Assessing Landslide Drivers in Social–Ecological–Technological Systems: The Case of Metropolitan Region of São Paulo, Brazil

Round 1

Reviewer 1 Report

Assessing landslides drivers is key to landslide disaster mitigation. This study used the socio-ecological-technological system (SETS) approach to investigate and quantify key variables in revelant to landslides using a logistic regression model. Overall, the manuscript is generally well written. Fundamentally, though, the manuscript suffers from some issues.

1. The current research gap for proposing the socio-ecological-technological system (SETS) should elaborate in Introduction. As desribed by authors, traditional susceptility analysis is often taken and related advances should added.

2. The landslide locations and the non-landslides landslides are critical for the training of logistic regression model. It lacks a map with terrain data for the key data.

3. Figure 5 can be changed to a Table for the independent variables.

4. They used the logistic regression only; however, some advanced machine learning algorithms (e.g., ensemble learning) show better performance than the single algorthm. I recommend authors compare some ensemble learning algorthms.

5. The probability of landslides can be predicted by the model. I suggest authors provide the map in the study area to further demonstrate the reliablity of the used model.

6. I cannot follow the idea for "2.038 events recorded" in line 16 and line 279.

7. Line 127 and 208, a citation starts at a sentence should be Author et al. [36].

8. Some related references should add in this paper. Zhang R, Chen Y, Zhang X, et al. Mapping homogeneous regions for flash floods using machine learning: A case study in Jiangxi province, China[J]. International Journal of Applied Earth Observation and Geoinformation, 2022, 108: 102717. Yao, Jing, et al. "Applications of Stacking/Blending ensemble learning approaches for evaluating flash flood susceptibility." International Journal of Applied Earth Observation and Geoinformation 112 (2022): 102932.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

In this research, A multi-step model approach was used to select the best set of variables related to landslides’ occurrence and assess their importance. The reviewer thinks the topic discussed in this paper is very important, which is of great significance for research of landslides. This reviewer sees that a minor revision will be needed before being accepted for possible publication. Here are the main comments for the revision.

(1) A key problem in this paper is the lack of introduction to the applicability of the proposed method. Is this model applicable to all landslides?

(2) The introduction section lacks some important references, such as:

10.1016/j.enggeo.2009.11.006; 10.1007/s00477-023-02394-4.

(3) Many of the figures in the paper have inconsistent fonts, and some of them are not clear (Fig.1,5,6,A2). Line 711 Fig.2 should be A2.

(4) How is the threshold of the Spearman correlation coefficients determined in Figure A1?

(5) It is suggested to supplement the basis for selecting each factors in Figure 5.

(6) The conclusion is not concise and innovative. I believe that the Authors should try to interpret and explain more clearly their results. Some key quantitative conclusions should be supplemented.

(7) In this study, the authors must improve the statements about what is new in their study and what are the contributions to the developments of landslides research.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

The article proposed for publication addresses highly topical issues of the occurrence and risk of landslides.

 Some well-known facts are confirmed

- "rainfall plays the most important role,  accounting for 70% of the predictive information of all variables combined. Terrain slope  yielded 20%, while subnormal housing conditions and housing density yielded slightly  lower contribution."

Research innovations have been added, looking at the relationships with population and income and the occurrence of landslides.

"Despite the contribution to the model, there is a dramatic difference of settlements conditions, expressed by the chance of a landslide occurrence: it is times higher in a subnormal settlement than in a regular one."

These results reinforce the role of local ordinances aimed to restrict occupation in steeper slopes and public policies to promote adequate housing to, making disasters caused by landslides a major source of risk.For me, this is a local factor, it depends a lot on the geographical location of the landslide.

The results achieved regarding "Vegetation cover" are debatable. I accept them and explain them with the specificity of the climate in the Brazilian region, different from the European

I highly recommend accepting the article for publication!

Author Response

Dear reviewer,

Thank you for your valuable feedback. We sincerely appreciate your insightful comments, that were considered together with the contributions from other reviewers to guide us towards improving the manuscript.

We have made significant revisions to the manuscript to provide a more comprehensive overview of the novelty and contributions of our study. We have also placed greater emphasis on clarifying the context of the SETS framework, enabling readers to better understand the significance and value of our research in the field of landslides.

In the attached document, we have gathered the main revisions done to the manuscript.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Authors addressed all my concerns and it's ready for publication.