This manuscript is a nice piece of analysis of three recent dangerous storm surges in the western Baltic Sea, an area famous for extensive variability of drivers of water level, their combinations and resulting properties of high water level events. The main outcome is that two of these events, even though seemingly severe, belong to the pool of relatively frequently occurring episodes in contemporary climate while the third one has at least one truly unusual feature in terms of wind direction. This outcome is in line with the general perception of the nature of climate change in the Baltic Sea region. Namely: storms have not become systematically stronger in this area. Instead, most severe events are driven by specific combinations of various drivers. Another important message is that wind direction may become the most critical feature in development of extreme events.

The manuscript is written professionally, with very good command of English. The setup of the problem is clear, the used methods are described properly and applied correctly, and statistical methods are employed adequately. The images are clear and informative. The conclusions are firmly backed up by the analysis. It is thus my pleasure to recommend this manuscript for publication, possibly with marginal technical revisions.

I have only very few, mostly technical, comments and mild suggestions.

Line 4: remove “of”.

Line 5: remove either “hindcast” or “simulation”.

Line 8: it makes sense to add a couple of words to “water level” to explain what is meant.

Line 15-16, 42, 48-49 and in some other occasions below: please check the sequence of references.

Line 19: it is recommended to use MWL.

Line 27 and in many occasions below: perhaps “south-western” is more traditional.

Line 28: remove “that” and “occurred” for brevity.

Line 37: I guess that the authors actually have in mind the publication [Soomere, T., Eelsalu, M., Kurkin, A., Rybin, A. 2015. Separation of the Baltic Sea water level into daily and multi-weekly components. Continental Shelf Research, 103, 23–32, doi: 10.1016/j.csr.2015.04.018]. The paper (Soomere and Pindsoo, 2016) made use of the 8-day time scale detected in the previous paper.

Line 43: perhaps “unfavourably” would sound better.

Line 60: consider replacing “last” by “finest” or similar.

Line 82: must be Degerby.

Caption to Figure 2: replace “and” by comma before ‘’Warnemünde”.

Lines 130-132 mostly repeat information presented on lines 115-120.

Line 198: something is wrong with the end of the sentence.

Line 221: “persist even after the winds have ceased” is not entirely correct as in many occasions the seiche is only launched when the wind starts to decay.

Line 247: “of severe storm surges”

Lines 247-249: it seems that that return value curves and values of parameters of extreme value distributions estimated from measured and simulated water levels deviate systematically in many locations of the Baltic Sea. Eelsalu et al. (2014) [Eelsalu, M., Soomere, T., Pindsoo, K., Lagemaa, P. 2014. Ensemble approach for projections of return periods of extreme water levels in Estonian waters, Continental Shelf Research, 91, 201–210, doi: 10.1016/j.csr.2014.09.012] hypothesize that wave set-up could be one of reasons while Soomere et al. (2018) [Soomere, T., Eelsalu, M., Pindsoo, K. 2018. Variations in parameters of extreme value distributions of water level along the eastern Baltic Sea coast. Estuarine, Coastal and Shelf Science, 215, 59–68, https://doi.org/10.1016/j.ecss.2018.10.010] demonstrate substantial mismatch of estimates of parameters of the generalized extreme value distribution retrieved from modelled and measured water level time series. Anyway, this is a minor and almost irrelevant aspect in the context of this manuscript.

Line 236: It might be mentioned that the some pairs of annual maximum water levels are not necessarily independent in the Baltic Sea because of possible prefilling covering December and January. This feature may slightly affect applicability of the generalised extreme value distribution in the interior of the Baltic Sea. For this reason some authors recommend using maxima over windy autumn and winter season that are definitely independent. However, the possible difference in the results apparently would be very minor.

Line 426: must be Küste.

Line 456: must be Leppäranta.

Lines 474, 486, 488: must be Suursaar, Ü.

Line 494: must be Dailidienė.

The paper is well-written and addresses the important issue of extreme water levels in the western Baltic Sea. The analysis examines the various components contributing to extreme water levels, with a focus on the role of wind speed and direction in driving storm surges and prefilling of the Baltic Sea. The separation and detailed analysis of these components provide valuable insights into the severity of individual events and the combined effects of extreme water level components. The study demonstrates that while the water levels in 2017, 2019, and 2023 were among the highest on record, the individual components were far from their potential extreme values (with the exception of Flensburg in 2023). This indicates that these events, despite their historical significance, did not reflect the full potential severity of the contributing processes.

The paper also effectively highlights the critical role of storm direction in driving high storm surges. For instance, although wind speeds in 2019 were higher at Travemünde and Warnemünde, water levels in these locations did not exceed those observed in 2017, likely due to small but significant differences in wind direction.

While the paper comprehensively addresses the key drivers of extreme water levels, the omission of wave-induced set-up is notable. Wave set-up can contribute significantly to extreme water levels, adding approximately 20 to 25% of the significant wave height to the water level depending on the method of estimations. During wind speeds of 15–20 m/s, wave heights were likely substantial, especially when the effective wind direction was near perpendicular to the shoreline. Although detailed wave-induced set-up analysis for these events is not expected, discussing this phenomenon in the context local features would contribute the study. Additionally, providing some information on wave conditions during these events would be beneficial.

Several methodological and technical details, such as return period estimations and effective wind speed calculations, are described/mentioned under the results section. It is recommended to move these descriptions to the methods section (Section 2) to improve the structure and clarity of the paper.

Specific remarks:

Line 15: References should be arranged in chronological order. Correct order in (Suursaar and Sooäär, 2007….etc).

Line 20: Provide the water level range for all three storm events, as was done for the 2017 storm (e.g., peak water levels of 1.6 to 1.8 m). Additionally, add "Fig. 1" after the first location mentioned in the text.

Line 35: Prefilling timescales of about 8 days was evaluated by T. Soomere et al. (2015) (https://doi.org/10.1016/j.csr.2015.04.018), while Soomere and Pindsoo (2016) demonstrated that the Baltic Sea's background water level might remain elevated for weeks or even months in some cases.

Line 75: Why modelled wind data points were chosen that far? Please provide some explanations. For example Flensburg, both observed and measured wind data points are quite far and modelled wind data point is located vicinity of one island, that may affect the wind fields in that data point? Please provide some explanations of the choice of the data points.

Line 85: Specify if the "full hour" data represents hourly mean values.

Paragraph 3.2.1 (Prefilling): Consider to move the methodological description for prefilling extraction to Section 2. Briefly explain the choice of the 7-day average.

Line 185: Were the contributions of 80% (wind) and 20% (prefilling) consistent across all locations?

Line 200: For the 2023 event, specify the wind's contribution to water levels at other locations for comparison.

Line 210–225:Among other contributions I would also expect a short discussion of the potential contribution of wave set-up, considering wave conditions (significant wave height and wave directions respect to the shoreline orientation) during these storms.

Line 235: I recommend to describe statistical methods (e.g., GEV assumptions, annual maxima) in Section 2. Are applied annual maxima values uncorrelated in GEV?

Line 265: Please provide more explanation of effective wind speed. I understand this is derived from the highest storm surge data and you have given the direction for each site that corresponds to the effective wind directions. What range was considered of extraction of effective wind from the given directions? Was it e.g. ±10°, 15° from the given directions at each site ?

Fig 7. Would be good to add effective wind speed for each panel or to the figure caption.

Line 310: Interesting to notice that nearly identical wind speeds and directions at Flensburg have led to different water levels in 2023 and 1958-1979. Was the prefilling component very small during the previous event in period 1958-1979 (yellow marker, Fig. 9)?

Fig. 9. please specify x-axis label ’ff’, that should mean wind speed?