A Methodology for Georeferencing and Mosaicking Corona Imagery in Semi-Arid Environments
Round 1
Reviewer 1 Report
The paper is overall well written and the idea is also interesting and useful for real world applications in land cover change and related applications.
My only comment is more of a suggestion for future development of the application using also free and open software as an alternative to ArcGis Pro which is a fantastic tool (I use it myself) but it comes with licence costs which could not be sustainable for developing countries who may want to implement the methodology proposed. Have you thought of a similar routine in QGIS?
I have spot one minor error in line 384 "Figure 12" should be instead "Figure 7"
Comments for author File: 
Comments.pdf
Author Response
Thank you very much for your review. Please see the attached document with our response to your comments.
Author Response File: Author Response.pdf
Reviewer 2 Report
This is a neat and useful paper which will contribute to efforts to make a unique and important dataset (Corona imagery) more useful to extend our timeseries of earth imagery. It is nearly ready for publication, but there are a few places where the authors could expand, elaborate, or more completely describe their methods and why they made certain choices. I would consider any efforts to address my comments a fairly minor revision. I will provide my comments in manuscript-order, with some being really minor, and others perhaps requiring a bit more effort (still minor) to address:
- In the introduction, it would further strengthen your arguments for using Corona imagery if you pointed out that pre-Landsat 5 our archive is very thin. While Landsat 1 started collecting data in 1972 our imagery from the 1972-1984 period is very sparse because of data-storage limitations. I would add this to the introduction.
- Expand your discussion of the "inherent distortions from the acquisition process" -- are these different than for other (non-Corona) imagery?
- On page three the last paragraph of the subsection "Georeferencing Corona Imagery" the authors state that older studies that use this imagery have "dated" georeferencing techniques. I have been georeferencing imagery since the late 1990s and the process is not fundamentally different today than it was then (we've improved the details, perhaps). I would suggest the authors either need to elaborate on what a "dated" georeferencing technique looks like, or strike this discussion point from the manuscript.
- While ArcGIS Pro 2.8 is good, I wasn't sure what criteria the authors used to decided that it was the "most efficient software" for this use. Please elaborate on this decision process, or just say "we liked the UI better."
- This might be my most major question: in the "2.2.2. Georeferencing" section the authors outline the process for selecting ground control points. They start with an origin strip (with the best features for referencing) and build their GCPs out from there. That's a fine approach, but... since distortions radiate out from the strip meridian (the center line of each image strip) what does this mean for position-error? I think some careful thought and discussion about what this protocol (building out from a primary strip) means for position-error propagation needs to be added. Distortion is likely most significant in the "overlap area" between strips since this area is at the edge of each strip, so linking strips together internally (rather than to base imagery) in this zone is going to result in more significant position error. I know this is unavoidable -- this isn't a method criticism so much as a recommended place in the manuscript to discussion position error propagation.
- The authors shoot for 20 GCPs per image strip. That's ok. But I would suggest in the paragraph where this is discussed the authors offer their perspective on an ideal number of GCPs in the absence of technical challenges. I've georeferenced imagery covering a much smaller area than a single Corona image strip with triple that number of GCPs before.
- Table 2 is good, I'd add another column showing the proportional number of points per category (just a column showing % of total).
- The first paragraph of the section 2.2.3 "Mosaicking Georeferenced Strips" states an accuracy of 15m. Is this measure just internal to Pro's georeferencing tools? Or does it incorporate ground verification?
- Line 338: "clipping" should be changed to "clipped"
- I was curious about the choice of bilinear interpolation against a source dataset that was scanned in 8-bit image format. Bilinear interpolation would turn the data into floating point, and I'm not sure why that would be a benefit given the integer status of the input data. Yes, nearest neighbor will be "blockier" but it would involve less modification of the input data. I always advise people working with 8-bit data to just keep it at 8-bit depth because the "smooth" nature of an alternative output is just visual (and we can achieve a smooth output using visualization techniques instead). I would ask the authors to discuss this decision from a data-standpoint in addition to a visual one.
- Well done including a Python script. Thank you!
Though this is a bit of a long list of things to adjust/edit/elaborate on, I do think the manuscript is in pretty good shape and that these edits are minor.
Author Response
Thank you very much for your review. Please see the attached document with our response to your comments.
Author Response File: Author Response.pdf
Reviewer 3 Report
The work proposal presented is a magnificent example of the necessary and essential union between "traditional" science and the digital world. I value very positively the work of "rediscovery" of a geodata source as relevant and little known as the CORONA project. Undoubtedly, these data will be the basis of one of the most relevant lines of work for the coming years in several lines, among which I highlight geography.
Therefore, very well-planned research is presented at the beginning, which makes very clear the research background and the objectives to be developed. In the study area section, it would be interesting to be able to clarify in greater detail why this specific area of study has been chosen in China. It is clear that in the choice there is an obvious landscape component, but do these characteristics not occur elsewhere on the Planet? Is there an interest in analysis in China on the part of researchers? These are details that would help to give greater precision to the text.
I congratulate the researchers for their good work, and I am looking forward to knowing more results and being able to operate with the source of work.
Author Response
Thank you very much for your review. Please see the attached document with our response to your comments.
Author Response File: Author Response.pdf
