Systematic Quantification and Assessment of Digital Image Correlation Performance for Landslide Monitoring

Round 1

Reviewer 1 Report

The study is to evaluate the performance of different types of images in detecting land displacement using DIC-FFT in order to determin the appropriate image resolution and windows size for landslide monitering. The article is generaly well-written and the experiment is well-designed. The results are presented generally clearly and the findings are useful for relevant studies on land displacement. Howevever, as the authors point out that the coarser resolution image like S-2 are not sufficient for the particular landslides studied here, there is a lack of discussion how these finding can benefit a larger audies who may study landslides of larger big size. Would those fomulas be also useful for them? Or what cares should be taken if they can be applied? In addition, it seems common sense to us that the higher the resolution of an image, the more accurate displacement information can be derived from it as these images (0.25 m orthophotos) contain information that will have more details compared to these lower resolution images. Then the novelty to validate this common sense needs to be reconsidered. But it is also interesting if the research includes the time efficiency when comparing the orthophoto, S-2, Rapideyes, and PlanetScope. Below are some general concerns.

1)      There is an issue with the introduction/description of the dataset used in the study – a bit of chaos in Figure 5 and the content related to Figure 5.

2)      The aims of the study should be stated more clearly, especially for the aim (i)- “the parameter combinations”- it needs to provide a clearer description.

3)      The core algorithm used in this paper is DIC-FFT, but a description of the DIC-FFT method is not sufficient, e.g., some pre-processing to the input images, such as what the method of co-registering for the images and what the Wallis Filtering and how the Wallis Filters affect the image quality, or does it have impacts to the subsequent image correlation calculations?

4)      Several control groups were designed to verify the performance of DIC in different scenarios. However, it is difficult to understand the setup of the different control groups in a straightforward way, and perhaps a bit more information is needed in the materials section to make it better understood by the audience.

5)      The text and labels on the figures are two small to be seen clearly. They need to be improved.

Detailed comments are as follows.

Line 29: “unknown slopes” pront to landslide or unkonw vulnerable slopes?

Line 46-47: The advantages of FFT are its speed…., “speed for what?”; Here should be a clearer statement for the advantages of FFT.

Line 81: “…suitable parameter combinations…”, what is the parameter combinations exactly mean here? Is it a combination of window size, spatial resolution, and temporal baseline? Maybe here it could be more straightforward by adding more information.

Line 84: ‘’DIC-user” – it should be DIC-FFT user because you were using the DIC-FFT toolkit for conducting the experiment. Would the results (window size, image resolution) also be applicable to other DIC methods? If yes, please note that in the text. Could be more specific about the parameters? Is it window size only? and what else?

Line 86 “landslides Driest (DST) and Moosfluh (MSF)”. Is there any specific reason to select these two locations at the Great Aletsch Glacier? Any differences between these locations? Here needs to be briefly clarified by providing more information.

Figure 1: Some text on the map are not clear. Try different colors to show them distinguishably. The north arrow and scale in the main figure are unclear. Orietate the sketch in the upper right corner the same direction as the main figure and put it in a rectangle and label it as a sub figure of Figure 1. Need a DEM as background for the sketch to show the elevation. Also label the Great Aletsch glacier in the sketch.

Line 100: what are “They”? “Thus” should be ‘”because of” or “due to”?

Line 177 “image errors” should not be categorized as meteorological constraints and what are the image errors you are expressing here? It needs to be clarified.

Figure 5: This figure needs to be improved as it is hard to get the datasets you have used. The exact start and end dates are not very clear for the two intervals from this figure.

Line 198: “we sampled the orthophotos…”, what resampling method was applied to your orthophotos to get different cell size raster datasets? Maybe clarifying the exact method you have used here could reduce ambiguity. Whether using different resampling methods would affect the preservation of information in the raster?

Line 208 “…..we first co-registered the images on sub-pixel….”. what is the method used in DIC-FFT to co-register the images? Maybe you need to provide more information about it as the accuracy of image co-register is highly linked to the final offset calculation.

Line 211 “…window sizes ranging from 10 x 10 to 8 192 x 8 192 pixels….”. Is there any specific reason to determine this window size range to conduct the DIC iterations? Why chose 10x10 to be the bottom limits and why chose 8192x8192 to be the top limits?

Line 214 “Pre-processing by applying Wallis filtering…..”. Does this filtering affect the image quality? This Wallis filtering can enhance the dynamic contrast for the images, so when you apply this filtering, what kind of features/information/textures will be highlighted/enhanced? It would be better to clarify whether this filtering will reserve some information or ignore some information.

Line 216 “Employing an oversampling factor of four……”. Why use the oversampling factor of four to improve image quality? Any advantages of using an oversampling factor of 4 instead of 1, 2, and 3?

Line 218 “…returned clearer results than the available TCI products….”. How did you evaluate them and determine which is better?

Line 226-228: Maybe this part needs to be moved to the Study area.

Line 229-230 “…..we selected five to six distinctive geomorphological features (e.g. boulder blocks, paths)……”. It is not clear to see the geomorphological features you mentioned in Figure 7. Maybe adding more close-up photos/images for these specific ground features would be more convincing.

230-231 “…..manually measured their displacements to define our ground truth….” How did you measure their displacements? Do you measure the displacement of the centers of two objects (e.g., boulder blocks) on the ground, or do you measure the displacement at a specific point on their surfaces? And how did you measure the displacement of a path? It is better to provide more information on this part.

Line 257: “from s study” > “from this study”

Figure 8: Why are the ranges of the two legend bars inconsistent? Why are there only two legend bars? Here needs an explanation for the legend ranges and links to the subfigures. The text including the lables and outlines on the figures are too small. In the apprarenthesis missing “are excluded”. The orthopho 5m in C and rapideye 5m are supposed to have the similar resolution why the displacement are represented in pixels of different sizes. Is it because fo the window size? Could you please explain in the text.

Figure 10: why the three markers of differences displacement from RapidEye are yellow inside? Should’t they be hollow triangles with green outlines?

Line: 359: something is missing before “high resolutions”. Is it “the displacements”?

Line 373: change “This loss of heterogeneity at a lower resolution reduces noise which could affect” to “This loss of heterogeneity whiche reduces noise at a lower resolution could affect …”  and further explain how it affect.

It is well-written in general, however some grammatic errors are detected and have been pointed out in the detailed comments. 

Author Response

Dear Reviewer,

We thank you for your valuable comments on our manuscript and appreciate the time you have invested. Your feedback has helped us to see and clarify ambiguous areas to further improve our work.

Please find below the following colour coding for the review and your comments in black; our responses to the review are in blue and the changes made to the manuscript are in green.

Author Response File: Author Response.pdf

Reviewer 2 Report

I am very interested in the article you proposed. After reading it, I feel that it is very practical and worth recommending. Here are some suggestions and questions for your reference.

1. Table1 & lines 165-166:  Why not resample to 0.1m but 0.25m?

2. Lines 198-200:  Which resampling method was used in this study? Will it effect the results of different scale resampling?  

3. Lines 208-210:  Even if the images are orthorectified, there is still the possibility of spatial misalignment between images. Please explain what method this study uses to perform spatial alignment to eliminate deviations caused by image processing?

4.Is there any special purpose for showing the same material in Figure. 3 and Figure. A4?

5. Figure. 7, Although the results of manual measurement are marked on the images with different dates, the difference between the two images cannot be clearly found.

6. Figure. 10(b) shows the result with the orthophoto resolution of 0.25 m. Does other resolution show the same/similar relationship?

7. Lines 402-476: Suggested slight adjustment for easier reading

No Comment.

Author Response

Dear Reviewer,

We thank you for your valuable comments on our manuscript and appreciate the time you have invested. Your feedback has helped us to see and clarify ambiguous areas to further improve our work.

Please find below the following colour coding for the review and your comments in black; our responses to the review are in blue and the changes made to the manuscript are in green italics.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

No more comments