The authors present a well-written, clearly structured paper on the usability of ESWD data with respect to hail events. Complementing previous publications, the question is addressed whether an extension of the data base over a longer period of time and to data with a lower quality level leads to comparable results. The core statement that a homogeneous basis for hail climatology has only been achieved over the last 20 years is clearly shown. The paper convinces with the evaluation regarding the relation of hail days and hail events, which can be well used to analyze national and temporal inhomogeneities in the hail reports of ESWD, which is well shown by the hail data from Poland before 1955.

One comment is directed to the better elaboration of this result in Section 7. It would be desirable if the data from Poland between 1947 and 1955 could be more clearly contrasted with those of more recent reports, in order to discuss indications of data inhomogeneities and doubtful entries for later investigations. In particular, these are figures of the seasonal distribution of hail events, the distribution of maximum hail diameter, and the time of day of the reports. In addition, the discussion of the results may be more open: While climatological trends may be considered as a possible but unlikely reason, there is no mention of the possibility of spurious entries that nevertheless meet the quality criteria. Data originating in official documents such as weather service yearbooks have a higher probability of receiving a higher quality level than reports originating from other sources. In the case in question, the data originated exclusively from the Stalin era, during which the Polish agricultural industry faced repression if it failed to meet ambitious agricultural aims. While the entries during this period provide little information on the weather events themselves, they almost exclusively contain very accurate estimates of losses for agricultural products. Although this investigation is beyond the scope of this paper, an open-ended discussion in this direction would be desirable with the caveat that historical data must be critically evaluated.

Apart from that, there are only a few minor comments. In section 7 results of the paper on hail reports at Polish weather stations are compared with those of the present paper. However, it is not written whether the criteria for large hail are the same. Thus, together with the large differences in the hail climatology, the question arises whether large hail is compared with sleet here. This possibility should be discussed.

Line 36: First supercell on July 27 and second one on July 28? Please check in ESWD

Lind 85: In recent years, ESSL also developed an mobile phone app to report severe weather (EWOB)

Line 165: The regression lines are not visible in figure 2.

Line 169: Numbers with plural -s instead of “number”?

Figure 7: You may consider to extent the data basis to years before 2000 to show the time of stabilization mentioned in lines 239 and 240.

Line 249: “suggests that the entries that the average hail size is” Delete “that the entries” or  “that the average”.

Figure 8: “Two red dots represent likely data-entry errors”: The red dots are not red in the figure.

Line 302: “would imply” instead of “would implies”?

Lines 302 and 303: “except for Germany which has a much greater number of reports proportional to the number of days.” And Poland as well according to the graph?

Review Climatology of Large Hail in Europe: Characteristics of the European Severe Weather Database

The paper presents a statistical analysis of large hail reports from the ESWD for Europe. Analyses performed for 120- and 20-year periods include time series of reports and hail days, diurnal and seasonal cycles, annual distributions of hail sizes, and trends in temporal accuracy. Additional emphasis is given to reports from Poland, which has very high numbers of reports in some 10-year periods since 1930.

In general, the paper is well written and clearly structured.

However, I have some major concerns, mainly about the quality and reliability of the data the analyses are based on, and the scientific content.

1. It is difficult for me to see new scientific results and profound conclusions that provide new insights into hail statistics. The paper is of course nice to read, but the scientific value seems to be low. Also, the conclusion section is more or less a summary rather than a presentation of conclusions and interpretations.
2. As far as I understood from the manuscript, the authors considered all ESWD reports in their analyses, irrespective of multiple reports from a single storm, or the country or region affected. Given the large differences of prevailing reports among European countries (as shown in the Table), it can be assumed that the results are dominated by individual countries (e.g. Germany, Russia, Poland), leading to large uncertainties in all estimated quantities.
3. In the same sense, hail reports are biased towards daytime and towards larger cities. This effect is difficult to estimate, but at least a profound statement is required (although a spatial analysis with respect to the distance of reports to larger areas may help to get an estimate of the latter effect).
4. Is a hail day one with at least one report across Europe (that would make no sense), or have you considered some threshold? For example, is a day with only one 2 cm report considered the same as a day with thousands of reports and hailstones larger than 10 cm? That would be strange. Furthermore, it makes no sense to define a hail day for the whole of Europe, with its wide variety of local climates. I would rather suggest limiting it to countries with a high number of reports, for example. I would also suggest considering different thresholds for both hail size and number of reports.
5. Point 4 also refers to the other analyses, such as the annual and diurnal cycles. It is mentioned that Púcik et al. (2019) divided the study area into at least two parts due to the different climates. Why did you not follow this?
6. A climatological period is usually defined as 30 years or more. It also includes spatial analysis. Neither is the case in this paper. Therefore, I suggest changing both the title and the wording in the manuscript.

Additional minor review points are those:

1. L21: 20-years is not a climatological period (major comment 5)
2. L30: “Large hail” for a diameter of > 2 cm is not a European definition, rather used by ESWD.
3. L44-45: You may add that most of the hail climatologies / statistics (e.g., those cited in Touvinen et al., 2009) are outdated
4. L50: It should be noted here that some pan-European hail hazard assessments are available, e.g. from Punge et al. 2014 or Punge et al. 2017 based on overshooting top detections, from Rädler et al. 2018 using reanalysis, or from Taszarek et al. 2018 using multiple data sources. In this sense, the statement in L60 “…their work shed the first light on” is not true.
5. I miss a better motivation and scientific objectives of the paper. “Increasing the size of the dataset through…extending the period of analysis” is too weak when only 2 additional years are considered.
6. P3, 2^nd paragraph: Why did you not use the most recent data until 2022? The analyses seem to be easily reproducible.
7. L98-105: This is the correct designation of the quality levels; in the later text they are incorrectly quoted.; L106: “…plausibly checked QC1…”, but this is “report confirmed”
   
   Also in L225-226 it should read “report confirmed”
8. L157: “…ability to detect reports linked to the same event, and hence have removed duplicate events from the dataset”. This would make no sense at all and is not the case. In the papers cited (e.g. Wilhelm et al., 2020) it is clear that a single streak is covered by several reports.
9. L77-79: Kunz et al. (2020) estimated annual and diurnal cycles not from ESWD data, but from radar-derived potential hail streaks (Z > 55 dBZ). These streaks were also combined with ESWD reports. The main difference is not the quality level of the ESWD reports considered because as written in Sect. 2, 70.4% were QC1 and 29% were QC+, leaving only 0.6% at Q0 level.
10. L188: Can you briefly describe how you converted UTC to LT?
11. L191-192: see comment (9); Although the diurnal cycles of Kunz et al. (2020) have a resolution of only 3 hours, there are some differences, which may be due to different study areas?
12. Fig 7: This figure is very interesting, but again not very valuable for the whole of Europe (and the under-reporting in most countries). I suggest that this type of figure be reproduced for countries where the number of reports is highest according to the Table.
13. L248 and L300: Did you use the Pearson product-moment or Spearman rank correlation coefficient? The latter would be more appropriate due to the obvious deviation from a normal distribution.
14. L287-289: The main reason for the high number of reports in Germany is obviously that ESSL was founded here. It should be mentioned that in some countries severe weather reports are collected by other institutions, e.g. KERAUNOS in France. Moreover, crowd-sourcing via meteo apps is well known and emerging in some countries, such as the MeteoSwiss app, which has collected >100,000 reports in recent years (compared to only 266 ESWD reports). So we should not blame spotters for being less enthusiastic.
15. L315 and others: I'm not sure about the comparability with the study by Suwala (2011), as they used station data over a period of 8 years. Station data often do not distinguish between hail diameters, but rather consider ice pellets or graupel in the same class as hail (see also the review by Punge and Kunz, 2018). This could at least explain the discrepancies with the large number of hail events in the cold season. However, this is frankly speculative, as there is no information available for the Polish station.
16. L324: Again, a separation by region would be desirable.
17. L374-377: The trend directions are not that clear. Eccel et al. (2012) or Manzato et al. (2023) found no positive trends in hailpad data in northern Italy, but fewer and larger hailstones. Dessens et al. (2015) found almost the same in their hailpad data in France. You may cite here the review paper by Raupach et al. (2021).

Edits/Typos:

L18: “…dataset for severe convective storms reports.” Otherwise it’s not true, as there are SCS statistics available from model data or overshooting top reports (see minor point X)

L20: “…to evaluate hail reports from… “ (you did not evaluate the database)

L38: “on 10 June”; you refer here to the supercell that hit the city of Munich on that day.

L43: “…intensity, and hailstone size.” On L23, you wrote “intensity, as measured by maximum hail size..”, but here you both intensity and size.

L54: “..which helps..”

L83: delete “insurance data information”: in the ESWD data, I see only 3 entries in 20 years that are from an insurance company; this is not worth mentioned here

L84: delete “organizations”; a large number of reports are not from organizations rather than from trained (and well-known) spotters

L93 “…also examined…” above and below you used past tense

L215: “…frequency of events…” for one event, the frequency is = 1;

L218: “…is more spherical…”

L224: decreased --> decreases

L273: delete from

L298: include a comma after Fig. 2

L302: implies --> imply

L330: mentions --> mentioned

This study explores some characteristics of hail in Europe based on hail reports from ESWD. A distinction is made between hail days and hail reports. The data are summarized in graphics. The seasonality and diurnal cycle are described and data with two different quality flags compared. The time evolution of the reported hail stone sizes is illustrated.

One major point that I would like to raise is that the paper does provide only very limited information about the data in the data base, the quality check procedures, and the methodology that is used to analyse the data. This information is crucial for the interpretation of the data and the discussion of the limitation and uncertainties of the data set (more detailed information on that point is listed below).

The second point is that the analyses are qualitative. This is ok but there should be no statements about changes over time in the abstract without underlying statistical analyses.

Major:

• Please provide (a lot) more information about the data sources of ESWD. What are all the data sources of ESWD (e.g. does it also contain insurance data?)? How big is the fraction of each data source (e.g. crowed-sourced vs. observers from weather services)? How do the different data sources change over time? This information is important as the uncertainty crowd-sourced data and insurance data is quite high

How large is the fraction of each data source in each quality class? How exactly are the quality classes assigned? What does plausibility checked mean exactly? A cross-check against radar information? A cross check against newspaper reports? Which data source provides typically, which types of variables (e.g. mean and max. size of the hail stones).  

Please include all of this information in the methods section.

Hail days: Please provide more information in the methods section how you identify hail reports and hail days? How can you have more hail days than hail reports (Figure 1)?

Hail events: Please explain how you remove duplicate dates. How do you define an event? What counts as a duplicate date? If the report is exactly at the same location?  In the same country? How much time difference do you allow for? How accurate are the report locations?

Time accuracy: If I understand it right, this information is self-declared? Has it every been verified against independent data (radar, satellite information)? How is this information obtained for historical data? Please expand the discussion of this variable to include these aspects in the methods section. How reliable is time information generally in crowd-sourced data?

Location accuracy: how is this parameter estimated? How is it verified?

2) How do your findings compare to hail climatologies based on radar/satellite and proxy indicators?

Minor points:

Abstract: The instensity as measued by ….  larger hailstones are rarer than smaller hailstones.

Introduction: there is a body of literature that discusses hail climatologies in Euope based on indirect observations (radar, satellite data) and proxy indicators (soundings, renalaysis). I recommend a qualitative comparison of the results with this body of literature and hence a brief discussion of this research branch in the introduction.

L86 please add: at the time of this study …

L106 please mention if these levels are inclusive, i.e. are all QC1 also included in Q0+?

L106 please mention that the quality control is strictly against reports from ESWD with a higher quality flag not with other data sources.

L179 “This data set” is referring to QC0+ correct? Please state so explicitly.

L205

L225 pausibly  plausibility?

L234 does not change dramatically  please compare distributions and check for stat. significance

L237 a period of stability   of what?

Figure 7: Can there be a bias towards larger hail stones in crowd-sourced data? People being eager to report large hail stones? How does it compare to ground observations from e.g. hail pads?

L276 This statement is a bit patronizing, I recommend removing it.

L281 suggest to add “People in countries …”

L287 Note that some countries such as Switzerland and Germany have national hail crowd-sourcing programs organized through the National weather services that might explain why there are fewer entries to ESWD

Section 7: It is not entirely clear why you dedicate such a detailed analysis to the data from Poland.

L353 the reported location accuracy

L359 I do not yet understand how you come to this conclusion about hail trends.

L377 consistent = homogeneous? If not, what do you mean exactly by consistent?

L396ff These statements need to be supported by statistical analyses

L399 the reported time accuracy