the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 13 Jan 2026
A protocol for vulnerability and exposure assessment in rapid extreme event attribution studies
Abstract. Over the past decade, the number of rapid extreme event attribution studies have increased substantially, both in frequency and speed of completion, often released in just a couple of weeks after an extreme weather event. Rapid analysis of vulnerability and exposure is a key complement to the hazard analysis in these studies in order to ensure a more holistic understanding of the drivers of observed impacts. In this paper, we present a method for rapid vulnerability and exposure assessment developed by the World Weather Attribution group. It focuses on the development and use of hazard-specific vulnerability and exposure assessment templates that are applied during a rapid literature review covering media, grey and academic literature. These templates are applied in conjunction with local expert judgement to elicit both breadth and depth of the assessment of potential drivers. This protocol supports the systematic integration of vulnerability and exposure into rapid attribution studies, strengthening their ability to inform the public about changing climate risks.
- Preprint
(494 KB) - Metadata XML
-
Supplement
(281 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-6568', Alexandre Pereira Santos, 13 Feb 2026
-
AC1: 'Reply on RC1', Roop Singh, 21 May 2026
Thank you for your feedback on the paper. We have noted the issues and suggestions you have made and responded to them individually:
- We will more explicitly outline the research question: what protocol can guide a rapid (within 1-3 week) vulnerability and exposure assessment as part of a rapid Extreme Event Attribution method that aims to inform the general public? We will also include literature that justifies the motivation behind including vulnerability and exposure as part of a rapid attribution service, referencing the IPCC risk framework evolution and importance for the general public and other audiences. In particular, WWA attribution studies are shared with the media in the weeks after an extreme event, and focusing only on changes to the hazard potentially risks reinforcing the “natural disaster” narrative. Therefore, including the wider drivers and attenuators of risk in the narrative is critical. We will update the text to more clearly point to this gap, its implications, and the motivation for this protocol to fill a gap in most rapid EEA services to date..
- We aim to explain observed impacts, contextualise the hazard analysis, and identify some amplifying or moderating factors that are apparent in the immediate aftermath of the event. We do not posit that these are the only factors that are driving vulnerability and exposure, as we know that more information about impacts and drivers becomes known in the weeks and months after the event. We think that there is a role for other vulnerability and exposure assessment methods to provide a full, and prioritized view of vulnerability and exposure drivers in the months and years after the event. However, our protocol attempts to fill a gap in vulnerability and exposure assessment and knowledge in the immediate aftermath of an event.
- Regarding the lack of external criteria for evaluating the effectiveness of the protocol. We have reviewed other methods of assessing vulnerability and exposure (e.g. structured expert judgement, disaster forensics) many of which require more than 1-3 weeks to do a full assessment, and therefore are not viable when the goal is to provide a rapid result. However, we do acknowledge and aim to include approaches from the DRR literature that focus on rapid assessments such as a Post-Disaster Needs Assessment and can use those approaches to provide points of comparison, and sufficiency criteria.
- We will go over the manuscript and ensure a more even length across sections.
- We will also include some sufficiency criteria which are not currently explicit in the manuscript. For example, the broad criteria used to select local experts includes geographic and thematic expertise, and availability in the 1-3 week timeframe.
- Regarding the case studies, the cases were chosen to provide a variety of applications of the protocol across geographies, hazards and data availability and we will make the selection criteria more explicit.
- Regarding the methods for section 4, we thank the reviewer for their suggestions, and in concert with the suggestions from reviewer 2, plan to mention this section in the introduction and merge it into section 2 to improve the flow. Further, we will make the review method more robust by re-doing the literature review in Scopus or Web of Science, screen a larger sample, clarify selection criteria, and re-synthesize. We will update the table and the supplementary material accordingly.
- In the discussion, we will more clearly reflect the gap in vulnerability and exposure in attribution studies, and include some examples of vulnerability and exposure factors captured in past studies through our methods that help improve decision-makers' understanding of drivers of impacts.
Citation: https://doi.org/10.5194/egusphere-2025-6568-AC1
-
AC1: 'Reply on RC1', Roop Singh, 21 May 2026
-
RC2: 'Comment on egusphere-2025-6568', Anonymous Referee #2, 26 Mar 2026
This paper presents a protocol for the rapid assessment of vulnerability and exposure within WWA attribution studies. It formalises the methods and approach used to do so over the past years, presents example case studies and outlines some of the limitations of their approach. The hazard-specific templates are a practical innovation which could also be useful to the broader research community. Broadly, the paper is well written. However, the contextualisation of this protocol within the broader literature is missing, conceptual clarity could be sharpened, and more detail about the scientific framing and methodological transparency are required for publication.
General comments
The paper would benefit from being more explicit about what scientific question the rapid assessment is designed to answer. Is the aim to explain observed impacts, contextualise the hazard analysis, identify amplifying or moderating factors, or inform recovery decisions? Clarity on this point would also help frame a more systematic discussion of what the approach can and cannot capture.
The protocol appears largely oriented toward identifying what exacerbated disaster impacts. It would be worth reflecting on how the assessment also captures what lessened impacts, including adaptation measures, early warning systems, and DRR interventions. The discussion section acknowledges a newsworthy bias toward what went wrong, but I believe that this deserves more systematic attention in the methodology itself.
A conceptual figure illustrating the workflow of the protocol would make the methodology easier to follow.
Section specific comments
Section 1: Introduction
- The introduction focuses almost exclusively on WWA and the extreme event attribution literature, but there is a wealth of literature on rapidly assessing vulnerability and exposure in post-disaster contexts within the DRR community (e.g. UNDRR, Sendai Framework monitoring, post-disaster needs assessments) that goes unacknowledged. Situating the paper within this broader landscape would strengthen the paper’s relevance beyond the attribution community (currently a large share of the references are WWA-affiliated publications).
- The exact definitions of exposure and vulnerability used here are missing, and a brief overview of how these concepts have been studied and measured in the past would introduce the methodology more clearly.
Section 2.1: Rapid evidence scan
- Please specify which websites, databases, or news aggregators are typically used. An indicative list would improve transparency and reproducibility of the method.
- A brief description of the event triggering and selection process (covered in Philip et al. 2020) would improve stand-alone readability.
Section 2.2: Engaging local partners
- More information on the disciplinary backgrounds of local partners and how varied perspectives are ensured would be helpful.
- The potential biases from relying on the Red Cross Red Crescent network should be explicitly acknowledged, including how conflicting perspectives are managed.
Section 2.3: Rapid literature review (currently mislabelled as a second section 2.1)
- Please specify the databases and search terms used. Even if this is for a rapid assessment, this information is important for methodological transparency.
- Suggestion to merge Section 4 with this section as the templates fit well here.
Section 2.4: Triangulation of evidence
- This section describes iterative gap identification rather than triangulation in the methodological sense (cross-validation of findings from independent sources). Consider renaming it or clarifying how triangulation is undertaken.
Section 2.5: Quality assurance
- It is not clear whether the quality of individual sources is formally evaluated. Are there criteria for assessing reliability? How are contradictory sources handled? Even a brief set of guiding principles would add rigour.
Section 2.6: Rapid quantitative exposure assessment
- Please justify the use of WorldPop 2020 rather than more recent releases, and acknowledge the limitations of gridded population datasets in data-sparse settings.
- It would be worth noting whether and how trends in exposure over time are considered.
- The paper is clear that quantitative rigour is beyond the scope of rapid assessment. A brief forward-looking reflection on what methodological or data developments might eventually enable more quantitative attribution of vulnerability and exposure contributions would add depth and help frame the paper’s place in a longer research agenda (including recent scientific advancements in earth observation/remote sensing/satellite data).
Section 3: Case studies
- Please briefly explain how these three case studies were selected. Were they chosen to represent a range of hazard types, geographies, or data environments?
Section 4: Updating and strengthening hazard-specific vulnerability templates
- This section appears somewhat abruptly after the case studies, the reader does not have too much context on what is being updated as we have not seen previous versions of the protocol. Consider merging this with Section 2 as mentioned above.
- “The first twenty English-language, peer-reviewed entries sorted by relevance were selected for review”: This does not allow for a systematic review, the relevance filter in Google Scholar does not work so well from my experience.
Section 5: Discussion : lessons learned and limitations
- The path dependency risk introduced by the templates is a valuable observation. The authors could reflect on how this could be monitored and mitigated over time.
- The discussion of media bias could usefully be extended to grey literature, which is also subject to systematic coverage gaps and biases.
- It would be useful to reflect more broadly on how this protocol can be relevant beyond WWA and how others in the scientific community might use it.
Citation: https://doi.org/10.5194/egusphere-2025-6568-RC2 -
AC2: 'Reply on RC2', Roop Singh, 21 May 2026
Thank you for your feedback on the paper. We have noted the issues and suggestions you have made and responded to them by section.
Overall: In this paper, we aim to discuss a protocol that can guide a rapid (1-3 week) vulnerability and exposure assessment and ensure consistency across studies. The intent is to explain observed impacts, contextualise the hazard analysis, and identify some examples of amplifying or moderating factors, while making clear that a rapid approach does not allow us to be comprehensive, nor do a full prioritization. However, in including these additional drivers of risk we are able to provide a fuller picture of how impacts materialize to the general public/other audiences, and also explore potential of levers that could reduce risks in future instances. Furthermore, one of the things that we check for when reviewing is a balance between what worsened impact and protective factors. While this did not come through clearly, we will ensure it’s reflected in the paper. Lastly, we like the suggestion of a conceptual figure, and plan to develop one.
Introduction: The focus on WWA comes from the operational use case of this method, however, we acknowledge the wealth of DRR literature on rapid assessments and will include a review of that work to properly situate this paper. We will also add the IPCC definitions of exposure and vulnerability to the manuscript.
Section 2.1: We have a methodology for triggering which is being prepared for a separate paper but can provide a brief overview here, and can include examples of the types of website, databases and news aggregators that are used.
Section 2.2: In selecting local experts, we aim to have a mix of relevant geographic and thematic expertise, and have developed a roster of external experts which we can expand on here. However, one of the largest constraints is availability during the rapid analysis period, since the timing of these assessments cannot be planned for in advance. The roster of experts does go beyond the Red Cross Red Crescent network, however, we do acknowledge the potential for bias can include that in the manuscript, including a method for managing conflicting perspectives which largely constitutes relying on the existing literature and acknowledging possible gaps, areas of disagreement or questions in the assessment text. We can include an example of an instance where this has occurred in the discussion section to deepen the analysis.
Section 2.3: Thanks for pointing out the mislabeling, we will update this. We will also add the search terms and databases. We will also update and merge section 4 into this section to improve the flow of the paper.
Section 2.4: We are happy to rename the section as suggested in order to improve its accuracy. We do sometimes triangulate information from news and government sources with literature, but we agree that the primary purpose is gap filling.
Section 2.5: Our approach focuses on using multiple lines of evidence, and we agree that we can include some guiding principles in the paper around this. We primarily focus on using high-quality, peer-reviewed literature, and make clear in the text if findings come from empirical sources. When we include news sources, it’s primarily to share the real-world impacts. We also review and include government sources, as well as other forms of grey literature, particularly in regions where there is a gap in the peer-reviewed literature and aim to acknowledge this in the text.
Section 2.6: WorldPOP was selected as the primary population data set due to its global coverage and high data quality, however, we plan to update this as new global datasets become available. Since the rapid exposure assessment needs to be replicable at every event location, a global data set was chosen instead of aiming to search for local data sets which often are scarce or outdated in event locations and not comparable across different locations. WorldPOP data has limitations in terms or resolution and local accuracy and can only be considered as an estimate of population and not exact numbers. Changes in exposure overtime are critical, and we do include that in the analysis where there is literature assessing this for the particular geography. We agree that there are some advancements in quantitative vulnerability and exposure assessments methods that could allow for more rapid quantitative assessments (including in the earth observation space), and we will add a discussion on this to the paper.
Section 3: Yes, we will update the text to share the motivation behind choosing these use cases which primarily help to provide insights into the applicability of this protocol for different hazards, and different regions (some with more data and literature and others with less).
Section 4: As mentioned above, we will sign post this in the introduction and merge into section 2.3. We will also make the review method more robust by re-doing the literature review in Scopus or Web of Science, screen a larger sample, clarify selection criteria, and re-synthesize. We will update the table and the supplementary material accordingly.
Section 5: It is a good suggestion to monitor and mitigate the potential path dependency over time and we can include a discussion of this in this section. We also will add on biases in grey and would also like to reflect potential biases in peer-reviewed literature. Lastly, we can add a discussion on how this protocol can be used by others to improve its applicability.
Citation: https://doi.org/10.5194/egusphere-2025-6568-AC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 1,474 | 670 | 120 | 2,264 | 244 | 83 | 109 |
- HTML: 1,474
- PDF: 670
- XML: 120
- Total: 2,264
- Supplement: 244
- BibTeX: 83
- EndNote: 109
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The preprint presents a protocol for rapid assessment of vulnerability and exposure in impact attribution studies. Its relevance lies in producing secondary evidence on a short time window around an extreme weather event, which can summarise otherwise dispersed evidence, foreground non-climatic drivers of exposure and vulnerability (e.g., socioeconomic or physical vulnerability), and direct future research. The draft does not include a clear research question, despite positioning itself fairly within the rapid attribution practice. Furthermore, this article sits in the science-policy or science-practice interface, which is evidenced by the pragmatic epistemic stance, the short time span for analysis, and the relevance aimed at humanitarian and disaster-response communities. The paper presents a brief context from the literature, the WWA rapid vulnerability and exposure assessment method, three case studies, a recent literature review, a short discussion, and an even shorter conclusion section.
The main merit of the work lies in presenting an extension to a long-practiced protocol (over 10 years of usage are mentioned) to the scrutiny of the scientific community. Adding this significance to the relevance of rapid attribution in shaping the public and scientific discussion afterwards gives a strong sense of relevance to the work. The general structure of the draft is not traditional, but it works fine. There is, however, uneven detail in the sections, leaving important gaps that we will address in detail. Additionally, the manuscript, in the current form, lacks stating its explicit research goal, external criteria for evaluating the effectiveness of the protocol (i.e., from previous research), and has a limited discussion of the added benefits and potential risks of adopting the protocol. We will expand on these topics next.
In terms of structure, the text is very well written and generally very clear. The main gaps lie in the uneven space devoted to the introduction (comparatively large), a short description of the methods (which should be the focus of the protocol), and a general discussion. The introduction does a good job of contextualizing the need for considering exposure and vulnerability in attribution studies in general. It does not acknowledge any conditions, previous studies, or lessons that point to the requirements for vulnerability and exposure protocols in this context. For instance, the preprint states that rapid assessments are scoping in nature, but is there a minimum quantity and quality of information that is necessary? Have other protocols been evaluated that could provide lessons and reasons for improvement? What changes when considering exposure and vulnerability (e.g., new requirements, challenges)? Overall, the introduction is too general to provide guidance to the methods presentation beyond the intrinsic motivations of making better assessments (which are abundant). To be systematic, however, this evaluation should not be limited to intrinsic criteria, but address extrinsic ones, which stem from the literature.
This lack of external references seems to reflect in the cursory nature of the method's presentation, which could benefit from stating what the sufficiency criteria are applied in each stage. In other words, how does the protocol guide in assessing the quality of evidence acquired in its application? As it is, the protocol's methodological considerations sound casual, which is probably not the case in reality. For example, in section 2.2, what are the criteria for selecting the local partners? By having "0-30" experts in different cases, we are left to interpret that either they are not important (when there is none), or one can never have enough of them (when there are 30). Qualitative studies have criteria for sufficiency and quality of evidence (e.g., sampling strategy and data or theoretical saturation) and can guide a better critique of the protocol by the authors. The overall structure of the protocol, however, is sound and makes sense. What is not clear is how its specific sequence, formulation, and guidance help achieve better results than other protocols could. This also hinders the impact of the manuscript, as the added value of the protocol becomes unclear.
Concerning the results, there are two issues. First, each case study has a different level of detail, and many questions remain unanswered in the first one, which is the shortest, for instance. A more even and systematic description of the cases relating to the phases and key steps of the protocol would make for a stronger presentation. The cases are there to test if the assumptions of the protocol hold against reality, and an honest presentation of good and bad performing steps would make for a more credible and grounded protocol presentation.
The second issue is the "update" section (number 4), which has structural, methodological, and content issues. While I appreciate that the traditional format of an academic paper is not set in stone, this section does not relate clearly to the problems raised in the introduction, nor does it respond directly to the methodological presentation. While chronologically it may have been necessary for the authors to go through with the review, that is not necessarily true for the readers. The academic paper has the burden of translating the process into a clean narrative. I am not convinced that this section could not become part of the introduction, strengthening the criteria that the protocol needs to respond to. Methodologically, it suffers from a worrying casual criteria for selecting the literature (i.e., the first 20 results). Again, finding sufficient literature is the absolute key for performing a good review, and even snowballing methods have a guiding line establishing when enough evidence has been collected. Google Scholar also lacks robust classification criteria beyond citations, and is not systematic enough. This issue becomes apparent in the table on page 7, which has many empty cells (e.g., no land use impact of extreme heat or cold) that seem to be an artifact of the search strategy. Moreover, the table is also very general and does little to clarify how the review improved the previous versions of the protocol. The corresponding supplementary material sits on the other extreme, as it is too verbose. My suggestion is to consider if this table could not provide a good framework for the protocol's requirements, sufficiency, and quality criteria in the introduction.
The discussion is good and addresses many of the themes raised in the introduction, with apt references to the case studies. There is still a lack of external criteria to assess the limitations of the protocol (e.g., expert selection criteria, potential confirmation bias). Additionally, there is no assessment of the validity of the results produced by the protocol in its 10-year history. Even if a systematic assessment is out of scope in this case, a summary foregrounding the insights that would not be available otherwise would greatly strengthen the case for the protocol adoption. Similarly, an assessment of mistakes, imprecisions, and reasons for improvement, for example, comparing rapid with longer and more robust attribution methods, would clarify the limitations and trade-offs of the approach. Minimally, readers would expect to know if the protocol reveals vulnerability and exposure that was otherwise ignored or underappreciated.
Overall, the preprint presents an extremely relevant topic but needs extensive review of its presentation to allow a wide readership to benefit from its (apparently) significant findings. The preprint has great potential, but the authors must also reveal the added value of their approach. This is currently only hinted at, but not explicitly stated. With extensive reviews, however, its quality may come to light.