Dear Editor,

Thank you for giving me the opportunity to review the article “Review article: Towards multi-hazard and multi-risk indicators – a review and recommendations for development and implementation”. In the enclosed manuscript, the authors provide a systematic review of recent literature addressing the topic of multi-hazard and multi-risk indicators. The authors clearly identify gaps in the literature and propose a series of recommendations to advance the uptake of indicators in the field of multi-hazard and multi-risk.

General comment:

Authors have undertaken a rigorous literature review, bringing together 194 scientific articles addressing a broad range of natural hazards, methods, and disciplines. The objective of the article is to build on the body of literature to provide recommendations on the development of indicators for multi-hazard and multi-risk for disaster risk management and assessment. Objectives are clear and the article is well written. However, I think that the article misses its announced objectives in its current state. The main findings of the article are gaps in the literature. These gaps indeed exist and are well summarized in Section 4. I believe that there is content in the article, and in the literature reviewed, that could shape recommendations that are more impactful. In particular, I see room for improvements on three broad aspects:

• Disconnection between recommendations (Section 4.3) and results from the literature review:

Recommendations provided in Section 4.3 do not seem to build on the literature reviewed in their current state. The four recommendations on Advancing research into multi-hazard and multi-risk frameworks may result from the fact that only a minority of the reviewed articles address hazard interrelations and impacts. These recommendations are instead very generic, and some have already been made in previous articles (e.g., Ward et al., 2022, Zscheischler et al., 2020). The following eight recommendations also lack connections with results from the literature review.

• Lack of clarity on the content of the reviewed literature:

The methodology, key words and filters used to compile the 194 article is clearly explained in Section 2. However, some information about the nature of articles collected are not provided, resulting in a difficult interpretation of the result part for the reader. In particular, information about the number of hazards represented, preferably from each category used in the article (meteorological, geophysical …) would be beneficial. The number of hazards/interrelations considered would also help to understand the state-of-the art in multi-hazard indicators.

• Presentation of results:

The alternation of usage raw numbers and percentages in section 3 is confusing. The classification chosen for Figure 3, Table 2 and Table 3 is also confusing, as we are seeing percentages of percentages. The Sankey diagram of Figure 3 is hard to interpret and could be improved. Figure 4 is great. The classification made between compound hazard and cascading hazard indicators in Section 3.3 is also not ideal in my opinion.

Finally, I think that the article has the potential to provide more specific recommendations, building on some existing indicators/indexes (especially for multi-hazard indicators). I therefore suggest major revisions, focusing on Section 3 and Section 4.

Specific comments:

1. Line 44-45 p1: “in other words, they did not include the interactions between hazards.” Hazard and also risks, no?
2. Line 203 p10. This division of “multi-hazard” literature in two groups is relevant and has already been done before (e.g. Tilloy et al., 2019), but it is not obvious to me in the context of the article. Could the authors provide a rational for the division?
3. Line 213-215 p11. It is not clear what is the “vulnerability and impact” category. Does that mean that 88% of all articles deal with hazards, 49% with vulnerability and 35% with impact?
4. Figure 3 p11. The figure is confusing as groups for hazards are “yes” and “no” while they correspond to methods used for vulnerability and impact. Another figure showing how groups of literature overlap would be more informative and connected to the text.
5. Table 3 p12. I do not understand why the “no” column only applies to hazard.
6. Line 259 p13: “The majority of geohazards were associated with multi-layer single hazard interactions (77%)” If it is a multi-layer approach, there are no interactions.
7. Line 261 p13. “When it came to multi-hazard interactions, geohazards were 261 almost equally distributed between compound (8%) and triggering and amplification interactions (12%).” That would be great to know which hazards are discussed here.
8. Line 267 p14. “We found that approximately 44% of the 194 articles reviewed were categorised as multi-layer single hazard studies, predominantly focusing on meteorological, hydrological, and geo hazards (Fig. 4).” Figure 4 show hazard count and not paper count.
9. Line 287-291 p14. The classification into five hazard groups mentioned implicitly takes into account hazard interrelations. For example: winter weather would include hazards such as extreme snowfall, avalanches, extreme cold, extreme wind… These hazards are interrelated (compound and triggering). Review the sentence accordingly.
10. Line 341 p15. Same as comment h)
11. Figure 5. Finally we see the hazards! Although I don’t understand how some hazards (e.g., Tsunamis) can be in both categories. The absence of scale make the figure prettier but less interpretable.
12. Line 374 p17. Drought is also a hydrometeorological hazard, it does not make much sense to separate in from flood here.
13. General p17. The separation of indication for compound and triggering interrelations is not very convincing in this context.
14. Line 388 p17. Same as comment h)
15. Line 484 p22. This first recommendation was already made in Kappes et al. (2012) more than a decade ago.
16. Line 487-491 p22. Why “interrelationship” is suddenly being used in these lines (more occurrences here than in the previous 21 pages) ?
17. Line 502 p22. You mean Section 4.2?
18. Line 586 p24. This is not new. It is even a requirement for EU funded projects.
19. Line 588 p24. Be aware that similar tools already exist. Examples: https://drmkc.jrc.ec.europa.eu/risk-data-hub#/atlas; https://climate-conflict.org/www/index/methodology
20. General conclusion: Please state that you identified gaps in the literature
21. Line 618-620 p25. It is not clear for me that recommendations are based on the literature review.

I look forward to reading a revised version if asked to by the editor 

References:

Tilloy, A., Malamud, B. D., Winter, H., & Joly-Laugel, A. (2019). A review of quantification methodologies for multi-hazard interrelationships. Earth-Science Reviews, 196, 102881.

Ward, P. J. et al. Invited perspectives: A research agenda towards disaster risk management pathways in multi-(hazard-)risk assessment. Nat. Hazard. Earth Syst. Sci. 22, 1487–1497 (2022).

Zscheischler, J., Martius, O., Westra, S. et al. A typology of compound weather and climate events. Nat Rev Earth Environ 1, 333–347 (2020). https://doi.org/10.1038/s43017-020-0060-z

Dear Editor,

Thank you for the invitation to review the article “Towards multi-hazard and multi-risk indicators – a review and recommendations for development and implementation” which I enjoyed reading and believe can make a valuable contribution to the literature with minor to medium level revisions. The paper outlines a robust literature review of articles that mention indicators, which are subsequently categorised by approach to multi-hazards and component of risk they address.

The introduction is strong and the recommendations for future work are clear, and I am confident the review has been undertaken rigorously. Most of my comments about the paper closely align with the comments of reviewer 1. I feel the three key areas where the paper could be enhanced are:

1. Presentation of the results

Similar to reviewer 1, the presentation of results in Section 3.1, Figure 3, Table 2 and 3 could be presented more clearly. It is not always clear when you are talking about a subset of the whole corpus of papers, or a subset of a subset, so I suggest using more words to make clear which subset you are referring to and using both raw numbers and percentages e.g., 49% of ‘ABC’ articles (n= XX) where ‘ABC’ might represent ‘all’ or ‘hazards-focussed’ etc.. I like the idea of Figure 3, but I found it somewhat overcomplicates quite a simple paragraph, and some of the percentages in-text do not clearly match the diagram (e.g., 29% of articles mentioned on line 219 seems to equate to 34% of articles in the Sankey diagram – perhaps I am misinterpreting something here, but this emphasises that the diagram is not easy to follow!). Similarly, both Table 2  and 3 would benefit from a more detailed figure caption which indicates what the percentages represent (i.e., % of all articles or % of a subset – I started to get a bit lost on whether rows/columns should add up to 100%) and further labelling of columns/rows to indicate what these represent.

2. Bringing the indicators upfront and further synthesis

I believe the strongest contribution to the literature can make is specific guidance about indicators, rather than further commentary on the complexity and/or absence of genuine multi-hazard indicators. At present, indicators are largely spoken in quite an ‘abstract’ sense until Section 3.2, and then this section is rather descriptive of individual papers rather than giving a strong sense of synthesis. Firstly, I would like to see more examples of specific indicators mentioned in Section 1 (beyond the box about the Sendai framework). I encourage the authors to consider if Table 4 could be expanded upon to really give a sense of the outputs of the systematic review, and brought right to the front of the results section in a brief overview (or if it is large, placed into supplementary material and a summary given in-text). Then, in Section 3.2, most paragraphs could do with a couple of sentences at the end to synthesise the findings of that paragraph (at present, most of the paragraphs start with a signposting sentence and then describe the methods/outputs of papers one by one).

3. Connection between the results and the discussion

Although the recommendations for further action are good, much like reviewer 1, it is sometimes hard to see how these are evidenced by the results section (e.g., the mention of dynamic risk is very limited up until the discussion). I believe that reviewer 1’s suggestions are an appropriate way forward for improving the link between the two sections.

Minor revisions:

Line 83 ‘For example, a precipitation indicator..’ – it would be helpful to add an example indicator here.

Line 136 Typo says ‘Section 33’

Paragraph starting line 148 Add a sentence indicating the process for arriving at this set of search terms to indicate the robustness of the method (i.e., that you are not missing specific terms).

Line 172 ‘Relevance was primarily assessed..’ consider adding the word ‘manually’ here – just to help distinguish that some parts of your methods are done in a semi-automated way using R.

Line 173 It is unclear why only geo and hydrometeorological hazards are considered here – this has not been mentioned up until this point.

Paragraph starting line 206. This paragraph needs expanding upon to explain how you arrived at these five classes and what these classes are (could go into a table).

Line 310 It would be useful to give a broad overview (one or two sentences) with some signposting of what the results section is going to cover .

Table 3 Figure caption – In text you refer to methods rather than indicators, so the word methods should appear in the figure caption.

Figure 4. This is a nice figure but the colour scheme makes it very hard to visualise the smallest categories on both screen and in print. Explore using a diverging colour scheme.

Sentence starting line 278 and ending on line 280 should be supported with a citation.

Figure 5A. There appear to be more hazards than there are circles, which means the labels are misaligned with the circles. This gets particularly hard to read around storm surge and soil salinity. If there is a hazard category that does not correspond to a circle, consider adding a cross or a lighter circle to make this clearer.

Table 4. I like this table and would like to see expansion of this. I would encourage the authors to experiment with fleshing this table out a little bit more and/or adding a visualisation (as mentioned above). A few thoughts for the authors to explore:

• At present, some indicators are listed in detail such as coastal flooding, and then others are spoken about quite broadly such as exposure sensitivity and resilience. The text for other indicators implies that the description in the table is finite, but actually there are many ways of measuring – for example landslides - % of area with steep slopes is rather an approximate indicator compared to others in the literature. I think what is needed is a bit more consistency in how the indicators are outlined. 
• At present, it is not really clear which citation relates to which indicator – for example, only 3 references are given for 7 indicators for hazard. This makes it difficult for the reader to look in further detail as it is not clear which citation relates to which indicator. It might result in a lot of repetition, but could you explore adding the citations in line with the indicators?
• At present, it is hard to get a sense of what ‘sets’ of indicators are appropriate or commonly used – I wonder if the authors could explore a network diagram or Sankey diagram to show this.