the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 08 Oct 2024
A New Database of Extreme European Winter Windstorms
Abstract. European windstorms pose a significant threat to people, infrastructure and the natural environment. Several windstorms in the recent past have caused substantial damages, and losses associated with extreme windstorms may increase with climate change. Characterizing the footprints of destructive windstorms is thus key to providing quantitative estimates of storm-related economic losses. To that end, we have developed a new, publicly available database of extreme European windstorm footprints for the extended winter season during 1995–2015. In contrast to previously compiled European windstorm databases, we include storm footprints derived from four different data sets, rather than a single source: the ERA5 reanalysis, the COSMO-REA6 reanalysis for Europe, the COSMO-Climate Limited-area Mode regional climate model driven by ERA5 on the EURO-CORDEX domain, and simulation output from the same model but on an enlarged Germany domain with higher horizontal resolution. The database includes both the footprints themselves, expressed as the relative daily maximum wind gusts associated with a storm event, and the daily maximum wind gusts in absolute magnitude associated with the footprints. We derived and included the storm footprints associated with the 50 most extreme storms, or Top50 storms, identified within each of the four input data sets. We applied a consistent methodology, the storm loss index, across input data sets for identifying storm footprints and assessing their severity. This enables a direct comparison between the footprints derived from the different input sources, eases future efforts to extend the time record of the database or to include additional input data sets, and enables assessment of uncertainty in the footprints. Moreover, since we derived the Top50 storms from each input on its native horizontal resolution, the database also allows to characterize the impact that horizontal resolution can have on footprint identification and severity assessment. Our database thus supports both the research community and the insurance industry in exploring the data set and resolution dependence of assessments of extreme storm hazards.
- Preprint
(2422 KB) - Metadata XML
-
Supplement
(1 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on essd-2024-298', Hugo Rakotoarimanga, 18 Feb 2025
General comments
The preprint is of good quality. The methodology is clearly exposed with appropriate references. The manuscript points towards applications to risk assessment and catastrophe modelling of a novel database of Extreme European windstorms. This dataset could specifically be useful to insurers with a European exposure, as the authors provide a clear understanding of the importance of reliable hazard data and the quantification of uncertainty which are current pain points in the practice of catastrophe modelling.
The dataset is novel as it aims at bridging a gap in the existing hazard sources concerning European winter windstorms by providing a consistent and robust approach in the modelling of extreme winds.
Contrasting with other available datasets such as XWS or C3S, one novelty of the dataset is the availability of several ‘realizations’ of each of the Top 50 European windstorms allowing for the quantification and exploration of uncertainties and their causes (such as horizontal resolution of the footprints).Specific comments
- The proposed novel database should be given a name.
- The dataset covers the period 1995-2015 where most of the major storms have occurred in Europe, and the authors’ claim is that the methodology could be easily applied to more recent storms. However, the data used to produce this new dataset seems to be available up to 2019. The reasons why the novel dataset only covers the period up to 2015 should be made explicit.
- The impact assessment of European windstorms is one of the main applications of the novel proposed dataset outlined by the manuscript. Hence the authors have chosen the well-established Loss Index to measure impacts, as there is no publicly available consistent economic or insured exposure database that could directly enable the computation of losses. However, I suggest that it may be more appropriate to explain that risk metrics such as OEP and AEP are proxy quantities, derived from the LI and not from actual economic or insured, for the sake of clarity. This comment is applicable to all mentions of losses.
- A general remark about the figures is that they are quite small, it is a bit difficult to assess visually the differences between the different data sources. For example, Figure 1 is quite clear, but the size of subplots in Figure 2 and the colorbar used for absolute wind gust makes visual appreciation of differences difficult. From an impact perspective it could be interesting to locate major cities (e.g. Cologne, Munich, Berlin…) on the maps.
- I suggest that Figure 4 and 5 should be instead bar plots for readability and as the data represented here is discrete on the x-axis.
- It would be interesting to the reader (especially with a focus on impacts) to represent the differences in terms of Loss Index between the different data sources on a figure, like the plots on Figure 2. This could be integrated at regional (e.g. administrative divisions) or country level.
- Figure 6 is difficult to read. It is informative indeed to keep this kind of figure to compare the differences in terms of normalized loss among the different data sources. Maybe this figure should only represent a fewer number of storms for readability (10?). I would suggest ordering the storms by decreasing normalized loss instead.
- While some storm names are obvious and well-known, some are quite obscure such as storm Fridtjof where a simple web search does not yield much information. It could be useful to always add the year of occurrence next to the storm name for the general reader.
- I suggest that a figure should represent the geographical extent of the different domains considered for clarity and their overlap. This is a very minor comment.
Technical corrections
- Line 103, we suggest replacing "integrated loss" by "integrated proxy of the losses" (see Specific comments).
- Line 416, there seems to be a grammar mistake, “point” should instead be “points”.
Citation: https://doi.org/10.5194/essd-2024-298-RC1 - RC2: 'Comment on essd-2024-298', Anonymous Referee #2, 05 Mar 2025
Data sets
Storm Database Files for A New Database of Extreme European Winter Windstorms Clare Marie Flynn, Julia Moemken, Joaquim G. Pinto, and Gabriele Messori https://zenodo.org/records/10594399
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 471 | 164 | 110 | 745 | 24 | 22 | 21 |
- HTML: 471
- PDF: 164
- XML: 110
- Total: 745
- Supplement: 24
- BibTeX: 22
- EndNote: 21
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1