Route Plans for UAV Aerial Surveys according to Different DEMs in Complex Mountainous Surroundings: A Case Study in the Zheduoshan Mountains, China

Round 1

Reviewer 1 Report (Previous Reviewer 2)

I would like to first acknowledge the authors efforts for this resubmission. I believe the mansucript has undergone a significant improvement.
The message if the paper is clearer and less biased in its current form, portraying how it can be a promising source of elevation data, but the error range does not justify its use for UAV flight planning.

Please see below general and specific comments  

General comments:

In the introduction, given the resturcuting undergone for this resubmission, I believe that some points have gotten lost between versions. The current aim of the study is to assess the viability of using different free DEMs and InSAR derived DEMs for flight planning. Given the known issues with obtaining accurate elevation from Sentinel-1 data, which is sustained by this research results, the introduction is missing the reason why this data source for DEM creation was used. I believe previous versions of this manuscript included reasons such as the opportunity to have up-to-date elevation data due to the continuous monitoring and activity of Sentinel-1. These and other reasons stated before should be included somewhere in the introduction as a justification of the work done. 

The general English style and phrasing has improved greatly from the previous version of the manuscript, thank you to the authors for imrpoving this. There are still several points that can be improved, and I recommend doing another check for this review round.

Please check throughout the document that the numbered references correspond to the in-text citations. Also, check the citation style, since now it is a mix and some references seem to have errors. 

All figures: please avoid using rainbow color ramps to show results, since these are misleading and not colorblind friendly. 

Specific comments:

L330-337 are confusing and can benefit from a better structuring and clearer terms, maybe even a schematic representation of the absolute error and the minimum route altitude.

L467-468, correct citation to: "This is consistent with the findings of Braun [47]". 

Figure 6: Please harmonize the legends between both subfigures, so that the resutls are comparable. That is, both color ramps should have the same minimum and maximum value, so that differences can be clearly observed. This is specially important if you wish to compare between different figures, e.g. L376 and 377 read: "In addition, comparing Figure 6 with Figure 5a, it can be found that there were some large and significant errors". 

Figure 6. There is a circle polygon in figure 6b not explained in the legend or figure caption. Please note that all figures should be self-explanable without the need to read the text in the manuscript. 

Figure 7. Please see comment above regarding harmonization of color ramps between subfigures to make the error assessment comparable. In addition for this figure, the role the normalized error index color ramp is not clear, since it does not seem to be depicted in any of the subfigures. Please use either the error or the normalized index error but not both. The sentence "The color bar of the errors and indexes correspond to each other." in the caption is not clear to me, please revise.

Author Response

Thank you for your review. We carefully read the useful comments and revised the manuscript. The revised parts are marked in blue.

Author Response File: Author Response.docx

Reviewer 2 Report (New Reviewer)

Dear Authors,

you will find my suggestions in the attached pdf. I find your work interesting, but the paper needs to be improved before publication. Increase the quality of the images and try to make the text more fluent.

Your sincerely,

 

the reviewer

Comments for author File: Comments.pdf

Author Response

Thank you for your review. The authors carefully read the comments and made modifications. The modifications are in blue. The attachment is the reply to the comments.

Author Response File: Author Response.pdf

Reviewer 3 Report (New Reviewer)

Dear Author, I have completed the review of your manuscript

The article is very intriguing and it makes an in-depth analysis about the satellite-derived DEM for further UAV investigation. Results are convincing and overall the manuscript is well balanced with a well written introduction.

Some sentences and parts of the written are still difficult to read so I am warmly suggesting to review the English style. Minor remark are reported on the pdf -file.

 

All the best

Comments for author File: Comments.pdf

Author Response

Thank you for your review. The authors carefully read the comments and corrected the draft. All corrected contents  are marked blue. The attachment is the reply to the comments.

Author Response File: Author Response.pdf

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Dear authors

 

The work presented here is interesting and relevant for photogrammetry-based altimetry (DEMs) of UAVs. The manuscript deserves a better presentation for the ease of the future reader, which will result in a greater impact of it.

Please follow the recommendation in the attached pdf.

Comments for author File: Comments.pdf

Author Response

Thank you and thanks for your useful comments. All the comment have been replied in the pdf attachment.

And all the revisions are marked red.

Author Response File: Author Response.pdf

Reviewer 2 Report

This article aims at generating DEMs based on Sentinel-1 and InSAR techniques to use as a basis for flight planning of UAV data. 

The methodology for DEM generation is very well explained with enough processing details that allow the reproducibility of this part of the study. Congratulations to the authors on that. 

However, I have several concerns regarding the aim of the paper to use such derived DEM for flight planning. I believe the authors do not do enough analysis of their results to backup their conclusions. Please see details below. I think that the authors should perform some extra accuracy assessments to be able to derive such conclusions, since the quality of the DEMs they obtained is not completely clear from the results presented. This, along with extensive English correction should be improved before resubmission. 

Major comments:

Extensive English language and style correction needed, some sentences seem to be machine translated, and in other cases thereare completely repeated phrases.

The perpendicular baselines chosen by the authors are very low. The authors do discuss the influence of such low perpendicular baselines on the results, which lead to high errors in particular in mountain areas, and they even refer to Braun 2021 who corroborates these findings. Braun 2021 mentions clearly that an ideal perpendicular baseline for DEM generation should lie betwen 150 and 300 m. Of course, this is not possible to find for every study area, and experiments to generate DEMs are welcome, I have serious doubts on the resulting DEMs when used for UAV flight planning as the authors intend to do. 

First of all, I am missing an accuracy assessment performed at the pixel level to identify locations where the S1-A DEMs differ greatly. This could be achieved by substracting the UAV DEM (which is the reference) to the S1-A DEMs and checking the errors locally. Aggregated measures such as mean average error and RMSE can be also computed from this simple difference. This is critical, specially for flight planning. For example, mountain areas with steep slopes are usually over or underestimated in InSAR DEMs with low perp. baselines. Given that the maximum altitude to fly a drone in China is 120m, and that the elevation differences fluctuate between 10 to 80 meters for the best result (ascending) on an aggregated scale (table 3), how much variation is accepted in such important planning to avoid collision risks and to still comply with fliying regulations? Even if the authors argue that the generated S1-A DEM is better than ALOS-PALSAR (lines 259 to 268), this does not mean that the S1-A DEM is a high-precision one, at least not with the presented evidence. 

Further, the authors make the case that this S1-A DEM is useful for flight palnning, but they do not show the results of the flight plan using this InSAR DEM. It would be interesting to compare the flight plan performed with high resolution UAV data to the InSAR DEM one and check the differences statistically. Are flight plans significantly different? Is there any risk of collision?

 

Minor comments:

- Why was only Sentinel 1A used? Just because there was no matching scene of Sentinel 1B for the ROI? If that is the case the text should be generalized to Sentinel-1 since S1-B could also be used.

- Reference list needs harmonization in a single style

- Figures: please avoid using misleading colorramps for continuous data (i.e. rainbow colors)

Author Response

Thank the reviewers for their contributions and excellent opinions. All the useful and significant comments are replied in the attachment. 

And all the corrections are red marked in the revised draft.

Author Response File: Author Response.docx

Reviewer 3 Report

Comparing DEMs from different remote sensing sources is needed, and this paper did a reasonably good experiment. However, the result of such a comparison differs according to the characteristics of the target topography.  In this sense, I do not understand why a relatively small area in China was selected for this experiment. The authors wrote nothing about it. Discussion and Conclusion seem to be reported as general facts, but they can be valid only for a specific Chinese area. In other words, the Discussion and Conclusion sections should be written more modestly.

The English used is understandable in most cases but needs improvements. For example, in the final paragraph of the Discussion section, the authors wrote, "The DEM data is relatively new compared to the free and open source". Here, "free and open source" are adjectives, so a noun is missing. Also, some DEMs have been free and open source for a long time, such as DEMs for the USA from the USGS. The first paragraph of the Conclusion section begins with "In this presentation",  but please write "In this paper". The following sentence starts with "And", but this should be avoided in formal scientific writing. The above examples are only a few of the linguistic problems found in the current draft.  

Author Response

Thank the reviewer for the efforts and time for our manuscript. All the significant and meaningful comments are answered in detail in the attachment.

In the revised draft, the corrections are marked red.

Author Response File: Author Response.docx

Reviewer 4 Report

In paragarph 4.1 it is not clear how you expect 0.01m accuracy, while the UAV data you collected is 0.1m. In general this paragraph is confusing and needs to be addressed with proper rephrasing. 

par. 4.1 - The claim 'The DSM and orthophoto products with a surface resolution of 0.01 m can be ob- 193 tained by processing the UAV aerial photos with high accuracy POS by Pix4Dmapper', is not valid when using 0.1m GSD UAV data. 

par 4.1 - Upscaling of data is not recomented. PLease justify properly why you need to do so. 

Fig. 4 - caption and figure should be on the same page

L 252 - Does 'input' means reference?

L 252- in which DSM did you use the raster to point tool?

Par 4.3 - there are much better techniques to compare DSMs with different GSD. Sampling only one pixel from a 0.1m DSM and comparing it with a 30x30m pixel  DSM, is not methodogically proper. Perhaps you could average 300x300 pixels in the 30x30m area and then make the comparison

Par 4.3 - Pix4D provides a very dense point cloud, from which the UAV DSM is created, by means of interpolation. This interpolation introduces yet another middle step where quality of original data (point cloud) decreases. For example comparing directly the original PC (a few million points) with the raster DSMs (easy done in any GIS such as ArcMAP or QGIS) would provide a much larger sample, while mean, std and RMS would be much more relevant.

Fig. 7 - Since you are directly comparing DSMs, you should scale the graph's Y axis (and all corresponding metrics) to difference from the reference DSM (UAV DSM i guess). Therefore the Y axis would be 0, plus or minus the difference from the reference, rather than absolute height.

Since most of the paper is about DSM somparison, I would suggest a change in title. The main issue of the paper is InSAR DSM comparision with UAV DSM and existing data bases, although this last part is totally uneccesary and with limited to no scientific interest. In order to justify the title, accuracy metrics for the UAV planning route, should be introduced and comparison of the DSMs should be made upon those accuracy criteria, and then categorise them as accepted or not accepted for UAV route planning.

Author Response

Thanks to the time and energy spent by the reviewer on our manuscript, all comments are important. The authors give a detailed reply to the comments in the appendix.

All the corrections are in red in the revised draft.

 

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

First of all, I appreciate your hard work in the acquisition, preservation of data and writing of this manuscript. But in its current form, it is so complex that it is not possible to identify its main contribution and main objective. I suggest that you subdivide it in such a way that it is easier to state the purpose of each resulting manuscript. Indeed, there are so many aspects that are addressed that they are difficult to follow and their true contribution is diluted. I do not consider its publication viable, however, this does not detract from the evident intensive work.

Reviewer 2 Report

Thank you to the authors for adding a per-pixel error analysis and the flight plans with the different DEMs.
I believe this has enhanced the manuscript greatly. However, the way the discussion and conclusions have been written do not represent the actual results from these analysis.
I still believe the authors try to push the use of Sentinel-1 DEM beyond its capabilities for UAV flight planning. I think the authors could present their results in a more unbiased manner, clearly showing that using Sentinel-1 DEM data poses too much risk of coalition as their results show. 

In addition, the added error maps and histograms are misleading, since the limits of the color ramps are different, this makes them incomparable.

Further, the readability of the manuscript is still very poor, even if an Englishcheck was supposedly performed. This makes the manuscript difficult to understand. 

Overall, I think the study is an interesting attempt to compare InSAR DEMs to other open source DEMs, however, the final aim of UAV flight planning might be too stretched. 

Reviewer 4 Report

My feeling is that the title doesn not reflect the content of the manuscript, as it is very optimistic to use satellite derived DSM for UAV flights, due to high inaccuracies which you also describe. Such inaccuracies, even in confined areas pose a thread to flight safety, hence we cannot rely on such DSMs. The manuscript in its core is a comparision among satellite and UAV derived DSMs, which is of some merit, althought not novel.