Channel Activity Remote Sensing Retrieval Model: A Case Study of the Lower Yellow River
Round 1
Reviewer 1 Report
Comments for author File: 
Comments.pdf
Author Response
Dear Reviewer，
We have modified the article according to your suggestions. Please see the response from attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
The study presents an interesting study on the GEE based analysis of the planform dynamics of the Yellow River in China. While the general analysis is sound, there are several minor issues which need to be solved prior to publication.
1) In the introduction, I miss Boothroyd et al. (2020) as a reference (https://0-doi-org.brum.beds.ac.uk/10.1002/wat2.1496). This article reviews applications of Google Earth Engine in fluvial geomorphology and should be cited in your study. This could be done in the paragraph from line 58 to line 88.
2) In the paragraph from line 89 to line 103, I agree with you that "traditional" methods for planform change assessments might have some shortcomings. However, you remain little vague here and I would like to see a more concise statement on the specific drawbacks of other methods. What are these traditional methods (e.g. digitizing from aerial imagery) and why do they perform worse compared to your remote sensing based approach? Please add some more information to this paragraph in the revised version of the manuscript.
3) I wonder, why you use 4 scences in 30 years only while using a cloud computing system. The major advantage of such system is the ability to process vast amount of data and I wonder why you did not use e.g. annual time steps for analysis? Annual time steps would be interesting in order to if rates of channel migration have changed over time, what is difficult to achieve with four observation points in time only. Please explain why you choose this low number of sampling points in time.
4) It would be great to have your algorithm available open access as it has become common in the community, e.g. via github. Anyway, I would like to see some code availability statement in a revised version of the manuscript.
5) I am little bit confused about Otsu thresholding for water mask creation and then in the next step using this combined MNDWI, NDVI, EVI method. Isn't that redundant? In the end, you use the centerlines for analyzing channel migration. This is done by using the MNDWI-NDVI-EVI method and Otsu-threshold derived water masks do not go into any further analysis? Please clarify the role of the Otsu derived water masks in your analysis of channel migration.
6) In lines 459-472, you discuss the advantages of your satellite data based algorithm. I generally agree on the benefits of satellite derived information of river systems. However, I think that some modification is needed here. First, you refer to provide sufficient temporal and spatial resolution while using decadal data only. In US or Central Europe, you could easily get decadal time steps of aerial images going further back in time with <1 m resolution, often even suitable for photogrammetry potentially providing 3D information. Thus, I suggest to highlight the specific advantages of the big satellite missions such as Landsat: global coverage, routine acquisition of data, high temporal resolution (even you don't make benefit from the last point it in your study).
7) In lines 473-485, in the discussion section, you give quantitative numbers of errors including a formula for calculation. However, this refers to general values only and not to a specific accuracy assessment in your study. I suggest to either carry out such accuracy assessment (in this case, it has to included in the methods and results section) or alternatively avoid a precise statement on error which does not arise from your own study. You could still give a general statement of error, however, I suggest to highlight that this refers to literature and not to your study.
8) I wonder, if the geomorphic/ecologic interpretation of the results tells the full story. In particular, I have some doubts about the term "stability". What does that mean in geomorphic terms? Is there no change at all? Is there no change anymore? Does the deficit in sediment supply caused by the reservoirs/dams cause an incision of the river rather than planform stage? Isn't the channel migration part of the natural system and channels can be in a geomorphic equilibrium while having a certain migration rate? Is there any ecologic consequence of channel stabilization? I would like to so see some reflection on these questions (not necessarily limited to these) in a revised section 5.2. Eventually, it should be renamed to "geomorphic interpretation of the results" if you follow my suggestion. In case, these points should also be reflected in a revised conclusion.
9) In section 5.3, you discuss "uncertainty analysis". However, no concise accuracy assessment has been carried out within this study and section 5.3 remains very general and vague giving no additional insights for the reader. Therefore, I suggest to remove it. To include your thoughts on potential future work, you might consider to include a brief outlook in the conclusion section.
Specific comments:
Line 138: I disagree with your definition of a single-channel river as for me single channel is opposed to multi-channel river types such as braided or anastomising rivers, see e.g. the definitions by Rinaldi et al. 2016 (https://0-doi-org.brum.beds.ac.uk/10.1007/s00027-015-0438-z. Simply say: "there are no relevant tributaries in the study areas and the river system is dominated by the main stem only." or something similar.
Fig. 2: "Vector map of study area" probably refers to the bounding box used for spatial filtering of the Landsat collections in GEE? I suggest to modify this or add an explanation to clarify.
Fig. 4: I suggest to use a more intuitive color scheme for the MNDWI images ranging e.g. from yellow (-1) to blue (1). This would result in an image where water occurs in blue and background landscape in yellow providing an improved cartography.
Fig. 9: Where do the thresholds for siltation, balance and scour come from? Please explain in the caption and/or text along with the appropriate reference.
Author Response
Dear Reviewer，
We have modified the article according to your suggestions. Please see the response from attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments for author File: Comments.pdf
Author Response
Dear Reviewer，
We have modified the article according to your suggestions. Please see the response from attachment.
Author Response File: Author Response.pdf
Reviewer 4 Report
The introduction should be supplemented with information on the river's potential for migration. Especially in the context of its confined (http://0-dx-doi-org.brum.beds.ac.uk/10.1007/s00027-015-0438-z).
line 38. Rosgen and Schumm should be quoted here (https://0-doi-org.brum.beds.ac.uk/10.1146/annurev.ea.13.050185.000253, https://0-doi-org.brum.beds.ac.uk/10.1016/0341-8162(94)90001-9)
line 72. In the beginning, it is necessary to mention the classic method of determining the centre line as the axis of the polygon resulting from the transformation of the boundary lines (https://0-doi-org.brum.beds.ac.uk/10.3390/rs13245147)
Figure 1. Part (a) In the legend, label both rivers
line 174. It is difficult to talk about the analysis of the channel morphology with the accuracy of the images of 30 m...
Table 1. To be supplemented with data on the discharge or water level on a specific day.
line 247. Why is this area value optimal?
line 261. Why is this pixel value optimal?
chapter 3.1. Extraction of channel boundary - How were the automatically obtained results verified?
chapter 3.2. Extraction and subsection of channel centerline - Why wasn't this solution used? - convert the channel boundary to a vector polygon and get the centre line from it (e.g. skeleton in GRAS). And then study the changes in the position of vector lines relative to each other.
line 319. If this central form is small but persistent it will affect the channel processes and should not be neglected.
There are too many "figure X. shows" statements in section 4.1. It reads wrong. Maybe it's worth making an annexe out of this information and comments and leaving only the most important results here.
chapter Results. A good illustration of the validity of your method would be to show satellite images /orthophoto maps with a central line.
line 403. These changes are huge! This needs to be commented on in more detail.
chapter Discussion. It is necessary to refer to the geomorphological verification of your results.
lines 473-480. What about the 30-meter accuracy of the image itself?
Author Response
Dear Reviewer，
We have modified the article according to your suggestions. Please see the response from attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 4 Report
I don't have any more comments.
