the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Identifying the drivers of private flood precautionary measures in Ho Chi Minh City, Vietnam
Thulasi Vishwanath Harish
Nivedita Sairam
Liang Emlyn Yang
Matthias Garschagen
Heidi Kreibich
Download
- Final revised paper (published on 16 Mar 2023)
- Preprint (discussion started on 17 May 2022)
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-171', Anonymous Referee #1, 09 Jun 2022
General Comments:
Incorporation of flood risk mitigation measures taken by the residents is an important part of flood risk analysis/management. This study aims to identify the drivers of the private sector to implement these measures in Ho Chi Minh City, Vietnam by analysing the survey data collected from 1000 at-risk households. For this purpose, the researchers used ‘Protection Motivation Theory’ in combination with the ‘Transtheoretical Model’ by accounting for both non-structural and structural measures based on both ‘Proactive’ and ‘Reactive’ behaviours. The article includes some interesting contents that are substantially practical in the real-world flood risk analysis. However, there lies a number of fundamental concerns, mostly related to the organisation and the flow of the information on the methodology and application.
Major Comments:
- It would appear that the majority of the contents are parts of a larger document that are put together without a decent amount of cohesion an linearity. This significantly disturbs the reader to follow the flow of the article and understand the novelty of the proposed method as well as the usefulness of its output in the context of flood risk management. More specifically:
- A bulk body of information in the ‘Introduction’ section constitutes the historical data on the flood events (lines 21-37), introduction to Flood Risk Management (lines 39-48) and the private precautionary measures (lines 49-65). Though important elements (and probably important for a thesis or dissertation), they are not specific to the research novelty and the method that the authors employed to “Identify the drivers of the private flood precautionary measures in Ho Chi Minh City, Vietnam”. However, Section 2, ‘Study area – Ho Chi Minh City’, appears to include the introductory information that the reader should obtain from an introduction section which is more focused on research itself. Also, the introductory information in Section 3 could aim the authors to rewrite the introduction section.
- Section 3, which is expected to convey the information on the ‘material and method’ in the research is not self-explanatory of what is specific to the research considering the aim and objectives. The authors could use the framework introductory information, explained in Section 2 (lines 99-124), in the rewritten ‘Introduction’ section and focus more on the materials and methods in the new section (such as lines 127-204).
- Section 4, Results and discussion, does not respect the standard of a decent academic article and therefore should be rewritten to comply those requirements (authors could use other examples published in NHESS). More specifically, the section and the subsections lack introductory sentences to acquaint the readers with appropriate preparatory information for what sort of outputs will be discussed with respect to a specific purpose in the coming subsections. Take the instance of line 206, which jumps directly to the specific outputs without any preparation to answer: ‘what’, ‘why’ and ‘how’ with respect to what was said before in the ‘materials and method’ section. Also, this section includes a large amount of discussions on the previous research other than specifying the key findings of the present research.
- Section 5, conclusion, should include the key findings of the research alongside the brief summary of the research. It should also include specific limitations in more details and suggestions for future research. It would also help the quality of the conclusion to provide information on the reproducibility of such outputs for other place/countries.
- The article does not weigh the privilege of the proposed research over other existing methods. For example, questionnaire surveys. In simple words, the article does not highlight the importance of using ‘Protection Motivation Theory’ and ‘Transtheoretical Model’ for conducting such research.
Technical/Minor Comments:
- The ‘Abstract’ should be revised so as to clearly elaborate on the method, what has been specifically analysed from the survey datasets, and what are the key findings of the research and what do they imply/show. Especially line 13 onwards, the flow of information does not seem to be correct; therefore, makes it difficult to understand. Also, avoid using long sentences as short ones would help the readability of the abstract and all your work.
- The ‘Introduction’ section has many repetitions, which could be made more concise with respect to the aim and objectives of the research. For example, doesn't line 24-26 convey similar meanings to the previous two lines?
- In-text citations does not seem to follow a uniform template. In some instances, a comma is used after et al. while in the others there is none. Compare for example, line 31, (Nguyen, et al. 2021(b)) and line 36, (Cao et al., 2021).
- Line 31-33, ‘Developing countries … ‘could be more specific. For example by specifying ‘What limited capacity?’
- Line 36. The authors have not reviewed any research yet; therefore, it would be better to provide some more explanatory information on the physical and environmental drivers of flood risk before providing such conclusion.
- Line 37. “To counteract the trend of increasing flood risk due to global change, improved flood risk management is necessary.” is a trivial piece of information and sounds unnecessary as it has neither related to what has been said before nor has been specifically in line with the flow and aim and objective of your research.
- Line 42 is a good place to explain about proactive measures in the context of the implementation of flood risk management strategies.
- Lines 42-45: the authors could elaborate more. Do you mean: because the flood hazard changes rapidly in urban areas amongst the household units, implementation of the conventional large-scale flood protection measures, such as dikes and retention basins, is challenging?
- Lines 45-48: is not comprehensible as the previous paragraph is not structured based on the research scope.
- Line 49: use 'Private precautionary measures have demonstrated to be effective in reducing flood damage.' instead.
- Line 53: use “There is a knowledge gap in ...” instead.
- Line 60: “Experiencing repeated flooding can change this attitude (Bubeck et al., 2018; Chinh et al., 2016).” Using such sentences from other sources requires mentioning further backup from other research. For instance, what did they specifically conclude in their research? Using such assertion for your purpose in the introduction does not provide the reader with the required clarity.
- Line 63: “These insights can guide the design of targeted risk communication campaigns and incentives to improve flood preparedness” The authors should elaborate more on such sentences.
- Lines 49-71: This paragraph should be more specific and concise to fit the purpose of this study. The authors should provide a summary of what has been done in the present research as well as its basis and novelty.
- Line 76: add reference for “The city's population is expected to grow even faster in the coming years.” Also, it is good to mention the population growth rate.
- Line 82: The flood risk is exacerbated by climate change, ongoing urbanization, increasing population and infrastructure density leading to a higher proportion of sealed surfaces.” Is a repetition. The authors should consider removing the repetitive sentences in the revised/rewritten manuscripts, especially in the ‘Introduction’ and ‘Materials and Method’ sections.
- Line 94-96: as last lines of your subsection should conclude your discussion on the issue. Please consider to follow a more linear approach in providing the reader with the required information before they reach the next section of your manuscript.
- As mentioned above, the majority of Section 3 could be used in the revised ‘Introduction’ sentence and this section should be more concise and specific to the researcher’s own work and method.
- Line 119: use “The present approach” or “The proposed framework” instead of “This framework”.
- Line 129: “The survey collected 1000 valid responses from local households who suffered from floods in the recent 10 years.“ here the authors are expected to mention the representation percentage of the selected number of household with respect to the total population (9 million + 2 million?). And also, how this rate would influence the validity of the research findings?
- Lines 131-132: Could be more specific. For example, what class of socio-economic profiles and what types of flood types have been investigated?
- Line 152: use “Each precautionary measure is categorised into … “.
- Figure 2. Is this figure necessary? The authors could use a simple table instead.
- Line 164: the authors should explain the reason that the lasso and elastic net regression models lead to identification of drivers. And in-detail explanation is required here.
- The authors should explain the notations in all the equations and avoid explaining the repeating ones. This can be done by providing a couple of lines below each equation/formula.
- Date shown in Figure 4: where did the authors derive/obtain the ‘Implementation cost of the private precautionary measures’? If it is a part of the present research, there has to be some explanation. If not, the authors should provide information on how they obtain them. Also consider it in the further discussion.
- Lines 240-245: Study worth analysing the difference in the socio-economical drivers between the countries that influence households to take flood mitigation measures at individual levels. What are the differences and how they might change according to each country socio-economical driver?
- Section 4.2 (starting from line 245): The variables discussed here are not mentioned/explained previously in the article, therefor the reader is not familiar with these terms. The authors can make a table and explain each before the reader reaches to Section 4.2.
- Lines 251-253: Good to know what Table 3 has. It should be explained before discussion on the results. Please do not mix the discussion of Table 3 with Figure 5. The authors should first explain Table 3, then discuss figure 5. The authors should ensure that they explain about the variables before jumping into the discussions. It can be in a few lines in the introductory paragraph of this section explaining what has been studied with what aim and how.
- Lines 253-258: the meaning of these sentences are not clear. For example, what is the difference between 'house damage' and 'house impact'? The authors should explain ‘house damage' and 'house impact' before reaching here to give the readers an idea of what they mean.
- Line 294: use comma after ‘Next’.
- Lines 297-299: There is no need for mentioning the findings of the previous research in your discussion if they are not related to the results’ discussion.
- Lines 301-305: require further elaboration.
- Lines 306-319: This is not a concluding paragraph for the discussion section. The authors can clearly explain what their discussion suggest with respect to the aim and objective of the reseach.
- In ‘Competing interests’, Is this necessary to mention that one of the authors is an executive editor at Natural Hazards and Earth System Sciences (NHESS) jounal? Please do check it with the editor. Also, make sure that the Appendix is located in a correct place.
- Titles of the sections and subsections should be more informative elaborating on their contents by also preserving the linearity in the revised manuscript.
- Also consider using more informative caption for the figures and the table 1.
Citation: https://doi.org/10.5194/egusphere-2022-171-RC1 - AC1: 'Reply on RC1', Thulasi Vishwanath Harish, 24 Aug 2022
-
RC2: 'Comment on egusphere-2022-171', Anonymous Referee #2, 10 Jun 2022
General comments
The drivers of individual choices in the context of flood protection and flood risk mitigation are not clear. This study takes an interesting and new approach into understanding such drivers, and it selects a very interesting local case for this scope, Ho Chi Minh City, in Vietnam, a city plagued by frequent flooding and with still lackluster government solutions on flood protection. The methodology is statistically advanced, and the size of the survey is impressive. The presentation of results is correct, and useful lessons can be drawn from the analysis. The study grapples with theoretical frameworks from the social and psychological sciences, and I commend the authors for explaining the key concepts and methods with sufficient care that someone without that background – like me – is still able to follow adequately. Terminology throughout the text is consistent. The article is quite concise, with the exception of some lengthy parts of the Introduction. The text is well written and generally clear, though I recommend that the authors revise it again to improve simplicity of some sentences and correct minor mistakes. The paper could be published in this special issue, pending careful revision on a number of aspects, both general and specific, as explained below.
The reference to the literature in the Introduction is largely inadequate. I include below specific comments on this issue, limited to the first lines of the introduction. It is necessary that the authors verify every statement and its supporting references carefully throughout the manuscript.
The Transtheoretical Model as implemented in the study distinguishes between households at two ‘risk reducing stages’: proactive and reactive. It is not clear to me, especially after seeing how this differentiation is carried out in the survey (lines 141-on), whether it is possible to determine whether the moment when the interviewee responds to the survey is before or after ‘the flood’. In a context where floods occur with remarkable frequency, are attitudes and behaviors influenced by thoughts of past floods or rather by expectations of future floods? And is it even possible to tell them apart? Can the authors clarify how they deal with this ambiguity, and how sensitive are the results with respect to this dubious point?
The key choice of aggregating responses according to whether the measure is structural or non-structural is not motivated. Even after reading the discussion of the results, I am not convinced that this is one of the two most relevant ways to discriminate among households or measures. I understand that the research is broadly framed in the context of a need for non-structural measures to also be implemented, next to structural ones, so that ‘integrated flood risk management’ is achieved. But it is not clear to me that this implies that structural vs non-structural is a key dimension along which the results of this behavioral survey should be analysed. I think that there is no clear a priori reason to assume that the type of measure matters heavily for the behavior of flood-prone actors, whereas it would seem more reasonable that factors like price (an hypothesis in fact disproven by this study) or familiarity with the measure should matter more, a priori. Please motivate this choice, or alternatively analyse and present results with the only differentiation of preventative vs reactive households, or other relevant differentiations.
In the Methods, there is no presentation of the explanatory variables that are taken into consideration in the survey and for the regressions. Nor is it stated whether households are surveyed about the cost of the measures, of whether costs are taken from other sources.
Specific comments
In the abstract we read that two seemingly contrasting beliefs both promote proactive implementation of private measures: “degree of belief that the government will implement effective flood protection measures and degree of belief that one has to deal with the consequences of flooding by themselves”. How is this possible?
L 24: whereas Botzen et al 2019a is a fine review of the trends and drivers of economic impacts of floods in the past and future, it doesn’t seem to support a statement specifically on the physical aspect of floods and climate change. There are several papers that can be picked for that, e.g., Winsemius et al. 2015 (10.1038/nclimate28930) for river flooding, or those referenced in the following sentence.
27: Whereas sea level rise obviously increases coastal flooding, it is not clear that it will bring more frequent or intense storm surges. E.g., Muis et al. 2020 https://doi.org/10.3389/fmars.2020.00263
L 30: Mendoza et al. does not seem an appropriate study to support that Vietnam is the country most vulnerable to climate change. Further, Dasgupta et al is a dated study. This is a bold statement that requires credible support.
31: The study of Nguyen et al. 2021b does not show that floods are the most damaging hazard in Vietnam, as it is concerned with a very different issue.
32: How can Hagedoorn et al. 2021 discriminate between the impacts of floods in developing and in other countries, when they only study flood adaptation behavior in Vietnam?
L 55: whereas Sairam et al 2019 (previously referenced) empirically verify the effectiveness of measures, the studies of Scussolini et el and Du et al are based on modeling, and undertake many assumptions. This has to be made clear in the manuscript, as the empirical and modeling approaches have different value when it comes to show-prove-support-report effectiveness of measures.
67: “evaluate how these drivers are associated with the willingness of households to adopt private flood precautionary measures”. Isn’t this a tautological sentence? Once the drivers of behavior are known, we also know what makes people willing to adopt the behavior. I might be missing something here.
L 68 “we develop an empirical datadriven approach complementing theoretical protection motivation theory and transtheoretical model frameworks”. I don’t think the meaning/content of this sentence will be clear to the reader. Further, the following sentence is repetition of previous sentences.
Section 2: I think much detail about physical/climatic aspects of HCMC can be shortened, as these are not highly relevant to this study. To any extent, if referring to the drivers of floods, this recent study rigorously looked at the key drivers and their dependencies: Couasnon et al. 2022 https://doi.org/10.1029/2021WR030002
88: “Protection of livelihood from flood events has a high priority and it is
leading to high investments in extensive flood defense systems (Kreibich et al., 2015; Weyrich et al., 2020).” This section is about HCMC, so why this statement with references on investments elsewhere?
Fig. 1. The figure is helpful in sketching the framework. Some questions: if ‘dependency on government influences the coping appraisal (and not the threat appraisal), wouldn’t it make sense for that block to link with the coping appraisal before its joining to the threat appraisal? Does the same apply to ‘household profile’ and threat appraisal? It is possible that I am wrong here. Why is ‘past flood experience’ not linked to the other arrows, and in a dashed-line box?
109: “Effectiveness of the PMT framework is limited as a household’s willingness to adopt protective measures in flood risk areas is not considered.” This is surprising, since the paragraphs above seemingly explained how PMT is precisely helpful to conceptualize households behavior towards flood protection measures. Please clarify, because otherwise the need for the transtheoretical model is not motivated.
128: It could be useful to know where those districts/wards are in HCMC, maybe via a map, so that an impression can be gathered of how their position relates to flood-prone areas of the city, potentially discriminating different types of flood, which you mention in the following sentence. This is only a suggestion, not a necessity.
Fig. 2. The figure is correct in principle, but I think it would be much clearer to the reader if a matrix format were chosen instead, with one level of classification along the columns and one along the rows. This is a much more common way to conceptualize the intersection of two classes.
Section 3.3: I suggest making more explicit the relationship between drivers and explanatory variables, and between ‘decisions’ and response variables. I suspect these coincide, respectively, but I am not sure.
L 176: “and when a group of variables have high pairwise correlation, then lasso randomly selects one variable from the group.” It is not clear that groups here consist possibly of more than 2 variables. In this case, there are several pairs of variables that can be correlated, and it is not clear how the variable(s) are selected in case of high pairwise correlations. Also, change to “before the lasso model saturates”.
L 178: it may be due to my lack of familiarity with these methods of regression, but it is not clear to me how terms L1 and L2 play a role in eq. 2, supposedly via hyperparameters alpha an lambda..
198: What do you mean by “aspects of the PMT-TTM framework”? are these the independent/explanatory variables, or classes of them? Linked to this, I cannot understand the following sentence starting with “Since the predictors…”: what are the predictors, again the explanatory variables?
L 206: Please consider whether it is appropriate to report results of the questionnaire as coinciding with information on the actual implementation of the results that respondents state to have implemented. Also, her please explain again what both events are, as the reader can’t be expected to memorize all aspects of the Methods.
232 and throughout the text: ‘dry-proofing’ is commonly a structural measure consisting of preventing water from entering the house. I think that what you mean by dry-proofing is what the literature commonly understands as wet-proofing, i.e., placing elements inside the house on higher ground, so that flood waters entering the house will cause less damage. Just one informal, arbitrary reference: https://www.coastal-management.eu/measure/wet-proofing-sealable-buildings.html
233: “Highest number of respondents have elevated their houses only after experiencing serious and recent flood events (Figure 3) because the flooding is getting worse in HCMC (Paulo and Rivai, 2021).” The causality in this sentence doesn’t seem clear. Also, if possible I would use a source different from a journalistic article to support the higher frequency of flooding in HCMC.
237: it is not clear what the contradiction is, here. Similarly, in the following sentence, the other contradiction needs to be made more explicit. Further, in the following sentence, a vast generalization is proposed about differences between developed and developing countries, on the bases on only three data points: HCMC, Denmark and Germany: this does not seem warranted, or should be played down. Lastly, note that ‘rapidly growing economy’ does not stand in contrast with ‘developed economy’.
251: what are house impacts, in contrast to house damage? Here the shortcoming of not having explained these variables becomes evident. Similarly, one is left to wonder about the meaning of other variables too, like ‘people’. Only later is the reader informed that there is an annex that supports this.
258: this counterintuitive effect is very interesting: could you try to explain the mechanism behind it briefly, also on the basis of the other studies that report it elsewhere?
300. You offer an explanation for the lack of results for this group of households. But it is not clear to me how the methods simply fails to produce any importance level above zero with this dataset: could you also offer an explanation of what happens here methodologically? Also, maybe I missed it, but how come for the non-structural proactive group the lasso and net-elastic models yield precisely the same results?
312: it doesn’t seem that ground elevation and precautionary savings (a concept that requires clarification) belong to dry- or wet-proofing.
326: I don’t think you can state that “costs do not restrict the implementation of precautionary measures in HCMC”. This is an exceptional claim that needs stronger evidence than the lack of correlation between the cost of measure and their rate of implementation.
Technical corrections
I have a number of suggestions regarding readability of the article.
L.13-on: This sentence is huge. It contains both methods and results, whereas methods were exposed already in the previous sentence. Also, for readability, I suggest reversing the sentence, like “Analysis reveals the factors that positively influence the proactive … : education; degree of …”.
L 16: “Households that experienced increasing… were more likely to implement measures reactively” or something similar seems more easy to understand.
L 18: I would leave to the reader to decide what is ‘important’.
25: “long-duration precipitation events”
59 “often not willing to take the responsibility and fail to implement”. Please revise for ambiguity
75: I am not sure, but I reckon HCMC is less than 80 km away from the sea. Please check.
76: ‘even faster’: it’s not clear what the reader should compare the faster future rate or growth to.
129: “households which suffered”
Fig. 3: Building elevation and Elevate are the same measure? Please stick rigorously to the same terminology to prevent ambiguity. Also, is there no respondent that did not answer any question?
211: sentence incomplete.
230: “lack of support to increase responsibility among households to implement other private measures”. It is not clear what this means. Also, if elevation is largely implemented, it is hard to argue that there is a general lack of responsibility regarding implementing measures privately.
232: eliminate ‘yet’.
Section 4.2: title is unclear: why not just “Drivers of…’
L 248: I suggest “except for the group of households that proactively undertook structural measures”. In general, throughout the text, you can turn around many sentences in this way, using verbs and active clauses instead of substantives, improving clarity.
249: again, I don’t think ‘importance’ is clear in this context. Also, there is probably no need to preemptively present what the manuscript section does, here.
325: change to “there is no correlation between the costs of a [type of] measure and its rate of implementation”
328: I would skip “identified a set of important aspects that motivates the implementation of precautionary measures” and straight away recap the key drivers of behavior. In the following sentence, ‘pragmatic’ does not seem the right word. Perhaps ‘activating communication’ or similar.
331: as for other sentences, this should also be turned around for clarity: “The analysis further shows that factor that positively influence the decision of proactive groups are…”. The same goes for the following sentence: “Therefore, to motivate proactive behavior of households …”. Last, I would skip the last sentence, as it doesn’t add anything meaningful.
Citation: https://doi.org/10.5194/egusphere-2022-171-RC2 - AC2: 'Reply on RC2', Thulasi Vishwanath Harish, 24 Aug 2022