Multivariate Analysis of Geological Data for Regional Studies of Geodiversity
, Tine Larsen Angvik 2 and Linda Wickström 3Round 1
Reviewer 1 Report
Thank you for the possibility to review this manuscript. This is an interesting study concerning the use of GIS-based geological data for comparing geodiversity properties across southern Norway and Sweden. In the study, ordination and cluster analysis methodologies have been used to calculate the importance and connections of a variety of geological/geodiversity variables across the study units. I very much like the general setting of the study, and consider this as an important and novel perspective for measuring geodiversity. This study gives the reader a nice, general overview of how quantitative, GIS-based geodiversity data can be obtained and analyzed in a regional context. The study setting, as well as the data building and the analysis methods, are topical and the study can interest a variety of readers with different backgrounds.
The general framework of the study is clear, but I suggest that the manuscript needs some re-writing and clarifying to be a logical and well-reasoned entity. My concerns and questions/comments regarding to the manuscript are detailed below row-by-row:
Rows 9-25: The abstract should be clearer, so I highly recommend the authors to re-write it. I suggest writing with a logical structure where first the theoretical background of the study is introduced, then the research aims/questions, then the data&methodology, results, and discussion.
Rows 29-33: A long sentence, please, clarify.
Row 34: Why are you citing here the first edition of Gray´s book, and not to newer, second edition?
Row 46: Please, rephrase the sentence so that it would not end to “etc”.
Rows 53-58: You could add here a sentence here about the context of this study (with what data, where is the study conducted). In addition, I suggest adding a more detailed aim of the study already in here (which is now described in rows 135-138).
Figure 1: I suggest placing this small map of the location of the study area as an index map included in which in now the Figure 2 (the bedrock map of the area). As such, it now would need the basic elements of maps included in the map: legend (what are the colours on the background of the map?), scalebar, north arrow.
Figure 2: I found it difficult to distinguish between some of the red and green colours of the rock types (especially between alkaline intrusive rock and acidic intrusive rock + quarz-feldspar-rich metamorphic rock; and, mica-rich sedimentary rock and mica-rich metamorphic rock).
Figure 3 (or actually, this should be Figure 4? now there are two Fig. 3s): It would be more reader-friendly if you put information of the different landscape major types and their colours as a legend in the map (instead of having it in the figure caption). I am also missing a scalebar and a north arrow from this map. You could also put a rectangle marking the location of the test area next to the arrow, so the reader would know, how big the test area actually is in relation to the whole study area.
Chapter 2.4. Terrain ruggedness, and Table 1: This chapter needs re-writing/clarifying. For me, it was unclear that what was actually done in here, in which units, and with which methods. How did you actually evaluate the performance of the indices (what was the method, how did you end up to the results of “weak”, “fairly strong”, “strong” etc)? What is the response that you are referring in Table 1?
Row 164: Could you explain a bit more here, what do Rough, Roughness and VRM mean/include?
Table 2: This table is quite overwhelming because in its current form, it includes a great number of variables just listed in some order below each other. Could they be ordered somehow better, e.g. in relation to which element of geodiversity are they referring (geology, terrain ruggedness, water elements…)? And, it would be useful for the reader to see a column where the measure unit of each variable is represented. Now it was unclear for me, what do the variables actually measure/represent (it is not listed in the table, or in the main text either). In the table caption, you say that these variables are used for the analysis in the test area – do you mean that these were only adopted to the test area shown in Figure 4? Related to this, you should be more clear in the main text, too, when telling that which analysis have been done to which area, and which variables are used in which analysis. You should also state clearly the unit of your study in the Table caption (landscape areas?).
Row 190: Could you open here, which polygons you are referring to in this context (landscape areas polygons, Fig. 3?)?
Row 210-211: You should put here more information on the methodology. E.g., which function of vegan did you use to conduct the analyses? What is the data structure (is there a distance matrix behind PCA ordinations)? Why have you done the ordination analysis with PCA ordination (not e.g. with NMDS)? Will the code for the analysis be available somewhere? How about the data?
Row 212-213: Why did you choose to examine the first four PCs? Where was this decision based on?
Rows 222-225: A long sentence, please, clarify.
Rows 218-230: This text (which describes the analysis process rather than results of it) belongs to the Materials and methods chapter, not to the Results.
Rows 234-237: A long sentence, please, clarify.
Rows 239-240: Please, re-phrase “tend to fall on the left side of the diagram”
Figure 5: Is there a mix-up of axis names here (now 4 and 3 -should they be 1 and 2)? Related to the ordination plots (Fig. 5 and Fig. 6) in general, these are not very reader-friendly. Would there be a way to e.g., distinguish different groups of variables by different colours/circles? Could the variable names be somehow better distinguished (now they are a bit of a mess in the middle…)? You could also add the information of the variation explained by each axis (e.g., in brackets after the axis name). In the figure text, you should tell the reader in more details, what do the variables and the axis indicate. Now, the figures cannot be read independently from the text (as I think that they should be – the figure and the caption should include all the information for the reader to be able to interpret these without scrolling further into text).
Rows 271-272: Where is this information based on?
Table 3. This could be moved to online appendices since especially the information on mean PCA values is not very important for the reader. The number/count of clusters could be added to Figure 7, e.g. in brackets after each cluster number in the map legend.
Figure 8: There is no need for a title inside the graph since there is a Figure caption. For me, it would have been clearer if the clusters were in order of 1-16, not in an order of importance.
Rows 372-430: I suggest that you would first write in the Discussion on the general results and discussion related to those (the text in these rows 372-430). Then you could write about the uncertainties related to the analysis (the text which is now in rows 338-371 and 431-443). I think that the Discussion chapter would be more logical this way, and the importance of the results would be highlighted by placing the interpretations of the results, and not the text about possible uncertainties, first in the text.
Conclusion –chapter: As in Discussion, I recommend re-ordering the text in the Conclusions. You could first write about the general findings, then maybe shortly go through the uncertainties (that are now in rows 448-459), and then in the end, draw some generalized conclusions/future perspectives.
Author Response
Thank you for the possibility to review this manuscript. This is an interesting study concerning the use of GIS-based geological data for comparing geodiversity properties across southern Norway and Sweden. In the study, ordination and cluster analysis methodologies have been used to calculate the importance and connections of a variety of geological/geodiversity variables across the study units. I very much like the general setting of the study, and consider this as an important and novel perspective for measuring geodiversity. This study gives the reader a nice, general overview of how quantitative, GIS-based geodiversity data can be obtained and analyzed in a regional context. The study setting, as well as the data building and the analysis methods, are topical and the study can interest a variety of readers with different backgrounds.
The general framework of the study is clear, but I suggest that the manuscript needs some re-writing and clarifying to be a logical and well-reasoned entity. My concerns and questions/comments regarding to the manuscript are detailed below row-by-row:
RESPONSE: Thank you for this assessment and for the detailed comments below. We have read the response carefully and done changes mostly in line of what has been suggested and think it represent an important improvement to the manuscript.
Rows 9-25: The abstract should be clearer, so I highly recommend the authors to re-write it. I suggest writing with a logical structure where first the theoretical background of the study is introduced, then the research aims/questions, then the data&methodology, results, and discussion.
RESPONSE: We have rewritten the abstract with this in mind.
Rows 29-33: A long sentence, please, clarify.
RESPONSE: This has been adjusted. We hope it is better now.’
Row 34: Why are you citing here the first edition of Gray´s book, and not to newer, second edition?
RESPONSE: The thinking was linked to illustrating that the definition develops and tend to get mor detailed. But this point is not important, so the reference to the first edition has been deleted.
Row 46: Please, rephrase the sentence so that it would not end to “etc”.
RESPONSE: Corrected
Rows 53-58: You could add here a sentence here about the context of this study (with what data, where is the study conducted). In addition, I suggest adding a more detailed aim of the study already in here (which is now described in rows 135-138).
RESPONSE: Some additional text has been added.
Figure 1: I suggest placing this small map of the location of the study area as an index map included in which in now the Figure 2 (the bedrock map of the area). As such, it now would need the basic elements of maps included in the map: legend (what are the colours on the background of the map?), scalebar, north arrow.
RESPONSE: Good suggestion! Has been done.
Figure 2: I found it difficult to distinguish between some of the red and green colours of the rock types (especially between alkaline intrusive rock and acidic intrusive rock + quarz-feldspar-rich metamorphic rock; and, mica-rich sedimentary rock and mica-rich metamorphic rock).
RESPONSE: We have adjusted colours, so it reads hopefully better now.
Figure 3 (or actually, this should be Figure 4? now there are two Fig. 3s): It would be more reader-friendly if you put information of the different landscape major types and their colours as a legend in the map (instead of having it in the figure caption). I am also missing a scalebar and a north arrow from this map. You could also put a rectangle marking the location of the test area next to the arrow, so the reader would know, how big the test area actually is in relation to the whole study area.
RESPONSE: Numbering of figures have been adjusted. Figures has been unified with legends.
Chapter 2.4. Terrain ruggedness, and Table 1: This chapter needs re-writing/clarifying. For me, it was unclear that what was actually done in here, in which units, and with which methods. How did you actually evaluate the performance of the indices (what was the method, how did you end up to the results of “weak”, “fairly strong”, “strong” etc)? What is the response that you are referring in Table 1?
RESPONSE: We have tried to explain this better and also to simplify somewhat. The aim was to have a better basis for selecting a ruggedness indicator that works well using DEM with high resolution. The assessment is visual. This has been clarified.
Row 164: Could you explain a bit more here, what do Rough, Roughness and VRM mean/include?
RESPONSE: More explanation has been added.
Table 2: This table is quite overwhelming because in its current form, it includes a great number of variables just listed in some order below each other. Could they be ordered somehow better, e.g. in relation to which element of geodiversity are they referring (geology, terrain ruggedness, water elements…)? And, it would be useful for the reader to see a column where the measure unit of each variable is represented. Now it was unclear for me, what do the variables actually measure/represent (it is not listed in the table, or in the main text either). In the table caption, you say that these variables are used for the analysis in the test area – do you mean that these were only adopted to the test area shown in Figure 4? Related to this, you should be more clear in the main text, too, when telling that which analysis have been done to which area, and which variables are used in which analysis. You should also state clearly the unit of your study in the Table caption (landscape areas?).
RESPONSE: We have considered this one as an appendix, but find it central so we have kept it in the text. Headlines and a new column as suggested has been added.
Row 190: Could you open here, which polygons you are referring to in this context (landscape areas polygons, Fig. 3?)?
RESPONSE: We have tried to clarify better in the text what these landscape areas are, hopefully it is better specified now.
Row 210-211: You should put here more information on the methodology. E.g., which function of vegan did you use to conduct the analyses? What is the data structure (is there a distance matrix behind PCA ordinations)? Why have you done the ordination analysis with PCA ordination (not e.g. with NMDS)? Will the code for the analysis be available somewhere? How about the data?
RESPONSE: This has been done and the R-script included in the appendix
Row 212-213: Why did you choose to examine the first four PCs? Where was this decision based on?
RESPONSE: It has been in the text clarified that the PCA yielded 4 interpretable axis. We did not chose to examine only the first four (of many).
Rows 222-225: A long sentence, please, clarify.
RESPONSE: We have done so.
Rows 218-230: This text (which describes the analysis process rather than results of it) belongs to the Materials and methods chapter, not to the Results.
RESPONSE: Have been moved.
Rows 234-237: A long sentence, please, clarify.
RESPONSE: Have done so.
Rows 239-240: Please, re-phrase “tend to fall on the left side of the diagram”
RESPONSE: Has been done. We have now referred to this has high and low PCA values.
Figure 5: Is there a mix-up of axis names here (now 4 and 3 -should they be 1 and 2)? Related to the ordination plots (Fig. 5 and Fig. 6) in general, these are not very reader-friendly. Would there be a way to e.g., distinguish different groups of variables by different colours/circles? Could the variable names be somehow better distinguished (now they are a bit of a mess in the middle…)? You could also add the information of the variation explained by each axis (e.g., in brackets after the axis name). In the figure text, you should tell the reader in more details, what do the variables and the axis indicate. Now, the figures cannot be read independently from the text (as I think that they should be – the figure and the caption should include all the information for the reader to be able to interpret these without scrolling further into text).
RESPONSE: Axis names has been corrected. We have introduced different colours and symbols in the figure and hope it helps. Variation explained has also been added. The figure text has been expanded.
Rows 271-272: Where is this information based on?
RESPONSE: I am not sure I understand the clusters. We have tried to cluster the PCA values and made a map that show the geographical distribution of them.
Table 3. This could be moved to online appendices since especially the information on mean PCA values is not very important for the reader. The number/count of clusters could be added to Figure 7, e.g. in brackets after each cluster number in the map legend.
RESPONSE: Have done so
Figure 8: There is no need for a title inside the graph since there is a Figure caption. For me, it would have been clearer if the clusters were in order of 1-16, not in an order of importance.
RESPONSE: Heading deleted. Order of clusters changed.
Rows 372-430: I suggest that you would first write in the Discussion on the general results and discussion related to those (the text in these rows 372-430). Then you could write about the uncertainties related to the analysis (the text which is now in rows 338-371 and 431-443). I think that the Discussion chapter would be more logical this way, and the importance of the results would be highlighted by placing the interpretations of the results, and not the text about possible uncertainties, first in the text.
RESPONSE: The chapter is restructured as suggested.
Conclusion –chapter: As in Discussion, I recommend re-ordering the text in the Conclusions. You could first write about the general findings, then maybe shortly go through the uncertainties (that are now in rows 448-459), and then in the end, draw some generalized conclusions/future perspectives.
RESPONSE: The same restruction has been adapted.
Reviewer 2 Report
-Abstract: needs to include a more incisive conclusion of the data;
-Figure 2 and 3, 7: Improve map organization. Include coordinates, projection and datum.
-include figure 3 context in the text before the figure presentation.
- it seems to be unclear how the GIS algorithm defines and constructs each landscape polygon.
-line 141: figure 3 or 4?
-name of figure 3 is repeated.
-second figure 3, no coordinates, no scale, orientation. Or is it not a map? Just a sketch illustration?
--the letters of figure 4 could be represented with the edition, even if manual of polygons that represent what the authors aim, in area.
-when using the concept of landscape and with application to geodiversity, before the methodological procedures, there is an expressive detailing in figure and text of the geological data, to the detriment of other variables that are part of the landscape.
-figures 5 and 6: suggestion to separate the elements of each group by different colors (geology, relief, ,...)
The methodology is very interesting and opens an important window in the organization of data for the analysis of the diversity of abiotic elements of the landscape (geodiversity), but care must be taken in the process of creating landscape polygons by mixing bases of very different scales or landscape units. of disproportionate sizes to truly identifiable features on a 1:250000 map, for example. A critical assessment of this at work, as a warning to future replicators, is needed.
Two concepts are important to be clear (partly in the methodology and partly in the conclusions): landscape units shelter areas in which abiotic, biotic and human characteristics are repeated and therefore homogenized in a territorial cut, while the geodiversity of a given area shelters the diversity among the existing abiotic elements.
Author Response
Thank you for your comments which we has been useful for us. We have tried to follow your comments as well as we can.
-Abstract: needs to include a more incisive conclusion of the data;
RESPONSE: The abstract has been rewritten.
-Figure 2 and 3, 7: Improve map organization. Include coordinates, projection and datum.
RESPONSE: All maps has got the required information
-include figure 3 context in the text before the figure presentation.
RESPONSE: Has been done.
- it seems to be unclear how the GIS algorithm defines and constructs each landscape polygon.
RESPONSE: Have tried to express this better.
-line 141: figure 3 or 4?
RESPONSE: Figure numbers have been changed and corrected.
-name of figure 3 is repeated.
RESPONSE: Corrected
-second figure 3, no coordinates, no scale, orientation. Or is it not a map? Just a sketch illustration?
RESPONSE: Changed
--the letters of figure 4 could be represented with the edition, even if manual of polygons that represent what the authors aim, in area.
RESPONSE: I am not entirely sure if I understand the comment. But text has been adjusted so we hope it is clearer now.
-when using the concept of landscape and with application to geodiversity, before the methodological procedures, there is an expressive detailing in figure and text of the geological data, to the detriment of other variables that are part of the landscape.
RESPONSE: The original landscape map and classification system include both physical, biological and human land use elements. It is mentioned in the text and indeed in the references. As the aim for this article is linked to geodiversity we have extracted the physical content as the basis of the analysis, which we think is correct. We have not aimed to link human activities or perception to the geodiversity profile, but if necessary it is easy to link the human land-use to the geodiversity profile, at least in Norway were this map exist as a complete dataset. See also comment below. In Sweden we have only extended the physical elements of it as this was our aim.
-figures 5 and 6: suggestion to separate the elements of each group by different colours (geology, relief, ,...)
RESPONSE: Done
The methodology is very interesting and opens an important window in the organization of data for the analysis of the diversity of abiotic elements of the landscape (geodiversity), but care must be taken in the process of creating landscape polygons by mixing bases of very different scales or landscape units. of disproportionate sizes to truly identifiable features on a 1:250000 map, for example. A critical assessment of this at work, as a warning to future replicators, is needed.
RESPONSE: it is clarified in the text that the polygons has been delineated based on an elevation database with the resolution 100m. All the rest of the data is collected and analysed as attributes within these polygons. The extension of the polygon structure into Sweden has shown good match and the algorithms used seem stable. This has been commented in the text.
“Two concepts are important to be clear (partly in the methodology and partly in the conclusions): landscape units shelter areas in which abiotic, biotic and human characteristics are repeated and therefore homogenized in a territorial cut, while the geodiversity of a given area shelters the diversity among the existing abiotic elements.”
RESPONSE: As commented above it is stated that the landscape gradients contain physical, biotic and human elements, but the landscape area polygons is a simple delimitation based on an elevation database, and the content of the polygons here is the physical ones, as the aim is to describe geodiversity profile. We have clarified this and an adjusted text in the discussion as follows:
“The original landscape map [19, 20] classification is based on landscape gradients containing physical, biological and human land uses attributes. The delimitation of the landscape types an landscape areas, however, are based on geomorphic criteria and thus show important structures related to geodiversity on a coarse scale. In this way the geodiversity can be analysed alone, but also in a wider landscape setting included both ecological and cultural aspects of the landscape.”
..and in the conclusion:
“……and to compare areas with each other with respect to their geodiversity profile. This can be done both alone as well as in a wider landscape setting.”
