The manuscript provides an analysis of the rainfall event on 17 August 2024 in Vienna Austria. To assess the extremeness of the event two datasets have been used: Rain gauge data from weather stations with long records but limited spatial representativeness and the radar-based INCA dataset with spatial coverage but short time series. Both datasets revealed that the event on 17 August 2024 was extraordinary with a rainfall of 107 mm/2h and return periods in the range of several 100 years. 

The study fits in the scope of the brief communication format of NHESS. It is well written, the goal and methods are clearly explained and the results are easy to follow and well illustrated. I recommend publishing the manuscript after addressing the following remarks:

P.3, L.54: What would happen in the unlikely case of an event around midnight? Is the independence of events still guaranteed?

P.3, L.63: Isn't the INCA dataset also shorter than 25 years (2004-2023)? Why were annual maxima used instead of POT?

P.3, L.68: Please explain the abbreviation HRV, because it is used here for the first time.

P.3, L.72: I guess, “trough” is meant here instead of “through”?

P.4, L.94: Inner Stadt may not be clear to a read who isn’t familiar with the city of Vienna. Maybe the authors could explain a bit more about the location.

P.4, L.108: Just out of curiosity: How would the return period of the event change when it would be included in the time series?

P.5, Fig. 1b: A line indicating the 25% and 75% percentile would be helpful to show because it is mentioned in the text.

This manuscript examines an extreme convective event that occurred in Vienna in 2024. It estimates the return period of the 2-hour rainfall in relation to the gauge location, to other gauges in the area and to a record of INCA model simulations.

I find the study interesting and worth publication, although I think some weaknesses need to be addressed beforehand. Please see my comments below

I'll be happy to discuss with the authors about any misunderstanding from my side. 

Kind regards,

Francesco Marra

• Line 5: this kind of accuracy on an estimate of a very long return period is misleading. I suggest to write something like “likely exceeds 600 years”.
• Line 65: please use “5%” or "five percent”
• The temperatures during this summer event were particularly high, and so was the atmospheric moisture content. This likely contributed to the development of such an extreme event (as the authors mention in the introduction in line 13). It is true that the timeseries does not reveal a significant trend (and this is perfectly normal even in presence of trends, e.g., see https://doi.org/10.1029/2010WR009798), but it is natural to imagine that the ongoing rise in temperature likely contributed to this extreme (and may contribute more in the future). I think this aspect cannot be neglected in the discussion and conclusions. Also, this is an important aspect to mention when assuming identical distribution in the extreme value analysis (we likely violate this assumption and this should be mentioned).
• Section 3.2: I think that here two aspects need some more attention
• The assumption of normality for MLE can in some occasions be a bit stretched. Why not using the same method to estimate uncertainty for both LM and MLE (i.e., the bootstrap)? This would grant consistency in the estimated confidence limits. Otherwise I don’t think you should compare them.
• given the importance of this section in the paper, I am a bit unsatisfied with only having estimates with these two parameter estimation methods (with uncertainties estimates in a rather inconsistent manner - see above). This is because 84 years can look like a long record but when it comes to extremes this is not really the case (you can just do some simple Montecarlo experiment from known distributions to see it).
  
  I don’t want to bias the authors toward a specific method or distribution so I’ll only give two classic examples. Many studies found that precipitation has extremes with GEV tail parameter close to 0.1. There are ways to include this information in the estimation (e.g., Martins & Stedinger https://doi.org/10.1029/1999wr900330). Another option is to use a regional framework such as the classic one based on LM by Hosking & Wallis (https://doi.org/10.1017/cbo9780511529443).
• One of the take-home is that the event was really anomalous also considering that no trends could be seen in the data. An aspect that would be interesting to explore is: “assuming your GEV estimated before the event is true, what is the probability of observing this event in a 84 year time interval?” This can be done using Montecarlo sampling and can provide some useful insights.
• The occurrence of such an extreme event over the city also raises questions about the possible impact of the urban area on extreme rainfall. Perhaps this aspect could be discussed - there is some literature on the topic.
• In line 169 you talk about possible temporal inhomogeneities in the INCA. I wonder if that is also a possibility in the 84 years gauge timeseries. I believe the instrument was changed over this time. Perhaps this aspect should be discussed.
• I found a bit of a mismatch between the rarity of the studied event (over 500 years estimated return period) and the 20-year level used for the INCA estimates. I wonder if using different extreme value methods (e.g., see above two examples) it would be possible to look for longer return periods in the INCA. In general, we expect many (~4) of the events in the gauge record to exceed the 20-year level, so this level is not that indicative for the 2024 event.