the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 17 Jun 2024
Review article: Research progress on influencing factors, data, and methods for early identification of landslide hazards
Abstract. The early identification of potential landslide hazards has always been a hot and difficult issue in the field of international landslide research. In recent years, many scholars have conducted extensive and beneficial explorations in this field, making significant contributions to the effective prevention of landslide disasters. However, until now, there are very few review documents on summarizing such valuable experience in the system, which makes it difficult to meet the ever-increasing demand of researchers in scientific documents. To address the gap, this paper systematically reviews 843 documents collected by the two data platforms of Web of Science (WOS) and Scopus from 1971 to 2023 by using the bibliometric analysis software. This paper first figures out the internal relationship between documents by analysing their spatial and temporal distribution characteristics, and then emphatically analyses the application, advantages and disadvantages of different early identification methods based on the influencing factors of landslide disaster formation and multi-source data acquisition links. And finally, this paper discusses the challenges and development trends in this field from four aspects of cooperative analysis, multi-source data, topic analysis and research trends, and puts forward some suggestions. This research can help researchers to use various early identification methods reasonably and provide summary and integration services of scientific document achievements for efficient research in this field.
This preprint has been withdrawn.
-
Withdrawal notice
This preprint has been withdrawn.
-
Preprint
(2198 KB)
Interactive discussion
Status: closed
-
RC1: 'Comment on nhess-2024-68', Anonymous Referee #1, 28 Jun 2024
Review articles are important for a research community, in the case of this article the landslide research community. Reviews help to prove that a community exists, which is a positive aspect in itself. They identify links between subjects and help the community to move ahead by identifying open problems, bottlenecks, or critical aspects of the state of knowledge and suggesting ways forward. For readers, review articles are a good way to keep abreast of recent -- and not so recent -- important advances in specific research fields. They contribute to transfer information from a specific field of research (landslides, in this case) to other fields, and -- importantly -- they represent relevant teaching aids for those entering a field, including e.g., master and PhD students.
For these reasons, I was pleased to be asked to review a submission to NHESS on “Research progress on influencing factors, data and methods for early identification of landslide hazards”, by Heng Lu and ten co-authors.
After reading the title, I was expecting a critical analysis of the current ability to predict landslides and their hazards, and I was hoping to find valuable recommendations and innovative suggestions on how to advance these capabilities, which are still limited. I have spent part of my career working on these topics that I consider crucial for the community of researchers and practitioners concerned with landslides, the problems they pose and the possible strategies to mitigate the related risks.
Unfortunately, my expectations were deluded. None -- or way too little -- of what I was expecting is in the article. Below, I describe the reasons for my, mostly negative, comments on the manuscript.
Building on research dating back to the early work of James D. Dana (1864) in the USA and Armin R. Baltzer (1875) in Switzerland (and probably by others before), over many years but particularly in the last four to five decades, scientists have established the study of landslides as a discipline in its own right. The breadth and complexity of the discipline is vast today and, like many other fields of science, is expanding. Unfortunately, the review does not capture that breadth and complexity in any way.
This has multiple reasons, all serious. I will focus on four main weaknesses, but others exist.
The first weakness refers to the subject the authors have embraced that is large (huge), perhaps too large for a single review. Landslides are very diverse phenomena, as are the hazards and risks they pose. This is well known in the literature, but it does not emerge in the review.
In fact, the breath of the research area (or areas) under review is not clearly defined in the article. Specifically, it is not clear what “early identification of landslide hazards” means to the authors, and what are the “influencing factors, data, and methods” that the authors have considered, and why they have considered these issues and not others. In other words, the article lacks a clear focus.
The ability for -- in the authors’ words -- an “early identification of landslide hazards” depends on the type of landslide, its inherent characteristics, and the meaning of the term “early”, which can refer to seconds, minutes, hours, days, weeks, months, seasons or more, depending on context. The ability to identify the landslide hazards varies for new individual landslides, for existing landslides, for clusters of a few landslides, or for large populations of new landslides. The review ignores this complexity.
The authors should have provided a detailed explanation of what they mean by "early identification of landslide hazards" and by "influencing factors, data, and methods" they have investigated. A clearer description of the scope of the article and the breath of the scientific subject(s) examined would have been beneficial. However, I am afraid that this would have only partially addressed the problems with the review article.
The second weakness I have identified has to do with the strategy used to search the bibliographic databases and obtain the list of references to review, and to perform the bibliometric analyses.
The authors have searched Clarivate Web of Science and the Elsevier Scopus bibliographic databases using the query << TS = “landslide*” AND “early” AND “identif*” >>. However, they do not explain to what fields the query was applied to. Even assuming it was applied extensively i.e., to the title, abstract, text, key words, the search does not guarantee that all (or even most) the relevant articles were selected, and significant articles were not missed. As an example, were articles dealing with e.g., rock falls, soil slips, debris flows, slumps, earthflows, mud flows, or deep-seated gravitational deformations considered, or not?
Indeed, it is possible that the search terms used in the query may not be present in many articles listed in the two databases, in which case they will be excluded. Furthermore, it is necessary to define precisely what is meant by the search term "identif*". It is likely that synonyms such as e.g., predisposing or triggering factors (or other terms) have been used in the articles, also preventing these articles from being selected by the query.
Given the size and complexity of the landslide literature, why have the authors opted for a basic – most possibly too basic – selection criteria? Why haven’t they used more search criteria i.e., multiple, more complex, and synergic queries, and then merged and seeded the results? Have the authors attempted an evaluation of the completeness and significance of the list of references they have obtained?
With this respect, I note here that the authors have not provided the list of articles they have selected and used for the review. For transparency and to allow others to repeat the queries and analyses, the authors should have provided the exact results of their search in the two original bibliographic databases e.g., as a list in the ancillary materials.
Overall, I have quite serious concerns that the 843 articles selected by the authors are truly representative of the landslide literature in the 52.5 year considered period. Other bibliographical analyses of the landslide literature have been attempted (e.g., https://doi.org/10.1007/s10346-015-0624-z). Why have the authors not compared their analyses to other existing analyses?
I will make four examples of papers that I consider relevant and – apparently – may not have been considered by the review. Two are relevant for early warning, namely the articles by Piciullo et al. (2018) https://doi.org/10.1016/j.earscirev.2018.02.013 and by Guzzetti et al. (2020), https://doi.org/10.1016/j.earscirev.2019.102973. The article by Intrieri et al (2019), https://doi.org/10.1016/j.earscirev.2019.03.019, reviewed the literature on the ability to forecast the time of failure of landslides at the slope-scale, and the article by Broeckx et al. (2020), https://doi.org/10.1016/j.earscirev.2019.102972, reviewed landslide mobilization rates, a very relevant topic for landslide hazard assessment. I emphasise that these four papers are all review papers, have all been published in the top-tier journal Earth-Science Reviews, and have received a substantial number of citations; and other review or key papers may not have been considered –- albeit this remain uncertain, as the list of the selected articles is not available.
The third weakness I have identified has to do with the scientific relevance, and the significance of the bibliometric analysis performed and the discussion of its results.
The authors seem to be convinced that such purely bibliometric analyses can help describe the “state” of a scientific field, and of the community that operates and (hopefully) attempts to advance that field. Although the idea, questionable by itself, in the recent years has sprouted a lively debate, the results of bibliometric or scientometric analyses can only be used in relative terms e.g., to compare different fields of science (see e.g., https://doi.org/10.1038/430311a), and not to measure or describe a single field. In this case the broad field of landslide studies.
As an aside, and with +respect to the suitability of the submission to the journal NHESS, the manuscript may be better suited for a different audience or journal focused on bibliometrics, sciencetometrics, or infometrics issues (e.g., the Jouurnal of Infometrics).
Commenting their bibliometric analysis, (e.g., Figure 5), the authors seem to embrace the idea that what matters is the number of papers published. That is, quantity as opposed to quality and impact. I am strongly convinced of the opposite. Nel Caine in 1980 and John L. Innes in 1983 published independently two fundamental papers on rainfall thresholds for the occurrence (and prediction) of rain-triggered landslides. Maybe my information is incorrect, but they both have not published other similar papers on rainfall thresholds. Thus, adopting the volumetric metrics used in the review, both Caine and Innes would rank at the bottom of the list. An unrealistic (and unfair) placement.
Although disseminating scientific results and ideas is clearly important (fundamental), and this is done (also) by publishing papers in peer-reviewed journals, what matters most is the impact that the publications have. Assessing the impact of single publications, and even more of research ideas, is known to be a difficult, complex, and uncertain task. However, there are better metrics than the number of entries in a bibliographic database to evaluate such impact. Among the many available metrics, the number of citations, the h-index, the g-index, and the composite index (C-score, Ioannidis et al. 2016, https://doi.org/10.1371/journal.pbio.1002501) provide more reliable results to evaluate the impact of single authors or of group of authors. For authors pertaining to the same research team, the statistics should be lumped together, for obvious reasons.
For other -- also obvious -- reasons, comparing authors working in large and small countries may not be the most appropriate ways of describing the geographical distribution of the studies – where “large” and “small” in this context may measure e.g., the size of the country or of the population, the number of universities and research centres, the number of scholars, the GDP or the percentage of the GDP invested in research, or the severeness of the landslide problem in each country. I expect the results to be different.
Indeed, it would have been more interesting to investigate where landslide studies have been made, and by whom (scholar, institutions, countries, etc.).
The time since publication, is also a variable that needs to be considered (see e.g. https://doi.org/10.1038/s41586-022-05543-x).
The last reason of concern I address stems from, and adds to, those already described, and it is the complete lack of consideration of the physical characteristics of landslides, their complexity, and their effects and impacts. This is reflected in the Conclusions, which largely state the obvious.
In addition to the problems related to the unclear scope and coverage of the review, and the doubtful completeness and representativeness of the articles selected from the two bibliographic databases, a crucial issue with the review is the lack of perspective. What are the problems that the relevant landslide literature has attempted to address over the more than 50 years covered by the review, and with what degree of success? What is the authors’ opinion on this research, on the different topics and subtopics they have considered (e.g., on “Landslide Formation Factors and Data Acquisition”, “Early Identification Methods” and their subdivisions)? What are the open problems and why are they have remained open? Because of the lack of data, models, or understanding? Or what else? What are the authors proposing specifically to fill the existing gaps, in terms of data, models, and understanding? Can the authors identify gaps that will be easier or more urgent to fill, than others that will be harder (or impossible), or less pressing to fill? Are there gaps that need to be addressed before than others? Based on the existing literature, are the current research paradigms adequate, or do we need different approaches and paradigms?
For a valuable review of the landslide literature, one would expect these questions to be posed and addressed, at least partially.
As an aside, the authors' statement at the end of the paper about data availability i.e., that "no data sets were used in this article" is incorrect, evidently. The entire paper is based on data. The use of data should have been acknowledged, and the data used to perform the bibliometric analyses should have been made available to the reader and the community at large, for the sake of transparency and reproducibility of the analyses.
In summary, given what I consider to be the serious shortcomings of the research behind the article, some of which I have outlined, the rather simplistic use of bibliometric statistics, and the questionable discussion of the results obtained, I do not think that this review would be a valuable addition to the landslide literature, and more generally to the natural hazards literature covered by the journal NHESS.
Citation: https://doi.org/10.5194/nhess-2024-68-RC1 -
RC2: 'Comment on nhess-2024-68', Anonymous Referee #2, 02 Jul 2024
The manuscript submitted by Heng Lu and 10 co-authors is intended to be a review article, titled: "Review Article: Research Progress on Influencing Factors, Data, and Methods for Early Identification of Landslide Hazards." The title is appealing, and the subject is of interest and falls within the scope of the journal NHESS.
However, the manuscript did not meet my expectations. Below, I outline my main concerns.
1. Lack of Focus and Clarity
The subject matter of landslides is broad and complex, and the review does not effectively capture this. The term "early identification of landslide hazards" is not well-defined, and the scope of "influencing factors, data, and methods" is unclear. This lack of focus undermines the review's effectiveness.
Section 4 of the manuscript (Early Identification Methods) seems to refer to landslide susceptibility and hazard assessment methods. It is unclear why the authors have chosen to label them as early identification methods. Within this topic, the manuscript offers a summary that does not contribute any significant new insights to the established scientific knowledge. Moreover, the examples chosen to illustrate each type of landslide susceptibility assessment method are presented without a uniform and clear criterion.
2. Bibliographic Search Strategy
According to the title of the work, it was assumed that the objective would be a critical literature review on the topic of early identification of potential landslide hazards, exploring the Web of Science (WoS) and Scopus databases within the time span from 1971 to 2023. But, the authors' bibliographic search using the query << TS = “landslide*” AND “early” AND “identif*” >> in the Clarivate Web of Science and Elsevier Scopus databases is inadequately described. It is unclear to which fields the query was applied (title? Keywords? Abstract?). Moreover, the query used to filter WoS and Scopus is restrictive, leaving many important works behind.
The lack of transparency in the selection process and the absence of a comprehensive list of selected articles further weaken the review. The subsequent analyses, using other words in the query within the databases, were conducted only on the original bibliometric database, thus retaining the initial bias identified.
This reviewer is not convinced that the authors' strategy in searching and screening allows them to make conclusive statements about the research on early identification of potential landslide hazards. In fact, the manuscript has a major drawback from the outset: it lacks a demonstration that using a single search strategy ("landslide*" AND "early" AND "identif*") would yield a comprehensive and representative list of scientific papers focused on the early identification of potential landslide hazards.
What if authors substitute "landslide" with terms like "debris flow," "rockfall," or other specific landslide types? What if authors replace the other terms (“early” AND “identif”) with "triggering"?
Table 3 lists the top 10 highly cited papers found by authors. Below, you can find a brief list of essential papers that seem to have been missed in the analysis. Each of these papers has received more citations than those listed in Table 3, in both the WoS and Scopus databases:
Keefer, D. K. (1984). Landslides caused by earthquakes. Geological Society of America Bulletin, 95(4), 406-421.
Iverson, R. M. (2000). Landslide triggering by rain infiltration. Water resources research, 36(7), 1897-1910.
Guzzetti, F., Peruccacci, S., Rossi, M., & Stark, C. P. (2008). The rainfall intensity–duration control of shallow landslides and debris flows: an update. Landslides, 5, 3-17.
Gariano, S. L., & Guzzetti, F. (2016). Landslides in a changing climate. Earth-science reviews, 162, 227-252.
3. Bibliometric Analysis
The reliance on purely bibliometric analyses to describe the state of landslide research is problematic. Such analyses are inherently limited and should not be used to measure or describe a single field comprehensively. It appears that the cleaning and preparation of documents for bibliometric analysis were not carried out thoroughly. This is evident, for example, in Figure 7, produced with the Vosviewer software, which aims to present a keyword cloud map of landslide influencing factors. However, the majority of listed keywords are not actually landslide influencing factors, making this analysis both useless and misleading.
4. Lack of Perspective and Critical Analysis
The review fails to address the fundamental issues and open problems in landslide research. There is a lack of critical analysis of the literature, and the authors do not offer substantial insights or recommendations for future research.
Overall, the manuscript appears to be an exercise in successive experimentation using a wide range of bibliometric analysis techniques. The main issue is the inadequacy of the initially constructed database. Additionally, there is a lack of in-depth analysis of the contents of the considered articles.
The presented work is entirely based on bibliometric analysis, lacking critical analysis, and the level of analysis and conclusions is very poor. For example, authors state "The total number of papers published by China ranked first, which is consistent with the number of papers published by high productivity institutions calculated in Fig. 5(d). The fundamental reason is that China is a country with frequent landslide disasters." The explanation is poor. What about considering the number of Chinese researchers and institutions currently studying and publishing on landslide topics?
Section 5 (Discussion and Research Trends) is not a discussion but rather an exploration of the database using an additional bibliometric tool: CiteSpace software.
5. Inadequate use of terms and concepts
The authors use the term "landslide disaster" extensively. A landslide is a natural process, it is not necessarily a disaster. A landslide disaster implies the occurrence of negative consequences resulting from the landslides. In other words, for a landslide to become a disaster, there must be exposure and vulnerability of the affected elements.
At the beginning of Section 4, the authors state, "There are many kinds of methods for early identification of potential landslide hazards, which can be divided into qualitative and quantitative methods according to the type of landslide inventory." According to the landslide literature (e.g. Corominas et al., 2014) the distinction between methods to assess landslide susceptibility is not based on the type of landslide inventory.
In the conclusion (which is a summary instead of a conclusion), shallow landslides are presented as an “important factor affecting the formation of landslides”, which is not acceptable.
6. Other comments
The authors often use indecipherable language. See below for a few examples:
Line 233 “a review paper entitled “A””
Line 240 “S.M and G are three high productivity authors.”
Line 246 “from high productivity institutions, such as U and C”
249-250 “high productivity institutions such as C1, C2, C3, I1, C4 and C5”
line 258: “bibliometric tools V and S”
The presentation of the landslide deformation stages is unclear. In line 47, the authors state that “Most of the landslides follow their evolution laws.” Which evolution laws are being referred to? In line 51, the authors state that landslides “have a uniform deformation process from deceleration to acceleration.” What is the meaning of deceleration in this context?
In Table 1, Authors refer to IoT (Internet of ThingS) and IOT (Internet of Things). What is the difference between these terms?
In section 2.2 authors describe the technical aspects of the bibliometric tools they have used. The reviewer does not think this is of interest for a NHESS reader.
The work of Bui et al. (2019) is presented has becoming a powerful tool to mitigate and manage landslide disasters in Lang Son, Vietnam. But this work deals (only) with landslide susceptibility. It is excessive to considerer such analysis as a powerful tool to mitigate and manage landslide disasters.
The expression “Spatial distribution” used in the title of section 2.3.2 and figure 3 caption is not appropriate.
In the Data Availability section, the authors state that “No data sets were used in this article,” which makes little sense, as the entire manuscript is constructed using a bibliometric database built based on WoS and Scopus.
7. Conclusion
Due to the serious shortcomings in the research methodology, the simplistic use of bibliometric statistics, and the lack of substantial discussion, I do not believe this review would be a valuable addition to the landslide literature or to the journal NHESS.
Citation: https://doi.org/10.5194/nhess-2024-68-RC2
Interactive discussion
Status: closed
-
RC1: 'Comment on nhess-2024-68', Anonymous Referee #1, 28 Jun 2024
Review articles are important for a research community, in the case of this article the landslide research community. Reviews help to prove that a community exists, which is a positive aspect in itself. They identify links between subjects and help the community to move ahead by identifying open problems, bottlenecks, or critical aspects of the state of knowledge and suggesting ways forward. For readers, review articles are a good way to keep abreast of recent -- and not so recent -- important advances in specific research fields. They contribute to transfer information from a specific field of research (landslides, in this case) to other fields, and -- importantly -- they represent relevant teaching aids for those entering a field, including e.g., master and PhD students.
For these reasons, I was pleased to be asked to review a submission to NHESS on “Research progress on influencing factors, data and methods for early identification of landslide hazards”, by Heng Lu and ten co-authors.
After reading the title, I was expecting a critical analysis of the current ability to predict landslides and their hazards, and I was hoping to find valuable recommendations and innovative suggestions on how to advance these capabilities, which are still limited. I have spent part of my career working on these topics that I consider crucial for the community of researchers and practitioners concerned with landslides, the problems they pose and the possible strategies to mitigate the related risks.
Unfortunately, my expectations were deluded. None -- or way too little -- of what I was expecting is in the article. Below, I describe the reasons for my, mostly negative, comments on the manuscript.
Building on research dating back to the early work of James D. Dana (1864) in the USA and Armin R. Baltzer (1875) in Switzerland (and probably by others before), over many years but particularly in the last four to five decades, scientists have established the study of landslides as a discipline in its own right. The breadth and complexity of the discipline is vast today and, like many other fields of science, is expanding. Unfortunately, the review does not capture that breadth and complexity in any way.
This has multiple reasons, all serious. I will focus on four main weaknesses, but others exist.
The first weakness refers to the subject the authors have embraced that is large (huge), perhaps too large for a single review. Landslides are very diverse phenomena, as are the hazards and risks they pose. This is well known in the literature, but it does not emerge in the review.
In fact, the breath of the research area (or areas) under review is not clearly defined in the article. Specifically, it is not clear what “early identification of landslide hazards” means to the authors, and what are the “influencing factors, data, and methods” that the authors have considered, and why they have considered these issues and not others. In other words, the article lacks a clear focus.
The ability for -- in the authors’ words -- an “early identification of landslide hazards” depends on the type of landslide, its inherent characteristics, and the meaning of the term “early”, which can refer to seconds, minutes, hours, days, weeks, months, seasons or more, depending on context. The ability to identify the landslide hazards varies for new individual landslides, for existing landslides, for clusters of a few landslides, or for large populations of new landslides. The review ignores this complexity.
The authors should have provided a detailed explanation of what they mean by "early identification of landslide hazards" and by "influencing factors, data, and methods" they have investigated. A clearer description of the scope of the article and the breath of the scientific subject(s) examined would have been beneficial. However, I am afraid that this would have only partially addressed the problems with the review article.
The second weakness I have identified has to do with the strategy used to search the bibliographic databases and obtain the list of references to review, and to perform the bibliometric analyses.
The authors have searched Clarivate Web of Science and the Elsevier Scopus bibliographic databases using the query << TS = “landslide*” AND “early” AND “identif*” >>. However, they do not explain to what fields the query was applied to. Even assuming it was applied extensively i.e., to the title, abstract, text, key words, the search does not guarantee that all (or even most) the relevant articles were selected, and significant articles were not missed. As an example, were articles dealing with e.g., rock falls, soil slips, debris flows, slumps, earthflows, mud flows, or deep-seated gravitational deformations considered, or not?
Indeed, it is possible that the search terms used in the query may not be present in many articles listed in the two databases, in which case they will be excluded. Furthermore, it is necessary to define precisely what is meant by the search term "identif*". It is likely that synonyms such as e.g., predisposing or triggering factors (or other terms) have been used in the articles, also preventing these articles from being selected by the query.
Given the size and complexity of the landslide literature, why have the authors opted for a basic – most possibly too basic – selection criteria? Why haven’t they used more search criteria i.e., multiple, more complex, and synergic queries, and then merged and seeded the results? Have the authors attempted an evaluation of the completeness and significance of the list of references they have obtained?
With this respect, I note here that the authors have not provided the list of articles they have selected and used for the review. For transparency and to allow others to repeat the queries and analyses, the authors should have provided the exact results of their search in the two original bibliographic databases e.g., as a list in the ancillary materials.
Overall, I have quite serious concerns that the 843 articles selected by the authors are truly representative of the landslide literature in the 52.5 year considered period. Other bibliographical analyses of the landslide literature have been attempted (e.g., https://doi.org/10.1007/s10346-015-0624-z). Why have the authors not compared their analyses to other existing analyses?
I will make four examples of papers that I consider relevant and – apparently – may not have been considered by the review. Two are relevant for early warning, namely the articles by Piciullo et al. (2018) https://doi.org/10.1016/j.earscirev.2018.02.013 and by Guzzetti et al. (2020), https://doi.org/10.1016/j.earscirev.2019.102973. The article by Intrieri et al (2019), https://doi.org/10.1016/j.earscirev.2019.03.019, reviewed the literature on the ability to forecast the time of failure of landslides at the slope-scale, and the article by Broeckx et al. (2020), https://doi.org/10.1016/j.earscirev.2019.102972, reviewed landslide mobilization rates, a very relevant topic for landslide hazard assessment. I emphasise that these four papers are all review papers, have all been published in the top-tier journal Earth-Science Reviews, and have received a substantial number of citations; and other review or key papers may not have been considered –- albeit this remain uncertain, as the list of the selected articles is not available.
The third weakness I have identified has to do with the scientific relevance, and the significance of the bibliometric analysis performed and the discussion of its results.
The authors seem to be convinced that such purely bibliometric analyses can help describe the “state” of a scientific field, and of the community that operates and (hopefully) attempts to advance that field. Although the idea, questionable by itself, in the recent years has sprouted a lively debate, the results of bibliometric or scientometric analyses can only be used in relative terms e.g., to compare different fields of science (see e.g., https://doi.org/10.1038/430311a), and not to measure or describe a single field. In this case the broad field of landslide studies.
As an aside, and with +respect to the suitability of the submission to the journal NHESS, the manuscript may be better suited for a different audience or journal focused on bibliometrics, sciencetometrics, or infometrics issues (e.g., the Jouurnal of Infometrics).
Commenting their bibliometric analysis, (e.g., Figure 5), the authors seem to embrace the idea that what matters is the number of papers published. That is, quantity as opposed to quality and impact. I am strongly convinced of the opposite. Nel Caine in 1980 and John L. Innes in 1983 published independently two fundamental papers on rainfall thresholds for the occurrence (and prediction) of rain-triggered landslides. Maybe my information is incorrect, but they both have not published other similar papers on rainfall thresholds. Thus, adopting the volumetric metrics used in the review, both Caine and Innes would rank at the bottom of the list. An unrealistic (and unfair) placement.
Although disseminating scientific results and ideas is clearly important (fundamental), and this is done (also) by publishing papers in peer-reviewed journals, what matters most is the impact that the publications have. Assessing the impact of single publications, and even more of research ideas, is known to be a difficult, complex, and uncertain task. However, there are better metrics than the number of entries in a bibliographic database to evaluate such impact. Among the many available metrics, the number of citations, the h-index, the g-index, and the composite index (C-score, Ioannidis et al. 2016, https://doi.org/10.1371/journal.pbio.1002501) provide more reliable results to evaluate the impact of single authors or of group of authors. For authors pertaining to the same research team, the statistics should be lumped together, for obvious reasons.
For other -- also obvious -- reasons, comparing authors working in large and small countries may not be the most appropriate ways of describing the geographical distribution of the studies – where “large” and “small” in this context may measure e.g., the size of the country or of the population, the number of universities and research centres, the number of scholars, the GDP or the percentage of the GDP invested in research, or the severeness of the landslide problem in each country. I expect the results to be different.
Indeed, it would have been more interesting to investigate where landslide studies have been made, and by whom (scholar, institutions, countries, etc.).
The time since publication, is also a variable that needs to be considered (see e.g. https://doi.org/10.1038/s41586-022-05543-x).
The last reason of concern I address stems from, and adds to, those already described, and it is the complete lack of consideration of the physical characteristics of landslides, their complexity, and their effects and impacts. This is reflected in the Conclusions, which largely state the obvious.
In addition to the problems related to the unclear scope and coverage of the review, and the doubtful completeness and representativeness of the articles selected from the two bibliographic databases, a crucial issue with the review is the lack of perspective. What are the problems that the relevant landslide literature has attempted to address over the more than 50 years covered by the review, and with what degree of success? What is the authors’ opinion on this research, on the different topics and subtopics they have considered (e.g., on “Landslide Formation Factors and Data Acquisition”, “Early Identification Methods” and their subdivisions)? What are the open problems and why are they have remained open? Because of the lack of data, models, or understanding? Or what else? What are the authors proposing specifically to fill the existing gaps, in terms of data, models, and understanding? Can the authors identify gaps that will be easier or more urgent to fill, than others that will be harder (or impossible), or less pressing to fill? Are there gaps that need to be addressed before than others? Based on the existing literature, are the current research paradigms adequate, or do we need different approaches and paradigms?
For a valuable review of the landslide literature, one would expect these questions to be posed and addressed, at least partially.
As an aside, the authors' statement at the end of the paper about data availability i.e., that "no data sets were used in this article" is incorrect, evidently. The entire paper is based on data. The use of data should have been acknowledged, and the data used to perform the bibliometric analyses should have been made available to the reader and the community at large, for the sake of transparency and reproducibility of the analyses.
In summary, given what I consider to be the serious shortcomings of the research behind the article, some of which I have outlined, the rather simplistic use of bibliometric statistics, and the questionable discussion of the results obtained, I do not think that this review would be a valuable addition to the landslide literature, and more generally to the natural hazards literature covered by the journal NHESS.
Citation: https://doi.org/10.5194/nhess-2024-68-RC1 -
RC2: 'Comment on nhess-2024-68', Anonymous Referee #2, 02 Jul 2024
The manuscript submitted by Heng Lu and 10 co-authors is intended to be a review article, titled: "Review Article: Research Progress on Influencing Factors, Data, and Methods for Early Identification of Landslide Hazards." The title is appealing, and the subject is of interest and falls within the scope of the journal NHESS.
However, the manuscript did not meet my expectations. Below, I outline my main concerns.
1. Lack of Focus and Clarity
The subject matter of landslides is broad and complex, and the review does not effectively capture this. The term "early identification of landslide hazards" is not well-defined, and the scope of "influencing factors, data, and methods" is unclear. This lack of focus undermines the review's effectiveness.
Section 4 of the manuscript (Early Identification Methods) seems to refer to landslide susceptibility and hazard assessment methods. It is unclear why the authors have chosen to label them as early identification methods. Within this topic, the manuscript offers a summary that does not contribute any significant new insights to the established scientific knowledge. Moreover, the examples chosen to illustrate each type of landslide susceptibility assessment method are presented without a uniform and clear criterion.
2. Bibliographic Search Strategy
According to the title of the work, it was assumed that the objective would be a critical literature review on the topic of early identification of potential landslide hazards, exploring the Web of Science (WoS) and Scopus databases within the time span from 1971 to 2023. But, the authors' bibliographic search using the query << TS = “landslide*” AND “early” AND “identif*” >> in the Clarivate Web of Science and Elsevier Scopus databases is inadequately described. It is unclear to which fields the query was applied (title? Keywords? Abstract?). Moreover, the query used to filter WoS and Scopus is restrictive, leaving many important works behind.
The lack of transparency in the selection process and the absence of a comprehensive list of selected articles further weaken the review. The subsequent analyses, using other words in the query within the databases, were conducted only on the original bibliometric database, thus retaining the initial bias identified.
This reviewer is not convinced that the authors' strategy in searching and screening allows them to make conclusive statements about the research on early identification of potential landslide hazards. In fact, the manuscript has a major drawback from the outset: it lacks a demonstration that using a single search strategy ("landslide*" AND "early" AND "identif*") would yield a comprehensive and representative list of scientific papers focused on the early identification of potential landslide hazards.
What if authors substitute "landslide" with terms like "debris flow," "rockfall," or other specific landslide types? What if authors replace the other terms (“early” AND “identif”) with "triggering"?
Table 3 lists the top 10 highly cited papers found by authors. Below, you can find a brief list of essential papers that seem to have been missed in the analysis. Each of these papers has received more citations than those listed in Table 3, in both the WoS and Scopus databases:
Keefer, D. K. (1984). Landslides caused by earthquakes. Geological Society of America Bulletin, 95(4), 406-421.
Iverson, R. M. (2000). Landslide triggering by rain infiltration. Water resources research, 36(7), 1897-1910.
Guzzetti, F., Peruccacci, S., Rossi, M., & Stark, C. P. (2008). The rainfall intensity–duration control of shallow landslides and debris flows: an update. Landslides, 5, 3-17.
Gariano, S. L., & Guzzetti, F. (2016). Landslides in a changing climate. Earth-science reviews, 162, 227-252.
3. Bibliometric Analysis
The reliance on purely bibliometric analyses to describe the state of landslide research is problematic. Such analyses are inherently limited and should not be used to measure or describe a single field comprehensively. It appears that the cleaning and preparation of documents for bibliometric analysis were not carried out thoroughly. This is evident, for example, in Figure 7, produced with the Vosviewer software, which aims to present a keyword cloud map of landslide influencing factors. However, the majority of listed keywords are not actually landslide influencing factors, making this analysis both useless and misleading.
4. Lack of Perspective and Critical Analysis
The review fails to address the fundamental issues and open problems in landslide research. There is a lack of critical analysis of the literature, and the authors do not offer substantial insights or recommendations for future research.
Overall, the manuscript appears to be an exercise in successive experimentation using a wide range of bibliometric analysis techniques. The main issue is the inadequacy of the initially constructed database. Additionally, there is a lack of in-depth analysis of the contents of the considered articles.
The presented work is entirely based on bibliometric analysis, lacking critical analysis, and the level of analysis and conclusions is very poor. For example, authors state "The total number of papers published by China ranked first, which is consistent with the number of papers published by high productivity institutions calculated in Fig. 5(d). The fundamental reason is that China is a country with frequent landslide disasters." The explanation is poor. What about considering the number of Chinese researchers and institutions currently studying and publishing on landslide topics?
Section 5 (Discussion and Research Trends) is not a discussion but rather an exploration of the database using an additional bibliometric tool: CiteSpace software.
5. Inadequate use of terms and concepts
The authors use the term "landslide disaster" extensively. A landslide is a natural process, it is not necessarily a disaster. A landslide disaster implies the occurrence of negative consequences resulting from the landslides. In other words, for a landslide to become a disaster, there must be exposure and vulnerability of the affected elements.
At the beginning of Section 4, the authors state, "There are many kinds of methods for early identification of potential landslide hazards, which can be divided into qualitative and quantitative methods according to the type of landslide inventory." According to the landslide literature (e.g. Corominas et al., 2014) the distinction between methods to assess landslide susceptibility is not based on the type of landslide inventory.
In the conclusion (which is a summary instead of a conclusion), shallow landslides are presented as an “important factor affecting the formation of landslides”, which is not acceptable.
6. Other comments
The authors often use indecipherable language. See below for a few examples:
Line 233 “a review paper entitled “A””
Line 240 “S.M and G are three high productivity authors.”
Line 246 “from high productivity institutions, such as U and C”
249-250 “high productivity institutions such as C1, C2, C3, I1, C4 and C5”
line 258: “bibliometric tools V and S”
The presentation of the landslide deformation stages is unclear. In line 47, the authors state that “Most of the landslides follow their evolution laws.” Which evolution laws are being referred to? In line 51, the authors state that landslides “have a uniform deformation process from deceleration to acceleration.” What is the meaning of deceleration in this context?
In Table 1, Authors refer to IoT (Internet of ThingS) and IOT (Internet of Things). What is the difference between these terms?
In section 2.2 authors describe the technical aspects of the bibliometric tools they have used. The reviewer does not think this is of interest for a NHESS reader.
The work of Bui et al. (2019) is presented has becoming a powerful tool to mitigate and manage landslide disasters in Lang Son, Vietnam. But this work deals (only) with landslide susceptibility. It is excessive to considerer such analysis as a powerful tool to mitigate and manage landslide disasters.
The expression “Spatial distribution” used in the title of section 2.3.2 and figure 3 caption is not appropriate.
In the Data Availability section, the authors state that “No data sets were used in this article,” which makes little sense, as the entire manuscript is constructed using a bibliometric database built based on WoS and Scopus.
7. Conclusion
Due to the serious shortcomings in the research methodology, the simplistic use of bibliometric statistics, and the lack of substantial discussion, I do not believe this review would be a valuable addition to the landslide literature or to the journal NHESS.
Citation: https://doi.org/10.5194/nhess-2024-68-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 398 | 115 | 23 | 536 | 22 | 19 |
- HTML: 398
- PDF: 115
- XML: 23
- Total: 536
- BibTeX: 22
- EndNote: 19
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1