Landslide Identification and Gradation Method Based on Statistical Analysis and Spatial Cluster Analysis

Round 1

Reviewer 1 Report

The article of Dai et al., dealing with the Landslide Identification method based on DInSAR data is interesting, clearly organized and sufficiently well written. I recommend the publishing on Remote Sensing, but I have the following minor comments and corrections that the authors need to make:

·       There is the final part of the introduction (from line 110 to line 121) and the opening part of 'Methods' (from line 132 to line 142) repeats itself. I suggest the authors to delete the final part in the intro and leave only the one in ‘Methods’;

·        Lines 218 and 219: the authors should provide more info on where and how the values of 0.99 and 2.58;

·        Figure 7: Add an inset of China with the location of the study area;

·        The figure 9 is indeed illegible as it is. I suggest dividing it into 3 figures and improving its quality and readability;

·       The figure 11 is also illegible; improve the quality.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper proposed a procedure to map landslides based on multi-temporal InSAR data, highlighting the combination of a few existing methods, a spatial hotspot analysis, and subsequent feature delineation and classification. I think the reported research is solid and it created interesting and significant contributions. The main concern of mine has to do with its writing. Specifically, the authors should consider working on the following issues:

1. English must be improved: much more efficient writing is needed and many minor English errors (or awkward use of the language) must be corrected to help readers understand the paper properly.

2. Some repetitions must be removed, which may require the authors' consideration of re-arranging the materials in the paper. For instance, the methods involved in deformation identification are repeated multiple times unnecessarily. Related to this problem, the way of presenting involved technical details should be improved: they should be given once, but thoroughly and clearly. 

 3. The authors should distinguish clearly in the paper between the adopted, existing methods and the authors' own, new contributions (e.g. combination of two methods, clustering analysis, LSM, LGM, etc.). Their own contributions must be explained in full detail, or be presented differently from the existing methods. 

4. Some key information is missing or unclear. The D-InSAR concept needs to be defined and cited the first time it appears in the paper. Figure 11 must clearly explain what the numbers 1 to 5 stand for. The way the authors explain the figure is incorrect. The authors must be more careful with things like this, as they are more about academic writing than about English usage.

I would recommend the authors find an experienced, English-speaking academic writer in the field to help them revise and improve the paper. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

1.       I recently saw tons of papers that applied machine learning models to generate landslide risk maps. Could you emphasize why your study is unique, innovative, and helpful to the community?

2.       Author should be describing the purpose of using stacking and the SBAS method for annual velocity. As mentioned, the stacking is vulnerable to disturbance by error. Is there any reason to use this purpose? Meanwhile, the time-series method has feature to reduce those error.

3.       How does the author determine the boundaries for the deformed region and the non-deformed region based on the histogram distribution? is there a reference to the determination of the threshold by including an explanation in it?

4.       Figure 12. For the recommendation, it more interesting to use different colour schemes for each data shown in this figure.

5.       Figure 14. Please add a legend in this figure specificity for the green-red scheme.

6.       Figure 15. Has mentioned 50 points for LOS characteristic, but there is no figure that show those points. Then, I suggest providing the figure with the selected LOS characteristic points.

7.       Despite the potential of this method for application in landslide cases. The analysis of accuracy should be provided by comparing to the other data such as field data or other references.

 

8.       Is there any reason to focus on the slope for the factor that analyzing of the landslide? Meanwhile, several environmental factors related to the topography may influence the risk of the landslide.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf