End Point Rate Tool for QGIS (EPR4Q): Validation Using DSAS and AMBUR

Round 1

Reviewer 1 Report

Well done for creating what will no doubt be a valuable, free and open-source tool for researchers and coastal managers. There is certainly a need for this tool and the kind of validation data you present is very important. However, I feel the manuscript needs considerable improvement before publication. I have outlined my suggestions below, those in bold are the most critical. I look forward to seeing this paper published (and trying  EPR4Q!). 

There are numerous, relatively minor, errors in grammar and the style and flow of the text. These make already complex subject matter more difficult to understand. I am also attaching a pdf version of the paper with grammatical errors and passages of awkwardly worded text highlighted in yellow.

68-71: I recommend removing the aim from here as you have not yet developed the case for it and it is more clearly worded below (lines 149-153).

88: In addition to the expense of a license for ArcGIS, DSAS is not supported for ESRI ArcGIS desktop versions beyond 10.6 (now they are up to 10.8) or ArcGIS Pro (which seems to be what ESRI is planning to replace the desktop version with). So, DSAS is not useable now unless you have an old version of ArcGIS.

92-93 I don’t think sub-headings are necessary for DSAS and AMBUR

117-120: These lines repeat information provided above. Remove.

127: What is meant by “a simple shoreline movement”? Is this a rate? Please provide some additional details.

It is up to you, but I suggest it would flow better to move section 1.4 describing EPR (and other shoreline change measures) up to section 1.1 then describe DSAS and AMBUR and their shortcomings.

154: Study Areas

156-167: These sentences should be re-worded to increase clarity and conciseness.

171: Replace exposition with exposure or if this is not what you mean, replace it with orientation.

174-176: When I think of exposure I think of the size and power of waves an area experiences rather than the direction it faces. If it was necessary to choose beaches with the ocean on alternating sides for the purpose of testing your tool please explain this. It would be good to add information on the ambient wave conditions at each site as well as the tidal range and level of human infrastructure.

Figure 1: I don’t think there is any need for the inset showing the location of the sites globally, the one focused on North America is sufficient. It is confusing to have only yellow and green dots on the nation-wide overview then orange and blue as well on the zoomed maps. Perhaps use extent indicators instead? Remove the heading “Study Site: ….” Add the full form of the State abbreviations to the caption for readers not from the USA.

190: Replace “both sides” with “on the landward and seaward sides of the shorelines”.

193: Ensure you are consistent with the tense chosen for the methods. (i.e., consists versus consisted)

190: First provide a simple overview of what is needed for the tool and what it does then get into the detail of the tools used and created to achieve it. E.g., The tool calculates the distance between the two shorelines by creating and measuring the length of transects between the inner and outer shoreline….

198-201: I can’t quite understand this step. How does extending the line of each shoreline allow you to split it into two parts? Please explain further and if possible make it clearer in Figure 2.

203-204: How does it know the direction and distance to “shift the location of the point”? Is that input by the user?

Figure 2. Great idea to have a graphical explanation of the steps. I think it would be impossible to understand without this. Some comments on each panel: a) I don’t think the large text in the bottom left corner saying “Inner Baseline, Outer Baseline”…etc is needed. It is clear from the annotated graphic. b) maybe this box can be called “Create transects”. Can you put the names of the tools used in boxes to make them more distinct from the labels? c) Not sure if the title of this box is correct as split with line tool comes in box d? Should it be “Extend and Buffer Shorelines”? d) this box confuses me a bit… How does it determine which way to shift the centroid points? It looks to be different direction for each shoreline? Perhaps hatch out or label the sections of buffer to be deleted. Maybe in brackets after each tool you can provide a little more detail, e.g. Split with Line (split xxxx with yyyy). f) As in c) maybe you could provide some more details about what the input for the tools are. g) Is “Results” the right title? maybe, calculate lengths of transects. “Transects calculed” should be Transect length calculated. Label advance and retreat lines separately.

Box e is missing. I do think that another box either between c and d or d and f would be helpful to break down the method and clearly explain it.

Section 3.1.1. Hmm… Can you make clear in this section how the transect metrics created/calculated in the tool fits into the formula? Perhaps a simple worked example would help?

Section 3.1.2. I find this section confusing, I’m not sure if it’s just because these concepts are difficult to describe without seeing the graphically (which Figure 3 helps with). Maybe it would help to provide a simple description of exactly what the output shows first (maybe why would also help) and then go into detail of how it was created including why each step was necessary.

247-248: What do you mean by “new shoreline” does this somehow apply the EPR up until a future date?

Figure 3. a) same comment as Figure 2a. hmmm I get lost from c onwards! f) the label “EPR Graphical Result – Retreat” is pointing to the area of shoreline which according to Figure 2 a has built out. Is that a mistake?

It would be useful to summarise somewhere, maybe a table, what the user defined parameters that are needed for the EPR4Q tool are. Do you have any tips on what values to use?

264: All other references are numbered, this one has author date.

268-273: This paragraph and Table 1 should move to the Study Areas section.

287-300: First provide a simple overview of the way you will compare them and explain why the steps explained in this paragraph were necessary. Why could you not use the same transects/baselines for all models? Explain this further.

308: Can you start the results with a sentence or two providing a broad overview of what you found before going into more detail?

Figure 3: Should be labelled Figure 4. The x-axis labels have been cut in half on my version. Don’t forget to label the x and y axes. Also try to keep the axes consistent for direct comparison between all figures. These points apply to all the matrix correlation plots presented.

Figure 4: Should be Figure 5. They are very close so it probably won’t change anything really but can you make the categories in the legend the same for each model?

333-336 – It seems strange that you say the results revealed an “advance” but the mean value was negative (which indicates retreat right?).

Does your tool also allow you to display the magnitude and direction of total change between the two shorelines? This would be helpful. It may be useful to include this in Table 2 as well (i.e., mean total change in shoreline position).

Figure 7: Describe in the caption what the white inset box is showing (areas of less agreement I presume?)

373-374: add the maximum rates for DSAS and EPR4Q in brackets.

394: That offset is weird! Any ideas why that happened? This makes it all the more important to explain in the methods why it was not possible to use the same transects/baseline for all the models.

Figure 11. Make sure the divisions in the legend are the same for each model and have the same number of decimal places. Can you show a closer zoom of the area in the white box. That way you should be able to make a judgement as to which model is the most correct (i.e. has generated the transects the best). This is important when it comes to people having faith in your new model.

449: Suitably comparable rather than “suitable”.

465-471: This paragraph is difficult to follow. Can you rephrase?

474: Generation of polygons how/from what and subtraction from what? How is this relevant?

502: The attention required to generate specific transects and achieve accurate is an important finding of you work but it is not made clearly enough in the discussion. It would also help to show in more detail how the different model transects looked in the areas of disagreement (as suggested for Figure 11).

 

Comments for author File: Comments.pdf

Author Response

Response to Reviewer 1 Comments

 

Dear Reviewer,

We acknowledge your insightful comments that have contributed to improve the original manuscript. Below we respond to your comments and we inform that all changes were done in the original manuscript using the option track-changes. Furthermore, English language and style were revised.

 

Point 1: There are numerous, relatively minor, errors in grammar and the style and flow of the text. These make already complex subject matter more difficult to understand. I am also attaching a pdf version of the paper with grammatical errors and passages of awkwardly worded text highlighted in yellow.

Response 1: The grammatical errors and passages of awkwardly worded text highlighted in pdf version were highlighted (yellow) and corrected in the manuscript.

 

Point 2: 68-71: I recommend removing the aim from here as you have not yet developed the case for it and it is more clearly worded below (lines 149-153).

Response 2: This sentence was removed.

 

Point 3: 88: In addition to the expense of a license for ArcGIS, DSAS is not supported for ESRI ArcGIS desktop versions beyond 10.6 (now they are up to 10.8) or ArcGIS Pro (which seems to be what ESRI is planning to replace the desktop version with). So, DSAS is not useable now unless you have an old version of ArcGIS.

Response 3: This information was added in lines 106 – 107.

 

Point 4: 92-93 I don’t think sub-headings are necessary for DSAS and AMBUR

Response 4: These sub-headings were removed.

 

Point 5: 117-120: These lines repeat information provided above. Remove.

Response 5: These lines were removed.

 

Point 6: 127: What is meant by “a simple shoreline movement”? Is this a rate? Please provide some additional details.

Response 6: This refers to a shoreline movement rate (mm.yr^-1). This information was included in line 141.

 

Point 7: It is up to you, but I suggest it would flow better to move section 1.4 describing EPR (and other shoreline change measures) up to section 1.1 then describe DSAS and AMBUR and their shortcomings.

Response 7: This suggestion was performed and now, the information about EPR description and other shoreline change measures appears before DSAS description.

 

Point 8: 154: Study Areas

Response 8: This was changed.

 

Point 9: 156-167: These sentences should be re-worded to increase clarity and conciseness.

Response 9: These sentences were rewritten.

 

Point 10: 171: Replace exposition with exposure or if this is not what you mean, replace it with orientation.

Response 10: The word ‘exposition’ was replaced by ‘orientation’.

 

Point 11: 174-176: When I think of exposure I think of the size and power of waves an area experiences rather than the direction it faces. If it was necessary to choose beaches with the ocean on alternating sides for the purpose of testing your tool please explain this. It would be good to add information on the ambient wave conditions at each site as well as the tidal range and level of human infrastructure.

Response 11: The word ‘exposure’ was replaced by ‘orientation’ since this was the right word. The required information about beaches was added in lines 189-199.

 

Point 12: Figure 1: I don’t think there is any need for the inset showing the location of the sites globally, the one focused on North America is sufficient. It is confusing to have only yellow and green dots on the nation-wide overview then orange and blue as well on the zoomed maps. Perhaps use extent indicators instead? Remove the heading “Study Site: ….” Add the full form of the State abbreviations to the caption for readers not from the USA.

Response 12: The map was redesigned according to these suggestions. The world map was removed and the dots on the nation-wide were better highlighted.

 

Point 13: 190: Replace “both sides” with “on the landward and seaward sides of the shorelines”.

Response 13: This was replaced.

 

Point 14: 193: Ensure you are consistent with the tense chosen for the methods. (i.e., consists versus consisted)

Response 14: The present tense was chosen for the methods, and thus, text was reviewed.

 

Point 15: 190: First provide a simple overview of what is needed for the tool and what it does then get into the detail of the tools used and created to achieve it. E.g., The tool calculates the distance between the two shorelines by creating and measuring the length of transects between the inner and outer shoreline….

Response 15: A simple overview was included in lines 206 - 207. The description of the tools was improved according to suggestions.

 

Point 16: 198-201: I can’t quite understand this step. How does extending the line of each shoreline allow you to split it into two parts? Please explain further and if possible make it clearer in Figure 2.

Response 16: The explanation was included in lines 251 - 252. Basically, the Split with Line tool needs a line crossing the polygon to cut it.

 

Point 17: 203-204: How does it know the direction and distance to “shift the location of the point”? Is that input by the user?

Response 17: No, it is an automatic input. It is supposed that a 200 meters distance, the 100% are in touch with the polygon created to extract transects. This was clarified in line 262.

 

Point 18: Figure 2. Great idea to have a graphical explanation of the steps. I think it would be impossible to understand without this. Some comments on each panel: a) I don’t think the large text in the bottom left corner saying “Inner Baseline, Outer Baseline”…etc is needed. It is clear from the annotated graphic. b) maybe this box can be called “Create transects”. Can you put the names of the tools used in boxes to make them more distinct from the labels? c) Not sure if the title of this box is correct as split with line tool comes in box d? Should it be “Extend and Buffer Shorelines”? d) this box confuses me a bit… How does it determine which way to shift the centroid points? It looks to be different direction for each shoreline? Perhaps hatch out or label the sections of buffer to be deleted. Maybe in brackets after each tool you can provide a little more detail, e.g. Split with Line (split xxxx with yyyy). f) As in c) maybe you could provide some more details about what the input for the tools are. g) Is “Results” the right title? maybe, calculate lengths of transects. “Transects calculed” should be Transect length calculated. Label advance and retreat lines separately. Box e is missing. I do think that another box either between c and d or d and f would be helpful to break down the method and clearly explain it.

Response 18: Figures 3 and 4 (number 2 and 3 in the first reviewed version) were redesigned to include these suggestions.

 

Point 19: Section 3.1.1. Hmm… Can you make clear in this section how the transect metrics created/calculated in the tool fits into the formula? Perhaps a simple worked example would help?

Response 19: Additional information was included in the text in lines 277 - 294 and a figure (Figure 2) was created to clarify this question.

 

Point 20: Section 3.1.2. I find this section confusing, I’m not sure if it’s just because these concepts are difficult to describe without seeing the graphically (which Figure 3 helps with). Maybe it would help to provide a simple description of exactly what the output shows first (maybe why would also help) and then go into detail of how it was created including why each step was necessary.

Response 20: Additional information was included in lines 314 – 337 and the result of the graphical visualization of EPR was included in Results section (Figure 5)

 

Point 21: 247-248: What do you mean by “new shoreline” does this somehow apply the EPR up until a future date?

Response 21: The term “new shoreline” was used to refer the predicted shoreline in a future date. A Figure (Figure 5) was created to see the prediction of shoreline in 2100 with EPR4Q and AMBUR.

 

Point 22: Figure 3. a) same comment as Figure 2a. hmmm I get lost from c onwards! f) the label “EPR Graphical Result – Retreat” is pointing to the area of shoreline which according to Figure 2 a has built out. Is that a mistake?

Response 22: Effectively, the label was wrong. Figure was changed according to suggestions and more information about the method was added in lines 364 – 389.

 

Point 23: It would be useful to summarize somewhere, maybe a table, what the user defined parameters that are needed for the EPR4Q tool are. Do you have any tips on what values to use?

Response 23: The EPR4Q requires the shorelines vectors and baselines. This information was reinforced in lines 228 - 230.

 

Point 24: 264: All other references are numbered, this one has author date.

Response 24: It was corrected.

 

Point 25: 268-273: This paragraph and Table 1 should move to the Study Areas section.

Response 25: Table 1 was moved to the Study Areas section.

 

Point 26: 287-300: First provide a simple overview of the way you will compare them and explain why the steps explained in this paragraph were necessary. Why could you not use the same transects/baselines for all models? Explain this further.

Response 26: It is possible to use the same transects, and in fact, preliminary tests were performed with the same transects and the obtained results were equal for all tools. Nevertheless, this analysis was not included in the present work because our aim was to assess the differences of transect creation of each tool. This was clarified in lines 607 - 611.

 

Point 27: 308: Can you start the results with a sentence or two providing a broad overview of what you found before going into more detail?

Response 27: An initial sentence with a broad overview of the results was included in lines 309 - 408. Furthermore, the EPR forecast graphical result (Figure 5) was added in this Section.

 

Point 28: Figure 3: Should be labelled Figure 4. The x-axis labels have been cut in half on my version. Don’t forget to label the x and y axes. Also try to keep the axes consistent for direct comparison between all figures. These points apply to all the matrix correlation plots presented.

Response 28: This was changed.

 

Point 29: Figure 4: Should be Figure 5. They are very close so it probably won’t change anything really but can you make the categories in the legend the same for each model?

Response 29:  This was changed. Figure was redesigned according to the suggestion.

 

Point 30: 333-336 – It seems strange that you say the results revealed an “advance” but the mean value was negative (which indicates retreat right?). Does your tool also allow you to display the magnitude and direction of total change between the two shorelines? This would be helpful. It may be useful to include this in Table 2 as well (i.e., mean total change in shoreline position).

Response 30: This was a mistake that was corrected. The tool allows displaying the magnitude and direction of total change between two shorelines. An example of magnitude and orientation can be seen in Figure 7.

 

Point 31: Figure 7: Describe in the caption what the white inset box is showing (areas of less agreement I presume?)

Response 31: The white box shows an area of irregular shoreline whose results were different for all tools. The meaning of white box was included in the manuscript. Furthermore, a zoom of the white box was included in Figure 9 (Figure 7 in the first reviewed version).

 

Point 32: 373-374: add the maximum rates for DSAS and EPR4Q in brackets.

Response 32: The maximum rates for DSAS and EPR4Q were added.

 

Point 33: 394: That offset is weird! Any ideas why that happened? This makes it all the more important to explain in the methods why it was not possible to use the same transects/baseline for all the models.

Response 33: The EPR4Q can produce transects with wrong angles as DSAS and probably AMBUR (AMBUR was the reference). When this happen, it is important to generate transects again using different baselines configuration until the result achieves acceptable transects. Other option, it is the edition of transects manually but it might take considerable time depending on the number of wrong transects. This information was added in line 617 – 620.

 

Point 34: Figure 11. Make sure the divisions in the legend are the same for each model and have the same number of decimal places. Can you show a closer zoom of the area in the white box. That way you should be able to make a judgement as to which model is the most correct (i.e. has generated the transects the best). This is important when it comes to people having faith in your new model.

Response 34: Figure was redesigned according to suggestions.

 

Point 35: 449: Suitably comparable rather than “suitable”.

Response 35: This was changed.

 

Point 36: 465-471: This paragraph is difficult to follow. Can you rephrase?

Response 36: The paragraph was rephrased.

 

Point 37: 474: Generation of polygons how/from what and subtraction from what? How is this relevant?

Response 37: This information was not relevant and for that reason, it was removed.

Point 38: 502: The attention required to generate specific transects and achieve accurate is an important finding of you work but it is not made clearly enough in the discussion. It would also help to show in more detail how the different model transects looked in the areas of disagreement (as suggested for Figure 11).

Response 38: More information about transects was included in Discussion Section in lines 612 – 625 and a zoom was added in all maps (Figures 7, 9, 11 and 13).

 

Reviewer 2 Report

Thank you so much for the opportunity to review the article entitled End Point Rate Tool for QGIS (EPR4Q): Validation 2 using DSAS and AMBUR by Lucas Terres de Lima, Sandra Fernández-Fernández, Jean Marcel de Almeida Espinoza, 4 Miguel da Guia Albuquerque, Cristina Almeida Bernardes.

The development of open source tools that are freely available is very important for the scientific community. This paper presents a tool to calculate the rate of erosion / accretion of shorelines along transects. The three methods have different approaches in drawing these transects that are aimed to be perpendicular to the shorelines, but they are not if the shorelines are very irregular / sinuous. From the rate calculated with the three methods it seems that the results are highly dependent to the location / orientation of the transect. When the shoreline is almost straight, the results are almost identical since the transects in all three methods have almost same location / orientation. For a highly irregular / sinuous shoreline, the results are vastly different because the transects have very different locations / orientations. So the difference between the three methods boils down to how and where the transects are drawn. Nowhere is any analysis to see which method performs better for highly irregular / sinuous shorelines, since for almost straight shorelines the three methods are practically indistinguishable.

The authors should pay attention to what is considered west and east USA coasts and where the two oceans, Atlantic and Pacific, are respective to the USA. Some attention is needed for statistics interpretation as well as explanation to different steps in the Methodology section. Please see comments on the highlighted text in the attached pdf file.

In conclusion I am recommending publishing with major corrections.

Comments for author File: Comments.pdf

Author Response

Response to Reviewer 2 Comments

 

Dear Reviewer,

We acknowledge your insightful comments that have contributed to improve the original manuscript. Below we respond to your comments and we inform that all changes were done in the original manuscript using the option track-changes. Furthermore, English language and style were revised.

 

Point 1: 164: Please replace with chosen.

Response 1: It was replaced.

 

Point 2: 174: Are you sure about that? California is on the west side of USA, and Pacific Ocean is west.

Response 2: It was a fool mistake that was corrected.

 

Point 3: 175: Hampton is on the east coast of USA and Atlantic Ocean is east of USA.

Response 3: This is absolutely right. It was changed.

 

Point 4: 186-187: Even if this is presented in Appendix A maybe it would be nice to have 1 or 2 sentences about the graphical modeler and its integration with QGIS.

Response 4: A paragraph was added about the QGIS Graphical Modeler in lines 212 - 225. Additionally, Figure about EPR4Q (Figure 2) (that was presented in Appendix A in the first reviewed version) was re-organized to QGIS 3.14 thinking in the migration for this version and it was included in the main text.

 

Point 5: 188: Maybe replace with "For".

Response 5: It was replaced.

 

Point 6: 189: Is the "inner" baseline towards the land and the "outer" baseline towards the ocean?

Response 6: Yes, this is correct. This information was better explained in lines 364 - 365.

 

Point 7: 191-192: Are the points equally spaced on each baseline? Are the same number of points on both baselines? How do you decide how many points to generate along the baseline? What if the shoreline is very sinuous and the outer base line is longer than the inner base line? Are always the baseline straight lines?

Also, when you mention a tool, don't mention only its "short" name, but actually state what the tool does and write its name in parenthesis e.g. generate n equidistant points along a line shapefile (v.to.points), or something similar.

Response 7: The user can add the spacing between transects GUI, and this procedure is performed with the v.to.points tool. The points are equally spaced on each baseline (in the present case study, 1 m) and the number of points is the same for both baselines. The number of points (i.e. distance between points) depends on the geomorphological features of the coast, if the coast shows a uniform shape transects can be more spaced whereas if the coast is highly irregular, transects need to be less spaced. Nevertheless, in this work we decided to use the same spacing in all shorelines studies to make comparisons between them. The case of small inner baseline and long outer baseline was not tested but probably will generate errors. It is important that the baselines have similarities in both sides, mainly in distances. The required information together with the description of the tools was added in lines 586 - 587.

 

Point 8: 193: The transects are perpendicular only if the two baselines are parallel with the direction of the real shoreline. Maybe it is more suitable to say the transects will be pseudo-perpendicular on the shorelines.

Response 8: This is right and for that reason, it was included in line 229 -230 that baselines must be parallel with the direction of the real shoreline.

 

Point 9: 195-200: This section is not really clear to the reader. Why are you doing a buffer, around which line are you doing the buffer, and why it is a single sided buffer? Also what are you splitting in two parts? From figure 2 it seems you split the buffer along the shoreline, but from the text this is not clear.

Response 9: It is necessary the single side buffer to extract the distance from baseline to the transects. For that reason, it is necessary to use a one side buffer, but this tool does not work in the QGIS Graphical Modeler. More information about the creation of transects was added in lines 246 – 270 and Figure 3 (Figure 2 in the first reviewed version) was redesigned to include this.

 

Point 10: 202-209: This paragraph needs re-wording because it is not clear. Figure 2 is very clear, though.

Response 10: This paragraph was corrected.

 

Point 11: 214-216: Do you mean that each transect is split in two shapefiles or there are two multiline shapefiles, one for each shoreline?

Response 11: Each transect with distance from baseline to newest shoreline and distance from baseline to oldest shoreline is split in two shapefiles. After this process, those transects (with the values in table attribute) are merged to perform the EPR equation process with Field Calculator tool. This information was included in lines 273 - 281.

 

Point 12: 224: Maybe replace with "represented" or "defined".

Response 12: It was changed.

 

Point 13: 232: In a straight line equation m = (Yn-Y1)/(Xn-X1). Please make sure equation 3 is correct.

Response 13: There was a mistake in equation 3 but it was corrected.

 

Point 14: 245-247: What are "diamond shapes", and what the rectangle, ovals and diamond tool does, and why do you need to generate diamond shapes? You should explain that the tool generates buffers around points in the shape of rectangles, ovals or diamonds.

Response 14: More information about the method was included in lines 317 - 319. Furthermore, an example of result of the EPR prediction visualization (Figure 5) was added in Results Section.

 

Point 15: 247-248: The diamond shape is actually a rhomb or parallelogram shape and has 4 corners, so what is the "tip" of the diamond shape you refer to since there are 4 of them?

Response 15: It is the height attribute of the diamond shape. The changes were performed in Figure 4 (Figure 3 in the first reviewed version), and in lines 320 - 321 to explain the idea of diamonds shape that represent the final position of shoreline in future.

 

Point 16: 248-250: What distance the baseline was shifted, in which direction was the baseline shifted? Why do you need to extract the area of the diamond shapes? Are not their area already equivalent to the EPR projection since you generated them that way to begin with?

Response 16: The baseline is shifted with extreme value (2Km) to cover the diamonds shapes. After this, it is added more one buffer with Alternative One Side Buffer tool to guarantee the coverage of the extension of EPR prediction. This information was included in lines 325 - 329 and Figure 5 was added to improve the explanation.

 

Point 17: 257: To have shoreline retreat or erosion, usually the shoreline moves inland. From the sequence 3 a to f, it seems that retreat is towards outer baseline. So should the reader infer that the outer baseline is towards land and inner baseline is towards ocean? If that is correct then this needs to be specified the first time when the inner and outer baselines are introduced. Also usually i would have thought that "inner baseline" would refer to a line towards land and "outer baseline" towards water / ocean and not reverse.

Response 17: This section was reformulated and additional information was included in lines 229 - 232. The term inner baseline is referred to land baseline and outer baseline is referred to ocean baseline.

 

Point 18: 259: The captions for Figure 2 and 3 need to be more detailed, in which each panel from a to f is also explained in short.

Response 18: Figure captions 3 and 4 (Figure captions 2 and 3 in the first reviewed version) were completed.

 

Point 19: 264: Is "uncertainty" here refers to root mean square errors (RMSE) or 95% confidence interval or accuracy (+/-1.96*RMSE)?

Response 19: It is the RMSE. It was changed.

 

Point 20: 282: This is not clear.

Response 20: This sentence was reformulated. Basically, we do not want favor one tool applying filters that take long time and advantage in the analysis, but in the same time, we do not want prejudice a tool that provide this improvement. So, default filters of AMBUR and DSAS were applied to mitigate this factor. Default filters used (DSAS – Smoothing Distance: 500 m; AMBUR - Windows Size: 5) seem adequate. Other values were tested during the research but those values created better transects.

 

Point 21: 291: Does that mean that each transect has a buffer?

Response 21: Each transect of reference transects (i.e. AMBUR transects) for the statistical analysis has a buffer.

 

Point 22: 293: Each side of what?

Response 22: Each side of AMBUR transects that were used for reference to validate EPR4Q results. This information was included in line 389.

 

Point 23: 298-299: Cannot generate identical transects to be used in all three different methods? Maybe differences in the 3 methods are only due to the slight different positions of the transects?

Response 23: It is possible to use the same transects, and this was assessed in preliminary tests obtaining equal results. Nevertheless, it is important to evaluate transects creation process with any kind of edition. In this paper, comparison between methods was performed only for transects that were almost in the same position. This information was included in lines 607 - 609.

 

Point 24: 305: Please give the version of R instance you used and reference the R itself (or the R CRAN site) and the two R packages you are using.

Response 24: The versions of R and packages were added in brackets.

 

Point 25: 317: This should be Figure 4.

Response 25: This mistake was corrected.

 

Point 26: 331: This should be Figure 5.

Response 26: This mistake was corrected.

 

Point 27: 347: What this means? California is on the west coast of USA, and if you speak of the shoreline, then land is to the east of the shoreline and the ocean is to the west.

Response 27: This was clarified in line 467.

 

Point 28: 352: Is the difference between 0.956 and 0.955 correlation coefficients statistically significant?

Response 28: No, it is not.

 

Point 29: 356: This should be Figure 6.

Response 29: This mistake was corrected.

 

Point 30: 372-374: How far apart / close the transects of the 3 methods are for the maximum value? Are the transects from AMBUR and DSAS closer to each other than either one to the respective transect in ERP4Q method? And if so maybe the difference is due to different transect positions respective to the shorelines, especially if the shoreline is very sinuous. Maybe this observation is also true for the minimum values obtained by the three different methods. 

Response 30: Transects of all tools are in a close position (please, see examples in Figures 7, 9, 11 and 13). Thus, obtained differences migth be sligthly related to this displacement but the maximum displacement between transects is of 0.3 m. This fact is due to the method described in section 3.2.2, which selects transects inside the buffer of reference transects (i.e. AMBUR transects) for statistical analysis. The selection of transects next to reference transects, reduces errors and removes distant and discrepant transects from statistical analysis. Therefore, the differences between the tools are more related to the creation of transects in terms of orientation of the base line and / or errors of geometry by algorithm. This information was added to lines 617-618.

Point 31: 374-375: At face value this is not true. The mean will indicate bias in the data. The mean can be 0 for normal distributions with no bias with ranges between -100 and + 100 as well as -1 and +1 for example, but that does not mean the two are similar in values, unless you refer that both have a normal distribution.

Response 31: This sentence was rewritten.

 

Point 32: 423-425: It seems that for irregular / sinuous coastlines the results are different in the 3 methods because each method generates the transects differently in different locations. The location difference between transects is insignificant when the shoreline is linear and the additional baselines are parallel with the shoreline for the 3 methods and the results are paralytically the same. If the transect locations are different, then the results will be different for a very irregular shoreline. In other words the difference between the 3 methods boils down to transect location.

 

Response 32: Each method might generate erroneous transects in automatic use (i.e. no edition after transects creation), but the analysis consider that both, DSAS and AMBUR being well validated method with worldwide application. The study tries to validate the EPR4Q comparing with AMBUR and DSAS. For that reason, the AMBUR was select to have reference transects, i.e. considering that AMBUR transects are correct. The analysis shows the tools generate wrong transects in sinuous shorelines, but the important information is that the EPR4Q generate similar results such as DSAS and AMBUR.

 

Point 33: 435-436: Also a negative bias might suggest erosion and a positive bias might suggest accretion.

Response 33: This was changed.

 

Point 34: 489: Instead of using straight or almost straight baselines around the shorelines, why not use baselines that are parallel with the shorelines, or a "median" shoreline within an envelope that covers the 2 most apart shorelines taken into consideration. If each baseline afterwards is transformed in an equal number of points then the transects will be guaranteed to be pseudo-perpendicular to the shoreline and will never intersect each other. Since in this case the 2 baselines could have different lengths, you can decide to get points 1 m apart on the shortest baseline and get the same number of points equally spaced on the longer shoreline. In this case the points on the longer shoreline will be a little bit farther away than 1 m.

Response 34: This is an interesting idea. In this research, different types of baselines as a buffer were tested but transects in the corners of rectilinear buffers baselines were very different in all the models, so this method was not selected to validate the EPR4Q.

Author Response File: Author Response.doc

Reviewer 3 Report

Dear Editor and Authors, the paper is of interest for IJGI journal. It is well written, Figures can be improved but overall are well worked. Introduction and Discussion can be improved with some more detailes and commments.

I believe the paper can be published after some minor changes. I leave below some suggestions and requirements to the authors, which I congrat for the work. 

....

Title. I suggest to revise the title. The presented one is very specific, and does not clearly represent the work.  A suggestion: "Comparison among tools for shoreline change analysis: EPR4Q, DSAS and AMBUR". Or similar., but the term "shoreline" must be present. As comment, in title you report the full name of EPR4Q, but only the acronym of DSAS and AMBUR. It would be maybe too long to report the whole names, but to be consistent, I would use just acronyms for all (but this, would be my choice!). 

Introduction. Besides the importance of shoreline monitoring/measurements, please insert at least a paragraph with a review of the techniques or images used to monitor/mark the shoreline. For instance, you could already introduce the general use of satellite images, as later you analyse/use those. This would help the reader to better understand the general topic. 

Fig.2 and Fig.3 please, try to enlarge the boxes, as they are too small. Infos inside the boxes are too small to be read. Besides, make smaller the arrows. The workflows are well done and fundamental in your paper, and therefore must be clear. 

In Methods, or Results,or Discussion, perhaps it is missing a figure showing how the three tools generate the transects. I have experience only with DSAS, and you can enable a function that controls the way in which transects are created in respect of the baseline. In this regard, I suggest to comment better in Discussion the influence/not influence of land-baseline : are/can be the tools differently affected by a different land-baseline? Can also a denser/shorter offset of the transects improve the shoreline analysis (and which tool seams to be more sensible to that)?

Discussion. I believe two more comments can be added. 1) is the computational time comparable among the tools? 2) DSAS returns several statistics for shoreline change analysis: besides transect generation, EPR4Q and AMBUR output the same number of statistics? 

 

Author Response

Response to Reviewer 3 Comments

 

Dear Reviewer,

We acknowledge your insightful comments that have contributed to improve the original manuscript. Below we respond to your comments and we inform that all changes were done in the original manuscript using the option track-changes. Furthermore, English language and style were revised.

 

Point 1: Title. I suggest to revise the title. The presented one is very specific, and does not clearly represent the work.  A suggestion: "Comparison among tools for shoreline change analysis: EPR4Q, DSAS and AMBUR". Or similar, but the term "shoreline" must be present. As comment, in title you report the full name of EPR4Q, but only the acronym of DSAS and AMBUR. It would be maybe too long to report the whole names, but to be consistent, I would use just acronyms for all (but this, would be my choice!). 

Response 1: We appreciate this suggestion, and we agree with being consistent with names. It is possible to use acronyms for DSAS and AMBUR because they are well-known (for example references number [5, 16] has DSAS as keyword) but in the case of End Point Rate for QGIS (EPR4Q), this is a new tool and the acronym is unknown for readers. For that reason, it is not suitable to use only the acronym. Furthermore, this work wants to provide a validation for the new tool (EPR4Q) and due to this, results are compared with those obtained using the current accepted tools (i.e. DSAS and AMBUR). The aim is not to determine what the best tool is.

 

Point 2: Introduction. Besides the importance of shoreline monitoring/measurements, please insert at least a paragraph with a review of the techniques or images used to monitor/mark the shoreline. For instance, you could already introduce the general use of satellite images, as later you analyse/use those. This would help the reader to better understand the general topic. 

Response 2: This information was included in lines 42 - 46, and thus, new references were added in Reference list.

 

Point 3: Fig.2 and Fig.3 please, try to enlarge the boxes, as they are too small. Infos inside the boxes are too small to be read. Besides, make smaller the arrows. The workflows are well done and fundamental in your paper, and therefore must be clear. 

Response 3: Figures 2 and 3 (in the first reviewed version, now Figures 3 and 4) were redesigned taking into account the suggestions.

 

Point 4: In Methods, or Results,or Discussion, perhaps it is missing a figure showing how the three tools generate the transects.

Response 4: Examples of transects generated by the three tools can be seen in zooms of Figures 7, 9, 11 and 14. These examples are from areas where tools drive to different results due to the orientation of transects and/or position (little displacement of 0.3 m between compared transects). It is necessary to take into account that in this research transects were not edited to assess the effectively of automatic use of the tools.

 

Point 5: I have experience only with DSAS, and you can enable a function that controls the way in which transects are created in respect of the baseline. In this regard, I suggest to comment better in Discussion the influence/not influence of land-baseline: are/can be the tools differently affected by a different land-baseline? Can also a denser/shorter offset of the transects improve the shoreline analysis (and which tool seems to be more sensible to that)?

Response 5:  The EPR4Q only works with two baselines defined and not allow different functions with land-baseline. The EPR4Q was created in Graphical Modeler of QGIS and this has its limitations. Furthermore, the objective of the study is not to compare the tools if not to validate the EPR4Q with tools already used and approved by scientific community.

 

Point 6: Discussion. I believe two more comments can be added. 1) is the computational time comparable among the tools? 2) DSAS returns several statistics for shoreline change analysis: besides transect generation, EPR4Q and AMBUR output the same number of statistics? 

Response 6: Information about the velocity of the analysis was included in lines 641-647. Current EPR4Q does not return all statistics that are present in DSAS but it is possible to include more in future versions.

 

Author Response File: Author Response.doc

Round 2

Reviewer 2 Report

Thank you for giving me the opportunity to read the revised manuscript End Point Rate Tool for QGIS (EPR4Q): Validation using DSAS and AMBUR. The manuscript is substantially improved, and I am recommending publishing after minor corrections. Besides few style/grammar instances, some of which I have highlighted in the manuscript, please check equation 5.

Also, I would suggest improving the Conclusion by stating clearly that your method gives identical results when the transects in all 3 methods are identical. You want this because you offer a new tool to do an analysis that it is done by other existing tools that had time to amass a bigger body of research and where tested and validated by a diverse research group. If you disregard the programing language the tools are created in and the fact that one needs a licensed software to run (ArcGIS), the main difference between the tools is the way the transects are generated. Beyond that, the tools analyze the transects using the same methodology. The main advantages of the new tool are: a. the complete open source nature of the tool; b. easier to implement than the other complete open source tool AMBUR; c. potential quicker run-time; and, d. no need to switch between software as with the case of QGIS and AMBUR (to keep all in an open software environment for example). 

Comments for author File: Comments.pdf

Author Response

Dear Reviewer,

We really appreciate your recent thoughtful comments that have contributed to improve the reviewed manuscript. Below we respond to your comments and we inform that all changes were done in the reviewed manuscript using the option track-changes.

 

Point 1: 53: Please change to 'shoreline changes'.

Response 1: It was changed.

 

Point 2: 119: Please replace with 'such as'.

Response 2: It was replaced.

 

Point 3: 126: Please delete and replace with your reference directly [50], otherwise the sentence seems incomplete.

Response 3: It was done.

 

Point 4: 193: Please replace with 'had'.

Response 4: It was replaced.

 

Point 5: 212: ‘2016’

Response 5: It was corrected.

 

Point 6: 225: Please change to 'the majority being'

Response 6: It was changed.

 

Point 7: Table 1- ‘west’

Response 7: It was changed.

 

Point 8: Table 1- ‘south’

Response 8: It was changed.

 

Point 9: 271:’consists of’

Response 9: It was changed.

 

Point 10: 355: Please check the equation (5). Should not it be: BEPR =Yn-mEPR * (X-Xn)

Response 10: It was corrected.

 

Point 11: 421: Does the shoreline orientation influence the analysis, or only the complexity of the shoreline itself?

Response 11: The shoreline orientation doesn’t influence the analysis. Nevertheless, it was necessary to test if EPR4Q could run properly using different orientations. The objective was to see if the transects creation and results can successfully perform on different places in the world. This information was included in lines 422-424.

 

Point 12: 437: it is

Response 12: It was included.

 

Point 13: 466: Please replace with 'performed successfully'.

Response 13: It was replaced.

 

Point 14: 507: a

Response 14: It was corrected.

 

Point 15: 675: It is very odd to start a sentence with a number in parenthesis, even if that is a reference. Maybe you can start the sentence with: Researchers [69] ... or maybe: Smoothing the distance ..... was proposed [69].

Response 15: It was changed according to the first suggestion.

 

Point 16: 677: should

Response 16: It was replaced.

 

Point 17: 687-692: This sentence needs a re-write, It is not very clear to the reader.

Response 17: It was re-written.

 

Point 18: 698: works

Response 18: It was changed.

 

Point 19: 738: to

Response 19: It was changed.

 

Point 20: 740: noticed

Response 20: It was changed.

 

Point 21: 744: What are low-code platforms? Do you mean tablets with small RAM and lower speed CPU, but extremely portable? I am not sure that the 'low-code' is the right terminology.

Response 21: The information about low-code development platforms was included in the first review in the lines 252-253. This low-code development platform consists in a development environment used to create application software without complex code programing. A new paragraph was included with the example of low-code development platform (QGIS Graphical Modeler).

 

Point 22: 745: fully

Response 22: It was changed.

 

Point 23: 746: What does this mean? I am not sure i understand it. You mean that on a desktop computer with more RAM and a higher speed CPU the difference in running time of the 3 methods might not be the same as for  Surface 3 notebook?

Response 23: The performed tests in the specified computer with the mentioned configuration revealed that EPR4Q is two times faster than DSAS and three times faster than AMBUR. There are no data about running time with other computers. Therefore, it is not suitable to say that the difference in running time is always two and three times. For that reason, computer specification was included. It was a mistake in the name of computer model but it was corrected in line 741.

 

Point 24: 760: Maybe replace with 'since it is'.

Response 24: It was replaced.

 

Point 25: Also, I would suggest improving the Conclusion by stating clearly that your method gives identical results when the transects in all 3 methods are identical. You want this because you offer a new tool to do an analysis that it is done by other existing tools that had time to amass a bigger body of research and where tested and validated by a diverse research group. If you disregard the programing language the tools are created in and the fact that one needs a licensed software to run (ArcGIS), the main difference between the tools is the way the transects are generated. Beyond that, the tools analyze the transects using the same methodology. The main advantages of the new tool are: a. the complete open source nature of the tool; b. easier to implement than the other complete open source tool AMBUR; c. potential quicker run-time; and, d. no need to switch between software as with the case of QGIS and AMBUR (to keep all in an open software environment for example). 

Response 25: These suggestions were included in lines 759-763 (identical results when same transects are used) and 773-777 (advantages of EPR4Q).