Next Article in Journal
Leveraging Classical Statistical Methods for Sustainable Maintenance in Automotive Assembly Equipment
Previous Article in Journal
Assessing the Feasibility of Eco-Industrial Parks in Developing Countries: A Case Study of Thang Long II Industrial Park in Vietnam
 
 
Article
Peer-Review Record

Assessing Endokarst Potential in the Northern Sector of Santo António Plateau (Estremadura Limestone Massif, Central Portugal)

Sustainability 2023, 15(21), 15599; https://doi.org/10.3390/su152115599
by Luís Reis 1,*, Luca Antonio Dimuccio 2 and Lúcio Cunha 2
Reviewer 2:
Reviewer 3: Anonymous
Sustainability 2023, 15(21), 15599; https://doi.org/10.3390/su152115599
Submission received: 16 May 2023 / Revised: 25 October 2023 / Accepted: 1 November 2023 / Published: 3 November 2023
(This article belongs to the Special Issue Geostatistics Applications in Resources and Environment)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The article aims to produce a cartographic model that identifies the areas most likely to find karstic caves (the endokarstic potential) in the northern sector of the Santo António plateau (extremadura limestone massif, central Portugal).

The analysis strategy consisted of collecting and processing geological, topographic, hydrogeological and vegetation cover data, as well as the location of the known entrances to the caves in the study area. Four conditioning factors were extracted from the collected data: lithostratigraphic units, fracture density, relief energy, and land cover. In a multi-criteria decision analysis framework, each previously chosen conditioning factor and the respective classes/categories were weighted using an Analytical Hierarchy Process (AHP).

The paper uses a geographic information system (GIS) to integrate geological, topographic, hydrogeological, and land cover data into a spatial database, which can be used for further research and analysis. The results of the GIS-built mapping model look promising, considering that the known cave entrances are mainly in areas classified as having high to very high endokarst potential.

The endokarstic potential map obtained can be used in operational and strategic environmental planning to help decision makers and local caving teams to more accurately and thoughtfully define the areas to be investigated, substantially reducing time and costs. field survey costs.

The article contributes to the understanding of the parameters that most contribute to endokarst formation and provides a more reliable model for a spatial representation of underground karstification potential.

The main criticism that can be made of the article is that the endokarst potential map produced in this study is based on the available data and the chosen constraints. Therefore, it may not be entirely accurate and may require further validation through field work. Furthermore, the applicability of the map may be limited to the study area and may not be generalizable to other regions.

On the other hand, the Analytical Hierarchy Process (AHP), a decision-making method developed by Thomas L. Saaty in the 1970s, is not without its challenges and potential problems. The main problems associated with are:

- Subjectivity: AHP relies heavily on subjective judgments and comparisons in which the interpretation and perception of the relative importance of criteria and alternatives may vary among different decision makers, leading to potential biases and inconsistencies.

- Inconsistent pairwise comparisons: It can be challenging for decision makers to consistently and accurately compare criteria and alternatives in pairs. If the pairwise comparisons are not done carefully, they can introduce errors and affect the reliability of the final priorities.

- Sensitivity to input data: AHP results may be sensitive to small changes in the pairwise comparison data. Inconsistencies or slight variations in judgments can have a significant impact on final priorities, which could lead to different decision outcomes.

- Consistency issues: AHP incorporates a consistency check to ensure the reliability of pairwise comparisons. However, decision makers may find it difficult to achieve the required level of consistency, and it may be subjective to determine an appropriate consistency threshold.

Despite these challenges, AHP is a valuable decision-making tool when used appropriately and with awareness of its limitations, as the authors of this article have shown.

It is a well-structured, well-written article that does not hide the methodological problems that arise and tries to solve them with different strategies. Among the recommendations and suggestions for the authors are the following:

-It is proposed to change the keywords cave to endokarts

-A thorough revision of the English must be carried out, since the translation seems very literal from the mother tongue.

-Each map should occupy one page so as not to lose the details that are represented.

-Figure 1 suggests that the elevation parameter be represented continuously and not discretely in the legend.

-Table 2 needs several improvements due to its size. It is suggested to take an annex and if not lower the line spacing and the size of the letters. Also try to synthesize the content as much as possible. The table should be presented in landscape as in the available preprint.

-The legend of figure 5 needs to be rearranged. It is suggested that the karstic spring and the hydrography be located under the relief energy parameter. Also, the Karstification Susceptibility variable in the legend does not match the map title, which is Lithostratigraphic Units.

Finally, authors are recommended to consult the following publication: Nhu, V.-H.; Rahmati, O.; Falah, F.; Shojaei, S.; Al-Ansari, N.; Shahabi, H.; Shirzadi, A.; Gorski, K.; Nguyen, H.; Ahmad, BB Mapeo del potencial de manantial de agua subterránea en el sistema acuífero kárstico utilizando modelos bivariados y multivariados de conjunto novedoso. Agua 2020, 12, 985. https://doi.org/10.3390/w12040985

Author Response

REVIEWER 1

The article aims to produce a cartographic model that identifies the areas most likely to find karstic caves (the endokarstic potential) in the northern sector of the Santo António plateau (extremadura limestone massif, central Portugal).

The analysis strategy consisted of collecting and processing geological, topographic, hydrogeological and vegetation cover data, as well as the location of the known entrances to the caves in the study area. Four conditioning factors were extracted from the collected data: lithostratigraphic units, fracture density, relief energy, and land cover. In a multi-criteria decision analysis framework, each previously chosen conditioning factor and the respective classes/categories were weighted using an Analytical Hierarchy Process (AHP).

The paper uses a geographic information system (GIS) to integrate geological, topographic, hydrogeological, and land cover data into a spatial database, which can be used for further research and analysis. The results of the GIS-built mapping model look promising, considering that the known cave entrances are mainly in areas classified as having high to very high endokarst potential.

The endokarstic potential map obtained can be used in operational and strategic environmental planning to help decision makers and local caving teams to more accurately and thoughtfully define the areas to be investigated, substantially reducing time and costs. field survey costs.

The article contributes to the understanding of the parameters that most contribute to endokarst formation and provides a more reliable model for a spatial representation of underground karstification potential.

The main criticism that can be made of the article is that the endokarst potential map produced in this study is based on the available data and the chosen constraints. Therefore, it may not be entirely accurate and may require further validation through field work. Furthermore, the applicability of the map may be limited to the study area and may not be generalizable to other regions.

Authors’ reply: Yes, it is true. The modelling caried out is based exclusively on data accessible and available from local authorities and from the bibliography, although its verification (non-validation) through the location of the known caves entrances demonstrates that it can be accepted. Of course, the future use of more robust objective analytical procedures, together with the results of the fieldwork prospecting by the local speleologists (guided by our cartographic prototype), possibly extendable to the whole karst massif, is desirable and in our intentions. To highlight these issues, a new final section has been added (please, see the new section 6).

On the other hand, the Analytical Hierarchy Process (AHP), a decision-making method developed by Thomas L. Saaty in the 1970s, is not without its challenges and potential problems. The main problems associated with are:

- Subjectivity: AHP relies heavily on subjective judgments and comparisons in which the interpretation and perception of the relative importance of criteria and alternatives may vary among different decision makers, leading to potential biases and inconsistencies.

- Inconsistent pairwise comparisons: It can be challenging for decision makers to consistently and accurately compare criteria and alternatives in pairs. If the pairwise comparisons are not done carefully, they can introduce errors and affect the reliability of the final priorities.

- Sensitivity to input data: AHP results may be sensitive to small changes in the pairwise comparison data. Inconsistencies or slight variations in judgments can have a significant impact on final priorities, which could lead to different decision outcomes.

- Consistency issues: AHP incorporates a consistency check to ensure the reliability of pairwise comparisons. However, decision makers may find it difficult to achieve the required level of consistency, and it may be subjective to determine an appropriate consistency threshold.

Authors’ reply: Yes, of all this is true. However, it is widely demonstrated in the literature how, with due care in the procedures and in the choice of validators, the used multi-criteria technique (i.e., the AHP) responds satisfactorily to the needs of a preliminary analysis of the phenomenon in question (in this specific case, the endokarst potential assessment), for then move (at a later stage of the investigation) to the application of more sophisticated and robust methodologies. Furthermore, the main objective of the present work is, above all, to produce a prototype of endokarst potential map that can contribute, in a practical sense, to a more efficient management of the karst territories under investigation. This work also aims to establish a relatively simple modelling procedure within reach of local professionals (speleologists, autarches, etc.). Being easier to apply, it’s also easier to generalize its use, it’s easy to know its weaknesses, and apply all necessary corrections to make it more robust and efficient. We classified the model with a good predictive capacity, regardless of the possibility of using, in the future, other more robust and sophisticated analytical methodologies (such as artificial intelligence). A comparison between different methodologies is beyond the scope of this work. These considerations have been added in the revised manuscript.

Despite these challenges, AHP is a valuable decision-making tool when used appropriately and with awareness of its limitations, as the authors of this article have shown.

It is a well-structured, well-written article that does not hide the methodological problems that arise and tries to solve them with different strategies. Among the recommendations and suggestions for the authors are the following:

-It is proposed to change the keywords cave to endokarts

Authors’ reply: done.

-A thorough revision of the English must be carried out, since the translation seems very literal from the mother tongue.

Authors’ reply: done (see certificate of English language edition by Elsevier).

-Each map should occupy one page so as not to lose the details that are represented.

Authors’ reply: done.

-Figure 1 suggests that the elevation parameter be represented continuously and not discretely in the legend.

Authors’ reply: done.

-Table 2 needs several improvements due to its size. It is suggested to take an annex and if not lower the line spacing and the size of the letters. Also try to synthesize the content as much as possible. The table should be presented in landscape as in the available preprint.

Authors’ reply: done.

-The legend of figure 5 needs to be rearranged. It is suggested that the karstic spring and the hydrography be located under the relief energy parameter. Also, the Karstification Susceptibility variable in the legend does not match the map title, which is Lithostratigraphic Units.

Authors’ reply: done.

Finally, authors are recommended to consult the following publication: Nhu, V.-H.; Rahmati, O.; Falah, F.; Shojaei, S.; Al-Ansari, N.; Shahabi, H.; Shirzadi, A.; Gorski, K.; Nguyen, H.; Ahmad, BB Mapeo del potencial de manantial de agua subterránea en el sistema acuífero kárstico utilizando modelos bivariados y multivariados de conjunto novedoso. Agua 2020, 12, 985. https://doi.org/10.3390/w12040985

Authors’ reply: done.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

There are a number of issues that must be resolved before publication can be considered.  If the following problems are well solved, then I think the paper has made a contribution.

1.      Although Endokarst data collection is somewhat part of Remote Sensing, the work heavily falls under Geomatics, GIS, and Data Science works As GIS is both mature technologies, this paper shows poor replication with very low implications compared to other literatures for such high efforts.

2.      The topic of this manuscript has good practical significance, but the academic innovation is not strong.

3.      The use of Analytic Hierarchy Process (AHP) in the manuscript is a relatively outdated method, which does not give new information for the broader scientific community, and this is not worth publishing.

4.      There is a lack of historical literature or empirical studies that show that temperature and rainfall do not have a significant effect on small-scale study areas, and the authors have not been able to give a criterion for delineating the size of the study area .(Line198-200)

5.      The training set here is only 1/3, and the test set is 2/3. Generally, the more data used for training, the more accurate the results. Why is the training set less than the test set here? It is recommended to give a reason for the data set allocation. (Line240-243)

6.      Are there problems with tables on consecutive pages that detract from the look and feel and are not intuitive to read? Is it possible to try to present it in a diagram? (Line323-325)

7.      It only mentioned modifying the relevant factors in the iterative process. It is suggested to introduce specifically how to modify the relevant factors.(Line452-454)

8.      Does the author lack persuasiveness in concluding that he has good predictive power simply by using the AHP method to obtain results, without comparing them with other methods? (Line532-533)

9.      this manuscript only uses the AHP method for modeling, and the model is too single. And other models can be added for comparison to draw more reliable conclusions.

10.  The AUC here is not considered the highest compared to other AUCs in other literature. Therefore, such replicated works should never be presented as academic works unless it is a breakthrough.

11.  The statistical indicators in this literature are too few, such as not discussing their accuracy to compare research results, resulting in a somewhat biased paper.

12.  It is best to provide corresponding formulas for AHP, AUC, and other methods to better explain them.

13.  The abstract needs improvement to the point.

14.  The discussion is little and needs improvement.

 

Author Response

reviewer 2

There are a number of issues that must be resolved before publication can be considered.  If the following problems are well solved, then I think the paper has made a contribution.

  1. Although Endokarst data collection is somewhat part of Remote Sensing, the work heavily falls under Geomatics, GIS, and Data Science works As GIS is both mature technologies, this paper shows poor replication with very low implications compared to other literatures for such high efforts.

Authors’ reply: it is widely demonstrated in the literature how, with due care in the procedures and in the choice of validators, the used multi-criteria technique (i.e., the AHP) responds satisfactorily to the needs of a preliminary analysis of the phenomenon in question (in this specific case, the endokarst potential assessment), for then move (at a later stage of the investigation) to the application of more sophisticated and robust methodologies. Furthermore, the main objective of the present work is, above all, to produce a prototype of endokarst potential map that can contribute, in a practical sense, to a more efficient management of the karst territories under investigation. This work also aims to establish a relatively simple modelling procedure within reach of local professionals (speleologists, autarches, etc.). Being easier to apply, it’s also easier to generalize its use, it’s easy to know its weaknesses, and apply all necessary corrections to make it more robust and efficient. We classified the model with a good predictive capacity, regardless of the possibility of using, in the future, other more robust and sophisticated analytical methodologies (such as artificial intelligence). A comparison between different methodologies is beyond the scope of this work. These considerations have been added in the revised manuscript.

 

  1. The topic of this manuscript has good practical significance, but the academic innovation is not strong.

Authors’ reply: see previous authors’ reply.

  1. The use of Analytic Hierarchy Process (AHP) in the manuscript is a relatively outdated method, which does not give new information for the broader scientific community, and this is not worth publishing.

       Authors’ reply: see the first authors’ reply.

  1. There is a lack of historical literature or empirical studies that show that temperature and rainfall do not have a significant effect on small-scale study areas, and the authors have not been able to give a criterion for delineating the size of the study area .(Line198-200)

Authors’ reply: new supporting bibliographic references have been added to make the ideas expressed clearer and more consistent. A justification about the chosen side of the study area was added.

  1. The training set here is only 1/3, and the test set is 2/3. Generally, the more data used for training, the more accurate the results. Why is the training set less than the test set here? It is recommended to give a reason for the data set allocation. (Line240-243)

Authors’ reply: The existence of distinct and heterogeneous territories into the study area – such as a typical karst terrain with many inventoried caves and other without inventoried caves, as well as non-karst territories - conditioned the delimitation of the training and test subsets of data. In particular, the train area was chosen considering the need to include a large number of known caves and, at the same time, covering all the distinct types of existing territories (karst and non-karst).

  1. Are there problems with tables on consecutive pages that detract from the look and feel and are not intuitive to read? Is it possible to try to present it in a diagram? (Line323-325)

Authors’ reply: done.

  1. It only mentioned modifying the relevant factors in the iterative process. It is suggested to introduce specifically how to modify the relevant factors.(Line452-454)

Authors’ reply: reformulated period.

  1. Does the author lack persuasiveness in concluding that he has good predictive power simply by using the AHP method to obtain results, without comparing them with other methods? (Line532-533)

Authors’ reply: see the first authors’ reply.

  1. this manuscript only uses the AHP method for modeling, and the model is too single. And other models can be added for comparison to draw more reliable conclusions.

Authors’ reply: see the first authors’ reply.

  1. The AUC here is not considered the highest compared to other AUCs in other literature. Therefore, such replicated works should never be presented as academic works unless it is a breakthrough.

Authors’ reply: see the first authors’ reply as well as the new section 6. In addition, considering that no similar investigation exist in the study area and in all of Portugal, this is a breakthrough.

  1. The statistical indicators in this literature are too few, such as not discussing their accuracy to compare research results, resulting in a somewhat biased paper.

Authors’ reply: see previous authors’ replies.

  1. It is best to provide corresponding formulas for AHP, AUC, and other methods to better explain them.

Authors’ reply: done.

  1. The abstract needs improvement to the point.

Authors’ reply: done.

  1. The discussion is little and needs improvement.

Authors’ reply: done. See also the new section 6.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The purpose of this manuscript is to develop a cartographic model that can identify areas with a high likelihood of containing karstic caves - known as endokarst potential - in the northern sector of Santo António Plateau. To achieve this, the authors collected, processed, and integrated geological, topographic, hydrogeological, and vegetation cover data into a spatial database using a Geographic Information System (GIS). Four conditioning factors were extracted from the data, including lithostratigraphic units, fracture density, relief energy, and land cover. In a multi-criteria decision-making analysis framework, each conditioning factor, along with its respective classes/categories, was weighted using the Analytic Hierarchy Process (AHP). The model's predictive capacity was verified using the Receiver Operating Characteristic (ROC) curves and the Area Under the Curve (AUC). This investigation is of significant importance to sustainability and merits publication. However, before publication, some minor to moderate revisions to the English language are required, along with some important issues.

Specific comments:

1.      Line 30: The focus of this article is on the assessment of Endokarst potential. As a reader, I expected the introductory paragraph of the article to provide some background information and context about the topic of Endokarst potential assessment. Additionally, i expected the paragraph to identify the research gap and the specific research problem that the study aims to address. This would help understand the significance and relevance of the research and how it contributes to the existing literature or practice.

 

2.      Line 40: The first sentence of each paragraph should present the author's own statement or argument, without the use of references or citations. This helps to establish the author's perspective and provides a clear and concise introduction to the main idea of the paragraph. However, subsequent sentences in the paragraph may include references or citations to support the author's argument or provide evidence for their claims.

 

3.      Line 84: A proper and stronger transition is needed to connect with the previous paragraph. This is not the proper way to do.

 

4.      Line 134: It is important that all figures in the article have an explanatory caption that uses a similar font and size. However, some of the figures appear dissimilar, which may cause confusion or inconsistency for readers. Therefore, it is recommended that the author revise the figures to ensure that they are consistent in their format and style, which will help to enhance the clarity and professionalism of the article.

 

5.      Line 148: same as Line 134

 

6.      Line 201: In your study, four karstification conditioning factors were utilized: lithostratigraphic units, fracture density, relief energy, hydrological data, and vegetation cover data. These factors were collected at varying scales or resolutions, specifically at 1:50,000 for lithostratigraphic units and fracture density, 1:10,000 for relief energy (topographic data), and 1:25,000 for vegetation cover data. To address the challenge of working with data of different scales or resolutions when building the cartographic model, could you please provide more information about the specific approach used in your study?

 

7.      Line 293: such us or such as?

 

8.      Line 317: The quantitative scale value is (0-1) or (0-10)? Please explain why?

 

9.      Line 323-325: Table-2 After reviewing the information provided, it appears that there is a discrepancy between the assigned weightings for FValues in the footnote of Table-2 and the qualitative assessment column in the same table. The footnote states that FValues from 0 to 4 with the assigned weighting: 0-1 (Very low), 1-1.5 (Low), 1.5-2.3 (Moderate), 2.3-3 (High), and 3-4 (Very high). However, in the qualitative assessment column) F, value of 0,9 was labeled as "very low," which does not match with the footnote information. It is possible that this discrepancy was an error and requires further clarification.

 

10.   Line 429: While Receiver operating characteristics (ROC) analysis and Area under the curve (AUC) are useful tools for evaluating the performance of binary classification models, there are some limitations to their use in endokarst potential assessment. Please explain how you overcome these challenges?

 

11.   Line 451: One of the main disadvantages of using ROC and AUC for endokarst potential assessment is that they are based on binary classification, which may not accurately represent the complex spatial and temporal variability of karst systems. Endokarst potential assessment involves evaluating the likelihood of karst development in a given area based on a range of factors, such as geology, topography, and hydrology. While a binary classification model can be useful for identifying areas with high or low karst potential, it may not capture the nuances of the karstification process or the variability of karst features within a given area.

 

12.   Line 506: Figure 7: Please check the comment number 4.

 

13.   Line 536: Figure 8: Please check the comment number 4.

 

14.   Extensive English correction and formatting are needed for this document. The font style is inconsistent across various sections, such as the reference section. Therefore, it is important to revise the document to ensure that the font style is consistent throughout. Additionally, there may be other formatting issues that need to be addressed, such as inconsistent spacing or indentation. A thorough review is necessary to ensure that the document meets the required standards of professionalism and clarity.

 

Comments on the Quality of English Language

Some minor to moderate revisions to the English language are required

Author Response

Reviewer 3

The purpose of this manuscript is to develop a cartographic model that can identify areas with a high likelihood of containing karstic caves - known as endokarst potential - in the northern sector of Santo António Plateau. To achieve this, the authors collected, processed, and integrated geological, topographic, hydrogeological, and vegetation cover data into a spatial database using a Geographic Information System (GIS). Four conditioning factors were extracted from the data, including lithostratigraphic units, fracture density, relief energy, and land cover. In a multi-criteria decision-making analysis framework, each conditioning factor, along with its respective classes/categories, was weighted using the Analytic Hierarchy Process (AHP). The model's predictive capacity was verified using the Receiver Operating Characteristic (ROC) curves and the Area Under the Curve (AUC). This investigation is of significant importance to sustainability and merits publication. However, before publication, some minor to moderate revisions to the English language are required, along with some important issues.

Specific comments:

  1. Line 30: The focus of this article is on the assessment of Endokarst potential. As a reader, I expected the introductory paragraph of the article to provide some background information and context about the topic of Endokarst potential assessment. Additionally, i expected the paragraph to identify the research gap and the specific research problem that the study aims to address. This would help understand the significance and relevance of the research and how it contributes to the existing literature or practice.

Authors’ reply: An intro has been added as suggested.

  1. Line 40: The first sentence of each paragraph should present the author's own statement or argument, without the use of references or citations. This helps to establish the author's perspective and provides a clear and concise introduction to the main idea of the paragraph. However, subsequent sentences in the paragraph may include references or citations to support the author's argument or provide evidence for their claims.

Authors’ reply: corrected.

  1. Line 84: A proper and stronger transition is needed to connect with the previous paragraph. This is not the proper way to do.

Authors’ reply: Changed as suggested.

  1. Line 134: It is important that all figures in the article have an explanatory caption that uses a similar font and size. However, some of the figures appear dissimilar, which may cause confusion or inconsistency for readers. Therefore, it is recommended that the author revise the figures to ensure that they are consistent in their format and style, which will help to enhance the clarity and professionalism of the article.

Authors’ reply: Improved.

  1. Line 148: same as Line 134

Authors’ reply: Improved.

  1. Line 201: In your study, four karstification conditioning factors were utilized: lithostratigraphic units, fracture density, relief energy, hydrological data, and vegetation cover data. These factors were collected at varying scales or resolutions, specifically at 1:50,000 for lithostratigraphic units and fracture density, 1:10,000 for relief energy (topographic data), and 1:25,000 for vegetation cover data. To address the challenge of working with data of different scales or resolutions when building the cartographic model, could you please provide more information about the specific approach used in your study?

Authors’ reply: To integrate the cartographic information (with various scales and resolutions) in the model we build a set of raster thematic layers, used as input variables, with a homogeneous spatial resolution of 5 m pixels.

  1. Line 293: such us or such as?

Authors’ reply: corrected to such as.

  1. Line 317: The quantitative scale value is (0-1) or (0-10)? Please explain why?
  2. Line 323-325: Table-2 After reviewing the information provided, it appears that there is a discrepancy between the assigned weightings for FValues in the footnote of Table-2 and the qualitative assessment column in the same table. The footnote states that FValues from 0 to 4 with the assigned weighting: 0-1 (Very low), 1-1.5 (Low), 1.5-2.3 (Moderate), 2.3-3 (High), and 3-4 (Very high). However, in the qualitative assessment column) F, value of 0,9 was labeled as "very low," which does not match with the footnote information. It is possible that this discrepancy was an error and requires further clarification.

Authors’ reply for comments 8 and 9: Added explanation in footnote of Table 2. “Each lithostratigraphic unit was evaluated according to the scale of values from 0 to 4, with the following weighting: 0-1 (Very low); 1-1.5 (Low); 1.5-2.3 (Moderate); 2.3-3 (High); 3-4 (Very high), which are the sum of the assessment made for each of the components under analysis (e.g. lithology, bed geometry), in turn analyzed on a scale of 0-1.”

  1. Line 429: While Receiver operating characteristics (ROC) analysis and Area under the curve (AUC) are useful tools for evaluating the performance of binary classification models, there are some limitations to their use in endokarst potential assessment. Please explain how you overcome these challenges?

       Authors’ reply: With a careful analysis of the ROC with the analysis of the accumulated frequencies, we were able to capture the variability of the modeled phenomenon (see Figure 8) and its interpretation.

       We classified the model with a good predictive capacity, regardless of the possibility of using, in the future, other more robust and sophisticated analytical methodologies (such as artificial intelligence). A comparison between different methodologies is beyond the scope of this work.

  1. Line 451: One of the main disadvantages of using ROC and AUC for endokarst potential assessment is that they are based on binary classification, which may not accurately represent the complex spatial and temporal variability of karst systems. Endokarst potential assessment involves evaluating the likelihood of karst development in a given area based on a range of factors, such as geology, topography, and hydrology. While a binary classification model can be useful for identifying areas with high or low karst potential, it may not capture the nuances of the karstification process or the variability of karst features within a given area.

Authors’ reply: idem. See also small improvements in the discussion and insertion of the chapter “Cartographic model challenges” for more explanation.  

  1. Line 506: Figure 7: Please check the comment number 4.

Authors’ reply: Improved.

 

  1. Line 536: Figure 8: Please check the comment number 4.

Authors’ reply: improved.

  1. Extensive English correction and formatting are needed for this document. The font style is inconsistent across various sections, such as the reference section. Therefore, it is important to revise the document to ensure that the font style is consistent throughout. Additionally, there may be other formatting issues that need to be addressed, such as inconsistent spacing or indentation. A thorough review is necessary to ensure that the document meets the required standards of professionalism and clarity.

Some minor to moderate revisions to the English language are required

Authors’ reply: Done.

 

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

I want to congratulate you on your clearly written manuscript.

I would like to emphasize a couple of points such as

-the problem of references (they should be given as numbers) as other mdpi journals.

-Font and quality of the papers (many of them are out of order)

-tiny misspellings

My other comments are embedded within the pdf file and can be downloaded.

 

Overall I have

 

Comments for author File: Comments.pdf

Author Response

REVIEWER 4

I want to congratulate you on your clearly written manuscript.

I would like to emphasize a couple of points such as

-the problem of references (they should be given as numbers) as other mdpi journals.

-Font and quality of the papers (many of them are out of order)

-tiny misspellings

My other comments are embedded within the pdf file and can be downloaded.

 

Overall I have

 

Authors’ reply: Thanks for the compliments and suggestions. All suggestions in the PDF were considered for improvement or correction of the manuscript.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

To the Author:

There are a number of issues that must be resolved before publication can be considered.  If the following problems are well solved, then I think the paper has made a contribution.

1.      The method used in the paper is still Analytic Hierarchy Process (AHP), which is a relatively old method and not very innovative for scientific research.

2.      There is a single method used in the paper, while AHP is more subjective and without another classical method to compare it with, its results are not very convincing.

3.      In the paper, the AUC of the training and testing areas are 78.3% and 86% respectively, in general, the results on the training set should be better than the testing set because the model minimizes the error or the results are optimal on the training set before the training process stops, but in the paper it's the other way around, please explain the reason for this.

4.      The article uses too few statistical indicators and does not consider the performance of accuracy, recall rate, kappa coefficient and other indicators on the results, and the conclusions obtained are biased.

5.    There are some typographical problems in this article, such as the fact that Figure 4 is mentioned in line 192 of this article, but it does not appear until line 321. 

Author Response

There are a number of issues that must be resolved before publication can be considered.  If the following problems are well solved, then I think the paper has made a contribution.

  1. The method used in the paper is still Analytic Hierarchy Process (AHP), which is a relatively old method and not very innovative for scientific research.
  2. There is a single method used in the paper, while AHP is more subjective and without another classical method to compare it with, its results are not very convincing.
  3. In the paper, the AUC of the training and testing areas are 78.3% and 86% respectively, in general, the results on the training set should be better than the testing set because the model minimizes the error or the results are optimal on the training set before the training process stops, but in the paper it's the other way around, please explain the reason for this.
  4. The article uses too few statistical indicators and does not consider the performance of accuracy, recall rate, kappa coefficient and other indicators on the results, and the conclusions obtained are biased.
  5.   There are some typographical problems in this article, such as the fact that Figure 4 is mentioned in line 192 of this article, but it does not appear until line 321. 

Authors’ reply: Improved.

Reviewer 3 Report

Comments and Suggestions for Authors

The authors have addressed my major concerns. 

Table 4 contains too many details. either should it be simiplied, or put it to appendix.

Comments on the Quality of English Language

English can be improved by a native speaker.

Author Response

The authors have addressed my major concerns. 

Table 4 contains too many details. either should it be simiplied, or put it to appendix.

Authors’ reply: Improved.

English can be improved by a native speaker.

Authors’ reply: Done.

Round 3

Reviewer 2 Report

Comments and Suggestions for Authors

Qualified job.

Back to TopTop