Chasing a millennium of European storms: a synthesis of three decades of multidisciplinary research

Pouzet, Pierre

doi:10.5194/egusphere-2026-3658

Preprints

https://doi.org/10.5194/egusphere-2026-3658

Preprints

15 Jul 2026

| 15 Jul 2026

Chasing a millennium of European storms: a synthesis of three decades of multidisciplinary research

Pierre Pouzet

Abstract. Understanding the long-term variability of coastal storms is essential for assessing future coastal risks in a context of accelerating climate change and sea-level rise. This study synthesizes eighty-one historical and geophysical storm chronologies from western Europe to investigate storm activity over the last millennium. To compare heterogeneous records, a decadal Storm Index (SI) was developed to quantify the relative occurrence of storm impacts within eight geographical regions and was visualized using a novel “storm stripes” approach designed to highlight both temporal and spatial variations in storm activity.

The synthesis reveals substantial regional differences in storm occurrence through time. The decrease in air and Sea Surface Temperatures (SST) since 1300 during the early LIA may have favoured higher storm activity in northern Europe, peaking around 1450–1470 during one of its coolest phases, characterized by weak northern European SST and a significant AMO trough. While the early LIA was mainly driven by NAO+ patterns, transitional NAO+ to NAO- phases may have led to the stormiest periods in the north. Conversely, the increase in air temperature and Mediterranean SST during the late LIA may explain the significant later stormy phase in the western Mediterranean, consistent with a large-scale NAO- phase but with its stormiest peak (1760–1850 AD) linked to a NAO- to NAO+ transitional period at a finer temporal scale. These contrasting patterns appear broadly consistent with variations in the North Atlantic Oscillation, although the influence of other modes of climate variability remains uncertain. By combining historical and geophysical evidence at a continental scale, this study provides a new perspective on the long-term evolution of European coastal storm activity. The results highlight the complexity of storm–climate interactions and demonstrate the value of multidisciplinary paleostorm reconstructions for improving our understanding of future climate adaptation and coastal mitigation strategies.

Received: 22 Jun 2026 – Discussion started: 15 Jul 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Pierre Pouzet

Status: final response (author comments only)

RC1:
'Comment on egusphere-2026-3658', Lisa Orme, 30 Jul 2026
The manuscript presents a compilation of records of storms spanning the last millennium, including documentary evidence and geophysical archives. An interesting storm index is calculated for different regions for each decade, allowing spatial differences in storminess over time to be estimated. The results indicate high storminess in the British Isles during the early LIA and then a southward shift during the mid and late LIA to higher storminess in southern Europe, which is an important finding. In the discussion these changes are related to sea temperature patterns and modes of atmospheric circulation (NAO and ENSO).
Main comments
The paper does not acknowledge some of the limitations within the geophysical datasets that may have an influence on the compilation. Bioturbation and sediment movement may blur the storm deposits over depths upwards and downwards, resulting in broader stormy periods than actually occurred. Some proxies may be linked with human disturbance, for example sand dune disturbance may have occurred from human actions not just storms, and flood deposits may also be reflecting human activity in the catchment. Temperature impacts on vegetation may also impact some of the reconstructions by enhancing aeolian/river erosion. More consideration of these challenges in section 1.3 is needed. I feel that your approach of combining many different types of proxies is one way that some of these proxy limitations can be minimised, as a storm signal across many types of record and locations is more likely to be reflecting storms. I would like to see this strength stated more explicitly and also the limitations considered. An explanation of why you have included flood records would also be beneficial to the manuscript (i.e. include modern evidence of the correlation between storm frequency and precipitation / floods) and could be added to section 2.2

In addition to the proxy limitations, the methods state at line 197 that where there was age uncertainty all decades within the range were considered. Does this mean that storm event layers with age uncertainty of 100 years (for example) were marked as stormy for each of these decades? Perhaps I have understood this wrong, but this could also create broader periods of high storminess within the results than realistically occurred. Many geophysical records would have quite wide age uncertainties and it would perhaps be better to include just the median or mean estimated ages of the stormy intervals instead of the full range.

In the discussion the relationship with the NAO, AMO and ENSO is discussed. It is interesting that the interplay between the NAO and SST’s could have caused the shift in storm track between northern and southern Europe over the LIA. I wonder however how far the prevalent storm track is influencing the ocean surface temperature rather than the reverse? Strong winds can create cooler ocean temperatures. Also how far are both SST’s and storminess being influenced by external drivers? They may change at the same time due to solar or volcanic drivers, without SST causing the storm changes or vis-versa.

The ENSO paragraph considering modern evidence could be made more succinct (lines 420-434) while the discussion would be improved with more explanation of the two proxy records (how they were made, why they differ so much, what granularity means).

Some additional proof reading of the manuscript is needed, I have highlighted some possible changes in my comments below.

Other points:
- should the years have C.E or A.D. throughout?
- Check that first mention of all acronyms (e.g. NAO, AMO) are written in full and abbreviated after.
- ‘millenary’ is an uncommon word for the last millennium, perhaps consider using millennium or millennia?
- Line 39-41 – quite a limited list of references for a broad topic, so perhaps put ‘e.g.’ Also are there any references for the last sentence about how palaeo records are used to refine models?
- Line 63 – references needed for European overwash deposits
- Line 65 – 74 – This paragraph doesn’t include the evolution of the peatland based aeolian reconstructions during the 2000’s (Bjorck and Clemmensen, 2004; De Jong et al, 2006; 2009). Similarly at line 77 the peatland aeolian records should be acknowledged alongside those studies using marine incursions.
- Line 80 – more references are needed in addition to Goslin to represent the different proxies now being used
Line 107 – ‘If European intense storms have always occurred throughout history’, consider removing this statement as rather general. ‘the variation… mostly depends on …NAO, AMOC or temperature variation’ this phrasing should be made less certain I think, as the literature debates the causes of storm variability.
Line 133 – consider putting some example references for each of these proxy types. I am unaware of the coral based storm records, and cannot find any listed in Table 1, so this would be beneficial. Alternatively refer to Table 1 and make the geophysical archives listed in Table 1 more clear (rather than having ‘Multiple types’ listed), which may improve the clarity.
Line 136 – Orme et al 2016 paper has river inwash deposits within a coastal lake, not marine overwash deposits
Line 173 - 174 – ‘data had to be critically assessed regarding authorship (e.g. whether the author was a direct witness…)’ why is this important?
Table 1 - Several storminess reconstructions are missing from the compilation. Were these removed through the screening process or missed? I have noticed the following ones are not included but please check for others:
> Orme et al., 2016 peatland records from Scotland in Quaternary Science Reviews
> Orme et al. 2017 a storm record from Spain in Geology
> Orme et al 2026 a storminess record from Ireland in Climate of the Past
> Hass (1996) an ocean sediment based storm reconstruction in Palaeogeography, Palaeoclimatology, Palaeoecology
> Vaasma et al. 2025, peatland storminess records in The Holocene
> Vandel et al. 2019 a storm record from Estonia in Quaternary Science Reviews
> Nielsen et al 2016 in Quaternary Science Reviews (although perhaps too far north for your study)
> Nielsen et al 2024 in The Holocene (also may be too far north)
Line 246-249 – put e.g. in front of these references as they are only a subset
Line 258 – ‘humid archives’ is an unusual description of peat and lake archives, what about wetland archives?
Line 274 and 286 – I am unsure about the inclusion of periods described as ‘climate instability’ or ‘climate deterioration’ as these are ambiguous terms and could refer to colder periods not storms. Please consider if these records/periods should be included in the compilation.
Line 296 – “vimers” I have never heard of this term related to storms and it needs an explanation or translation in brackets
Line 363 – the A3 and A4 areas are not at lower latitudes than the British isles area (although they don’t extend as far north). Would it be more accurate to say the storm track may have shifted south and east? So storms may have crossed France and tracked NE over southern Scandinavia?
Line 370 – ‘These two most significant and clear chronologies’ the focus on these two studies needs more justification and it is not clear why these are the most significant. Perhaps say they are geographically distinct to capture the southern and northern storm track positions? Or the areas most influenced by the NAO currently?
Figure 6 – mark on graph with arrows/labels which direction is el nino and la nina, or say in caption that shaded areas show El Nino

Minor corrections
- Line 33 - ‘of’ not ‘off’
- Line 34 – remove ‘As’
- Line 45 – not sure what coastal sensible systems are?
- Line 56 – Three not tree
- Line 73 – rather than macrotidal, coastal may be more accurate
Line 98 – ‘recent decades’ not ‘the last decades’ (and remove repetition of recent decades from next line)
Line 124 – perhaps ‘temporal approach to understand’ rather than ‘temporal depth to apprehend’
Line 141-142 – consider rephrasing the first sentence to explain more clearly and concisely what is meant
Line 296 – change ‘opposite’ to ‘other hand’.
Line 306 – change ‘excepted for the ‘ to ‘Aside from’
Line 352-357 – this is a long sentence, please revise and use fewer semicolons
Line 411 – think you mean early not late LIA.
Line 412 and 425 – no need to have the reference in both the sentence and in brackets
Line 422 – unclear what symmetric means in this context
Line 437 – use a different word than ‘infirm’, perhaps ‘contradict’.
Line 463 – perhaps ‘test’ rather than ‘precise’
Citation: https://doi.org/10.5194/egusphere-2026-3658-RC1
- AC1:
  'Reply on RC1', Pierre Pouzet, 03 Nov 2026
  The manuscript presents a compilation of records of storms spanning the last millennium, including documentary evidence and geophysical archives. An interesting storm index is calculated for different regions for each decade, allowing spatial differences in storminess over time to be estimated. The results indicate high storminess in the British Isles during the early LIA and then a southward shift during the mid and late LIA to higher storminess in southern Europe, which is an important finding. In the discussion these changes are related to sea temperature patterns and modes of atmospheric circulation (NAO and ENSO).
  I first want to thank the reviewer for taking the time to examine this work and to propose a constructive and detailed review of this manuscript. As requested by the journal, I am first providing responses to the reviewer’s comments to facilitate the discussion. The manuscript will then be revised considering these valuable comments, which will help to greatly improve the paper.
  Main comments
  The paper does not acknowledge some of the limitations within the geophysical datasets that may have an influence on the compilation. Bioturbation and sediment movement may blur the storm deposits over depths upwards and downwards, resulting in broader stormy periods than actually occurred. Some proxies may be linked with human disturbance, for example sand dune disturbance may have occurred from human actions not just storms, and flood deposits may also be reflecting human activity in the catchment. Temperature impacts on vegetation may also impact some of the reconstructions by enhancing aeolian/river erosion. More consideration of these challenges in section 1.3 is needed. I feel that your approach of combining many different types of proxies is one way that some of these proxy limitations can be minimised, as a storm signal across many types of record and locations is more likely to be reflecting storms. I would like to see this strength stated more explicitly and also the limitations considered. An explanation of why you have included flood records would also be beneficial to the manuscript (i.e. include modern evidence of the correlation between storm frequency and precipitation / floods) and could be added to section 2.2
  
  Thank you for this relevant and important comment. Firstly, it is important to remind that the aim of this paper is to synthesize existing storm chronologies rather than to reassess the sedimentological interpretations of each individual study. However, I agree that the limitations of the underlying proxies and their potential influence on the compilation are not sufficiently discussed in the current manuscript. Section 1.3 will therefore be expanded to address these issues, including the potential effects of bioturbation, sediment redistribution and human activity on some records that you mention. The benefits of combining different types of archives will also be made more explicit, as agreement between independent records can provide stronger evidence of past storm activity.
  Regarding the fluvial records, thank you for this suggestion as well. The flood terminology for these studies reflects how the events are described by the authors of the original studies, while the aim of the compilation is to include only events identified as being related to storms. The selection will therefore be reviewed to ensure that this criterion has been consistently applied, and the methodology will be expanded to better explain the rationale for including these records. If they are not relevant, we may exclude flood terminology related records in the second version of this work. The revised manuscript will also clarify the distinction between the terminology used in the source studies and the criteria applied in this compilation, with concrete examples of included and excluded flood terminology related records.
  In addition to the proxy limitations, the methods state at line 197 that where there was age uncertainty all decades within the range were considered. Does this mean that storm event layers with age uncertainty of 100 years (for example) were marked as stormy for each of these decades? Perhaps I have understood this wrong, but this could also create broader periods of high storminess within the results than realistically occurred. Many geophysical records would have quite wide age uncertainties and it would perhaps be better to include just the median or mean estimated ages of the stormy intervals instead of the full range.
  
  Yes, this is an important methodological challenge. Precise dates directly linked to storm events are used whenever available. When no precise date is available, the full age uncertainty range is considered to make sure that the event – or the stormy phase- is captured. Using only the median or mean estimated age could help limit the temporal spread of the reconstructed storm signal, but it could also result in some events being missed or assigned to the wrong decade. Over the last 1,000 years, many chronological constraints have relatively limited uncertainties: for example, some records dated using Pb/Cs have uncertainties of only a few decades, while radiocarbon dates often have uncertainties of less than a century, although this varies between records. This is one of the reasons why a decadal rather than an annual resolution was chosen. That said, it is true that this approach can affect the results by potentially extending periods of high storminess, and this limitation and possible bias should therefore be clearly acknowledged in the manuscript ; and it will be addressed more specifically in the revised version.
  In the discussion the relationship with the NAO, AMO and ENSO is discussed. It is interesting that the interplay between the NAO and SST’s could have caused the shift in storm track between northern and southern Europe over the LIA. I wonder however how far the prevalent storm track is influencing the ocean surface temperature rather than the reverse? Strong winds can create cooler ocean temperatures. Also how far are both SST’s and storminess being influenced by external drivers? They may change at the same time due to solar or volcanic drivers, without SST causing the storm changes or vis-versa.
  
  Thank you for this very interesting comment. I agree that the relationships between the NAO, SSTs and storm-track variability need to be interpreted with caution. The discussion will be revised to take into account the fact that these relationships may work in both directions, as storm activity can also influence SSTs. The possible role of common external drivers, such as solar and volcanic forcing, will also be considered. The aim will be to make it clearer that these co-variations do not necessarily indicate a direct causal relationship, and that the relative contributions of these different mechanisms remain uncertain.
  The ENSO paragraph considering modern evidence could be made more succinct (lines 420-434) while the discussion would be improved with more explanation of the two proxy records (how they were made, why they differ so much, what granularity means).
  
  This section will be revised for the next submission.
  Some additional proof reading of the manuscript is needed, I have highlighted some possible changes in my comments below.
  
  Other points:
  - should the years have C.E or A.D. throughout?
  Yes, thanks for your vigilance, we will use C.E to follow the editorial requirements of COTP.
  - Check that first mention of all acronyms (e.g. NAO, AMO) are written in full and abbreviated after.
  This will be checked in the entire manuscript.
  - ‘millenary’ is an uncommon word for the last millennium, perhaps consider using millennium or millennia?
  Thank you for the suggestion. The term will be replaced with “millennium” where appropriate.
  - Line 39-41 – quite a limited list of references for a broad topic, so perhaps put ‘e.g.’ Also are there any references for the last sentence about how palaeo records are used to refine models?
  Yes it’s true that it’s a bit restrictive; this sentence and the references linked will be enhanced.
  - Line 63 – references needed for European overwash deposits
  Thanks for pointing this out. The missing references will be added in the revised version.
  - Line 65 – 74 – This paragraph doesn’t include the evolution of the peatland based aeolian reconstructions during the 2000’s (Bjorck and Clemmensen, 2004; De Jong et al, 2006; 2009). Similarly at line 77 the peatland aeolian records should be acknowledged alongside those studies using marine incursions.
  Thanks for this reminder. This approach may have been unintentionally underrepresented in the current version of the manuscript, and more information will be added in the revised version.
  - Line 80 – more references are needed in addition to Goslin to represent the different proxies now being used
  Yes it’s true that more details can be offered here, more references will be added to be more representative of this literature.
  Line 107 – ‘If European intense storms have always occurred throughout history’, consider removing this statement as rather general. ‘the variation… mostly depends on …NAO, AMOC or temperature variation’ this phrasing should be made less certain I think, as the literature debates the causes of storm variability.
  That's true, the drivers of storm variability are still debated, this section will be written with more cautious in the revised version.
  Line 133 – consider putting some example references for each of these proxy types. I am unaware of the coral based storm records, and cannot find any listed in Table 1, so this would be beneficial. Alternatively refer to Table 1 and make the geophysical archives listed in Table 1 more clear (rather than having ‘Multiple types’ listed), which may improve the clarity.
  Thanks for this suggestion. The references of this paragraph will be reviewed, and the distinction between geophysical and historical archives will be clarified in Table 1. The “Multiple types” category will also be specified more clearly to avoid ambiguity ; and it is possible that this category will be withdrawed.
  Line 136 – Orme et al 2016 paper has river inwash deposits within a coastal lake, not marine overwash deposits
  Sorry for this mistake, this will be corrected in the new version of the database
  Line 173 - 174 – ‘data had to be critically assessed regarding authorship (e.g. whether the author was a direct witness…)’ why is this important?
  This is important for the analysis of historical archives because the authorship of the historical document (and particularly whether he/she was a direct witness to the event), helps assess the reliability of the source. First-hand accounts may provide more direct evidence, whereas second-hand accounts may involve a possible reinterpretation of information. This assessment therefore helps to evaluate the strength of the evidence supporting the documented storm event. These details will be also added in the paper to be more specific.
  Table 1 - Several storminess reconstructions are missing from the compilation. Were these removed through the screening process or missed? I have noticed the following ones are not included but please check for others:
  Thank you very much for this suggestion that will improve the database. Orme et al., 2016 was already present (ref 6), but the other listed below will be added (to be confirmed for the two refs of Nielsen depending on their location; that may be integrated in a new Boreal Europe area with Iceland records).
  > Orme et al., 2016 peatland records from Scotland in Quaternary Science Reviews
  > Orme et al. 2017 a storm record from Spain in Geology
  > Orme et al 2026 a storminess record from Ireland in Climate of the Past
  > Hass (1996) an ocean sediment based storm reconstruction in Palaeogeography, Palaeoclimatology, Palaeoecology
  > Vaasma et al. 2025, peatland storminess records in The Holocene
  > Vandel et al. 2019 a storm record from Estonia in Quaternary Science Reviews
  > Nielsen et al 2016 in Quaternary Science Reviews (although perhaps too far north for your study)
  > Nielsen et al 2024 in The Holocene (also may be too far north)
  Line 246-249 – put e.g. in front of these references as they are only a subset
  Ok, this will be added.
  Line 258 – ‘humid archives’ is an unusual description of peat and lake archives, what about wetland archives?
  Yes, I agree that it’s a better description.
  Line 274 and 286 – I am unsure about the inclusion of periods described as ‘climate instability’ or ‘climate deterioration’ as these are ambiguous terms and could refer to colder periods not storms. Please consider if these records/periods should be included in the compilation.
  These studies were initially selected because they discuss storm-related impacts (other studies have not been included in the compilation). However, I agree that terms such as “climate instability” or “climate deterioration” can be ambiguous. All chronologies based on such terms will therefore be reassessed to ensure that their inclusion in the compilation is justified by evidence of storm activity. Other studies that do not provide sufficient evidence of storm-related impacts will be excluded.
  Line 296 – “vimers” I have never heard of this term related to storms and it needs an explanation or translation in brackets
  It’s a regional French ancient spelling for storms and marine floodings (Jean-Luc Sarrazin, 2012), you can have more information here : https://journals.openedition.org/norois/4034 ; More information will be added about this subject in the manuscript to be more specific.
  Line 363 – the A3 and A4 areas are not at lower latitudes than the British isles area (although they don’t extend as far north). Would it be more accurate to say the storm track may have shifted south and east? So storms may have crossed France and tracked NE over southern Scandinavia?
  Thanks for this suggestion. This interpretation will be considered in the revised version of the manuscript, with particular attention to the possible southward and eastward shift in storm tracks (it will also depend on the new results obtained after the current update of the database).
  Line 370 – ‘These two most significant and clear chronologies’ the focus on these two studies needs more justification and it is not clear why these are the most significant. Perhaps say they are geographically distinct to capture the southern and northern storm track positions? Or the areas most influenced by the NAO currently?
  It is right, the current sentence is not very clear. These two chronologies were selected because they represent geographically distinct northern and southern areas, allowing a comparison of storm activity under different NAO regimes. This aspect will be clarified in the revised manuscript.
  Figure 6 – mark on graph with arrows/labels which direction is el nino and la nina, or say in caption that shaded areas show El Nino
  OK, more information will be added for this part of the Figure.
  Minor corrections
  Thank you very much for these helpful comments and suggestions. All the minor corrections proposed below will be considered in the next version of the manuscript. I am also very grateful for the care and attention you have devoted to reading and reviewing this work. Your perspective as a specialist in geophysical archives is particularly valuable for this interdisciplinary work, and I am very pleased to have had the opportunity to benefit from your expertise.
  - Line 33 - ‘of’ not ‘off’
  - Line 34 – remove ‘As’
  - Line 45 – not sure what coastal sensible systems are?
  - Line 56 – Three not tree
  - Line 73 – rather than macrotidal, coastal may be more accurate
  Line 98 – ‘recent decades’ not ‘the last decades’ (and remove repetition of recent decades from next line)
  Line 124 – perhaps ‘temporal approach to understand’ rather than ‘temporal depth to apprehend’
  Line 141-142 – consider rephrasing the first sentence to explain more clearly and concisely what is meant
  Line 296 – change ‘opposite’ to ‘other hand’.
  Line 306 – change ‘excepted for the ‘ to ‘Aside from’
  Line 352-357 – this is a long sentence, please revise and use fewer semicolons
  Line 411 – think you mean early not late LIA.
  Line 412 and 425 – no need to have the reference in both the sentence and in brackets
  Line 422 – unclear what symmetric means in this context
  Line 437 – use a different word than ‘infirm’, perhaps ‘contradict’.
  Line 463 – perhaps ‘test’ rather than ‘precise’
  
  Citation: https://doi.org/10.5194/egusphere-2026-3658-AC1
RC2:
'Comment on egusphere-2026-3658', Tim Soens, 14 Oct 2026
This is an extremely relevant paper: the impact of climate change on extra-tropical storms remains highly uncertain, and the awareness of this among scientists, but also a wider public, is limited (a lot of research still automatically assumes that the present warming automatically triggers more storm activity everywhere). At the same time, it is highly challenging to reconstruct storminess for the past 1K years. For the pre-instrumental period there are different proxies, but these are far less homogenous than for instance for temperature or precipitation reconstruction. This paper hence calibrates in a way the different reconstructions for different macro-regions within Europe.
While the paper is innovative, timely, and of tremendous importance for a broad scholarship, there are still several issues I would invite the authors to comment on, and/or to clarify in the paper:
Source selection: I doubt whether it makes sense of including pure ‘flood’ reconstructions, for instance 28, Peng et al., which offers a reconstruction of river paleofloods in the Meuse basin, with no relation to storms.

Sources: For the British Isles and the Southern North Sea I miss the work and index compiled by Adriaan de Kraker (for instance Storminess in the Low Countries, 1390-1725, Environment and History 2013 Vol. 19 Issue 2 Pages 149-171 DOI: 10.3197/096734013X13642082568570) and Mark Bailey et al. (The transformation of the Suffolk coast 1200 to 1600: From Orford Ness to Gosefor, The Proceedings of the Suffolk Institute of Archaeology & History 2021 Vol. 45 Pages 86-111), who use an important historical series by Sears, for the East Anglian coast

Analysis:
The aggregation and tabulation of the different series helps to detect patterns. But how reliable is the calculation of the Storm Index per decade, when the number of studies is limited? The Storm Index, if I understand it correctly, compares the number of studies mentioning an impactful storm in a decade compared to the number of studies covering this area in this decade. But for some regions only two, three or four studies are available? So, an omission in one source, easily impacts the index?

And what about the local versus the regional indices? Are these treated at the same level? An isolated storm record in one locality and a regional analysis covering the entire Southern North Sea? If only the latter, but not the former indicates a major storm in a decade, is the index then 0.5? This would be an underestimation, as local conditions might always impact the storm signal? Perhaps there is no way of avoiding this, but the uncertainties could be discussed.

I noticed a possible problem, when confronting the observation that record density increases markedly after 1300 on the one hand and the conclusion that this period also sees higher storm activity, which is explained by the transition from warm medieval climate to LIA (or NAO+ conditions to NAO- conditions). Should we not be more careful, given the increase in record density? Would it be worthwhile to confront more explicitly the reconstructions based on historical data and those solely based on geophysical proxies? Because it is mostly the historical data which are not evenly spread in time, and more importantly, only become dense enough after 1300 CE. One of the conclusions (at least for the British Isles) is that storminess also increases after 1300, which is linked to the transition from the MWP to the LIA. But, there always is the possibility that the historical sources in particular underestimate storm activity in the pre-1300 period, so that the ‘increase’ is rather the result of a bias in the sources. Disambiguation between historical sources and geophysical ones could solve this?

Can we control for tidal activity? Many of the proxies used for storminess are subject to interference by tidal conditions: the ‘storm surges’. If a storm coincides with high tide, the impact is self-evidently larger, and it impacts many of the signals (notably those via floods or waves). For older storms there is no way of controlling for the tide, but if the storm is accurately dated, this becomes possible.

On p. 20 it is suggested that “climatic instability” is itself taken a proxy for storminess ("Climatic instability”, “Climatic deterioration episodes”, “Cold and windy climate phases” describe broader climatic conditions operating over longer temporal scales, which may favour increased storm activity without directly referring to specific storm, wind, or flood parameters). If formulated this way, there seems to be a risk of circular reasoning: because the climatic deterioration is both used to reconstruct storminess and to explain it? (while the climate reconstructions are mostly used as explanation, in a second stage, which is more interesting).

Interpretation/discussion:
Overall I think the observation of a shift in storm activity from North to South is an important one, and in line with the expectations from the literature.

Can NAO+ to NAO- be considered as homogenous periods, with a transition from one to another? Are the signals homogenous enough?

The frequency versus intensity problem is mentioned briefly in the conclusions, but deserves further elaboration. What do the data capture? See the observation by Trouet et al. (2012), that NAO + is associated with higher storm frequency, but NAO- higher storm intensity: so, less frequent, but more intense storms. In the Southern North Sea the 1950s are now pointed out as an extremely ‘stormy’ decade, but this probably is the result of the 1953 storm, a highly intense (and catastrophic) storm, but not necessarily an indication of a decade with a lot of storm activity? Inversely, the 1450-70 period is now conceived as the most stormy period along the English coast, but is this the result one or two intense storms or rather an increase in the frequency of storms?

Storm trajectories are only mentioned as possible follow-up, but both in previous historical reconstructions (Van Vliet-Lanoë, 2014), and in discussions on the impact of climate change on storminess today (Calafat et al. 2022 in Nature; Karwat et al. 2022 JGR Atmospheres) there is already the suggestion of different regional storm trajectories forged by changes in NAO and AMOC explaining for instance why storm activity might be decreasing in the Southern North Sea and along the Atlantic coast of France, while increasing in the north of the British Isles. This literature could be interesting to inteprete the shifting patterns in the past as well?

In their 2014 study on Skagen, Clemenssen et al. notice a clear cyclical pattern in storm activity from 1850 onwards, without clear long-term signal, apart from a marked higher storminess between 1860-75. Are both results relevant for the present discussion? And why is their chronology so different from the present one? (on Figure 5 1860s and 1870s in the Kattegat region do NOT stand out as particularly stormy?).

I hope these comments are useful, and can help to further improve the paper into a cornerstone of future storminess research!
Citation: https://doi.org/10.5194/egusphere-2026-3658-RC2
- AC2:
  'Reply on RC2', Pierre Pouzet, 03 Nov 2026
  This is an extremely relevant paper: the impact of climate change on extra-tropical storms remains highly uncertain, and the awareness of this among scientists, but also a wider public, is limited (a lot of research still automatically assumes that the present warming automatically triggers more storm activity everywhere). At the same time, it is highly challenging to reconstruct storminess for the past 1K years. For the pre-instrumental period there are different proxies, but these are far less homogenous than for instance for temperature or precipitation reconstruction. This paper hence calibrates in a way the different reconstructions for different macro-regions within Europe.
  I would first like to sincerely thank the reviewer for this very positive and encouraging assessment of the manuscript. I am particularly grateful for these thoughtful comments and for the recognition of the challenges of this work. As requested by the journal, I am first providing responses to the reviewer’s comments to facilitate the discussion. The manuscript will then be revised considering these valuable comments. I will do my best to address the reviewer’s interesting suggestions as thoroughly as possible to then improve the manuscript.
  While the paper is innovative, timely, and of tremendous importance for a broad scholarship, there are still several issues I would invite the authors to comment on, and/or to clarify in the paper:
  Source selection: I doubt whether it makes sense of including pure ‘flood’ reconstructions, for instance 28, Peng et al., which offers a reconstruction of river paleofloods in the Meuse basin, with no relation to storms.
  
  Thank you for raising this important point. The terminology used in these studies reflects how the events are described by the original authors, while the aim of our compilation is to include only events identified as being related to storm activity. Sedimentary flood records can provide complementary evidence of past hydroclimatic variability, but this does not necessarily imply a direct link with storms. After reviewing Peng et al. (2019), I agree that this study reconstructs flooding phases in the Meuse basin without establishing a clear link to storm activity, and it will therefore be excluded. By contrast, Hess et al. (2024) provide an example of how coastal sedimentary records can be used to reconstruct storm-induced overwash events. In this case, the terminology relates to flooding impacts, but the link with storm activity is explicitly established. All records described using flood-related terminology will be reviewed to ensure that this criterion has been consistently applied. If some studies are finally not relevant, they will be withdrawed from the database in the second version. The distinction between the terminology used in the original studies and the criteria applied in our compilation will also be clarified in the revised manuscript.
  Hess and al. 2024 : https://onlinelibrary.wiley.com/doi/10.1002/jqs.3568
  Sources: For the British Isles and the Southern North Sea I miss the work and index compiled by Adriaan de Kraker (for instance Storminess in the Low Countries, 1390-1725, Environment and History 2013 Vol. 19 Issue 2 Pages 149-171 DOI: 10.3197/096734013X13642082568570) and Mark Bailey et al. (The transformation of the Suffolk coast 1200 to 1600: From Orford Ness to Gosefor, The Proceedings of the Suffolk Institute of Archaeology & History 2021 Vol. 45 Pages 86-111), who use an important historical series by Sears, for the East Anglian coast
  
  Thank you for these very useful suggestions. These studies will be reviewed and incorporated into the database where relevant, to ensure that these historical sources are properly represented in the entire compilation.
  Analysis:
  
  The aggregation and tabulation of the different series helps to detect patterns. But how reliable is the calculation of the Storm Index per decade, when the number of studies is limited? The Storm Index, if I understand it correctly, compares the number of studies mentioning an impactful storm in a decade compared to the number of studies covering this area in this decade. But for some regions only two, three or four studies are available? So, an omission in one source, easily impacts the index?
  
  I fully agree with the reviewer that this is an important limitation. The Storm Index may be sensitive to quantitatives differences in documentation in regions with fewer available sources, where a single record can have a greater influence on the result (for instance in the Southern Nort Sea area). This is one of the reasons why the climatic comparison in the final section of the paper focuses primarily on regions with a relatively large number of sources (i.e. A2 British isles vs A7 mediterranean area ; with a bit of A3 Kattegat and Baltic sea secondarily). Nevertheless, I agree that this limitation should be made more explicit, and futher explanation will be added to the manuscript to discuss this limitation.
  And what about the local versus the regional indices? Are these treated at the same level? An isolated storm record in one locality and a regional analysis covering the entire Southern North Sea? If only the latter, but not the former indicates a major storm in a decade, is the index then 0.5? This would be an underestimation, as local conditions might always impact the storm signal? Perhaps there is no way of avoiding this, but the uncertainties could be discussed.
  
  Thank you for this important comment. In the compilation, each chronology contributes equally to the Storm Index, regardless of its spatial resolution. This is a deliberate methodological choice, as a local record should not be considered less valuable simply because it documents storm impacts at a limited number of sites.
  Although the documented impacts may be localised, the storm itself may have affected a wider area, even where no direct evidence of impacts is available. Local records therefore provide valuable evidence of storm impacts, even if these impacts cannot necessarily be extrapolated to the entire macro-region.
  
  On the opposite, the regional categories do not necessarily refer to studies covering the entire region. Rather, they group storm impacts across one or several areas within the region, without necessarily covering the whole territory. For some studies, such as Lamb and Frydendahl (whose chronology has already been subdivided), representing the documented impacts as a set of individual locations would have been difficult without losing the coherence of the original chronology.
  
  However, giving equal weight to all chronologies means that local and regional records may differ in their spatial representativeness. The Storm Index should therefore be interpreted as an indicator of the decadal frequency/occurency? (see in a future answer) of documented storm impacts in the available chronologies within each macro-region ; rather than a direct measure of actual storm frequency, or of a spatially homogeneous measure of storm activity. To make these limitations more transparent, I will explore how the number of chronologies contributing to the Storm Index could be incorporated into the figures (and particularly the storm stripes), allowing readers to better assess the documentary basis of the observed patterns. The implications of differences in spatial coverage and the sensitivity of the index to the number of available sources will also be discussed more explicitly in the revised manuscript.
  I noticed a possible problem, when confronting the observation that record density increases markedly after 1300 on the one hand and the conclusion that this period also sees higher storm activity, which is explained by the transition from warm medieval climate to LIA (or NAO+ conditions to NAO- conditions). Should we not be more careful, given the increase in record density? Would it be worthwhile to confront more explicitly the reconstructions based on historical data and those solely based on geophysical proxies? Because it is mostly the historical data which are not evenly spread in time, and more importantly, only become dense enough after 1300 CE. One of the conclusions (at least for the British Isles) is that storminess also increases after 1300, which is linked to the transition from the MWP to the LIA. But, there always is the possibility that the historical sources in particular underestimate storm activity in the pre-1300 period, so that the ‘increase’ is rather the result of a bias in the sources. Disambiguation between historical sources and geophysical ones could solve this?
  
  Thank you for raising this important point. I agree that the marked increase in record density after 1300 CE may influence the observed increase in documented storm impacts, particularly in regions where historical sources become more abundant during this period. Indeed, historical records are scarce before 1200 CE in several regions, making it difficult to determine whether the apparent increase reflects changes in storm activity, improved documentary coverage, or a combination of both. However, historical studies account for 2 of the 14 references compiled for the British Isles and 5 of the 21 references for the Mediterranean region. More importantly, a preliminary examination of the British Isles records indicates that the increase in documented storm impacts after 1300 CE is also apparent in the geophysical archives. This suggests that the observed pattern cannot be attributed solely to the increasing availability of historical sources, although the influence of changes in record density cannot be entirely ruled out. To address this issue more explicitly, the distinction between historical and geophysical archives will be made clearer in Table 1, and the consistency of their respective temporal patterns will be discussed in the revised manuscript. The potential influence of documentary coverage on the interpretation of the results, particularly the proposed relationship with the transition from the MWP to the LIA, will also be addressed. Finally, the database underlying this study will be made openly accessible upon completion of the manuscript, allowing readers to examine the compiled chronologies and assess the basis of the results.
  Can we control for tidal activity? Many of the proxies used for storminess are subject to interference by tidal conditions: the ‘storm surges’. If a storm coincides with high tide, the impact is self-evidently larger, and it impacts many of the signals (notably those via floods or waves). For older storms there is no way of controlling for the tide, but if the storm is accurately dated, this becomes possible.
  
  Thank you for this important comment. I agree that tidal conditions can significantly modulate coastal storm impacts, particularly through the interaction between tides, storm surges and/or waves. However, systematically accounting for tidal conditions is not feasible across our compilation, given the heterogeneity of the sources and their variable temporal resolution. This limitation will be clarified in the manuscript, noting that tidal influences depend on the type of archive and are particularly relevant to several records of coastal flooding and wave impacts. And as we said before, this Storm Index should therefore be interpreted as an indicator of documented storm impacts/occurency rather than a direct measure of storm frequency or intensity.
  On p. 20 it is suggested that “climatic instability” is itself taken a proxy for storminess ("Climatic instability”, “Climatic deterioration episodes”, “Cold and windy climate phases” describe broader climatic conditions operating over longer temporal scales, which may favour increased storm activity without directly referring to specific storm, wind, or flood parameters). If formulated this way, there seems to be a risk of circular reasoning: because the climatic deterioration is both used to reconstruct storminess and to explain it? (while the climate reconstructions are mostly used as explanation, in a second stage, which is more interesting).
  
  Thank you for raising this important point. The terms “climatic instability”, “climatic deterioration episodes” and “cold and windy climate phases” are retained from the original studies for the sake of transparency. However, these studies are not included solely because they describe periods of climatic deterioration, but because they report geomorphological changes that the authors interpret as potentially related to increased storm activity. For example, Wilson et al. (2004) discuss widespread dune instability during episodes of climatic deterioration, which they suggest may be associated with an increased frequency of storms and storm surges. I agree, however, that these records need to be interpreted with caution. And particularely these dune archives that often have relatively imprecise chronologies ; and linking phases of dune activity to specific storm events may be sometimes difficult. This is one of the reasons why combining geomorphological evidence with historical records is valuable. The distinction between evidence used to identify past storm impacts and broader climatic reconstructions used to discuss possible drivers of storm variability will be clarified in the revised manuscript to avoid any circular reasoning.
  Wilson et al 2004 : https://journals.sagepub.com/doi/10.1191/0959683604hl716rp
  Interpretation/discussion:
  Overall I think the observation of a shift in storm activity from North to South is an important one, and in line with the expectations from the literature.
  
  Thank you for this positive feedback and for highlighting this important result presented in the manuscript.
  Can NAO+ to NAO- be considered as homogenous periods, with a transition from one to another? Are the signals homogenous enough?
  
  You are right to raise this important point. Three different NAO reconstructions (Trouet, Proctor and Ortega’s works cited in the paper) were cross-compared to identify periods characterised by predominantly positive or negative NAO tendencies. The agreement between these three independent scientific sources adds some robustness to the observed trends. However, these periods should be interpreted as reflecting dominant tendencies rather than homogeneous, stable phases, given the internal variability of the NAO. The transitions between positive and negative tendencies should therefore not be considered as sharp boundaries. This point will be clarified in the manuscript, with explicit acknowledgement of the potential for shorter-term fluctuations within these broader periods.
  The frequency versus intensity problem is mentioned briefly in the conclusions, but deserves further elaboration. What do the data capture? See the observation by Trouet et al. (2012), that NAO + is associated with higher storm frequency, but NAO- higher storm intensity: so, less frequent, but more intense storms. In the Southern North Sea the 1950s are now pointed out as an extremely ‘stormy’ decade, but this probably is the result of the 1953 storm, a highly intense (and catastrophic) storm, but not necessarily an indication of a decade with a lot of storm activity? Inversely, the 1450-70 period is now conceived as the most stormy period along the English coast, but is this the result one or two intense storms or rather an increase in the frequency of storms?
  
  Thank you for raising this important methodological point. We can conclude from this global discussion that the compilation primarily captures the occurrence of documented storm impacts at the decadal scale, rather than directly measuring storm frequency or intensity. The Storm Index could therefore be more accurately described as a sort of "decadal storm-impact occurence index" and should not be interpreted as a direct measure of storm frequency. A decade identified as particularly stormy may reflect one or more highly destructive events, a higher frequency of storms, or a combination of both. For example, the high index for the 1950s in the Southern North Sea may largely reflect the catastrophic 1953 storm, rather than necessarily indicating a particularly high frequency of storms throughout the decade. Similarly, the available data do not always allow us to determine whether the high index for the English coast during 1450–1470 reflects one or two particularly intense events or a genuine increase in storm frequency. It is really important to take caution about these interpretation, so as the binary coding does not systematically distinguish between these possibilities, this limitation will be clarified in the manuscript, particularly in the discussion and conclusions.
  Storm trajectories are only mentioned as possible follow-up, but both in previous historical reconstructions (Van Vliet-Lanoë, 2014), and in discussions on the impact of climate change on storminess today (Calafat et al. 2022 in Nature; Karwat et al. 2022 JGR Atmospheres) there is already the suggestion of different regional storm trajectories forged by changes in NAO and AMOC explaining for instance why storm activity might be decreasing in the Southern North Sea and along the Atlantic coast of France, while increasing in the north of the British Isles. This literature could be interesting to inteprete the shifting patterns in the past as well?
  
  Thank you for this valuable suggestion. I agree that previous historical reconstructions and studies of recent storminess provide a useful framework for interpreting the contrasting regional patterns observed in our compilation. The discussion will be expanded to incorporate the suggested references and consider whether changes in storm trajectories, potentially associated with variations in the NAO and AMOC, could help explain these differences. However, these mechanisms will be discussed as possible explanations rather than established causes. Our compilation documents storm impacts rather than directly reconstructing storm tracks, but it may provide a useful basis for future investigations into historical storm trajectories. Furthermore, relationships identified for the recent period cannot necessarily be extrapolated to earlier centuries. This distinction will be made clearer in the revised manuscript.
  In their 2014 study on Skagen, Clemenssen et al. notice a clear cyclical pattern in storm activity from 1850 onwards, without clear long-term signal, apart from a marked higher storminess between 1860-75. Are both results relevant for the present discussion? And why is their chronology so different from the present one? (on Figure 5 1860s and 1870s in the Kattegat region do NOT stand out as particularly stormy?).
  
  First of all, we did not include Clemmensen et al. (2014) in our database because their chronology covers the period 1860–2012, which falls outside the temporal scope of our compilation. Nevertheless, their results provide a useful independent comparison with our reconstruction. Clemmensen et al. identified exceptionally high storminess between 1860 and 1875, followed by substantial multidecadal variability. This peak is not clearly reproduced in our Kattegat compilation for the 1860s and 1870s. However, the Storm Index for the 1860s is 0.36 (4 stormy on 11 actives chrono ; before the current database update), indicating a “low to moderate” occurrence of documented storm impacts rather than an absence of storm activity. The two chronologies are not directly equivalent: Clemmensen et al. reconstructed storminess from wind observations at Skagen Fyr, combining historical observations and instrumental measurements, whereas our compilation is based on documented storm impacts across the broader Kattegat region. As discussed in the manuscript, chronological uncertainties associated with isotopic dating make it difficult to identify robust trends in storm activity during the most recent part of our compilation. The discrepancy may therefore reflect differences in data type and spatial coverage, as well as chronological uncertainties in our reconstruction. I think that there is consequently not necessarily a true contradiction between the two records.
  I hope these comments are useful, and can help to further improve the paper into a cornerstone of future storminess research!
  Thank you very much for these extremely encouraging words. I am truly honoured by your positive assessment of the paper and sincerely grateful for the time you have devoted to reading and commenting on this work. The perspective of a historian is particularly valuable for this research, which relies heavily on historical sources and aims to bring together different disciplinary approaches. I am very pleased that you took the time to review this work, and these comments will undoubtedly help to improve it further.
  
  Citation: https://doi.org/10.5194/egusphere-2026-3658-AC2

Pierre Pouzet

Viewed

Total article views: 539 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
346	154	39	539	56	55

HTML: 346
PDF: 154
XML: 39
Total: 539
BibTeX: 56
EndNote: 55

Views and downloads (calculated since 15 Jul 2026)

Month	HTML	PDF	XML	Total
Jul 2026	115	32	13	160
Aug 2026	25	14	0	39
Sep 2026	75	45	7	127
Oct 2026	97	44	9	150
Nov 2026	34	19	10	63

Cumulative views and downloads (calculated since 15 Jul 2026)

Month	HTML	PDF	XML	Total
Jul 2026	115	32	13	160
Aug 2026	25	14	0	39
Sep 2026	75	45	7	127
Oct 2026	97	44	9	150
Nov 2026	34	19	10	63

Viewed (geographical distribution)

Total article views: 474 (including HTML, PDF, and XML) Thereof 474 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 14 Nov 2026

Short summary

Coastal storms have shaped European coastlines and societies for centuries, yet their long-term evolution remains poorly understood. By combining evidence from historical documents and environmental records, this study reconstructs storm impacts across Europe over the last thousand years. The results reveal contrasting periods of increased storm activity in northern and southern Europe, improving our understanding of past climate variability and coastal hazards.


Total:	0
HTML:	0
PDF:	0
XML:	0