. Something like this is not available in the Andes mountains, unfortunately.

In section 3.3, the indicator RMSE, is quite different from the RMSE traditionally used in accuracy assessment. This is not common. The authors also made a detailed discussion in section 5.1, stating that the RMSE does not represent error but represent variability. However, this is quite misleading. I suggest not to use the terminology RMSE, as its traditional definition is error, not variability. Consequently, the discussion in 5.1 is not necessary, just change the terminology.

OUR ANSWER: Very much agreed that the RMSE has been discussed differently, compared to other studies. In section 3.3 though, we believe the RMSE is still described as a normal metric for accuracy estimation of – in our case – the SLE and the mean deviation from that SLE, that is present in a catchment. In the discussion section, we then tried to explain why sometimes, the RMSE is becoming comparably large, and we had a long discussion among the authors whether we should do that. The RMSE in the context of SLEs gives us a precise understanding of the variability of the (local) SLEs, which in their sum add up to the overall SLE, and which can vary significantly within a catchment. In the end, each local “sub-SLE” is subject to sometimes quite large deviations from the mean SLE. We need the mean SLE as we want to derive catchment-wide metrics, statistics, and projections of the possible future. At the same time, we want to inform the reader about the variability und uncertainty. We have modified the parts in the discussion where we try to explain the RMSE and label it as variability. Section 3.3 stayed as is, because the metric is – in the end – a regular RMSE.

In section 3. Materials and methods, I suggest a flow chart to illustrate the methods, for better readability.

OUR ANSWER: We have been thinking to include a flow chart, and we were debating for and against several times. Because there is a flow chart in the MDPI study about the Alps (Koehler et al. 2022) we decided against it in the end. Anybody who is interested in more details about the procedure can refer to that publication, which is open access as well and should therefore pose no challenges.

Reviewer 3 Report

Comments and Suggestions for Authors

After reviewing the manuscript (remotesensing-3255646), some comments and suggestions are as follows.

(1) In line 280 and 281, the median accuracy of the estimated SLE as measured by the RMSE is 453 m. How did authors calculate RMSE of 453 m ?

(2) In line 283, how to define and calculate RI, EI and RMSE in Figure 2. How to define and calculate red vertical line mark in Figure 2. It is better to use a Landsat data and illustrate the procedure of data processing.

(3) In Figure 4, it is very clear that positive SLE anomalies are generally more frequent at recent years and anomalies tend to be greater in Spring and Fall. Is it due to warming climate recently? Could you explain it in terms of climate change.

(4) In line 287, why is it between 76% and 92% of all observations remained? Is there any reason for this threshold ?

(5) What is the main drivers of snow cover or Snow Line Elevation (SLE) trends in the study area ?

(6) It is suggested to add some major findings and outlook of the study in the end of Abstract. The abstract seems a little short.

(7) In line 92-93, both catchments are W-E-oriented and reach elevations from sea level to 6954 m a.s.l. However, the highest elevation is 6936 Meters m.s.l. in Figure 1 and 6548 Meters a.s.l. in Figure 6. Three highest elevations are different in the study area. It should be corrected and same units should be used. The unit of elevation is m a.s.l.

(8) In line 117, the river basin has a total area of 7 322 km2 [20], whereas in Table 1 the total area of river catchment of Aconcagua is 7 318 km2. Two total areas are different.

(9) In line 419, “of the Maipo catchment, respectively (Error! Reference source not found.) should be corrected .

(10) It is suggested to draw a flowchart to show the procedures of snow classification, snow line elevation retrieval and time series data generation. Currently, it is quite difficult to understand the data processing procedures for reviewers and other readers.

Author Response

After reviewing the manuscript (remotesensing-3255646), some comments and suggestions are as follows.

OUR ANSWER: Before going into the specific comments, we want to thank the reviewer for their time and effort to provide us with valuable feedback. We know that it always takes considerable time to do these manuscript reviews. It is only thanks to the reviewers that we are able to publish our research, and we greatly appreciate it!

In line 280 and 281, the median accuracy of the estimated SLE as measured by the RMSE is 453 m. How did authors calculate RMSE of 453 m ?

OUR ANSWER: For each scene, the SLE is calculated and defined as one single elevation threshold between snow covered and snow-free area. Every false pixel (snow-covered pixel below the SLE, snow-free pixel above the SLE) is then analyzed for its elevation and the distance of the pixel’s elevation to the SLE. From all these values, the RMSE is then calculated. The median RMSE is then calculated from all RMSEs from all Landsat scenes.

In line 283, how to define and calculate RI, EI and RMSE in Figure 2. How to define and calculate red vertical line mark in Figure 2. It is better to use a Landsat data and illustrate the procedure of data processing.

OUR ANSWER: RI is the percentage of pixels in a catchment within a scene that can be used for SLE estimation, i.e., pixels that contain information about whether snow is present. From the 100% of all potential pixels in a catchment, the "unknown" pixel values are subtracted, depending on the capture. Unknown are the pixels from which it is not clear what surface they represent, i.e., pixels without data (SLC error), cloud-covered pixels, or pixels in terrain shadow. The median RI over all used scenes indicates how much % of pixels were available on average for SLE estimation. In general, the higher the RI, the more reliable the SLE estimation. This could be illustrated with two Landsat scenes used for snow classification: One scene is cloud-free, while the other is partially cloud-covered. The corresponding RI in the SLE raw data for the cloud-covered scene should be lower

EI is the proportion of "incorrectly" classified pixels ("incorrectly" in the sense discussed in the remark on RMSE) out of the total number of pixels in the catchment/scene.

The RMSE gives a measure of the accuracy of the derived snow line by calculating the distance of all falsely classified pixels (i.e., “snow” below the SLE, “clear land” above the SLE) from the estimated SLE.

In line 287, why is it between 76% and 92% of all observations remained? Is there any reason for this threshold ?

OUR ANSWER: There is no fixed threshold. The 76% to 92% are the result of varying cloud cover, SLC-gaps, and therefore data availability for each scene. This statement simply means that between 76% and 92% of the Landsat data going into the processing were available after all processing to derive the SLEs – depending on the Landsat tile.

What is the main drivers of snow cover or Snow Line Elevation (SLE) trends in the study area ?

OUR ANSWER: the main drivers are temperature and precipitation, with ENSO affecting both and therefore also playing a role. We did not analyze any time series of temperature or precipitation data, as this study was only focused on the Landsat data. That is why we cannot clearly state which parameter has which concrete effect.

Climate records from e.g. world bank show a steady increase in temperatures for the Metropolitan Region of Santiago de Chile (attached you find a screenshot from the climate records) and Valparaiso (Chile - Climatology | Climate Change Knowledge Portal ). The grid size of these temperature datasets can also be seen from this figure, explaining why we did not want to include it, but at the same time suggesting that this temperature together with the megadrought is a prime reason for the receding SLEs. We have added additional sentences to the discussion to emphasize this, without going into too much detail about the temperature data itself.

In the end, what we present here is what we see from the satellites. It could be a next step to combine these observations with climate models (ideally a local model) to understand the reasons better, or we might try to perform a causal analysis

(6) It is suggested to add some major findings and outlook of the study in the end of Abstract. The abstract seems a little short.

Our answer: we totally agree to this statement and have extended the abstract by adding more of the results and findings.

Our Answer: the differences in elevations came through information taken from literature and a 30 m DEM. We made sure that there is a consistent elevation given throughout the manuscript

In line 117, the river basin has a total area of 7 322 km2 [20], whereas in Table 1 the total area of river catchment of Aconcagua is 7 318 km2. Two total areas are different.

OUR ANSWER: Similar to the elevation error before, this mistake happened because one information is taken from literature, the other was taken from 30 m datasets inside the GIS. We corrected this.

In line 419, “of the Maipo catchment, respectively (Error! Reference source not found.) should be corrected .

OUR ANSWER: Has been fixed. Was referring to figure 3

OUR ANSWER: we have discussed this repeatedly. In the end we decided against it, as there is a figure illustrating exactly this workflow in Koehler et al. 2022, which is in “Remote Sensing” and therefore also open access. We hope it is ok to refer to this other publication rather than creating an new flowchart which would illustrate the same.

Author Response File: Author Response.docx

Reviewer 4 Report

Comments and Suggestions for Authors

The authors employed a well-established regional snowline elevation (RSLE) determination method to analyze snow cover variation trends in the Chilean Andes using Landsat data. The study provides valuable insights into snow cover dynamics through monthly snowline elevation time series data. While the manuscript is well-structured and informative, I have a few minor comments that should be addressed before acceptance:

Line 266: Please clarify why the median snowline elevation was chosen for the variation analysis instead of the mean or other statistical measures? A brief justification for this choice would strengthen the methodology section.
The projection of snowline elevation in 2050 appears to be based on the assumption that snow cover variation follows a linear trend. This assumption should be explicitly stated in the manuscript, along with a discussion of its limitations and potential uncertainties.
The RSLE method developed by Krajci et al. (2014) assumes that snowline elevation is uniform across all slope directions within mountainous watersheds, with calculations based on the statistical analysis of snow-covered and snow-free pixels. This raises questions about the interpretation of RMSE in Lines 384-385, where it is stated that a high RMSE reflects “high variability” rather than “high error.” This interpretation may not hold true given the methodological assumptions. I recommend revising this statement and considering recent research on snowline uncertainty, such some studies which highlights potential errors arising from snow cover mapping algorithms. Incorporating a discussion of these uncertainties would enhance the manuscript.
Please verify the citation in Line 419 for accuracy and consistency with the reference list.

Author Response

Line 266: Please clarify why the median snowline elevation was chosen for the variation analysis instead of the mean or other statistical measures? A brief justification for this choice would strengthen the methodology section.

OUR ANSWER: We chose the median to limit the effect of strong outliers, which can occur in extreme years. These would then have a significantly stronger impact on the mean, and we want to have a stable baseline for the projection of the 2050 results. This is why we chose the median. We have added a sentence stating this now in the methods section

The projection of snowline elevation in 2050 appears to be based on the assumption that snow cover variation follows a linear trend. This assumption should be explicitly stated in the manuscript, along with a discussion of its limitations and potential uncertainties.

OUR ANSWER: it is true that the linear trend is only an assumption. We tested linear trends with our SLE results and obtained significant results, but it could also be that there trends are – for example - somewhat exponential. We have added a comment to the discussion section to make sure that the readers are aware of this. It might be interesting to mention here though that the other studies about SLE/snow cover trends in mountain regions, which are cited in the discussion of our manuscript, also relied on linear trends

The RSLE method developed by Krajci et al. (2014) assumes that snowline elevation is uniform across all slope directions within mountainous watersheds, with calculations based on the statistical analysis of snow-covered and snow-free pixels. This raises questions about the interpretation of RMSE in Lines 384-385, where it is stated that a high RMSE reflects “high variability” rather than “high error.” This interpretation may not hold true given the methodological assumptions. I recommend revising this statement and considering recent research on snowline uncertainty, such some studies which highlights potential errors arising from snow cover mapping algorithms. Incorporating a discussion of these uncertainties would enhance the manuscript.

OUR ANSWER: We are aware of the assumption that the SLE is uniform within the catchment. And this point of the discussion was also repeatedly discussed among the authors. The fact that the Andes Mountains stretch N-S, and the majority of the precipitation is coming from the West significantly increases the influence on the variability of the snow line within a catchment. West-facing slopes receive substantially more snowfall than east-facing. That is the same for the Alps or the Carpathian Mountains, where the Krajci method was developed and applied to in the first place, but here it is much more significant. An RMSE of more than 400 m means of coarse that the SLE derived for a catchment has this error. It is an error and not a variability, mathematically/statistically speaking. That is because the assumption of a uniform snowline is not reflecting the reality very well, but is useful for statistical analyses such as trends and future projections. We made the attempt to address this problem in the discussion, but we agree that the way we described it was not ideal. That is why we modified this part in the discussion, making clear that yes, it is an error, but we are aware of the source of the error, and that it is related to the variability within the catchment. In a future study we will try to derive separate SLEs for the various slopes and try to see the influence. The problem here will be – for visualization, that we might end of with not-so-smooth transitions between the different slope directions. That might be acceptable though if the RMSE can be turned down significantly.

Please verify the citation in Line 419 for accuracy and consistency with the reference list.

OUR ANSWER: Has been fixed. Was referring to figure 3

Reviewer 5 Report

Comments and Suggestions for Authors

Thank you to the authors for what appears to be a very well written, and thought out manuscript. While I do find that the authors could go into a bit more depth, discussing their results, I think that what is presented is satisfactory for publication. I can see the value in this work, for hydrological modeling and resource assessments that rely on calculations of SWE from the SLE, particularly into the future. The figures are basic, but I like what is presented for OND 2050.

Please check a broken reference on line 419

Please proof read and do a careful grammar check on your proofs prior to publication.

Author Response

OUR ANSWER: Has been fixed. Was referring to figure 3

Please proof read and do a careful grammar check on your proofs prior to publication.

OUR ANSWER: we have given the manuscript to an English native speaker and made sure that a few of those bumpy-sounding sentences have been modified.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The count and elevation in figure 3 to 5 need to be capitalized.

The sub-figures need to explain in the captions.

Author Response

Again, we want to thank the reviewer for investing additional time into this review. We apprectiate the support a lot.

We addressed the last comments and made the respective changes, namely:

The count and elevation in figure 3 to 5 need to be capitalized.

Our Answer: We changed the figures accordingly

2. The sub-figures need to explain in the captions.

Our Answer: We added a short explanation about what the subfigures illustrate

Reviewer 2 Report

Comments and Suggestions for Authors

I appreciate the author's meticulous revisions, which have effectively addressed most of the major concerns. One remaining recommendation would be to enhance the introduction section by more explicitly articulating the paper's innovative contributions and broader academic significance.

Author Response

Again, we want to thank the reviewer a lot for the additional time they put into revieweing our manuscript. We appreciate it a lot!

We addressed the last comment, namely:

Our Answer: We added another short paragraph into the introduction, stating the novelty of our study and manuscript more clearly:

"The presented study showcases the successful application of the SLE method to the Andes Mountains, a region where it had not been previously applied. Building on its proven efficacy in the Carpathian Mountains [17] and European Alps [10,18], this study breaks new ground by deriving SLE statistics seasonally for the first time. This innovative approach provides more nuanced insights into the impact of climate change on the development of seasonal snow cover, revealing distinct seasonal trends."

Reviewer 3 Report

Comments and Suggestions for Authors

I cannot understand the capital C is used in snow cover in the title of the paper? “Snow Cover” in the title should be “Snow cover”.
In line 65, Cordero et al. [12] calculated Snow Cover extends for 22 zones in the

In this sentence, I think the “Snow Cover” should be “snow cover” also.

Author Response

We want to thank the reviewer again very much for their time revieweing our manuscript. We really appreciate it a lot!

About the last comments: We addressed them both, changing the capital "C" in the titel to a lower "c" and also changing the "Snow Cover" in line 65 to "snow cover"