General comments

The manuscript displays an interesting analysis of the impact of precipitation clustering on the duration and frequency of high discharge events in Switzerland and Europe. It uses appropriate data sets and makes a courageous attempt to stratify the analysis along the large number of degrees of freedom that govern the relationship between (clustered) extreme precipitation and high discharge. Overall this analysis and stratification are useful, but the authors seem to address the topic from a pretty methodological point of view, thereby presenting quite a large number of figures and results that are not always directly meaningful to the reader. In particular the large number of spatial maps don’t convey a very clear message of spatial structure in the findings, and some reduction in figures and a more concise display of results would be appreciated. On the other hand, some potentially relevant physical processes are discussed somewhat in the discussion section and annex figures, and some of these would have been interesting to elaborate in somewhat more detail. Particularly the notion of (seasonally dependent) soil moisture memory and the impact of catchment size deserve a more explicit discussion and interpretation of the results.

The paper is very well written and (most of) the statements are well supported by the results shown, but a small change in the emphasis on the presentation of results would probably make the paper even more informative for the audience. So this leads me to recommend to accept after minor corrections.

Specific comments

• L30-31: also Khanal et al (2019) address the role of atmospheric clustering and soil moisture memory on Rhine discharge explicitly
• L49-50: Kew et al (2013) looked at the effect of clustering on the probability to have compounding discharge and coastal surge peaks for the Rhine river. This is an interesting application domain of studying temporal clustering/compound events in this context
• L81: given the relatively small effect of removing this baseflow component (L265-269), and the question of whether the resulting runoff data can be interpreted well I would suggest to leave out this baseflow correction and work with total discharge instead
• L101-102: maybe add explicitly that these factors display a seasonal cycle
• L116-117: the fact that discharge impact is characterized more by absolute than anomalous discharge also applies to the impact of clustered precipitation on discharge. So I find this methodological inconsistency of using absolute or anomalous values not convincing
• L117: “smaller” à “lower”
• L126: the 30 and 60 days are not clear to me, I compared them to the max 8 weeks of clustering and couldn’t find the logical match. Please explain somewhat better
• L147: what motivated this combination of N and L? It came somewhat unannounced
• L166: “less variability”: less than what?
• L208: please refer forward to discussion section when introducing “karst effects” (and in the discussion section: please explain in some detail what this effect is about)
• L222-224: this should be discussed in the section on Swiss results, not here
• L225-228: also this feels that it belongs to the methods section, not to the results section
• L236-237: “Cumulative precipitation … periods.”: I don’t know what you want to convey with this statement
• Fig 11: the set of panels don’t convey a very clear message of a spatially meaningful structure. Also it takes me a long time to make up my mind of what is actually shown in the different rows and columns. Also: what does "non-significant" (grey shading) mean in these panels? I don't really understand the grey shading in all basins of panel 11-d. I feel this display of information is too extensive.
• L237: “increase”: of what relative to what?
• L245: events are classified by percentile, by persistence, by significance. It's not easy to disentangle all these attributes. What's the key point you want to convey here?
• L246: “often result from TCEP.”: how can I deduce this statement from these panels? They look quite similar to panels 11d-f. And how does “cluster frequency” shown in Fig 11 translate to TCEP?
• L249: “Overall, the connection to TCEP is weaker for less persistent high discharge periods.”: how can I see this from the figure? What should I compare to what to understand this statement? Same for statement in L254
• L272-273: I didn’t read the Bevacqua reference, but doesn’t this combination of arguments imply that it is the change of the wet-day frequency and R95p that is responsible for this, without a significant change in clustering?
• L274-275: this soil moisture memory plays a smaller role in winter/spring time, where the highest discharge occurs, I would reckon. A seasonal instead of a spatial analysis would have been more interesting (see also statement in L306)
• L282: figures 11 and 12 don’t show soil moisture impacts
• L307: “largest events”: discharge or precipitation events?
• L317: double use of “high”
• L324: 1000 km2 is pretty small. I would assume that clustered precipitation extremes can give high discharge in much larger basins
• L330: “vary” à “varies”
• L336: this seasonal signature is of great interest and could be promoted to the main text
• L351-352: both “antecedent soil moisture” and “timing of precipitation” refer to (temporal and spatial) clustering, so are attributes that are within scope of the current analysis
• L355: do “cross-catchment analyses” imply that you would show more results like fig 14? That would be great!

References

Kew, S. F., Selten, F. M., Lenderink, G., and Hazeleger, W.: The simultaneous occurrence of surge and discharge extremes for the Rhine delta, Nat. Hazards Earth Syst. Sci., 13, 2017–2029, https://doi.org/10.5194/nhess-13-2017-2013, 2013.

Khanal, S., A. Lutz, W. Immerzeel, H. Vries, N. Wanders and B. van den Hurk (2019): The impact of meteorological and hydrological memory on compound peak flows in the Rhine river basin; Atmosphere 2019, 10(4), 171; https://doi.org/10.3390/atmos10040171

Review for manuscript “On the links between sub-seasonal clustering of extreme precipitation and high discharge in Switzerland and Europe”

Authors: Alexandre Tuel, Bettina Schaefli, Jakob Zscheischler, and Olivia Martius

Journal: Hydrology and Earth System Sciences

Summary

This study assesses to which degree temporal clustering of extreme precipitation influences high flow magnitude and persistence using precipitation and streamflow data from Swiss and European catchments. The authors study discharge events following seasonal precipitation anomalies and try to establish a link between precipitation clustering and discharge magnitude and frequency. They show that temporally clustered precipitation leads to more persistent discharge responses than non-clustered events.

General remarks

I think that this is a well-written and –structured study which highlights the importance of looking at precipitation events beyond the ones in immediate temporal proximity to high flow events. I have one methodological concern though regarding the event definition procedure and its potential impacts on the results/conclusions. In addition, I suggest to reconsider figure choice as the results of the Swiss and European analyses are very similar (one could maybe even say redundant). I think that narrowing down the selection might help to better focus the reader’s attention on the main points.

Major points

1. I suggest to clearly highlight the research gap after the paragraph ending on l.54. What remains to be investigated given the results presented in the authors’ previous study?
2. An overlap of only 10 years between precipitation and discharge events (l.72) seems very little given that the study focuses on extreme events.
3. Please specify how the baseflow filter parameter (l.82) was determined and why it does not vary in dependence of catchment properties.
4. Precipitation extremes are defined as anomalies (l.97-100) while discharge extremes are defined using a fixed threshold (l.115). While I understand the desire to consider precipitation events potentially co-occuring with wet antecedent conditions, I think this is not necessarily achieved by using a seasonally varying precipitation threshold (some events relevant in terms of discharge may still be missed). I think that choosing events based on the discharge rather than the precipitation events would be more consistent with the aim of the study. I think that the effect of choosing one over the other event identification procedure on the results of the analysis should be demonstrated in a small sensitivity analysis (e.g. on a small subset of catchments). Specifically, it would be important to know what effect (a) choosing events based on precipitation has compared to choosing events based on discharge and (b) what an effect choosing a seasonal rather than a fixed precipitation threshold has.
5. Were the percentiles computed empirically (l.155)?
6. What about uncertainty in percentile computation because of small event sample sizes (l.159)? How does the analysis deal with cases where there are only very few events of the same length and what event sample sizes are we generally talking about?
7. I think that the European analysis is nice to put the Swiss analysis into broader perspective. However, I also think that it leads to the presentation of slightly too much material. I therefore suggest to just show a subset of the European results. I would try to focus the reader’s attention to the most important information.

Minor points and editing suggestions

l.3: ‘this question’ lacks a reference in the previous sentence.

l.10: the influence of temporal clustering on what? I understand that the statement refers to high flows and think that this could be made clearer.

l. 310: specify what ‘this bias’ refers to.

Figure 1: suggest to reconsider color scale choice as a continuous variable is presented using a non-continuous color scheme.

Figure 2: I like figure 2. However, the horizontal lines in panel b should be explained in the legend.

Figure 11: legend entry for grey color needed.

See Supplement