The manuscript entitled “Cost estimation for the monitoring instrumentalization of Landslide Early Warning Systems” develops a cost-effective method for low-cost and easy-to-use EWS instrumentalization in landslide-prone areas identified based on data-driven methods. In general, the manuscript contains an interesting topic that is considered one of the important stages in landslide mechanism assessment; but there are several modifications that have to be considered. In this regard, the following comments are requested to be addressed by the authors:

C1: The English of the paper is readable; however, I would suggest the authors have it checked, preferably by a native English-speaking person, to avoid any mistakes.

C2: The necessity & novelty of the manuscript should be presented and stressed in the “Introduction” section.

C3: Provide a literature of the methods developed/applied on landslide mechanism assessment and modeling in “Introduction”. The use of a table to demonstrate the advantage-disadvantage of these methods can be useful. Towards the end, mention the superiority & repeat the novelty of your work.

C4: Please add a subsection clearly articulating the main limitations, wider applicability of your methods, and findings in the “Discussion” section.

C5: The authors should deepen the discussion.

C6: As a suggestion, the following articles could be useful for improving this manuscript.

1. Nikoobakht, S., Azarafza, M., Akgün, H., & Derakhshani, R. (2022). Landslide susceptibility assessment by using convolutional neural network. Applied Sciences, 12(12), 5992. https://doi.org/10.3390/app12125992
2. Fathani, T.F., Karnawati, D., Wilopo, W., Setiawan, H. (2023). Strengthening the Resilience by Implementing a Standard for Landslide Early Warning System. In: Sassa, K., Konagai, K., Tiwari, B., Arbanas, Ž., Sassa, S. (eds) Progress in Landslide Research and Technology, Volume 1 Issue 1, 2022. Progress in Landslide Research and Technology. Springer, Cham. https://doi.org/10.1007/978-3-031-16898-7_20
3. Nanehkaran, Y. A., Licai, Z., Chengyong, J., Chen, J., Anwar, S., Azarafza, M., & Derakhshani, R. (2023). Comparative Analysis for Slope Stability by Using Machine Learning Methods. Applied Sciences, 13(3), 1555. https://doi.org/10.3390/app13031555
4. Yang, F.-Y.; Zhuo, L.; Xiao, M.-L.; Xie, H.-Q.; Liu, H.-Z.; He, J.-D. A Statistical Risk Assessment Model of the Hazard Chain Induced by Landslides and Its Application to the Baige Landslide. Appl. Sci.2023, 13, 3577. https://doi.org/10.3390/app13063577
5. Gariano, S.L., Melillo, M., Brunetti, M.T., Kumar, S., Mathiyalagan, R., Peruccacci, S. (2023). Challenges in Defining Frequentist Rainfall Thresholds to Be Implemented in a Landslide Early Warning System in India. In: Sassa, K., Konagai, K., Tiwari, B., Arbanas, Ž., Sassa, S. (eds) Progress in Landslide Research and Technology, Volume 1 Issue 1, 2022. Progress in Landslide Research and Technology. Springer, Cham. https://doi.org/10.1007/978-3-031-16898-7_27
6. Segoni, S., Serengil, Y. & Aydin, F. A prototype landslide early warning system in Rize (Turkey): analyzing recent impacts to design a safer future. Landslides20, 683–694 (2023). https://doi.org/10.1007/s10346-022-01988-3
7. Han, Min, et al. "An Early Warning System for Landslide Risks in Ion-Adsorption Rare Earth Mines: Based on Real-Time Monitoring of Water Level Changes in Slopes." Minerals13.2 (2023): 265. https://doi.org/10.3390/min13020265

Technological advances in recent years, and in particular the emergence of the Internet of Things paradigm, low-cost programmable electronics and sensors, ubiquitous (wireless) communication infrastructure and a wide range of data-based services and resources available on the Internet, bear the great potential for filling a much needed gap of monitoring high-risk geographic areas with sufficient density in order to allow for effectiv early warning of disasters of natural origin. The potential merits of such technological solutions are nevertheless challenged by still hard to quantify costs and benefits, therefore slowing down their adoption by decision makers. The paper deals with this relevant topic and offers some new insights into the cost of implementing an early warning system for landslides. The work is based on data gathered from field work, experience with a real-life sensor network deployment, and analysis of various databases of historical landslide events. This is a clear strenght of the work.



Despite the above, I have three main concerns regarding the article in its current form:

Concern no. 1: Section 2 on material and methods provides a description of a number of tasks undertaken for merging and filtering databases and geographic information systems about historic landslides, field observation and demographic information in order to identify the most suitable candidate areas for deploying monitoring sensor networks. Regretfully, however, the description lacks enough detail for the work to be reproducible. To give a few examples on p. 8: Statements such as "...we use topographic, geological and precipitation factors..." (What are these factors? How are they used?), "...we rely on socio-demographic factors..." (same as before), "...we tested several methods..." (Which methods?) "...and 500 pixels proved to be the most appropriate..." (By what criterion?), are not sufficient in a scientific article.

Concern no. 2.: For the cost evaluation the authors emphasize that only the costs of the implementation of the wireless geosensor network are included in the cost function, while cost of aspects such as risk evaluation, social interventions, social work and network maintenance are not included. While it is valid to isolate the cost of the technology (the geosensor network) form other costs, is is also the case that a number of cost factors considered for the technology in this work are missing. In particular, the cost of operating the network over time is an essential aspect that decision-makers must know. Securing budget for just deploying the network is not enough: it's wasted money if not accompanied by proper maintenance. Maintenance costs include replacement parts, the cost of vandalized or stolen sensor nodes, cost of human resources in the field for doing maintenance and in the lab for repairs, plus tools for fieldwork, transport to and from sites, etc. Other operational costs include the cost of Internet access to and from the gateways, cost of databases and severs in the Cloud and possibly software tools. The cost of manufacturing the sensor nodes should include a yield factor for the various parts (purchased off the shelf or manufactured in house). Finally, a realistic cost estimate should include some rental cost for office and workshop sapce and utilities (electricity at least). Otherwise, who pays for that once the decision-maker places an order?

Concern no. 3: The argumentation about cost-effectiveness in Section 3 centers on various ways of prioritizing how to choose the deployment sites under budget restrictions. While considering these aspects does make sense, there is a more fundamental prior question that is not addressed nor discussed: what is the benefit of deploying the technology at each site? What is the cost-benefit relationship for each site, is it worth the cost? Intuition says it is, but only if the answer is yes, then the pressented prioritization comes into play. Similarly, the argument on p. 18 "if the city would use the same budget to instrumentalize 9 EWS" is hard to follow: the city probably does prioritize based on cost-benefit and (correctly) decides to prioritize other expenses. The argument about re-balancing budget only holds water if the cost-benefit relationships are understood.

On a related matter, a number of potential sites are ruled out because it is complicated to deploy networks there. What is the proportion of population that could benefit from the monitoring networks if unlimited budget was available? How much population would need other kinds of solutions?

There are a number of specific comments and technical corrections that should be addressed. The following list is not comprehensive:

1. P. 2: "among many other things" is somewhat too colloquial for a research paper. "Untapped potentials", should read "untapped potential".
2. P. 7: "30.200 report of potential mass movements": how many of those are unique and distinct events?
3. P. 9: "measurements from 215 stations"? Are these all stations available in the Aburrá valley, is this a subset? How were they chosen? "Root mean square error of 506 mm". Seems a lot. Can you put it in perspective, give some insight? What is a number that could be expected?
4. P. 10: "Training a statistical model". Which one? Can you give a reference?
5. P. 11: The word "sensor" is used in place of "node" (one node can have several sensors). This confuses the reader.
6. P17.: "Amout of people". Better word: population. The word "cheap" should be avoided. In-expensive, low-cost.
7. Table S1 is cited several times, but it is not available.
8. Acronyms not defined: CSM-EXT, AOI
9. All pages: phrases beginning with "we" are by far too many, this should be revised.

I encourage the authors to revise their paper and improve it, because this kind of work, while complex and difficult to do, is very much necessary.