the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 23 Jan 2026
Learning from the past to inform flood risk management: Analysis of public survey data in Belgium on flood early warning and response during the July 2021 flood
Abstract. In July 2021 an intense rainfall event resulted in severe flooding in Belgium as well as neighbouring countries. The Walloon Region of Belgium was severely affected with 39 fatalities reported; in the aftermath, the warning system was criticised. In this paper we assess the flood forecasting, warning and response system in the Walloon Region of Belgium for the July 2021 flood event. The analysis is based on an online survey of affected residents (n = 550) and investigates the reception of official warnings, the interpretation and trust in those warnings, and subsequent response behaviour. We find that among the respondents in the Walloon Region 33 % reported not having received any warning, and 56 % did not know how to respond effectively. We analyse the most important influencing factors across the warning chain using the Protective Action Decision Model as a theoretical framework and test influencing factors using logistic and linear regression models. We find that those who were most severely affected at the household level were less likely to receive an official warning. Additionally, flood experience, the level of perceived surprise at the household level, and perceived flood severity significantly influenced whether individuals knew how to respond to the upcoming flood. Despite the flood’s huge magnitude, those who took damage-reducing actions were more likely to report an actual reduction in flood damage. This analysis highlights the need to improve Belgium's flood warning system by ensuring timely issuance of clear warnings and underscores the benefits of enhanced flood risk awareness for damage reduction.
- Preprint
(935 KB) - Metadata XML
-
Supplement
(275 KB) - BibTeX
- EndNote
Status: open (until 19 Mar 2026)
- CC1: 'Comment on egusphere-2025-6376', Julien Magana, 23 Jan 2026 reply
-
RC1: 'Comment on egusphere-2025-6376', Anna Goodden, 20 Feb 2026
reply
‘Learning from the past to inform flood risk management’ surveyed 550 residents in Belgium to assess the flood warning system in Walloon in response to the major floods in July 2021. Overall, this was a really enjoyable read and a well thought through study! The description of your methods were really clear, and your study made coherent sense.
To improve it further, there are a few tweaks that could be made to improve the framing, and overall recommendations presented by the paper:
Major comments:
- The framing could be slightly adjusted in Section 1. The introduction and background explains the PADM model and WMO four pillar framework for early warning systems. But, I think the PADM model (and the significance) of it could have been explained more, as at times it seemed like the paper was going in the direction of focusing on the WMO rather than PADM. This could be addressed by adding a few sentences of why PADM matters before explaining the components of it, and then elaborating further about other studies that have used it, and how/why it’s helpful to understand people’s responses to hazards. You could also reference the WMO more throughout the paper, to keep that link going throughout.
- Similar to this, Figure 3 is great! And should be in the results rather than the conclusion, as it is presenting what the paper has found. I also think this can be used to draw in the PADM model literature referenced in Section 1, and why all this theory matters for your results. Because PADM itself was not mentioned much in your results, I think having a final results paragraph linking all your results together and to the wider PADM literature would be helpful to tie everything together.
- The policy recommendations of the paper for future Belgium flood policies could be clearer throughout Section 4. You use the phrase ‘probably help’ on line 634, which could be tweaked to be more decisive on how (and if) flood hazard maps can impact responses to flood events.
Minor comments:
- Minor referencing error on line 227, where the references are not merged into one list and in the PDF pre-print it says ‘Click or tap here to enter text’.
- Some more small referencing errors on lines 394/5, 489, and 529/30.
- The research questions were very clear, and made Section 3 clear and very easy to follow. Though, sub-sections would benefit from another layer of numbering (e.g 3.1.1, 3.1.2, 3.2.1), as occasionally I lost track of which specific section I was reading.
Citation: https://doi.org/10.5194/egusphere-2025-6376-RC1
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 230 | 67 | 17 | 314 | 39 | 12 | 23 |
- HTML: 230
- PDF: 67
- XML: 17
- Total: 314
- Supplement: 39
- BibTeX: 12
- EndNote: 23
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Very interesting study and I had a very good time reading it, I find it very approachable in the sense that even without prior knowledge we can understand fully the study and its results ! Thank you for this it's very valuable !
In Section 1.1 you indicate that you expand the PADM, maybe it can be nice to explain a bit further on the reasons of your expansion, for example go in more depth on why you are separating pre-decision process and perceptions. Then also what can help you is after your figure when you explain why PADM is suitable for your study explain also why it's limited, as this would pave you the way to justify your expansion (which is a very nice addition by the way).
For Section 1.2, I think you have a lot of visual elements with a clear timeline and where critical decisions must be taken and when, maybe a simple timeline figure could be a nice addition to this section which is rich in information, it would wrap it up and help the reader understand the stakes for the points further discuted in the paper ! "For an early warning system to be effective, warning messages need to be received by users with a sufficient lead time allowing them to 190 take protective action (called response in Fig. 1)", following up my previous comment, this can be very nicely visually represented in the timeline I described I think.
I think your methodology section is solid, valid and well justified, I really liked how it was split in RQs as well, it gives a very clear overview of how you did the study and makes it replicable.
I really like Figure 3 as it wraps up perfectly your results section, maybe you could have intermediary versions of this figure with focus points on the results section and at the end have the overall framework expanded, that way maybe it would make your result section less heavy in text ? A summary table could also be a nice addition at the end of the result section with the main trends found in your results and what recommendations would they lead to potentially? To have this clearer overview of your findings I think that would be a nice addition to work.
I would add also limitations in terms of methodology, I think one critisicm you could receive (which is very easy and common with survey studies) is that you didn't try to have follow-up interviews to confirm at least part of your findings, this is a classic, so maybe it can be nice to mention it in the limitations !
Thank you for the very enjoyable read, great study !