Submit to this Journal Review for this Journal Propose a Special Issue

Article Menu

Share Help Cite Discuss in SciProfiles

Open AccessArticle

Peer-Review Record

Geospatial Analysis of Mass-Wasting Susceptibility of Four Small Catchments in Mountainous Area of Miyun County, Beijing

Int. J. Environ. Res. Public Health 2019, 16(15), 2801; https://doi.org/10.3390/ijerph16152801

by Chen Cao¹, Jianping Chen¹, Wen Zhang^1,*, Peihua Xu¹, Lianjing Zheng² and Chun Zhu³

Reviewer 1: Anonymous

Reviewer 2: Anonymous

Int. J. Environ. Res. Public Health 2019, 16(15), 2801; https://doi.org/10.3390/ijerph16152801

Submission received: 6 June 2019 / Revised: 31 July 2019 / Accepted: 3 August 2019 / Published: 6 August 2019

Round 1

Reviewer 1 Report

1. CONTENT

The current study presents two statistical modelling approaches, i.e. the "frequency ratio" (FR) and the "information value" (IV) methods, to map flash flood susceptibility in a small (about 10 km2) mountainous area (near Beijing), which consisted of four catchments. In this area, 50 flash flood events were identified (during 2012-2015). The major part (70%) of these events was selected to fit the models, and the remaining part was used for model validation. Nine semi-quantitative parameters were chosen as input for both models (elevation, slope, curvature, stream power index, topographic wetness index, lithology, land use, soil type and a criterion about heavy rains). Both models were found to provide satisfactory results, although the FR model was performing slightly better.

2. MAIN COMMENTS

2.1. Objetive of the study: what is new in the proposed methodology ?

Lines 75-80: it should be clearly explained what is new in the study, compared to similar ones (e.g., Youssef et al. 2016).

Lines 62-80: it should be also clearly justified why the FR and the IV methods have been chosen.

2.2. Methodology: how did you define "flash floods" ?

Line 110: please, be more specific about what is meant by "flash flood" in the study: was it any flood observed in the small mountainous catchments, or a flood with more specific characteristics (or impact) ?

Line 110: in addition, did you define the flood "location" as the upper part of a stream where a flood ocurred, or did you use another criterion ?

Line 110: do you have data about the intensity of the studied 50 flash floods (such as: estimates of the peak discharge or of the maximum stage) ? If so, please report these data.

2.3. Discussion: were all the "contributing factors" useful ?

Line 223-228: the usefullness of each "conditionning factor" in the statistical modelling of the studied area should be discussed.

Line 124 (Table 1): considering that flash flood are produced by local storms, a special attention should be paid to the representativity -or not- of the heavy rains recorded at Beijing (how far is it from the studied area ?) and during 2006-2010 (whereas the studied flash floods ocurred in 2012-2015).

Line 161: it should be observed that the two "classes" of heavy rain are very similar in practice.

Lines 313-314: at first glance, I do not see any effect of the chosen "heavy rain" criterion on the ocurrence of flash floods in the studied area (see Table 2); do you disagree ? Otherwise, it would be irrelevant to mention the category of "12-14 short-term heavy precipitation frequency" as a conditionning factor of flash flooding in the studied area (see: Line 360).

Lines 337-338: the question arise to know if the "better performance" of the FR model compared to the IV model is significant or not.

Suggestion: instead of using rain data (that is: the number of heavy rains per year in Beijing) which are probably not representatve of the local storms that have produced the studied flash floods, why not using instead the "month of the year" as input for your statistical modelling (assuming that there is a marked rainy season in the studied area) ?

2.4. Discussion: avoid comments which are not about the obtained results

Line 292-298: this paragraph is not a part of the Discussion, and should be moved to the Results section.

Line 304-307: how usefull are the comments about the gneiss and mica properties for the study ?

Line 308-31: comments about the risk of destruction of farmland terraces is not a part of the Discussion, but of the Introduction (i.e., it is a motivation for the study).

Line 316-335: comments about the need to protect farmland terraces and to take mitigation decisions is not a true part of the Discussion.

Lines 337-351: this paragraph should be shortened, because it contains many repetitions and obvious statements (e.g., the sentence "Because of FR’s strong predicting ability, its result would be better" at Line 338 is obvious).

Lines 361-363: the statment "the existing farmland terraces in the four catchments could retain water contributing to flash flood hazard alleviation. In turn, it also suffers damages from flash flood, local residential should pay more attention on theses farmland terraces" is not a finding of the study.

2.5. Presentation: improve the manuscript organization and check data

Line 96: please, check the numbering of all sections (e.g., "1.1" instead of "2.1").

Line 162 (Figure 3): there is no comment in the text about "land use".

Line 207: comments about Figures 4 and 5 should belong to the "Methodology" section.

Line 213: do you mean "691-846 m" instead of "391-846 m" ?

Line 358: do you mean "846 m" instead of "749 m" ?

2.6. Presentation: avoid exaggerations, repetitions and obvious statements

Line 13: avoid exagerations (i.e., "vital" instead of "very vital")

Line 35: avoid exagerations (i.e., "large" instead of "uncountable")

Line 109: avoid exagerations (i.e., omit "trustworthy").

Line 368: avoid exaggerations ("can be useful" instead of "can be very useful")

Line 372: avoid exaggerations ("it could be useful" instead of "it could be very vital")

Line 36: omit the obvious sentence "It is not easy to predict the flash flood because of its complex condition".

Lines 40-45: please, be more concise (i.e., the importance of topography is repeated several times).

Lines 372-373: avoid repetitions (i.e., omit the last sentence of the Conclusion)

3. MINOR COMMENTS

Line 7: "Technology" instead of "Techonlogy".

Line 18: do you mean "randomly" instead of "stochastically" ?

Line 36: "floods are" instead of "floods is".

Lines 46: omit "would".

Line 94: please, add to the title: "and location of the flash floods identified during 2012-2015".

Line 101: omit "Based on".

Line 124 (Table 1): "soil" instead of "soli".

Line 136: "the water infiltrates easier into the soil" instead of "the water is easier to infiltrate into the underground".

Line 136: what do you mean by "surface run-off infiltration" ?

Line 241: omit "As".

Line 243: omit "as shown in Table 2" (it was said at the beginnig of section).

Line 270-276: please, only express the value of the AUC-criterion as a decibal number (that is: do not use percentages, which is redundant).

Line 281: what do you mean by "The main channel gradients of small catchments are mostly large in this study area" ? (e.g., do you refer to the stream slope ?)

Line 282: do you mean "stream power" ?

Line 286: "flash floods mainly ocurred" instead of "there mainly occurs flash flood"... What do you mean with this statement ?

Line 302: what do you mean by "in principle" ?

Lines 315-316: what do you mean by "the bottom of the four channels are eliminated as very high and high possibility to be inundated" ?

Line 357: what do you mean by "evidences showed that there mainly occurred flash flood in the four catchments" ?

Line 409: what is "Giscience" ?

Author Response

Dear Reviewer:

Thank you very much for your comments concerning our manuscript entitled “Spatial Analysis of Flash Flood Susceptibility Assessment in Small-catchment Scale: Case of Mountainous area in Miyun County, Beijing” (IJERPH-532946). Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our researches. We have studied comments carefully and have made correction which we hope meet with approval. We also have sent the manuscript to MDPI English service for extensive English editing. Revised portion are marked in red in the paper. The main corrections in the paper and the responses to your comments are as following:

1. CONTENT

The current study presents two statistical modelling approaches, i.e. the "frequency ratio" (FR) and the "information value" (IV) methods, to map flash flood susceptibility in a small (about 10 km2) mountainous area (near Beijing), which consisted of four catchments. In this area, flash flood events were identified (during 2012-2015). The major part (70%) of these events was selected to fit the models, and the remaining part was used for model validation. Nine semi-quantitative parameters were chosen as input for both models (elevation, slope, curvature, stream power index, topographic wetness index, lithology, land use, soil type and a criterion about heavy rains). Both models were found to provide satisfactory results, although the FR model was performing slightly better.

2. MAIN COMMENTS

2.1. Objetive of the study: what is new in the proposed methodology?

Comment：Lines 75-80: it should be clearly explained what is new in the study, compared to similar ones (e.g.,Youssef et al. 2016).

Response：Thank you for your valuable suggestion. We have added contents in section Introduction:

“Farmland terraces are abundant in catchments, and their structural strengths are very low. In the event of a flash flood, small landslides will form, and large numbers of terraces have a high risk of being damaged. Thus, it is very important to highlight flash flood protection and adaptation approaches for agricultural areas to minimize the consequences of flash flood hazards due to different human activities and climate change conditions. On the other hand, technical measures, such as farmland terraces, can be used for soil and water conservation. To a certain extent, these terraces could be an approach to intercepting an oncoming flash flood event; in turn, the terraces would sustain damages. Terrace stone walls continued to be reconstructed by local residents when erosions or landslides occurred in the past. Thus, field surveys have found that most farmland terraces have been well maintained. Studies have been conducted in middle and low farmland areas to establish relationships between farmland maintenance and floods on a sub-catchment scale [48]. Because forests can prevent the appearance of a flash flood, protecting forestland is imperative [49]. Farmland terraces, which can retain water in catchments, also contribute to the alleviation of flash flood hazards. Since farmland terraces also suffer damages from flash floods, local residents should pay more attention to them. Considering that there are many farmland terraces in mountainous areas in Beijing, especially the intermediate- and low-elevation areas, appropriate flash flood management plans for these areas are vital.

This study aims to determine the spatial probability of flash flood-affected hazard occurrence in four catchments. The correlation between influence factors and flash flood-affected hazard occurrence is identified, and the accuracy is evaluated. Furthermore, the present work also conducts a comparative assessment of two statistical models used for flash flood hazard susceptibility mapping: the frequency ratio (FR) model and information value model (IVM). The FR and IVM methods were selected for their mathematical simplicity, their ability to extract data in a limited time period, and their effectiveness. The prediction accuracy and performance of each method were assessed using four catchments in a mountainous area. Nine flash flood hazard susceptibility factors were used in the two models using GIS software. The results were validated using the area under the receiver operating characteristic curve (ROC) method.”

Comment：Lines 62-80: it should be also clearly justified why the FR and the IV methods have been chosen.

Response：Thank you for your valuable suggestion. We have added contents in section Introduction:

“The FR method has been proven to be effective, and it has been successfully applied to flash flood hazard susceptibility mapping and landslide susceptibility mapping [22,44,45]. In view of the effectiveness of the FR method, it was selected as a statistical method in the present study to better explore the effect of different mapping units on the susceptibility mapping of debris flow. Furthermore, the information value model, which has been proved to be a very useful method for measuring the degree of influence of each causative factor, is a simple probabilistic bivariate statistical method whose accuracy is acceptable [46,47].”

2.2. Methodology: how did you define "flash floods"?

Comment：Line 110: please, be more specific about what is meant by "flash flood" in the study: was it any flood observed in the small mountainous catchments, or a flood with more specific characteristics (or impact)?

Response：Thank you very much for your suggestions. We have added contents in this part.

“2.1 Identifying locations of flash flood hazards in 2012–2015

The hazards caused by a future flash flood can be estimated by analyzing past records. An inventory map can show the distribution and characteristics of hazards caused by a flash flood in the study area [50]. The mapping of hazards caused by flash flooding in the four catchments is necessary for depicting the relationship between susceptibility ranges and influence factors. Extensive field investigation and observations were conducted to produce a comprehensive and reliable inventory map. The flash flood hazard inventory map shows the spatial distribution of hazards that were caused by flash flooding in the study area. This was used as a base map to generate the flash flood hazard susceptibility map. We analyzed records of flash flood hazards to identify sensitive areas that were prone to destruction by flash flooding. The 21 July 2012 flash flood, which damaged roads, slopes, houses, farmland terraces, etc., was an important previous flash flood event in this study. Between 2012 and 2015, some flash floods occurred in the four catchments. The inventory map was first created by locating flash flood hazards in the four catchments using documents and detailed field surveys. A good source of information includes interviews with local residents, which were conducted to identify destroyed houses and public facilities damaged by the flash floods that occurred between 2012 and 2015. Because the storm and flash flood on 21 July 2012 left local residents with significant impressions, they give this event special attention. Field surveys confirmed landslides, collapses, and erosion due to flash flooding. The farmland terraces are mainly located at the bottom of the catchment and partly on the hillside, so they are highly vulnerable to flooding and easily damaged. Identifying the effects of flash flooding is fairly straightforward.

Comment：Line 110: in addition, did you define the flood "location" as the upper part of a stream where a flood ocurred, or did you use another criterion?

Response: Thank you very much for your suggestions. Actually, the hazard locations means hazards damaged by flash flood. Thus, the accurate defining of “locations” should be “flash flood-affected hazard locations”, we have revised theses phrases in the whole manuscripts. And we also added the defining of “flash flood-affected hazard locations”.

Comment：Line 110: do you have data about the intensity of the studied 50 flash floods (such as: estimates of the peak discharge or of the maximum stage)? If so, please report these data.

Response：Thank you for your valuable comments. In this study, we focus on a study of the areas prone to be damaged by the occurrence of flash floods. We are sorry for the misleading of the title of the study. And we have revised the title as “Geospatial Analysis of Flash Flood Hazard Susceptibility in Small-catchment Scale: Case of Four Catchments in Mountainous area, Miyun County, Beijing”

2.3. Discussion: were all the "contributing factors" useful?

Comment：Line 223-228: the usefullness of each "conditionning factor" in the statistical modelling of the studied area should be discussed.

Response： Thank you for your reminding. We have added contents in this part: “For the lithology factors, the highest IV value is 0.482 for gneiss. The IVs of the other lithologies, including quartzite, diorite, and acid rock, are 0.062, −0.0114, and −1.127, respectively. The IV of farmland is 0.613, which is the only positive value for the land use conditioning factor. Farmland is prone to hazard exposure. The IV of forest and construction land are −0.038 and −1, respectively. In terms of soil type, the IVs of cinnamon soil and brown soil are 0.118 and −0.091, respectively. Higher flow accumulation has higher IVs, i.e., 1.218 (351651–622719 subclass) and 0.377 (117217–351651 subclass). ”

Comment：Line 124 (Table 1): considering that flash flood are produced by local storms, a special attention should be paid to the representativity -or not- of the heavy rains recorded at Beijing (how far is it from the studied area ?) and during 2006-2010 (whereas the studied flash floods ocurred in 2012-2015).

Response：It is true as you suggested. Using rainfall data during 2006-2010 is not correct. And we have checked the whole paper and replaced the short-time heavy rain (STHR) by flow accumulation. And we have also added the picture of flow accumulation in Figure.4 and detail data in Figure 5, Table 1 and Table 2. Besides, we updated the results of the susceptibility map of Figure 7 and Figure 8, as well as validation results of Figure 9.

We added contents:

“This study applied flow accumulation as an influence factor. The basic idea is that the DEM represented by regular grids has a unit of water at each point. Natural water flows from a high point to a low point, and the amount of water that flows through each point depends on the flow direction. The convergence of each grid shows the flow accumulation and reflects the amount of water in each grid in the area. The flow accumulation map is shown in figure 4i. Flow accumulation was categorized into nine subclasses: (1) 0–2442, (2) 2442–4884, (3) 4884–9768, (4) 9768–14652, (5) 14652–26862, (6) 26862–46398, (7) 46398–117217, (8) 117217–351651, and (9) 351651–622719.”

We have added the results description of flow accumulation in Section 4.1 and Section 4.2.

Section 4.1 :“Higher flow accumulation has higher FR values, i.e., 3.382 (351651–622719 subclass) and 1.458 (117217–351651 subclass). ”.

Section 4.2 : “Higher flow accumulation has higher IVs, i.e., 1.218 (351651–622719 subclass) and 0.377 (117217–351651 subclass).”

Comment：Line 161: it should be observed that the two "classes" of heavy rain are very similar in practice.

Response：It is really true as what you commented. We have deleted the influence factor of short-time heavy rain. And we use flow accumulation as one of the influence factors. Detailed revision can be seen in last response.

Comment：Lines 313-314: at first glance, I do not see any effect of the chosen "heavy rain" criterion on the ocurrence of flash floods in the studied area (see Table 2); do you disagree ? Otherwise, it would be irrelevant to mention the category of "12-14 short-term heavy precipitation frequency" as a conditionning factor of flash flooding in the studied area (see: Line 360).

Comment：Lines 337-338: the question arise to know if the "better performance" of the FR model compared to the IV model is significant or not.

Response：We have added contents in this section: “Their AUCs are very close, with the frequency ratio method slightly more accurate and more applicable than the information value method for defining flash flood hazard susceptibility classes.”

Comment：Suggestion: instead of using rain data (that is: the number of heavy rains per year in Beijing) which are probably not representative of the local storms that have produced the studied flash floods, why not using instead the "month of the year" as input for your statistical modelling (assuming that there is a marked rainy season in the studied area) ?

Response：It is really true as what you commented. Instead of using rain data, the “month of the year” is really a good choice. It may be adequate for large-scale susceptibility mapping. In this study, the four catchments are not so large. The four catchments are small, where there is no marked rainy season. The rain mainly uniform in the whole catchment. Thus, we have deleted the influence factor of short-time heavy rain according to your suggestion. And we use flow accumulation as one of the influence factors. Detailed revision can be seen above.

2.4. Discussion: avoid comments which are not about the obtained results

Comment：Line 292-298: this paragraph is not a part of the Discussion, and should be moved to the Results section.

Response：We have moved this part to Results section.

Comment：Line 304-307: how usefull are the comments about the gneiss and mica properties for the study?

Response：We have deleted this sentence. “In study area, the area formed with gneiss was more likely to occur hazard. Gneiss is formed by deep metamorphism of magmatic rocks or sedimentary rocks. When mica content is high, compressive strength of rock decreases. Shear strength is lower along the gambling direction.”

Comment：Line 308-31: comments about the risk of destruction of farmland terraces is not a part of the Discussion, but of the Introduction (i.e., it is a motivation for the study).

Response：Thank you for your constructing advice. It really true that risk of destruction of farmland terraces is a motivation for the study. And we have moved this part to Instruction section.

The revised paragraph is “Farmland terraces are abundant in catchments, and their structural strengths are very low. In the event of a flash flood, small landslides will form, and large numbers of terraces have a high risk of being damaged. Thus, it is very important to highlight flash flood protection and adaptation approaches for agricultural areas to minimize the consequences of flash flood hazards due to different human activities and climate change conditions. On the other hand, technical measures, such as farmland terraces, can be used for soil and water conservation. To a certain extent, these terraces could be an approach to intercepting an oncoming flash flood event; in turn, the terraces would sustain damages. Terrace stone walls continued to be reconstructed by local residents when erosions or landslides occurred in the past. Thus, field surveys have found that most farmland terraces have been well maintained. Studies have been conducted in middle and low farmland areas to establish relationships between farmland maintenance and floods on a sub-catchment scale [48]. Because forests can prevent the appearance of a flash flood, protecting forestland is imperative [49]. Farmland terraces, which can retain water in catchments, also contribute to the alleviation of flash flood hazards. Since farmland terraces also suffer damages from flash floods, local residents should pay more attention to them. Considering that there are many farmland terraces in mountainous areas in Beijing, especially the intermediate- and low-elevation areas, appropriate flash flood management plans for these areas are vital.”

Comment：Line 316-335: comments about the need to protect farmland terraces and to take mitigation decisions is not a true part of the Discussion.

Response：We have moved this part to Instruction section.

Comment：Lines 337-351: this paragraph should be shortened, because it contains many repetitions and obvious statements (e.g., the sentence "Because of FR’s strong predicting ability, its result would be better" at Line 338 is obvious).

Response：Your suggestion is valuable. And we have deleted obvious statements (e.g., (a) In this study, it was concluded that the performance of FR method outperformed IV model. (b) Because of FR’s strong predicting ability, its result would be better. (c) Both of the two models for the estimation of flash flood hazard areas can be useful tools for the mitigation of the devastating impact of flash floods.)

This paragraph has been revised as “The results show that FR performs better than IV. Their AUCs are very close, with the frequency ratio method slightly more accurate and more applicable than the information value method for defining flash flood hazard susceptibility classes. Researchers have explained that the FR can be used as a supporting method to determine the importance sequence of factors in modeling [75]. However, because the AUC values are close, it cannot be definitively concluded that one model should be selected over the other. The approach to flash flood hazard susceptibility mapping should be applicable for a specific catchment. There is no consensus on the general guideline for selecting flash flooding susceptibility influencing factors. Therefore, in this study, the selection and the number of types of flood susceptibility factors were determined from the characteristics of the geological environment in the four catchments. The influence factors to include in the study should be characteristic of the study area and are easily selected. Thus, it is worth trying different combinations of influence factors in future works, with the hope that more practical and precise susceptibility maps for flash flood hazard management can be achieved.”

Comment：Lines 361-363: the statement "the existing farmland terraces in the four catchments could retain water contributing to flash flood hazard alleviation. In turn, it also suffers damages from flash flood, local residential should pay more attention on theses farmland terraces" is not a finding of the study.

Response：Thank you for your suggestion. We have deleted this sentence.

2.5. Presentation: improve the manuscript organization and check data

Comment：Line 96: please, check the numbering of all sections (e.g., "1.1" instead of "2.1").

Response：We have checked the numbering of all sections: “1 Introduction; 2 2 Study area and inventory maps; 3. Methodology; 4. Results; 5. Discussion; 6. Conclusion”.

Comment：Line 162 (Figure 3): there is no comment in the text about "land use".

Response：Thank you for you suggestion. We have added “The land use type influences infiltration, water convergence, and the relationship between the surface water and groundwater. Different vegetation types have different capacities of rainfall interception and water storage. The type of vegetation also affects the time and size of water confluence.”.

Comment：Line 207: comments about Figures 4 and 5 should belong to the "Methodology" section.

Response：It is really true as you suggested. We have moved comments about Figures 4 and 5 to the “Methodology” section.

Comment：Line 213: do you mean "691-846 m" instead of "391-846 m" ?

Response：Thank you for your kind reminding. We have revised this.

Comment：Line 358: do you mean "846 m" instead of "749 m" ?

Response：Thank you for your kind reminding. We have revised this.

2.6. Presentation: avoid exaggerations, repetitions and obvious statements

Comment：Line 13: avoid exagerations (i.e., "vital" instead of "very vital")

Response：We have used “vital” instead of “very vital”.

Comment：Line 35: avoid exagerations (i.e., "large" instead of "uncountable")

Response：We have used “large” instead of “uncountable”.

Comment：Line 109: avoid exagerations (i.e., omit "trustworthy").

Response：We have omitted “trustworthy”.

Comment：Line 368: avoid exaggerations ("can be useful" instead of "can be very useful")

Response：We have used “can be useful” instead of “can be very useful”.

Comment：Line 372: avoid exaggerations ("it could be useful" instead of "it could be very vital")

Response：We have used “it could be useful” instead of “it could be very vital”.

Comment：Line 36: omit the obvious sentence "It is not easy to predict the flash flood because of its complex condition".

Response：We have omitted the obvious sentence.

Comment：Lines 40-45: please, be more concise (i.e., the importance of topography is repeated several times).

Response：We have deleted repeated part, and revised it as: “Topography plays an important role in the flash floods through a basic interaction involving elevation across multiple spatial and temporal scales [3,4].”

Comment：Lines 372-373: avoid repetitions (i.e., omit the last sentence of the Conclusion)

Response：We have omitted the last sentence of the Conclusion.

3. MINOR COMMENTS

Comment：Line 7: "Technology" instead of "Techonlogy".

Response：We have corrected "Technology" instead of "Techonlogy"..

Comment：Line 18: do you mean "randomly" instead of "stochastically" ?

Response：We have used “randomly” instead of “stochastically” in Line 18. We also have corrected this error in Line 110.

Comment：Line 36: "floods are" instead of "floods is".

Response：We have used "floods are" instead of "floods is".

Comment：Lines 46: omit "would".

Response：We have omitted this word.

Comment：Line 94: please, add to the title: "and location of the flash floods identified during 2012-2015".

Response：We have added this in the title of section 2.1.

Comment：Line 101: omit "Based on".

Response：We have omitted “Based on”

Comment：Line 124 (Table 1): "soil" instead of "soli".

Response：We have corrected this error.

Comment：Line 136: "the water infiltrates easier into the soil" instead of "the water is easier to infiltrate into the underground".

Response：We have used "the water infiltrates easier into the soil" instead of "the water is easier to infiltrate into the underground".

Comment：Line 136: what do you mean by "surface run-off infiltration" ?

Response：It has been revised as “The surface run-off and water velocity are controlled by the slope angle”.

Comment：Line 241: omit "As".

Response：We have omitted “As”.

Comment：Line 243: omit "as shown in Table 2" (it was said at the beginnig of section).

Response：We have omitted “as shown in Table 2”.

Comment：Line 270-276: please, only express the value of the AUC-criterion as a decibal number (that is: do not use percentages, which is redundant).

Response：We have corrected the expression format according to your comment.

Comment：Line 281: what do you mean by "The main channel gradients of small catchments are mostly large in this study area" ? (e.g., do you refer to the stream slope ?)

Response：We have revise this sentence as: “The slopes of the main channels are mostly large.”

Comment：Line 282: do you mean "stream power" ?

Response：We have revise this sentence as: “The impact of a flash flood is strong enough to destroy vegetation, roads, farmland terraces, etc.”

Comment：Line 286: "flash floods mainly ocurred" instead of "there mainly occurs flash flood"... What do you mean with this statement ?

Response：We have revised it as: “Using the results of the field investigation, we find that flash floods tend to occur in the study area rather than debris flows.”

Comment：Line 302: what do you mean by "in principle" ?

Response：Sorry for the mistake. We have deleted this phrase.

Comment：Lines 315-316: what do you mean by "the bottom of the four channels are eliminated as very high and high possibility to be inundated" ?

Response：We have revised it as: “It can be seen in figure 7 and figure 8 that the channels of the catchments are divided into areas with very high and high susceptibility, and these are areas prone to damage by flash floods.”

Comment：Line 357: what do you mean by "evidences showed that there mainly occurred flash flood in the four catchments" ?

Response：We have revised this sentence as: “Field surveys confirmed the previous occurrence of landslides, collapses, and erosion due to flash flooding. Farmland terraces are mainly located at the bottom of catchments and partly on the hillside, making them vulnerable to flooding and damage.”

Comment：Line 409: what is "Giscience" ?

Response：It should be GIScience & Remote Sensing, whose ISSN is 1548-1603

We tried our best to improve the manuscript and made some changes in the manuscript. And we marked the changed parts in red in revised paper. We appreciate for Editor and Reviewers’ warm work earnestly, and hope that the correction will meet with approval. Once again, thank you very much for your comments and suggestions.

Yours sincerely,

Wen Zhann, Ph.D.

College of Construction Engineering, Jilin University

938 Ximinzhu Road, Changchun 130026, China

Phone number: +86 13604402994 E-mail: [email protected]

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript deals with an interesting subject such as flash floods. However, I think that to start the title of the manuscript does not reflect well the work done. For the title of the manuscript proposes an analysis of susceptibility to the occurrence of flash-floods, while the analysis performed (in my opinion) focuses on a study of the areas prone to damage by the occurrence of flash-floods. Something that in my opinion is not similar.

Other minor aspects have been indicated in the text of the attached document.

However, there is a main aspect that raises doubts about the analysis made by the authors, and that in my opinion questions or questions the subsequent Discussion and Conclusions obtained from this manuscript. This weak point of the analysis, in my opinion, is the field evidence used by the authors to generate the inventory map, which in my opinion is not always correct. For example, the authors propose the use of evidence of breakage of retaining walls and also of small superficial landslides, as events associated with the occurrence of flash floods, and in my opinion this does not have to be correct; There is no direct and univocal relationship between these evidences in the field and the occurrence of flash floods. In fact, these evidences can occur perfectly if the occurrence of flash-floods. In addition, the spatial distribution of inventory map data can very significantly condition the results obtained. This is therefore an aspect in which the authors should work seriously and devote sufficient effort to define and locate another series of field evidences that can be directly and unequivocally related to the occurrence of flash-floods.

Comments for author File: Comments.pdf

Author Response

Dear Reviewer:

Thank you very much for your comments concerning our manuscript entitled “Spatial Analysis of Flash Flood Susceptibility Assessment in Small-catchment Scale: Case of Mountainous area in Miyun County, Beijing” (IJERPH-532946). Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our researches. We have studied comments carefully and have made correction which we hope meet with approval. Revised portion are marked in red in the paper. We also have sent the manuscript to MDPI English service for extensive English editing. The main corrections in the paper and the responses to your comments are as following:

Comments and Suggestions for Authors

1. Comment：The manuscript deals with an interesting subject such as flash floods. However, I think that to start the title of the manuscript does not reflect well the work done. For the title of the manuscript proposes an analysis of susceptibility to the occurrence of flash-floods, while the analysis performed (in my opinion) focuses on a study of the areas prone to damage by the occurrence of flash-floods. Something that in my opinion is not similar.

Response: Thank you for your comments. It is really as what you commented. We focus on a study of the areas prone to be damaged by the occurrence of flash floods. We are sorry for the misleading of the title of the study. And we have revised the title as “Geospatial Analysis of Flash Flood Hazard Susceptibility in Small-catchment Scale: Case of Four Catchments in Mountainous area, Miyun County, Beijing”

2. Comment：Other minor aspects have been indicated in the text of the attached document.

Response: Thank you very much for your suggestions. We have revised the minor aspects you suggested carefully. The details are shown as follows.

3. Comment：However, there is a main aspect that raises doubts about the analysis made by the authors, and that in my opinion questions or questions the subsequent Discussion and Conclusions obtained from this manuscript. This weak point of the analysis, in my opinion, is the field evidence used by the authors to generate the inventory map, which in my opinion is not always correct. For example, the authors propose the use of evidence of breakage of retaining walls and also of small superficial landslides, as events associated with the occurrence of flash floods, and in my opinion this does not have to be correct; There is no direct and univocal relationship between these evidences in the field and the occurrence of flash floods. In fact, these evidences can occur perfectly if the occurrence of flash-floods. In addition, the spatial distribution of inventory map data can very significantly condition the results obtained. This is therefore an aspect in which the authors should work seriously and devote sufficient effort to define and locate another series of field evidences that can be directly and unequivocally related to the occurrence of flash-floods.

Response: Thank you for you valuable suggestions, we have added contents in the manuscript:

2.1 Identifying locations of flash flood hazards in 2012–2015

From an inventory map, a flash flood hazard susceptibility map can be produced. A flash flood hazard map was generated using a previous inventory map and remote sensing images. 71 flash flood hazard locations were surveyed in the four catchments and were used in further analysis (Fig. 1). 50 flash flood locations were randomly selected to build and train the models. The remaining 21 flash flood locations were used as validation data.

We also have replaced the figure 2 using new photos.

We also have added contents in Discussion section:

“Pierson [75] stated that damage to vegetation due to debris flow has distinctive characteristics, such as severe destruction in thalwegs and mud coatings. On the other hand, flash flood damage to vegetation has other characteristics: typically light and irregular erosion of tree bark, concentration of damage near beds (from saltating bed material) and near (sometimes above) the maximum stage (from floating debris), and finer branches that are commonly bent but not broken or stripped. ”

Thank you very much for your kind reminding. Your suggestions give us a good guide in the future work. We will work seriously and devote sufficient effort to define and locate another series of field evidences that can be directly and unequivocally related to the occurrence of flash-floods.

4. Comment：：Line 38: “Flood flash” should be flash floods

Response: Thank you for your kind reminding. We have revised this error.

5. Comment：Line 42: I'm not sure about the meaning of this phrase and what the authors want to express

Response: We are sorry for the mistake. Thank you for your kind reminding. We have deleted this sentence according to Comment 6.

6. Comment： Line 44: Review the wording of the full paragraph. It gives the sensation of being unconnected phrases that are sustained in many cases on bibliographical references that have nothing to do with the theme of flash floods.

Response: Thank you for your kind reminding. We have deleted sentences that have nothing to do with the theme of flash floods.

7. Comment： Line 49: I'm not sure about the meaning of this phrase and what the authors want to express

Response: Thank you for your reminding. We have revised this sentence as “Thus, land use managers should be able to identify all aspects of landscape vulnerability in a potential flash flood.”

8. Comment： Line 51: “3S” You must define the meaning of 3S

Response: we have defined the meaning of “3S technology” according to your comment.

“Remote sensing (RS), geographic information systems (GISs), and the Global Positioning System (GPS) are now widely applied as so-called “3S technology”.”

9. Comment： Line 51-61: This part of the paragraph suffers from any continuity in the wording with the above, and also is not based on any bibliographic reference. Correct wording

Response: Thank you for your comment. We have rewrite this paragraph, and added several bibliographic references. The revised paragraph is:

“Remote sensing (RS), geographic information systems (GISs), and the Global Positioning System (GPS) are now widely applied as so-called “3S technology”. RS and GIS techniques have been applied for flash flood modeling [8,9], and appropriate assessment methods should be similarly applied for flash flood hazard susceptibility mapping [10]. In the 1990s, GIS-based flash flooding assessments were used for small catchments [11]. The use of GIS has greatly progressed in the field of environmental science in applications such as landslide and groundwater susceptibility mapping and flash flood hazard susceptibility mapping. These technologies can provide a good perspective for flash flood assessment research. Various flash flood hazard susceptibility maps have been created in different countries. The use of RS and GIS has increased significantly in response to the need for rapid data collection and improved flood bitmaps for commercial satellite products. Further, GIS is a useful tool for studying events with multidimensional behavior; for example, flash floods are investigated using a variety of spatial–temporal models. In order to obtain accurate results from these models, it is vital that the input factors retain their spatial associations [12].”

The added bibliographic references are:

8. Biswajeet, P.; Mardiana, S. Flood Hazrad Assessment for Cloud Prone Rainy Areas in a Typical Tropical Environment. Disaster Adv 2009, 2, 7-15.

9. Pradhan, B.; Hagemann, U.; Tehrany, M.S.; Prechtel, N. An easy to use ArcMap based texture analysis program for extraction of flooded areas from TerraSAR-X satellite image. Comput Geosci-Uk 2014, 63, 34-43, doi:10.1016/j.cageo.2013.10.011.

10. Cloke, H.L.; Pappenberger, F. Ensemble flood forecasting: A review. J Hydrol 2009, 375, 613-626.

11. Kim, E.S.; Choi, H.I. A method of flood severity assessment for predicting local flood hazards in small ungauged catchments. Nat Hazards 2015, 78, 2017-2033.

12. Sanyal, J.; Lu, X.X. Application of remote sensing in flood management with special reference to monsoon Asia: A review. Nat Hazards 2004, 33, 283-301.

10. Comment： Line 62-74: This paragraph is a summary of bibliographical references without a clear sense. Many techniques are proposed, but in many cases they have not been or are not applicable in the field of flash floods, but especially in the field of surface landslides, or in the field of debris flow.

Response: It is really true as you mentioned. We have revised the bibliographical references as what you suggested. The added references are as follows:

13. Tehrany, M.S.; Pradhan, B.; Jebur, M.N. Flood susceptibility mapping using a novel ensemble weights-of-evidence and support vector machine models in GIS. Journal of hydrology 2014, 512, 332-343.

14. Shafapour Tehrany, M.; Shabani, F.; Neamah Jebur, M.; Hong, H.; Chen, W.; Xie, X. GIS-based spatial prediction of flood prone areas using standalone frequency ratio, logistic regression, weight of evidence and their ensemble techniques. Geomatics, Natural Hazards and Risk 2017, 8, 1538-1561.

15. Costache, R.; Zaharia, L. Flash-flood potential assessment and mapping by integrating the weights-of-evidence and frequency ratio statistical methods in GIS environment–case study: Bâsca Chiojdului River catchment (Romania). Journal of Earth System Science 2017, 126, 59.

16. Jothibasu, A.; Anbazhagan, S. Flood susceptibility appraisal in Ponnaiyar River Basin, India using frequency ratio (FR) and Shannon’s Entropy (SE) models. Int J Adv Rem Sens GIS 2016, 5, 1946-1962.

17. Wang, Z.; Lai, C.; Chen, X.; Yang, B.; Zhao, S.; Bai, X. Flood hazard risk assessment model based on random forest. Journal of Hydrology 2015, 527, 1130-1141.

18. Lee, S.; Kim, J.-C.; Jung, H.-S.; Lee, M.J.; Lee, S. Spatial prediction of flood susceptibility using random-forest and boosted-tree models in Seoul metropolitan city, Korea. Geomatics, Natural Hazards and Risk 2017, 8, 1185-1203.

19. Muñoz, P.; Orellana-Alvear, J.; Willems, P.; Célleri, R. Flash-flood forecasting in an Andean mountain catchment—Development of a step-wise methodology based on the random forest algorithm. Water 2018, 10, 1519.

20. Nandi, A.; Mandal, A.; Wilson, M.; Smith, D. Flood hazard mapping in Jamaica using principal component analysis and logistic regression. Environmental Earth Sciences 2016, 75, 465.

21. Rasyid, A.R.; Bhandary, N.P.; Yatabe, R. Performance of frequency ratio and logistic regression model in creating GIS based landslides susceptibility map at Lompobattang Mountain, Indonesia. Geoenvironmental Disasters 2016, 3, 19.

22. Cao, C.; Xu, P.; Wang, Y.; Chen, J.; Zheng, L.; Niu, C. Flash flood hazard susceptibility mapping using frequency ratio and statistical index methods in coalmine subsidence areas. Sustainability 2016, 8, 948.

23. Kazakis, N.; Kougias, I.; Patsialis, T. Assessment of flood hazard areas at a regional scale using an index-based approach and Analytical Hierarchy Process: Application in Rhodope-Evros region, Greece. Science of the Total Environment 2015, 538, 555-563.

24. Stefanidis, S.; Stathis, D. Assessment of flood hazard based on natural and anthropogenic factors using analytic hierarchy process (AHP). Natural hazards 2013, 68, 569-585.

25. Vojtek, M.; Vojteková, J. Flood Susceptibility Mapping on a National Scale in Slovakia Using the Analytical Hierarchy Process. Water 2019, 11, 364.

26. Dano, U.L.; Balogun, A.-L.; Matori, A.-N.; Wan Yusouf, K.; Rimi Abubakar, I.; Mohamed, S.; Ahmed, M.; Aina, Y.A.; Pradhan, B. Flood susceptibility mapping using GIS-based analytic network process: A case study of Perlis, Malaysia. Water 2019, 11, 615.

27. Bui, D.T.; Ngo, P.-T.T.; Pham, T.D.; Jaafari, A.; Minh, N.Q.; Hoa, P.V.; Samui, P. A novel hybrid approach based on a swarm intelligence optimized extreme learning machine for flash flood susceptibility mapping. Catena 2019, 179, 184-196.

28. Wu, J.; Liu, H.; Wei, G.; Song, T.; Zhang, C.; Zhou, H. Flash Flood Forecasting Using Support Vector Regression Model in a Small Mountainous Catchment. Water 2019, 11, 1327.

29. Ahmadlou, M.; Karimi, M.; Alizadeh, S.; Shirzadi, A.; Parvinnejhad, D.; Shahabi, H.; Panahi, M. Flood susceptibility assessment using integration of adaptive network-based fuzzy inference system (ANFIS) and biogeography-based optimization (BBO) and BAT algorithms (BA). Geocarto International 2018, 1-21.

30. Khosravi, K.; Pham, B.T.; Chapi, K.; Shirzadi, A.; Shahabi, H.; Revhaug, I.; Prakash, I.; Bui, D.T. A comparative assessment of decision trees algorithms for flash flood susceptibility modeling at Haraz watershed, northern Iran. Science of the Total Environment 2018, 627, 744-755.

11. Comment： Line 75-80: I think it is necessary to significantly improve the introduction to the subject of study, putting more emphasis on those works that are really focused on the study of flash floods, their dynamics, their genesis and the hazards and risk associated with them.

Response: We have rewrite section Introduction.

“1 Introduction

12. 1 Introduction

13. Flash flooding is a common disaster that has occurred throughout anthropogenic development [1]. A flash flood often happens in a short time, with a high peak discharge due to rainfall occurring within an hour [2]. In addition to the people who suffer from these disasters, artificial structures can be destroyed by floods. Thus, the extent of flash flood damage is extensive.

14. Flash floods are always accompanied by secondary disasters, such as landslides, collapses, and erosions. These are known as flash flood-affected hazards, and they have negative consequences for people. Flash flood hazard maps, which are a very important component of catchment management, are mostly dependent on the topographic and geomorphic features of the region. Topography plays an important role in flash floods through basic interactions involving elevation across multiple spatial and temporal scales [3,4].

15. Anthropogenic activities, including farmland expansion and timber harvesting, change natural conditions, which increases the risk of flash flood hazard occurrence. Although flash flood hazards are inevitable, it is important to identify areas that are likely to be easily damaged in the event of a flash flood [5]. Thus, land use managers should be able to identify all aspects of landscape vulnerability in a potential flash flood. [6]. Flash flood hazard susceptibility mapping has been recognized as the first necessary step in flood prevention and its management [7].

16. Remote sensing (RS), geographic information systems (GISs), and the Global Positioning System (GPS) are now widely applied as so-called “3S technology”. RS and GIS techniques have been applied for flash flood modeling [8,9], and appropriate assessment methods should be similarly applied for flash flood hazard susceptibility mapping [10]. In the 1990s, GIS-based flash flooding assessments were used for small catchments [11]. The use of GIS has greatly progressed in the field of environmental science in applications such as landslide and groundwater susceptibility mapping and flash flood hazard susceptibility mapping. These technologies can provide a good perspective for flash flood assessment research. Various flash flood hazard susceptibility maps have been created in different countries. The use of RS and GIS has increased significantly in response to the need for rapid data collection and improved flood bitmaps for commercial satellite products. Further, GIS is a useful tool for studying events with multidimensional behavior; for example, flash floods are investigated using a variety of spatial–temporal models. In order to obtain accurate results from these models, it is vital that the input factors retain their spatial associations [12].

17. Various technologies involving GIS and RS have been developed by researchers in the field of environmental science. Among the different approaches, natural hazard zoning has applied the most popular and widely used models, including weights of evidence [13-15], Shannon's entropy [5,16], random forest [17-19], logistic regression [14,20,21], the statistical index [22], and the analytical hierarchy process [23-26]. For example, Kazakis et al. [23] used an index-based approach to assess the regional flood hazard. Nowadays, machine learning algorithms, including artificial neural networks [27-29], support vector machines [13], and decision trees [30], are also widely used. Flash flood hazard susceptibility mapping has been carried out by diverse methods, including the statistical index (SI) [31], fuzzy theory [32,33], and the artificial neural network (ANN) [34-36]. Frequency ratio (FR) and information value (IV) models have already been frequently used for landslide susceptibility mapping [37,38]. Similar models have been used in gully erosion [39], landslide susceptibility mapping [40,41], and forest fire susceptibility [42,43]. The FR method has been proven to be effective, and it has been successfully applied to flash flood hazard susceptibility mapping and landslide susceptibility mapping [22,44,45]. In view of the effectiveness of the FR method, it was selected as a statistical method in the present study to better explore the effect of different mapping units on the susceptibility mapping of debris flow. Furthermore, the information value model, which has been proved to be a very useful method for measuring the degree of influence of each causative factor, is a simple probabilistic bivariate statistical method whose accuracy is acceptable [46,47].

18. Farmland terraces are abundant in catchments, and their structural strengths are very low. In the event of a flash flood, small landslides will form, and large numbers of terraces have a high risk of being damaged. Thus, it is very important to highlight flash flood protection and adaptation approaches for agricultural areas to minimize the consequences of flash flood hazards due to different human activities and climate change conditions. On the other hand, technical measures, such as farmland terraces, can be used for soil and water conservation. To a certain extent, these terraces could be an approach to intercepting an oncoming flash flood event; in turn, the terraces would sustain damages. Terrace stone walls continued to be reconstructed by local residents when erosions or landslides occurred in the past. Thus, field surveys have found that most farmland terraces have been well maintained. Studies have been conducted in middle and low farmland areas to establish relationships between farmland maintenance and floods on a sub-catchment scale [48]. Because forests can prevent the appearance of a flash flood, protecting forestland is imperative [49]. Farmland terraces, which can retain water in catchments, also contribute to the alleviation of flash flood hazards. Since farmland terraces also suffer damages from flash floods, local residents should pay more attention to them. Considering that there are many farmland terraces in mountainous areas in Beijing, especially the intermediate- and low-elevation areas, appropriate flash flood management plans for these areas are vital.

19. This study aims to determine the spatial probability of flash flood-affected hazard occurrence in four catchments. The correlation between influence factors and flash flood-affected hazard occurrence is identified, and the accuracy is evaluated. Furthermore, the present work also conducts a comparative assessment of two statistical models used for flash flood hazard susceptibility mapping: the frequency ratio (FR) model and information value model (IVM). The FR and IVM methods were selected for their mathematical simplicity, their ability to extract data in a limited time period, and their effectiveness. The prediction accuracy and performance of each method were assessed using four catchments in a mountainous area. Nine flash flood hazard susceptibility factors were used in the two models using GIS software. The results were validated using the area under the receiver operating characteristic curve (ROC) method.

20. Comment： Line 103-104: Here the most critical aspect of the manuscript appears. The field criteria used for the generation of the inventory map. In my opinion, the first two criteria mentioned in Figure 2 need not be related directly or univocally to the occurrence of flash floods. And this aspect is so important that it condones the rest of the manuscript, because it conditions the results obtained, and conditions the discussion and the conclusions derived from the study carried out. Tanto las roturas de diques de contención, como la presencia de pequeños deslizamientos superficiales pueden producirse sin que tenga que haber ocurrido un flash flood. En este sentido, y para ser utilizados en análisis multicriterio (parecidos o relacionados al que se presenta en este manuscrito), serían de mayor utilidad parámetros de forma de la cuenca o de caracteristicas de la misma (relieve, desnivel, esfericidad, densidad de drenaje, ...). Esto es así, porque los parámetros utilizados por los autores parecen más acordes para el estudio de un proceso que se desencadena en un punto determinado de la cuenca, y esto no es así en el caso de los flash floods. Un flash flood no se desencadena en un punto determinado de la cuenca, sino que se desencadena a partir del comportamiento general del conjunto de la cuenca ante unas condiciones externas (como sería la ocurrencia de precipitaciones de alta intensidad en periodos de tiempo pequeños).

Response: Thank you for you valuable suggestions, we have added this paragraph:

“The hazards caused by a future flash flood can be estimated by analyzing past records. An inventory map can show the distribution and characteristics of hazards caused by a flash flood in the study area [50]. The mapping of hazards caused by flash flooding in the four catchments is necessary for depicting the relationship between susceptibility ranges and influence factors. Extensive field investigation and observations were conducted to produce a comprehensive and reliable inventory map. The flash flood hazard inventory map shows the spatial distribution of hazards that were caused by flash flooding in the study area. This was used as a base map to generate the flash flood hazard susceptibility map. We analyzed records of flash flood hazards to identify sensitive areas that were prone to destruction by flash flooding. The 21 July 2012 flash flood, which damaged roads, slopes, houses, farmland terraces, etc., was an important previous flash flood event in this study. Between 2012 and 2015, some flash floods occurred in the four catchments. The inventory map was first created by locating flash flood hazards in the four catchments using documents and detailed field surveys. A good source of information includes interviews with local residents, which were conducted to identify destroyed houses and public facilities damaged by the flash floods that occurred between 2012 and 2015. Because the storm and flash flood on 21 July 2012 left local residents with significant impressions, they give this event special attention. Field surveys confirmed landslides, collapses, and erosion due to flash flooding. The farmland terraces are mainly located at the bottom of the catchment and partly on the hillside, so they are highly vulnerable to flooding and easily damaged. Identifying the effects of flash flooding is fairly straightforward. ”

We also have replaced the figure 2 using new photos.

21. Comment：Line 159: “short-term heavy precipitation” What are the thresholds used?

Response: Flash flood is caused by heavy or excessive rainfall in a short period of time, generally less than six hours. Brooks considered that flash flooding is frequently associated with heavy precipitation occurring over a short period of time. He believed that the hourly precipitation dataset (HPD) could be useful for observing and defining the threat of flash flooding. HPD is used to develop a climatology of heavy rains on timescales of 3 h or less. Spatially, in the western mountainous area of Beijing, the frequency of short-term heavy rains (STHR) is defined as ≥20 mm/h.

However, in this study, we have checked the whole paper and replaced the short-time heavy rain (STHR) by flow accumulation. And we have also added the picture of flow accumulation in Figure.4 and detail data in Figure 5, Table 1 and Table 2. Besides, we updated the results of the susceptibility map of Figure 7 and Figure 8, as well as validation results of Figure 9.

We added contents:

We have added the results description of flow accumulation in Section 4.1 and Section 4.2.

Section 4.1 :“Higher flow accumulation has higher FR values, i.e., 3.382 (351651–622719 subclass) and 1.458 (117217–351651 subclass). ”.

Section 4.2 : “Higher flow accumulation has higher IVs, i.e., 1.218 (351651–622719 subclass) and 0.377 (117217–351651 subclass).”

22. Comment： Line 166: “FFIM”: I think that this acronym is not previously defined.

Response: Thank you for your kind reminding, we have revised it as “flash flood hazard susceptibility mapping”.

23. Comment： Line 166-184: The writing of this section should be revised and improved, because it is really difficult to understand what the authors are trying to say and therefore understand the meaning of each of the parameters.

Response: We have revised the part：

3.1 Frequency ratio

The FR method is an accurate and effective technique that is based on the observed relationships between the distribution of debris flows and related factors. In this study, the FR method was used to perform flash flood hazard susceptibility mapping. The FR is defined as the ratio of the probability of the occurrence of a flash flood hazard to the probability of a nonoccurrence for a given attribute [62,63]. The larger the FR, the stronger the effect of the given factor on the debris flow [64]. This approach reveals the correlation between flash flood hazard susceptibility areas and the influence factors in the catchment. First, the FR for each factor type or range was calculated by using Eq (3):

(3)

where C is the number of cells with hazards caused by a flash flood in each conditioning factor subclass; D is the total number of cells with hazards caused by a flash flood in the four catchments; M is the cell number of each conditioning factor subclass; N is the total cell number of the four catchments. FR values greater than 1 indicate higher densities of flash flood hazards in the category compared with the density of hazards in the four catchments, and it translates to a higher correlation between the category and the occurrence of flash flood hazards. FR values less than 1 indicate a lower correlation [65]. The flash flood hazard susceptibility index (FFHSI) was calculated using Eq (4):

(4)

where FR is the weight of the FR model, and N is the number of influence factors. The greater the FFHSI, the higher the possibility that flash flood-affected hazards will occur.

3.2 Information value model

The information value model (IVM) is a quantitative analysis method developed from information theory. The information value method is a bivariate statistical approach to deriving data for flash flood-affected hazard areas, as well as the unaffected areas. With this method, the probability of flash flood-affected hazard occurrence in the study area can be quantified in the flash flood-affected hazard classes. Yin and Yan [66] proposed this method, and Van Westen [67] modified it. It involves the computation of (1) the cell number of total flash flood-affected hazards for each influence factor subclass and (2) the cell number of total pixels of flash flood-affected hazards in the study area. Recently, this method has become increasingly favored by scholars and has been applied to geological hazard assessment and environmental evaluation [68-70].

The information value I (x_i, H) of each conditioning factor x_i is

24. （5）

where N_iis the number of cells with hazards caused by flash flooding in each conditioning factor subclass x_i, N is the total number of hazards caused by flash flooding in the study area, S_i is the area of each conditioning factor subclass x_i, and S is the total number of cells in the four catchments. The information value of each conditioning factor subclass is calculated as

25. （6）

where I_i is the total information value of each conditioning factor subclass, and n is the number of conditioning factor subclasses.

26. Comment： Line 187: CE Shannon ???

Response: We have revised this as: “Yin and Yan [66] proposed this method, and Van Westen [67] modified it.”

66. Yin, K. Statistical prediction model for slope instability of metamorphosed rocks. In Proceedings of Proceedings of the 5th International Symposium on Landslides; pp. 1269-1272.

67. Van Westen, C.J. Application of geographic information systems to landslide hazard zonation. 1993.

27. Comment： Table 2: Information content: Ii?????Check

Response: We have checked Table 2. The Information content should be “Information value model (IVM)”.

28. Comment： Line 205: I do not enter to evaluate the correction of the exposed results, which may be correct. However, I believe that the results are affected by what was previously indicated regarding the selection of field evidence for the generation of the inventory map.

Response: It is really true as you suggested. We have explained it in section 2.1

2.1 Identifying locations of flash flood hazards in 2012–2015

29. Comment： Line 213: 391-846m,should be “691-846”

Response: We have revised this error.

30. Comment： Line 239: Same that for section 4.1: I do not enter to evaluate the correction of the exposed results, which may be correct. However, I believe that the results are affected by what was previously indicated regarding the selection of field evidence for the generation of the inventory map.

Response: Response: It is really true as you suggested. We have explained it in section 2.1

31. 2.1 Identifying locations of flash flood hazards in 2012–2015

32. The hazards caused by a future flash flood can be estimated by analyzing past records. An inventory map can show the distribution and characteristics of hazards caused by a flash flood in the study area [50]. The mapping of hazards caused by flash flooding in the four catchments is necessary for depicting the relationship between susceptibility ranges and influence factors. Extensive field investigation and observations were conducted to produce a comprehensive and reliable inventory map. The flash flood hazard inventory map shows the spatial distribution of hazards that were caused by flash flooding in the study area. This was used as a base map to generate the flash flood hazard susceptibility map. We analyzed records of flash flood hazards to identify sensitive areas that were prone to destruction by flash flooding. The 21 July 2012 flash flood, which damaged roads, slopes, houses, farmland terraces, etc., was an important previous flash flood event in this study. Between 2012 and 2015, some flash floods occurred in the four catchments. The inventory map was first created by locating flash flood hazards in the four catchments using documents and detailed field surveys. A good source of information includes interviews with local residents, which were conducted to identify destroyed houses and public facilities damaged by the flash floods that occurred between 2012 and 2015. Because the storm and flash flood on 21 July 2012 left local residents with significant impressions, they give this event special attention. Field surveys confirmed landslides, collapses, and erosion due to flash flooding. The farmland terraces are mainly located at the bottom of the catchment and partly on the hillside, so they are highly vulnerable to flooding and easily damaged. Identifying the effects of flash flooding is fairly straightforward.

33. From an inventory map, a flash flood hazard susceptibility map can be produced. A flash flood hazard map was generated using a previous inventory map and remote sensing images. 71 flash flood hazard locations were surveyed in the four catchments and were used in further analysis (Fig. 1). 50 flash flood locations were randomly selected to build and train the models. The remaining 21 flash flood locations were used as validation data.

34. Comment： Line 264: You should explain better the validation process.

Response: Thank you for your suggestion. We have added the validation process.

“Validation of the flash flood hazard susceptibility results is one of the most important tasks [74]. In this study, the results of flash flood hazard susceptibility mapping were validated by the receiver operating characteristic (ROC) technique. In the ROC curve, the vertical axis represents the true positive rate, and the horizontal axis represents a false positive rate. The AUC was used to evaluate the validity of the four models. From the training and testing data, the success and prediction rates of six models were calculated by using the AUC. The value of the AUC varies from 0.5 to 1, and the accuracy of the model is high if the value of the AUC is close to 1.”

35. Comment： Line 280: As in the case of the results, I consider that this discussion is affected by what was previously indicated regarding the selection of field evidences for the generation of the inventory map.In addition, at this point an aspect mentioned at the beginning is revealed. I think there is no agreement between the title of the manuscript and the analysis carried out. This is because at no time do I see an analysis of the susceptibility of the terrain to the occurrence of flash floods, but is trying to perform an analysis on which areas are most likely to occur damage associated with the occurrence of flash floods. With the addition that the criteria of evidence used do not consider that they are the most correct.

Response: Thank you for your comments. It is really as what you commented. We focus on a study of the areas prone to be damaged by the occurrence of flash floods. We are sorry for the misleading of the title of the study. We perform an analysis on which areas are most likely to occur damage associated with the occurrence of flash floods. And we have revised the title as “Geospatial Analysis of Flash Flood Hazard Susceptibility in Small-catchment Scale: Case of Four Catchments in Mountainous area, Miyun County, Beijing”

We also have explained how the hazard evidences we investigated.

36. Comment： Line 291: In Figure 9, are the tree scars into the channel? This evidence may be related to debris flow process or landslides process too.

Response: Thank you for your suggestions. The tree scars are located in the bottom of the channel. We have added sentences and references:

37. Comment： Line 352: I believe that the conclusions, based on the study in its current form, should be updated based on a better and more correct choice of field evidence that allows the inventory map to be generated. In such a way that this new inventory map can modify the obtained results, and that therefore these variations are transferred towards the possible conclusions obtained from the present study.

Response: It is really true that the results rely on the inventory map. And we have explained how we investigated the hazards in comment 3. Thank you very much for your kind reminding. Your suggestions give us a good guide in the future work.

Yours sincerely,

Wen Zhann, Ph.D.

College of Construction Engineering, Jilin University

938 Ximinzhu Road, Changchun 130026, China

Phone number: +86 13604402994 E-mail: [email protected]

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

As indicated in the first revision, the manuscript deals with an interesting subject such as flash floods. However, I believe that the title of the manuscript still does not reflect well the work done. From the title of the manuscript, the authors propose an analysis of the susceptibility to the occurrence of flash floods, while the analysis carried out (in my opinion) focuses on a study of the areas prone to damage by the occurrence of flash floods. Something that in my opinion is not similar. Moreover, the serious question arises whether the authors are really analyzing flash floods, or debris-flows.
After the first review, I still do not agree with important aspects of the manuscript. Mainly because I consider that the evidences used by the authors to generate the inventory map are not always correct. The authors continue using as evidence the breakages of retaining walls and also of small superficial landslides (indicating that these processes are always associated with flash floods, something that is not supported by bibliographic references), as events associated with the occurrence of flash floods, and in my opinion this is not correct. There is no direct and univocal relationship between these evidences in the field and the occurrence of flash floods. In fact, these evidences can happen perfectly without flash floods occur. Landslides, collapses are noted by authors as process that always occur at the same time that flash floods, but there are not evidences about it. What happen in bed rock basins?
The variables used by the authors for the analysis of susceptibility to "flash floods" seem more appropriate for the study of processes such as debris flow or shallow landslides, in which there are areas that trigger the process. While in the case of flash floods, there is not an activation area, but a response from the basin (which can be favored by its characteristics such as slope, roundness, drainage density, ...) to high intensity rainfalls in short periods of time. In fact, in the review process that the authors have carried out, they have eliminated the only variable that referred to the rain intensity, substituting it for the accumulated flow value in each point of the basin. How you relate the flow accumulation to the ocurrence of flash floods?. This change, I believe, points again towards an analysis of debris flows, rather than a flash flood analysis.
Not are other variables of general use taken into account in the analysis of flash floods, such as the time of concentration of the basin (Tc), or the time lag (T_lag). See the following bibliographic references for flash floods definition:
- Borga, M., Stoffel M., Marchi, L., Marra F., Jakob, M. 2014. Hydrogeomorphic response to extreme rainfall in headwater systems: Flash floods and debris flows. J. Hydrol. 518, 194–205. DOI: 10.1016/j.jhydrol.2014.05.022
- Creutin, J.D., Borga, M., Gruntfest, E., Lutoff, C., Zoccatelli, D., Ruin, I., 2013. A space and time framework for analyzing human anticipation of flash floods. J. Hydrol. 482, 14–24. DOI: 10.1016/j.jhydrol.2012.11.009
Then, all the subsequent Discussion, and Conclusions can be questioned.
In addition, the parts of the text that are marked as modified (in first review process) are not real, since the modification of a pair of words within a paragraph has led the authors to indicate the full paragraph as modified. There is no correct editing of the changes made, which has significantly hindered the task of revising the manuscript.
In general, I believe that the authors have not answered (or correctly solved) the comments indicated in the first review. And therefore, the manuscript continues to suffer from the same problems that had already been mentioned previously.

Author Response

Dear Reviewer:

Thank you very much for your comments concerning our manuscript entitled “Spatial Analysis of Flash Flood Susceptibility Assessment in Small-catchment Scale: Case of Mountainous area in Miyun County, Beijing” (IJERPH-532946). We have considered carefully and revised our manuscript according to your valuable comments. Revised portion are marked in red in the paper in details. Details can be seen in the file of “REVISED MANUSCRIPT”. The main corrections in the paper and the responses to your comments are as following:

Comment I: As indicated in the first revision, the manuscript deals with an interesting subject such as flash floods. However, I believe that the title of the manuscript still does not reflect well the work done. From the title of the manuscript, the authors propose an analysis of the susceptibility to the occurrence of flash floods, while the analysis carried out (in my opinion) focuses on a study of the areas prone to damage by the occurrence of flash floods. Something that in my opinion is not similar. Moreover, the serious question arises whether the authors are really analyzing flash floods, or debris-flows.
After the first review, I still do not agree with important aspects of the manuscript. Mainly because I consider that the evidences used by the authors to generate the inventory map are not always correct. The authors continue using as evidence the breakages of retaining walls and also of small superficial landslides (indicating that these processes are always associated with flash floods, something that is not supported by bibliographic references), as events associated with the occurrence of flash floods, and in my opinion this is not correct. There is no direct and univocal relationship between these evidences in the field and the occurrence of flash floods. In fact, these evidences can happen perfectly without flash floods occur. Landslides, collapses are noted by authors as process that always occur at the same time that The variables used by the authors for the analysis of susceptibility to "flash floods" seem more appropriate for the study of processes such as debris flow or shallow landslides, in which there are areas that trigger the process. While in the case of flash floods, there is not an activation area, but a response from the basin (which can be favored by its characteristics such as slope, roundness, drainage density, ...) to high intensity rainfalls in short periods of time. In fact, in the review process that the authors have carried out, they have eliminated the only variable that referred to the rain intensity, substituting it for the accumulated flow value in each point of the basin. How you relate the flow accumulation to the ocurrence of flash floods?. This change, I believe, points again towards an analysis of debris flows, rather than a flash flood analysis.

Not are other variables of general use taken into account in the analysis of flash floods, such as the time of concentration of the basin (Tc), or the time lag (T_lag). See the following bibliographic references for flash floods definition:

- Borga, M., Stoffel M., Marchi, L., Marra F., Jakob, M. 2014. Hydrogeomorphic response to extreme rainfall in headwater systems: Flash floods and debris flows. J. Hydrol. 518, 194–205. DOI: 10.1016/j.jhydrol.2014.05.022

- Creutin, J.D., Borga, M., Gruntfest, E., Lutoff, C., Zoccatelli, D., Ruin, I., 2013. A space and time framework for analyzing human anticipation of flash floods. J. Hydrol. 482, 14–24. DOI: 10.1016/j.jhydrol.2012.11.009. Then, all the subsequent Discussion, and Conclusions can be questioned.

Response I: It is really true as you commented that the title of the manuscript did not reflect well the work done. And the title of the manuscript has led a misunderstanding of our manuscript. Thus, we have revised the title of our manuscript as: “Geospatial Analysis of Mass-wasting Susceptibility of Four small Catchments in Mountainous Area, Miyun County, Beijing”. The reason we revised the tile of our manuscript is basically according to your comments of the following main aspects:

(1) The two literatures are very valuable for guiding us to understand the definition and basic principle of flash flood. It also helps distinguish the different between flash flood and debris flow according to the aspects of forecasting, geomorphic responses, rainfall estimation. According to the literatures you suggested, we find that our study area may not easy to occur flash flood, instead, it is more likely to occur debris flow. Due to that we have revised the title and main concerns of our manuscript. We also deleted/revised the all description related to flash flood.

(2) These influencing factors are selected according to the characteristics of the study area and previous literatures. It is true that our change of adding flow accumulation as one of the influencing factor means that we don’t perform the flash flood analysis. Driven by the pull of gravity, mass wasting comprises all the sedimentary processes related to remobilization of sediments deposited on slopes including creep, sliding, slumping, flow, and fall. It can basically cover surface deformations such as shallow landslides, collapses, breakage of retaining walls and farmland terraces breakage, etc. Thus, identification and mapping of mass-wasting prone areas need proper understanding of the main influencing factors. A variety of thematic data layers have been used in susceptibility assessment such as geomorphology, geology, hydrology, and anthropogenic impacts.

(3) What we should focus is the hazards, such as debris flow, collapses or the original landslides, even the breakages of retaining walls. We considered these hazards could be called as mass-wasting. Mass-wasting is a natural phenomenon by which rock, soil or debris move downwards due to the action of gravity. It describes all the processes that act continuously with varied intensity on all type of slopes to lower the ground surface. The mass-wasting process is controlled by the interaction of geological agents and processes with the geo-materials. The degree and type of movements depend upon a few aspects of geology, environment, geomorphology, hydrology, and some additional environmental stress factors, including biotic factors.

(4) As you commented, it is not easy to identify the occurrence time of each hazard point and it is not easy to identify whether the hazard is caused by flash flood or debris flow. Thus, the field evidences we used could be able to generate mass-wasting inventory map, not a flash flood-affected hazard inventory map. The breakage of retaining walls and small superficial landslides may be caused by other reason, or perhaps caused by flash flood or debris flow. We don’t consider the occurrence time of each hazard point, we only consider the hazards positions and environment. We don’t consider the hazard occurrence time neither. In the mass-wasting inventory map, spatial distribution of the mass-wasting hazards were investigated in the field with the Handheld GPS and remote sensing method. This inventory map is what we have tried our best to obtain.

Thus, we revised the title of our manuscript as: “Geospatial Analysis of Mass-wasting Susceptibility of Four small Catchments in Mountainous Area, Miyun County, Beijing”.

If you have any more suggestions, we will adopt it.

Comment II: In addition, the parts of the text that are marked as modified (in first review process) are not real, since the modification of a pair of words within a paragraph has led the authors to indicate the full paragraph as modified. There is no correct editing of the changes made, which has significantly hindered the task of revising the manuscript.
Response II: We are sorry for the modification of a pair of words within a paragraph has led the authors to indicate the full paragraph as modified. We should mark the revised part in details. After we revised the manuscript, we sent the manuscript for English editing. The modified version has been revised a lot, which is much different from former version.

In this revised version, we marked the modified part in details, avoiding misleading.

Comment III: In general, I believe that the authors have not answered (or correctly solved) the comments indicated in the first review. And therefore, the manuscript continues to suffer from the same problems that had already been mentioned previously.

Response III: We are sorry for not answering or correctly solving the comments indicated in the first review. Based on the newest version of our manuscript, we answer your comments in the first review. The answers are as follows:

Response: Thank you for your comments. It is really as what you commented. We focus on a study of the areas prone to occur mass-wasting in the study area. We are sorry for the misleading of the title of the study. And we have revised the title as “Geospatial Analysis of Mass-wasting Susceptibility of Four small Catchments in Mountainous Area, Miyun County, Beijing”

2. Comment：However, there is a main aspect that raises doubts about the analysis made by the authors, and that in my opinion questions or questions the subsequent Discussion and Conclusions obtained from this manuscript. This weak point of the analysis, in my opinion, is the field evidence used by the authors to generate the inventory map, which in my opinion is not always correct. For example, the authors propose the use of evidence of breakage of retaining walls and also of small superficial landslides, as events associated with the occurrence of flash floods, and in my opinion this does not have to be correct; There is no direct and univocal relationship between these evidences in the field and the occurrence of flash floods. In fact, these evidences can occur perfectly if the occurrence of flash-floods. In addition, the spatial distribution of inventory map data can very significantly condition the results obtained. This is therefore an aspect in which the authors should work seriously and devote sufficient effort to define and locate another series of field evidences that can be directly and unequivocally related to the occurrence of flash-floods.

Response: Thank you for you valuable suggestions. As you commented, it is not easy to identify the occurrence time of each hazard point and it is not easy to identify whether the hazard is caused by flash flood or debris flow. Thus, the field evidences we used could be able to generate mass-wasting inventory map, not a flash flood-affected hazard inventory map. The breakage of retaining walls and small superficial landslides may be caused by other reason, or perhaps caused by flash flood or debris flow. We don’t consider the occurrence time of each hazard point, we only consider the hazards positions and environment. We don’t consider the hazard occurrence time neither. In the mass-wasting inventory map, spatial distribution of the mass-wasting hazards were investigated in the field with the Handheld GPS and remote sensing method. This inventory map is what we have tried our best to obtain. We have revised the contents in part 2.1 as:

2.1 Identifying locations of mass-wasting inventory

The future mass-wasting can be estimated by analyzing past records. An inventory map can show the distribution and characteristics of mass-wasting in the study area [19]. Mass-wasting events come in many shapes, sizes and speeds. Typically, the steeper the angle of a slope, the faster will be the down-slope movement of rock and sediment. Also, water can play a significant role in mass wasting, sometimes acting as the key component to a mass-wasting event, or serving as a lubricant within a mass of sediment and rock, enabling it to travel faster and further than it would otherwise. Types of mass-wasting mainly contain rock fall and rock avalanche, rock slide and slump, rock avalanche, debris flow, earth flow, and creep.

The mapping of mass-wasting in the four catchments is necessary for depicting the relationship between susceptibility ranges and influencing factors. Extensive field investigation and observations were conducted to produce a comprehensive and reliable inventory map. The mass-wasting inventory map shows the spatial distribution of mass-wasting in the study area. This was used as a base map to generate the mass-wasting susceptibility map. We analyzed records of mass-wasting to identify susceptible areas that were prone to occur new mass-wasting. The inventory map was first created by locating mass-wasting in the four catchments using documents and detailed field surveys. A good source of information includes interviews with local residents, which were conducted to identify destroyed houses and public facilities damaged by mass-wasting that occurred before. The storm and flash flood on 21 July 2012 left local residents with significant impressions, they give this event special attention. Field surveys confirmed landslides, collapses, and erosion, which were regarded as mass-wasting. The farmland terraces are mainly located at the bottom of the catchment and partly on the hillside, so they are highly vulnerable to flash flooding or debris flow and easily damaged. Identifying the locations of mass-wasting is fairly straightforward.

From an inventory map, a mass-wasting susceptibility map can be produced. A mass-wasting susceptibility map was generated using a previous inventory map and remote sensing images. 71 mass-wasting locations were surveyed in the four catchments and were used in further analysis (Figure. 1). 50 mass-wasting locations were randomly selected to build and train the models. The remaining 21 mass-wasting locations were used as validation data.”

3. Comment：：Line 38: “Flood flash” should be flash floods

Response: We have deleted this.

4. Comment：Line 42: I'm not sure about the meaning of this phrase and what the authors want to express

Response: We have deleted this.

5. Comment： Line 44: Review the wording of the full paragraph. It gives the sensation of being unconnected phrases that are sustained in many cases on bibliographical references that have nothing to do with the theme of flash floods.

Response: Thank you for your kind reminding. We have deleted sentences that have nothing to do with the theme of flash floods.

6. Comment： Line 49: I'm not sure about the meaning of this phrase and what the authors want to express

Response: Thank you for your reminding. We have revised this sentence as “Thus, land use managers should be able to identify all aspects of landscape vulnerability [3]”

7. Comment： Line 51: “3S” You must define the meaning of 3S

Response: we have defined the meaning of “3S technology” according to your comment.

“Remote sensing (RS), geographic information systems (GISs), and the Global Positioning System (GPS) are now widely applied as so-called “3S technology””

8. Comment： Line 51-61: This part of the paragraph suffers from any continuity in the wording with the above, and also is not based on any bibliographic reference. Correct wording

Response: Thank you for your comment. We have rewrite this paragraph, and added several bibliographic references. The revised paragraph is:

“Remote sensing (RS), geographic information systems (GISs), and the Global Positioning System (GPS) are now widely applied as so-called “3S technology”. RS and GIS techniques have been applied for different mass-wasting hazard susceptibility modeling [5,6], and appropriate assessment methods should be similarly applied for mass-wasting susceptibility mapping [7-9]. The use of GIS has greatly progressed in the field of environmental science in applications such as landslide and groundwater susceptibility mapping and flash flood hazard susceptibility mapping. These technologies can provide a good perspective for mass-wasting assessment research. Various landslide susceptibility maps, debris flow susceptibility maps and rockfall susceptibility maps have been created in different countries[10,11]. The use of RS and GIS has increased significantly in response to the need for rapid data collection and improved mass-wasting bitmaps for commercial satellite products. Further, GIS is a useful tool for studying events with multidimensional behavior; for example, mass-wasting hazards are investigated using a variety of spatial–temporal models. In order to obtain accurate results from these models, it is vital that the input factors retain their spatial associations [4]..”

The added bibliographic references are:

4. Morjani, Z.E.A.; Ebener, S.; Boos, J.; Ghaffar, E.A.; Musani, A. Modelling the spatial distribution of five natural hazards in the context of the WHO/EMRO Atlas of Disaster Risk as a step towards the reduction of the health impact related to disasters. International journal of health geographics 2007, 6, 8.

5. Zinck, J.A.; López, J.; Metternicht, G.I.; Shrestha, D.P.; Vázquez-Selem, L. Mapping and modelling mass movements and gullies in mountainous areas using remote sensing and GIS techniques. International Journal of Applied Earth Observation and Geoinformation 2001, 3, 43-53.

6. Pardeshi, S.D.; Autade, S.E.; Pardeshi, S.S. Landslide hazard assessment: recent trends and techniques. SpringerPlus 2013, 2, 523.

7. Cloke, H.L.; Pappenberger, F. Ensemble flood forecasting: A review. J Hydrol 2009, 375, 613-626, doi:10.1016/j.jhydrol.2009.06.005.

8. Akgün, A. An integrated mass wasting susceptibility assesment by geographical information systems and remote sensing applications: Example from North Turkey. In Proceedings of EGU General Assembly Conference Abstracts.

9. Rowden, K.W.; Aly, M.H. A novel triggerless approach for mass wasting susceptibility modeling applied to the Boston Mountains of Arkansas, USA. Natural Hazards 2018, 92, 347-367.

10. Zhang, Y.; Yue, P.; Zhang, G.; Guan, T.; Lv, M.; Zhong, D. Augmented Reality Mapping of Rock Mass Discontinuities and Rockfall Susceptibility Based on Unmanned Aerial Vehicle Photogrammetry. Remote Sensing 2019, 11, 1311.

9. Comment： Line 62-74: This paragraph is a summary of bibliographical references without a clear sense. Many techniques are proposed, but in many cases they have not been or are not applicable in the field of flash floods, but especially in the field of surface landslides, or in the field of debris flow.

Response: We have added the bibliographical references that related to mass-wasting hazard susceptibility, such as landslide susceptibility, debris flow susceptibility, sinkhole susceptibility, rockfall susceptibility, etc. And we have also deleted the bibliographical references related to flash flood that has nothing to do with our manuscript.

10. Comment： Line 75-80: I think it is necessary to significantly improve the introduction to the subject of study, putting more emphasis on those works that are really focused on the study of flash floods, their dynamics, their genesis and the hazards and risk associated with them.

Response: We have rewrite section Introduction.

“Mass-wasting is common occurred throughout anthropogenic development [1]. Mass-wasting is a natural phenomenon by which rock, soil or debris move downwards due to the action of gravity. It describes all the processes that act continuously with varied intensity on all type of slopes to lower the ground surface. The mass-wasting process is controlled by the interaction of geological agents and processes with the geo-materials. The degree and type of movements depend upon a few aspects of geology, environment, geomorphology, hydrology, and some additional environmental stress factors, including biotic factors. Thus, the extent of mass-wasting damage is extensive. Mass-wasting are related to hazards caused by gravity, such as landslides, collapses, and debris flow. Mass-wasting maps, which are a very important component of catchment management.

Anthropogenic activities, including farmland expansion and timber harvesting, change natural conditions, which increases the risk of mass-wasting occurrence. Although mass-wasting hazards are inevitable, it is important to identify areas that are likely to be easily occur mass-wasting events [2]. Thus, land use managers should be able to identify all aspects of landscape vulnerability [3]. Mass-wasting susceptibility mapping has been recognized as the first necessary step in hazards prevention and its management [4].

Remote sensing (RS), geographic information systems (GISs), and the Global Positioning System (GPS) are now widely applied as so-called “3S technology”. RS and GIS techniques have been applied for different mass-wasting hazard susceptibility modeling [5,6], and appropriate assessment methods should be similarly applied for mass-wasting susceptibility mapping [7-9]. The use of GIS has greatly progressed in the field of environmental science in applications such as landslide and groundwater susceptibility mapping and flash flood hazard susceptibility mapping. These technologies can provide a good perspective for mass-wasting assessment research. Various landslide susceptibility maps, debris flow susceptibility maps and rockfall susceptibility maps have been created in different countries[10,11]. The use of RS and GIS has increased significantly in response to the need for rapid data collection and improved mass-wasting bitmaps for commercial satellite products. Further, GIS is a useful tool for studying events with multidimensional behavior; for example, mass-wasting hazards are investigated using a variety of spatial–temporal models. In order to obtain accurate results from these models, it is vital that the input factors retain their spatial associations [4].

Various technologies involving GIS and RS have been developed by researchers in the field of environmental science. Among the different approaches, natural hazard zoning has applied the most popular and widely used models, including weights of evidence [12-14], Shannon's entropy [15], random forest [16-18], logistic regression [19-22], the statistical index [23,24], and the analytical hierarchy process [25,26].. Nowadays, machine learning algorithms, including artificial neural networks [27-30], support vector machines [31], and decision trees [32], are also widely used. Frequency ratio (FR) and information value (IV) models have already been frequently used for landslide susceptibility mapping [33,34]. Similar models have been used in gully erosion [35], landslide susceptibility mapping [36,37], and forest fire susceptibility [38,39]. The FR method has been proven to be effective, and it has been successfully applied to flash flood hazard susceptibility mapping and landslide susceptibility mapping [24,40,41]. In view of the effectiveness of the FR method, it was selected as a statistical method in the present study to better explore the effect of different mapping units on the susceptibility mapping of debris flow. Furthermore, the information value model, which has been proved to be a very useful method for measuring the degree of influence of each causative factor, is a simple probabilistic bivariate statistical method whose accuracy is acceptable [42,43].

Farmland terraces are abundant in catchments, and their structural strengths are very low. Small landslides always form, and large numbers of terraces have a high risk of being damaged. Thus, it is very important to highlight protection and adaptation approaches for agricultural areas to minimize the consequences of mass-wasting due to different human activities and climate change conditions. On the other hand, technical measures, such as farmland terraces, can be used for soil and water conservation. To a certain extent, these terraces could be an approach to intercepting an oncoming hazardous event; in turn, the terraces would sustain damages. Terrace stone walls continued to be reconstructed by local residents when erosions or landslides occurred in the past. Thus, field surveys have found that most farmland terraces have been well maintained. Studies have been conducted in middle and low farmland areas to establish relationships between farmland maintenance and rainfall on a sub-catchment scale. Because forests can guarantee less soil erosion and keep the slope more stable [44], protecting forestland is imperative. Farmland terraces, which can retain water in catchments, also contribute to the alleviation of mass-wasting. Since farmland terraces also suffer damages from different conditions, local residents should pay more attention to them. Considering that there are many farmland terraces in mountainous areas in Beijing, especially the intermediate- and low-elevation areas, appropriate mass-wasting management plans for these areas are vital.

This study aims to determine the spatial probability of mass-wasting occurrence in four catchments. The correlation between influencing factors and mass-wasting inventory is identified, and the accuracy is evaluated. Furthermore, the present work also conducts a comparative assessment of two statistical models used for mass-wasting susceptibility mapping: the frequency ratio (FR) model and information value model (IVM). The FR and IVM methods were selected for their mathematical simplicity, their ability to extract data in a limited time period, and their effectiveness. The prediction accuracy and performance of each method were assessed using four catchments in a mountainous area. Nine mass-wasting susceptibility factors were used in the two models using GIS software. The results were validated using the area under the receiver operating characteristic curve (ROC) method.

11 Comment： Line 103-104: Here the most critical aspect of the manuscript appears. The field criteria used for the generation of the inventory map. In my opinion, the first two criteria mentioned in Figure 2 need not be related directly or univocally to the occurrence of flash floods. And this aspect is so important that it condones the rest of the manuscript, because it conditions the results obtained, and conditions the discussion and the conclusions derived from the study carried out. Tanto las roturas de diques de contención, como la presencia de pequeños deslizamientos superficiales pueden producirse sin que tenga que haber ocurrido un flash flood. En este sentido, y para ser utilizados en análisis multicriterio (parecidos o relacionados al que se presenta en este manuscrito), serían de mayor utilidad parámetros de forma de la cuenca o de caracteristicas de la misma (relieve, desnivel, esfericidad, densidad de drenaje, ...). Esto es así, porque los parámetros utilizados por los autores parecen más acordes para el estudio de un proceso que se desencadena en un punto determinado de la cuenca, y esto no es así en el caso de los flash floods. Un flash flood no se desencadena en un punto determinado de la cuenca, sino que se desencadena a partir del comportamiento general del conjunto de la cuenca ante unas condiciones externas (como sería la ocurrencia de precipitaciones de alta intensidad en periodos de tiempo pequeños).

Response: Thank you for you valuable suggestions, we have revised this paragraph. As you commented, it is not easy to identify the occurrence time of each hazard point and it is not easy to identify whether the hazard is caused by flash flood or debris flow. Thus, the field evidences we used could be able to generate mass-wasting inventory map, not a flash flood-affected hazard inventory map. The breakage of retaining walls and small superficial landslides may be caused by other reason, or perhaps caused by flash flood or debris flow. We don’t consider the occurrence time of each hazard point, we only consider the hazards positions and environment. We don’t consider the hazard occurrence time neither. In the mass-wasting inventory map, spatial distribution of the mass-wasting hazards were investigated in the field with the Handheld GPS and remote sensing method. This inventory map is what we have tried our best to obtain. We also have replaced the figure 2 using new photos.

2.1 Identifying locations of mass-wasting inventory

12 Comment：Line 159: “short-term heavy precipitation” What are the thresholds used?

We added contents:

We have added the results description of flow accumulation in Section 4.1 and Section 4.2.

Section 4.1 :“Higher flow accumulation has higher FR values, i.e., 3.382 (351651–622719 subclass) and 1.458 (117217–351651 subclass). ”.

Section 4.2 : “Higher flow accumulation has higher IVs, i.e., 1.218 (351651–622719 subclass) and 0.377 (117217–351651 subclass).”

13 Comment： Line 166: “FFIM”: I think that this acronym is not previously defined.

Response: We have deleted it. We revised it as “mass-wasting susceptibility index (MWSI)”.

14 Comment： Line 166-184: The writing of this section should be revised and improved, because it is really difficult to understand what the authors are trying to say and therefore understand the meaning of each of the parameters.

Response: We have rewrite the methodology part：

3.1 Frequency ratio

The FR method is an accurate and effective technique that is based on the observed relationships between the distribution of debris flows and related factors. In this study, the FR method was used to perform mass-wasting susceptibility mapping. The FR is defined as the ratio of the probability of the occurrence of a mass-wasting to the probability of a nonoccurrence for a given attribute [52,53]. The larger the FR, the stronger the effect of the given factor on the debris flow [54]. This approach reveals the correlation between mass-wasting susceptibility areas and the influence factors in the catchment. First, the FR for each factor type or range was calculated by using Eq (3):

(3)

where C is the number of cells with mass-wasting in each influencing factor subclass; D is the total number of cells with mass-wasting in the four catchments; M is the cell number of each influencing factor subclass; N is the total cell number of the four catchments. FR values greater than 1 indicate higher densities of mass-wasting in the category compared with the density of hazards in the four catchments, and it translates to a higher correlation between the category and the occurrence of mass-wasting. FR values less than 1 indicate a lower correlation [55]. The mass-wasting susceptibility index (MWSI) was calculated using Eq (4):

(4)

where FR is the weight of the FR model, and N is the number of influencing factors. The greater the MWSI, the higher the possibility that mass-wasting will occur.

3.2 Information value model

The information value model (IVM) is a quantitative analysis method developed from information theory. The information value method is a bivariate statistical approach to deriving data for mass-wasting areas, as well as the unaffected areas. With this method, the probability of mass-wasting occurrence in the study area can be quantified in the mass-wasting classes. Yin and Yan [56] proposed this method, and Van Westen [57] modified it. It involves the computation of (1) the cell number of total mass-wasting for each influence factor subclass and (2) the cell number of total pixels of mass-wasting in the study area. Recently, this method has become increasingly favored by scholars and has been applied to geological hazard assessment and environmental evaluation [58-60].

The information value I (x_i, H) of each influencing factor x_i is

（5）

where N_iis the number of cells with mass-wasting in each influencing factor subclass x_i, N is the total number of mass-wasting in the study area, S_i is the area of each influencing factor subclass x_i, and S is the total number of cells in the four catchments. The information value of each influencing factor subclass is calculated as

（6）

where I_i is the total information value of each influencing factor subclass, and n is the number of influencing factor subclasses.

15 Comment： Line 187: CE Shannon ???

Response: We have revised this as: “Yin and Yan [56] proposed this method, and Van Westen [57] modified it.”

56. Yin, K. Statistical prediction model for slope instability of metamorphosed rocks. In Proceedings of Proceedings of the 5th International Symposium on Landslides; pp. 1269-1272.

57. Van Westen, C.J. Application of geographic information systems to landslide hazard zonation. 1993.

16 Comment： Table 2: Information content: Ii?????Check

Response: We have checked Table 2. The Information content should be “Information value model (IVM)”.

17 Comment： Line 205: I do not enter to evaluate the correction of the exposed results, which may be correct. However, I believe that the results are affected by what was previously indicated regarding the selection of field evidence for the generation of the inventory map.

Response: It is really true as you suggested. We have explained it in section 2.1

18 Comment： Line 213: 391-846m,should be “691-846”

Response: We have revised this error.

19 Comment： Line 239: Same that for section 4.1: I do not enter to evaluate the correction of the exposed results, which may be correct. However, I believe that the results are affected by what was previously indicated regarding the selection of field evidence for the generation of the inventory map.

Response: Response: It is really true as you suggested. We have explained it in section 2.1

20 Comment： Line 264: You should explain better the validation process.

Response: Thank you for your suggestion. We have added the validation process.

“Validation of the mass-wasting susceptibility results is one of the most important tasks [66]. In this study, the results of mass-wasting susceptibility mapping were validated by the receiver operating characteristic (ROC) technique. In the ROC curve, the vertical axis represents the true positive rate, and the horizontal axis represents a false positive rate. The AUC was used to evaluate the validity of the four models. From the training and testing data, the success and prediction rates of six models were calculated by using the AUC. The value of the AUC varies from 0.5 to 1, and the accuracy of the model is high if the value of the AUC is close to 1.”

21 Comment： Line 280: As in the case of the results, I consider that this discussion is affected by what was previously indicated regarding the selection of field evidences for the generation of the inventory map. In addition, at this point an aspect mentioned at the beginning is revealed. I think there is no agreement between the title of the manuscript and the analysis carried out. This is because at no time do I see an analysis of the susceptibility of the terrain to the occurrence of flash floods, but is trying to perform an analysis on which areas are most likely to occur damage associated with the occurrence of flash floods. With the addition that the criteria of evidence used do not consider that they are the most correct.

Response: It is really true as you commented that the title of the manuscript did not reflect well the work done. And the title of the manuscript has led a misunderstanding of our manuscript. Thus, we have revised the title of our manuscript as: “Geospatial Analysis of Mass-wasting Susceptibility of Four small Catchments in Mountainous Area, Miyun County, Beijing”. The reason we revised the tile of our manuscript is basically according to your comments of the four aspects, details can be seen in Comment/Response I.

2.1 Identifying locations of mass-wasting inventory

22 Comment： Line 291: In Figure 9, are the tree scars into the channel? This evidence may be related to debris flow process or landslides process too.

Response: The two literatures you suggested are very valuable for guiding us to understand the definition and basic principle of flash flood. It also helps distinguish the different between flash flood and debris flow according to the aspects of forecasting, geomorphic responses, rainfall estimation. According to the literatures you suggested, we find that our study area may not easy to occur flash flood, instead, it is more likely to occur debris flow.

23 Comment： Line 352: I believe that the conclusions, based on the study in its current form, should be updated based on a better and more correct choice of field evidence that allows the inventory map to be generated. In such a way that this new inventory map can modify the obtained results, and that therefore these variations are transferred towards the possible conclusions obtained from the present study.

Response: It is really true that the results rely on the inventory map. And we have explained how we investigated the hazards in Comment/Response I and 2 comment. Thank you very much for your kind reminding. Your suggestions give us a good guide in the future work.

Yours sincerely,

Wen Zhann, Ph.D.

College of Construction Engineering, Jilin University

938 Ximinzhu Road, Changchun 130026, China

Phone number: +86 13604402994 E-mail: [email protected]

Author Response File: Author Response.pdf

Article Menu

Geospatial Analysis of Mass-Wasting Susceptibility of Four Small Catchments in Mountainous Area of Miyun County, Beijing

Further Information

Guidelines

MDPI Initiatives

Follow MDPI