The paper “Floods in the Pyrenees: A global view through a regional database” shows the analysis on the first systematic dataset of flood episodes referred to the Pyrenees massif, named PIRAGUA_flood, concerning the period 1981-2015. The topic is very important, and the paper fill a gap for a wide area including regions of different countries.

Nevertheless, I think that the paper is not ready for publication, it must be strongly reviewed and better addressed. As is, it seems a folder containing scattered notes without a clear goal to reach. I suggest to select a series of goals, declare them at the beginning, and try to reach them, even if this mean that some of the paragraph must be eliminated. Currently there are a series of points but no one of them is analyzed deeply.

The database is updated to 9 years ago! It seems very strange, especially because the Authors are interested to analyze temporal distribution of the events, and an old series does not allow to evaluate the recent tendencies, especially in the light of climate change. I think that this is the large obstacle to the publication of the paper

From the point of view of the structure, the paper needs to be homogeneised. The paragraphs are short and not always contain what the title says. Some attributes, available for subsection of the study area or sub-periods, should be eliminated because they are useless if not available in a homogeneous way (see table 2: “the compensation paid by the CCS to the municipalities in the 230 Spanish Pyrenees for floods that took place between 1996 and 2015, adjusted to 2015” for example).

Introduction should be enlarged and improved, mainly by quoting more recent papers published in the latest years of this century. Lines from 41 to 53 describe the study area. Why this part is included in the introduction instead of be in the STUDY AREA section? Figure 1: the size is large with respect to the information contained; a European map must be included to allow the reader to understand where the study area is.

A table clearly reporting the main physical characteristics of each study region (and the abbreviation used, possibly being the same throughout the paper…) and the information sources must absolutely be included. As is, this section is very verbose and not understandable, and it is not clear what are the regions, where they are and what country they belong. The sentence “L 105 In some specific cases, precipitation maps were created from rainfall reanalyses provided by SAFRAN (Quintana et al., 2016), which allowed us to detect some municipalities that suffered flood damage where there was no other record” is unclear. It is the methodology used for some specific region? What region? This must be included in the abovementioned table.

In Database structure and methodology, it is necessary to put a figure to show the structure of the database because in this way is not understandable. I don’t understand what is the meaning of the two tables described and what is the relation between them. How these tables are linked?

Line 111: there is a typo. Table 1: put a legend to explain the meaning of the colors

Figure 5-6-7-8: it is unclear where we are in the study region. It is almost useless to put the name of some municipality in very big characters, because readers living outside Europe are not required to know those municipalities. Instead, as in this scientific sector is a common practice, a small framework of study area and sub section must be used (the same size for all the figures).

Table 3. I don’t understand why in this table the authors used the nations. I suggest to maintain both the names of the countries and the regions, in two lines of the table. The same for table 4.

Appendix A: must be completely rearranged in a schematic way, in form of a table, because as is it is useless

This paper presents a detailed inventory of floods having caused significant damages and/or disruptions in the Pyrenean mountains for the 1981 – 2015 period, and proposes some analyses in terms of frequency, geographical distribution, temporal distribution, trends, and weather types related to these flood events. The presented database is rich (181 events, coverage of the whole Pyrenean region, information combining intensity of damages, victims, and compensation costs) and brings interesting information about the occurrence of floods in the considered region.

However, I think this paper should be significantly improved on the following aspects (see also my detailed comments hereafter):

• Some possible limits affecting the comprehensiveness of the inventory should be better stated and discussed. Particularly, obtaining a very low number of ordinary events in some specific regions is certainly caused by some limits in the inventory, provided that the sources of information involved are different for each region. For this reason, I think that most of the proposed analyses should rather focus on extraordinary and catastrophic events, than on the total number of events
• Some methodological details would require some additional explanations. This concerns both the procedure used for flood inventory (consultation of newspapers limited to pre-identified dates? How the 0.5 and 5 codes representing cars swept away and victims ca be combined with the 1 to 3 codes representing flood severity?), and the methods used for analysis of trends (definition of the variable used) and weather types (definition of the “daily averaged” fields).
• Some sections (4.2 and 6) do not bring significant information in my opinion, and could be highly summarized based on a much more limited number of figures, to put the emphasis on other (more informative) sections.
• The links between floods and weather types (section 7) are not presented in an optimal manner. I think this section should include additional information to better illustrate the relative weights of the different weather types causing floods, and the temporal and geographical repartition of these weights.  

Considering this, my opinion is that this paper would require major revisions before publication.

Detailed comments :

l.10  “the results of the analysis of the first”

l.13 I suggest to replace “adapting to” with “and adapted to”

l.16 Replace “paid out made” with “paid out”

l.20 Please provide an explanation here for the meaning of «million₂₀₁₅ » 

l.29 “unnoticed in the databases” Which databases? Do you mean here databases providing inventories of damaging flood events which are detailed in the next sentences, and/or more generally hydrometeorological databases that may also fail to capture heavy rainfall events and floods because of the too coarse density of observation networks. Could you please provide more details?

l.30 Could you provide references or URLs for these two databases?

l.34-35 “there are very few such episodes that have affected the Pyrenees Mountain region on the databases.” Please indicate which databases are concerned: databases from reinsurance companies? For enhanced clarity, maybe this sentence should be rather placed just after the sentence pointing out the limits of databases from insurers.

l.38-39 please formulate more explicitly the regions concerned in each country, and if the whole Greece is concerned: “the Catalonia region and the Balearic Islands in Spain, the former Languedoc-Roussillon, Midi-Pyrénées and PACA (Provence-Alpes-Côte d’Azur) regions in France, the Calabria region in Italy, and the whole Greece.”

l.46 GDP: please detail the acronym

l.50 please specify the country (Spain I can imagine)

l.51 where did this september 2023 event occur?

l.54 do you mean a comprehensive database of damaging flood events?

l.86 “For events that were not known, it was even more time-consuming to identify new events on the CCS database” I don’t understand this, could you please reformulate?

l.87 could you indicate here if the newspapers were systematically consulted for the whole period or just for the dates pre-identified from other source of data ? (I imagine the second option is the right one)

l.92 rather “included in the “?

l.95-97 please provide references and/or URLs for these sources.

l.100-104 again here, a reference and/or URL would be useful.

l.118 Please add here a sentence to clarify the meaning of “level” in table 1  “In table 1 the levels 1 to 3 correspond to the level of damage observed for different categories of assets: level 1 refers to possible ..”

Table1: I think it is not necessary here to repeat four times this table. The explanation provided in the caption for the classification in the four categories is sufficient, and could be placed directly in the text.

l.123 It is not clear here how these additional categories (car swept away and fatalities) are combined with the four initial categories (ordinary – extraordinary – catastrophic – major catastrophic). For instance is a catastrophic event with fatalities classified in the category 2 or 5 ? Could you clarify this ?

l.134-136 If possible, please provide references or URL for these sources of data.

l.139-140 Could you please describe here in more details the variables used for trend analysis: number of events per year?  Number of events per season and per year? Number of events per category and per year? ..

l.153 “was then averaged on a daily basis” Do you mean here that the fields were temporally averaged for each calendar day ? This seems surprising since the considered fields may largely vary within one day. I think a clarification and a justification are necessary here.  

l.150 – 171 I think a figure would be useful here to illustrate how many categories were used, and how the mslp and z500 fields differ within each class (for instance through a reference to fig 16).

l.178-179 The high difference in the number of events depending on the considered country is rather surprising, particularly the low number of ordinary events recorded for France. This difference may be related to differences in the sources of data used for each country, with possibly a different level of detail and comprehensiveness in each.  But this does not seem to be confirmed by fig.2 that shows a relatively similar frequency of events at the municipality scale for France and Spain (but a higher frequency for Andorra). I think this question of representativeness of inventory should be discussed here.

l.182 “It means that if the numbers of events that have affected each Pyrenean region are added, the result is 242 events.”

l.183 A more simple formulation can be used here “In this article, it is considered that ..”

l.192 To clarify please mention explicitly which municipality was hit by the 2013 Garonne river catastrophic flood.

l.202-203 Figure 3 and table 2 seem to confirm that the inventory of ordinary floods is incomplete in Occitanie, Nouvelle Aquitaine, and Aragon regions (see my remark above). I think this could be stated more explicitly here, and maybe some reasons for this could also be advanced country (for France, I think events are recorded only if considered as having exceeded a 10-year return period).

l.204 I think focusing on extraordinary and catastrophic floods is the good choice here, provided the doubts on the inventories of ordinary floods.

l.223 I do not see this information of 39 victims in November in table 2 or other figures. Maybe an histogram showing the seasonality of floods and victims would complement usefully the information provided in table 2 (this could be grouped with figure 12).

l.241-242, 252-253, 264-265 Please mention here that the compensation amounts stand for the whole regions.

Figures 2, 3 and 5-6-7: the color scales used in the maps showing the number of floods per municipality could be homogeneous among the different figures. This would confirm the consistency between the figures and facilitate the comparisons.

l.274 and fig. 8 please keep the same name for Toulouse or Tolosa

Section 4.2 and figures 5-11 : this section (and the related figures) is rather linear and repetitive, and brings very limited new information if compared with table 2 and figures 2-4. I think adding a general figure of the CCS compensations at the same geographical scale as figures 2-4, and providing further comments in section 4.1 about the differences between regions based on table 2 and figures 2-4, would be largely sufficient.

l.306 “The events are distributed..”

l.315-317 and figure 12: I think showing distributions based on the total number of events is rather misleading here because of the possible heterogeneity between regions in the inventory of ordinary events. This is probably the reason why the joint distribution is unimodal close to the distribution of rainfall events in Spain.  I think showing statistics based only on the extraordinary and catastrophic events would have been more relevant and representative here (i.e. the bimodal distribution).

l.320 theses results are rather related to Table 2 and fig.13 (not fig.12)

Section 6 and tables 3-4 bring very low added value if compared with section 5, and I do not see the interest to develop the cross-border episodes. I would suggest to remove this, or maybe to present just a table with the regions affected by the 41 “transregional” events (rather than cross border). I also think the development about the 1982 (figures 14 and 15) could be rather presented in the methods section (section 3.1) to illustrate the results of the flood inventory.

Section 7: this section shows the temporal distribution of each of the weather types causing floods (figure 17), and some examples of the floods caused by each weather type are provided. I think some information illustrating the weights of each weather type in the generation of floods is missing here. It could be for instance a table similar to fig.17 but showing the distribution (or the numbers) of floods related to each weather type for each month of the year, and also the global distribution of weather type having caused floods. Also, some maps showing the number of floods in each municipality related to one weather type (or to groups of weather types occurring mostly in summer or autumn)  would probably be informative about the regions affected for each category of weather type.

l.405-409 Again here the possible links with the limits of the flood inventory are not mentioned.

l.444 It is mentioned in table 2 and section 5 that both trends for the entire Pyrenean are not significant at 90%. This is in contradiction with what is stated here (conclusion).