Heavy Rainfall Triggering Shallow Landslides: A Susceptibility Assessment by a GIS-Approach in a Ligurian Apennine Catchment (Italy)
Round 1
Reviewer 1 Report
This paper is a very interesting and necessary study.
Manuscripts are also complete enough to be published.
I give you the following minor comment.
Comment. Compared with some existing statistical landslide prediction methods using sensitivity analysis, the authors said that the originality of this study is the analysis from relatively unstructured data such as newspapers articles, interviews, and damage reports. General statistical methods are covered in the introduction, but sensitivity analysis using unstructured data, which is represented to originality in this study, has not been addressed. It is necessary to supplement Chapter 1 or Section 2.2. Please, describe what are limitations of existing studies, and that the authors did to overcome the limitations.
Author Response
In Section 2.2 we further implemented consideration about accuracy of unstructured input data and uncertainty
associated with the GIS-based approach adopted, including the modeling of the predisposing parameters in the acquired
regional thematic maps, the resolution of the grid adopted in the spatial analysis, the data combination and the
extrapolation of the punctual value to the corresponding cell-grid and the assessing of the criteria weights attributed to
each predisposing elements.
We clarified limitations of existing studies and novelty of our work in Section 1.
Furthermore, limitations and assets of the simplified approach adopted in the study to asses landslide prone areas
starting heterogeneous and unstructured data have been highlighted in the Section 5.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
I think that in the introduction part, authors could briefly discuss the topic about climate changes and using stationary or nonstationary models, in order to provide a more general overview of the main natural trigerring factor (rainfall), also in a framework of possible climate changes
With this aim, authors could insert the following references
Acero, F.J.; Parey, S.; García, J.A.; Dacunha-Castelle, D. Return Level Estimation of Extreme Rainfall over the Iberian Peninsula: Comparison of Methods. Water 2018, 10, 179, doi:10.3390/w10020179.
Koutsoyiannis D, Montanari A. Negligent killing of scientific concepts: the stationarity case. Hydrol. Sci. J. 2014, 60(7-8), 1174-1183, http://dx.doi.org/10.1080/02626667.2014.959959
De Luca, D.L.; Galasso, L. Stationary and Non-Stationary Frameworks for Extreme Rainfall Time Series in Southern Italy. Water 2018, 10, 1477.
Brunetti, M.; Maugeri, M.; Nanni, T. Changes in total precipitation, rainy days and extreme events in northeastern Italy. Int. J. Climatol. 2001, 21, 861–871, doi: 10.1002/joc.660.
Rodrigo, F.S.; Trigo, R.M. Trends in daily rainfall in the Iberian Peninsula from 1951 to 2002. Int. J. Climatol. 2007, 27, 513–529, doi: 10.1002/joc.1409.
EEA – European Environment Agency. Climate Change adaptation ad disaster risk reduction in Europe. 2017; Volume 15, pp. 1–171, ISSN 1725–9177 (https://www.eea.europa.eu/publications/climate-change-adaptation-and-disaster/at_download/file).
Ban, N.; Schmidli, J.; Schär, C., Heavy precipitation in a changing climate: Does short-term summer precipitation increase faster? Geophys Res. Lett. 2015, 42, 2014GL062588, doi:10.1002/2014GL062588.
Lehmann, J.; Coumou, D.; Frieler, K. Increased record-breaking precipitation events under global warming. Clim. Change 2015, 132, 501–515, doi:10.1007/s10584-015-1434-y.
Author Response
We are thankful to the reviewer for his suggestions and the list of possible references to implement. However, we think
that an analysis of climate changes and their effects on the rainfall regime, one of the most debated and studied topic
nowadays, lies outside the aims of our study, in particular the purpose of giving to local authorities and land-planning
administrators a simplified, reproducible but effective approach based on input data easy to find and update in order to
detect areas characterized by high proneness to instability.
Author Response File: Author Response.pdf
Reviewer 3 Report
This manuscript presents a landslide susceptibility assessment in an area in Italy, which is based on the weighted scoring of considered predisposing factors. The used landslide information is collected from different sources including newspapers and unpublished local archives. In general this manuscript is more like a technical report.
It is hard to tell the research issue of this study from main body of Section Introduction, which is actually no relation with the innovation stated in the last paragraph of this section. This makes this manuscript is very similar to a report rather than a scientific paper.
It should be evaluated if current result from the proposed method and settings (especially the innovation point, because the method framework is normal) is more reasonable/better than results from other settings or other often-used methods. This part in current manuscript is very weak.
Other specific comments are below:
Lines 35-37: "...[8-16]" -- It is improper (even ridiculous) to cite so many references after a sentence about broad background. Similar problems in some other places in the manuscript.
Lines 307-309: More detailed evidence are needed for this presentation. Fig. 9 does show this point.
Fig 9d (incorrect figure number on Figure) and the corresponding discussion are not clearly matched (e.g., type index is used in Figure whereas type names are used in the text).
Table 5: Rationality of the reclassification, especially for aspect, is not convincing.
Lines 407-409: The landslide susceptibility map result, as well as the classification parameter-settings, has not been convinced by current evaluation.
Author Response
Comments and Suggestions for Authors
This manuscript presents a landslide susceptibility assessment in an area in Italy, which is based on the
weighted scoring of considered predisposing factors. The used landslide information is collected from
different sources including newspapers and unpublished local archives. In general, this manuscript is more
like a technical report.
It is hard to tell the research issue of this study from main body of Section Introduction, which is actually no
relation with the innovation stated in the last paragraph of this section. This makes this manuscript is very
similar to a report rather than a scientific paper.
It should be evaluated if current result from the proposed method and settings (especially the innovation
point, because the method framework is normal) is more reasonable/better than results from other settings
or other often-used methods. This part in current manuscript is very weak.
Landslide susceptibility assessment is one of the most debated issue in the geological field. To deal with a problem as
wide and complicated is a bit like describing into the history of the Second World War, which many historians and
authors have been written about: what can be written about the great conflict again?In our case, we are facing a similar
problem: many authors have argued about landslides susceptibility and performed different approaches and algorithms
to assess it, but no one has dealt with this issue in this area of Italy. Furthermore, no existing model has been
implemented using this kind of input data, i.e., information from newspapers, chronicle notes from local and social
media, etc., unpublished local archives, etc.: they are probably not so complete and structured as data usually used to
assess conventional susceptibility, but they are very useful to analyze landslide prone areas and to update regularly
them.So, this study appears as a small piece, but very important as it adds data to previous knowledge: for thisreason, it
is probably not so innovative from a methodological point of view, but it is innovative because it analyzed unpublished
data and proposed a simplified approach based on data easy to find and to update, that will surely be useful to
stakeholders.
Regarding the consideration of this paper as a technical report, the authors absolutely do not agree with the reviewer’s
opinion. For the structure and the setting, the paper cannot be considered a technical report because it has different
characteristics. Moreover, several science journals have published papers that present a landslide susceptibility
assessment in different areas in Italy and worldwideas a research article, e.g.:
Gokceoglu C., Sonmez H., Nefeslioglu H.A., Duman T.Y., Can T. (2005) - The 17 March 2005 Kuzulu landslide
(Sivas, Turkey) and landslide susceptibility map of its near vicinity. Eng. Geol., 81, 65-83.
Ruff, M., Czurda, K. (2008) – Landslide susceptibilkity analysis with a heuristic approach in the Eastern Alps
(Voralberg, Austria). Geomorphology, 94, 314-324.
Poiraud A. (2014) – Landslide susceptibility-certainty mapping by a multi-method approach: a case study in the
Tertiary basin of Puy-en-Velay (massif central, France). Geomorphology, 216, 208+224.
Oliveira S.C., Zezere J.L., Lajas S., Melo R. (2017) – Combination of statistical and physically based methods to assess
shallow slide susceptibility at the basin scale. Nat. Hazards Earth Syst. Sci., 17, 1091-1109.
Bera A., Mukhopadhyay B.P., Das D. (2019) - Landslide hazard zonation mapping using multi-criteria analysis with
the help of GIS techinique: a case study from Eastern Himalayas, Namchi South Sikkim. Nat. Haz., 1-25.
Aims and innovations of the study have been further highlighted both in Section 1 (Introduction) and in Section 5
(Conclusion).
Other specific comments are below:
Lines 35-37: "...[8-16]" -- It is improper (even ridiculous) to cite so many references after a sentence about
broad background. Similar problems in some other places in the manuscript.
We revised and reduced the number of references, as suggested.
Lines 307-309: More detailed evidence are needed for this presentation. Fig. 9 does show this point.
As described in section 2.2, we gathered landslide information from different sources, first of all newspapers articles and
chronicles notes from local and social media, damage reports and catalogues compiled by different local and regional
authorities, etc. Consequently, data included in the analysis are heterogeneous, incomplete and relatively unstructured.
First of all, we have no areal information about landslide size and we georeferenced landslides as punctual features. By
evaluating the spatial distribution of punctual data on the lithological map, we observed that in 35% of cases punctual
features were localized at the contact between different rock formations or at the boundaries of thick slope deposits.
Having no quantitative data about areal size of landslides and their source areas, and detailed evidences to support the
qualitative presentation (ex L307-309), we prefer to eliminate the paragraph, also to avoid a misunderstanding about
the type of gathered data.
Fig 9d (incorrect figure number on Figure) and the corresponding discussion are not clearly matched (e.g.,
type index is used in Figure whereas type names are used in the text).
Figure number (D) and the corresponding labels (type names in place of type indexes) have been corrected.
Furthermore, we preferred to use a different kind of graph (pie charts in place of histograms) in order to optimize the
layout of the figure and to highlight categories with very low percentages.
Table 5: Rationality of the reclassification, especially for aspect, is not convincing.
As described in section 2.2, we reclassified predisposing geological, morphological and environmental factors into four
relevant classes, from low to very high, based on the statistical distribution of punctual instability processes in each
category of the corresponding thematic layers (lithology, aspect, steepness and land-use). To define score ranges in each
class, we analyzed the distribution of the percentage values obtained from the spatial analysis and divided them into
four classes according to the incidence of landslide in the different categories.
Lines 407-409: The landslide susceptibility map result, as well as the classification parameter-settings, has
not been convinced by current evaluation.
As described in section 2.2 and underlined in the following sections, we have no information about areal size and
spatial distribution of the landslides but only punctual and heterogeneous data due to the peculiarity of the source
information (mainly qualitative notes or reports).Consequently, we cannot assess a conventional susceptibility analysis
and map: we produced a simplified, reproducible but effective GIS-based approach to detect areas characterized by high
proneness to landsliding, performed by combining punctual landslides and damage data to predisposing factors, such as
lithology, slope morphology, land-use, proximity of drainage pattern and anthropic elements (roads, buildings, manmade
structures, etc.). Classification of the parameter-settings is based on the statistical analysis of the spatial
distribution of landslides in each category of the thematic layers, in order to evaluate the incidence of the predisposing
factors on slope failures.
Limitations of the resulting map of the landslide prone areas are the lower accuracy than the conventional susceptibility
map and the incomplete information about the instability processes (size, type, state of activity, volume of involved
material, etc.).However, the adopted method, based on data easy to find and a simplified statistical approach, has
significant assets: compared to the landslides classified in the land-planning, the resulting map includes updated
information and it represents a very dynamic system, which can be easily andregularly updated every time it occurs a
rainfall event able to trigger shallow landslides.
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
The manuscript has been revised according to comments on the former version, to a certain extent.
Note that in the response to my main comment, "...but no one has dealt with this issue in this area of Italy" is not a persuasive reason for this study as a scientific contribution. The use of a new data source for landslide inventory could be. Under this situation, the presentation of the method design should be focused on how to create the inventary from this data source, and how to control the inventary quality. And the result discussion should correspondingly focus on analyzing if above processing really worked in this study area. This part in current, very long manuscript is very weak. For example, only the first paragraph of Section "2.2 Methods" is about how to create the inventary from this data source. This is what I actually said in my main comment.
Author Response
The manuscript has been revised according to comments on the former version, to a certain extent.
Note that in the response to my main comment, "...but no one has dealt with this issue in this area of Italy" is not a persuasive reason for this study as a scientific contribution. The use of a new data source for landslide inventory could be. Under this situation, the presentation of the method design should be focused on how to create the inventory from this data source, and how to control the inventory quality. And the result discussion should correspondingly focus on analyzing if above processing really worked in this study area. This part in current, very long manuscript is very weak. For example, only the first paragraph of Section "2.2 Methods" is about how to create the inventory from this data source. This is what I actually said in my main comment.
We highlight the use of new data sources to create a landslides inventory in the study area as a novelty and a scientific contribution, particularly in section 4 (discussion) and 5 (conclusion).
In Section 2.2 we implemented the presentation of the method used to create the inventory from new data sources and to test the quality of the inventory.In particular, most of the landslide news have been acquired from online networks using both manual search and automatic web feed aggregators of news, e.g., Google News or Google Alert. Search has been performed by combining specific keywords, such as “landslide”, “shallow landslide” related to other specific arguments, such as “rainfall” or “damage”, and their Italian synonyms and words in both singular and plural form. Each news has been accurately checked because the term “frana” (Italian word for landslide) can be used with a different meaning. For example, “A scuolaerounafrana” means “I was rubbish at numbers at school”. Furthermore, we exclude news occurred out of the considered temporal range (2000-2017), because some news can be focused on mitigation actions relate to older landslides. Regard geolocating the landslide news, when accurate information about location were available, we used latitude and longitude values quoted in technical reports to georeferenced the slope failures; otherwise, when toponyms, road marks, physiographic elements or sub-municipality localities were included in newspaper articles or damage reports, we identified and manually georeferenced the approximate location of the instability processes using regional technical maps or modern cartographical platforms (e.g., Google Earth, Google Maps and Google Street View), resulting in a poor localization accuracy.
Furthermore, a briefly analysis of the geodatabase has been described in the section 3 (results). We clarified that when landslide news from newspapers and other online networks match to the the georeferenced and more accurate notifications from local authorities, we preferred the latter at the expense of the online news.
Author Response File: Author Response.pdf
