This paper presents an analysis of some high-quality rainfall and streamflow data collected in two adjacent catchments in Luxembourg (the Ernz Blanche basin). The objective, as reflected in the title of the manuscript, was to understand better the hydrological mechanisms resulting in flash flooding in this catchment. The paper is generally clear and straightforward to read, though I think that the main focus should have been more strongly on large rainfall events than on hydrologic response under more usual events.

            Oddly, though the authors mention the occurrence of several historical flash floods, including one in 2016 and another in 2018, they do not describe those events in any detail. They provide no discharge data, no runoff coefficients, and no rainfall event data. In order to find something of these events, I consulted an EGU Abstract (Iffly et al. 2018) by some of the same authors. There, I was able to learn that the 2016 event had much more intense rainfall than anything that the authors investigate in the present ms., recording 20 mm in 10 minutes (=120 mm/h), 50 mm in 1 hour, and up to 70 mm in 6 hours (=almost 12 mm/h). In contrast, in the present paper the most intense event reported had a maximum rainfall rate of ~ 27 mm/h. All but one of the remaining events listed in Table 2 had maximum intensities of < 10 mm/h. These seem unlikely to be responsible for flash floods. I was not able to locate information on the 2018 flash flood event for additional comparison. I think that it would help readers place the results of the current ms. in context, if some information on the historical flash floods could be provided, at least in summary.

            I think that the focus on 'ordinary' events needed some comment. How can a study of much more ordinary rainfall events shed light on what occurs during the seemingly far more intense rainfalls that seemingly accounted for the historical flash floods? Did these, for instance, occur when the soil had been thoroughly wetted by antecedent rainfalls? Does surface runoff overtop ground surface roughness elements when the rain is sufficiently intense (above some threshold?), allowing a smoother and more direct path downslope? What was the nature of the precipitation? I assume that the flash floods were the result of shorter, more intense, convective events, and therefore were likely to have occurred in summer (this information is missing from the current ms.). I imagine that these were late afternoon events, but this would also be relevant information. Were there very local runoff sources located close to the stream channels, perhaps? Iffly et al. refer to lag times to runoff peak of just 90 minutes, whilst in the present study these lags extend to many hours. Could the movement of convective cells parallel to the long, narrow catchment be significant? Did that occur (perhaps Doppler radar might shed some light on this)? Catchment response to intense convective cells might be quite different from that in stratiform rain, for instance, and different parts of the catchment might show altered hydrologic responses under those different rainfall inputs.

            The study is weakened by the assumption of a constant runoff coefficient through the duration of rainfall (mentioned in line 200 and elsewhere). This seems particularly inappropriate for long events of several days duration, such as were examined in this ms., and even for events of a few hours duration, when breaks in rainfall (e.g. shown in Figure 6 and Figure 7) allow soil drainage and the re-invigoration of soil infiltrability. It would have been interesting and informative to have seen at least some preliminary sensitivity testing to see how important an effect a changing runoff coefficient might have been to the hydrologic modelling. Perhaps the authors have done such tests and could comment?

            It would also strengthen the argument of the paper if the authors could present some data on hydrophobicity in the forested areas, that they appeal to as a mechanism to account for more runoff there. Was hydrophobicity actually present, or was this not investigated? If present, does it dissipate in longer events, so that perhaps it differentially affects runoff behaviour in short convective events in summer?

            The authors identify LAI as an important factor in the hydrologic response (TTD) (lines 491-492). Though without comment, the authors appear to use LAI data from 2002-2006, many years prior to their field data collection. This warrants some comment. Further, the LAI seems to be very small, to judge from Figure 9 (left panel), seemingly the only data presented on this variable. The authors only appear to link LAI to the speculation about litter layers and wettability, evidence for which is not provided. Could the authors offer a fuller comment on why LAI might relate to TTD? Do they consider this to be a real, physical effect, or merely a chance statistical correlation (for instance, via some other seasonally-varying parameter)? Their comments and thoughts would be helpful. They could also perhaps consider presenting LAI data for their catchments (as a map) if they have it available. It would appear to be very variable among fields, forests, etc.

            Finally, I wondered whether there is a role for roofs, roads, drains, culverts, etc., in the catchment response. I do not know this area, but Figure 2 suggests that, at least locally, the villages may have impervious areas that are efficiently drained. The main stream channels also warrant at least some description. Have they been modified, perhaps to flow between artificial banks or walls? How significant is the channel travel time from the upper to the lower catchment? In the same way, landuse could helpfully be described, especially whether fields are tilled seasonally.

            Overall, this is a solid study, containing some interesting results. However, I am not sure to what extent these actually bear on the factors accounting for flash flooding.

Minor errors:

line 13: should be 30 km² (space is required between numerals and symbol for unit of measurement)

line 46 and throughout the paper: 'et al' should be 'et al.' (as a contraction of et alia)

line 93: should end sentence with a question mark

line 120: omit the parentheses

line 145: Captions are reversed (left to right)

line 162: should say 'Figure 2 left', not right

line 180: it would be preferable to refer to time-aggregated data as rainfall rates (they are equivalent mean rainfall rates, not true intensities)

line 253: again, space required following numerical quantity

Figure 9: there are two dashed lines, only one is listed in the legend

line 500: Hortonian (capital H after the family name of Robert Horton)

David Dunkerley

School of Earth, Atmosphere and Environment

Faculty of Science

Monash University

General comments:

The occurrence of extreme events like flash floods are usually be linked to extraordinary catchment system states or precipitation characteristics. They may be a matter of threshold-behaviour. The analysis of the catchment runoff reaction on ordinary rainfall events with a linear model therefore does not necessarily contribute to the understanding of extreme events. Thus, I would strongly recommend to adapt the frame of your manuscript and agree with Prof. Dunkerley’s comments.

It is clear that extreme events are very difficult to measure and often do not occur in the short measurement periods available in projects. However, the rainfall-runoff dataset generated from August 2019 to July 2020 is valuable for understanding the rainfall-runoff reaction of ordinary rainfall and maybe this should be the focus of the paper.

In order to be able to better classify the measured events, the following information would be very helpful:

1. a) information about the flash flood events in 2016 and 2018
2. b) information about the probability of occurrence of the precipitation intensities. I am not aware whether information on design precipitation with defined return periods is available in Luxembourg. Even if such evaluations are of course subject to uncertainties, they can nevertheless provide guideline values for the classification of the measured precipitation events.
3. c) Presumably there is a difference in the runoff response between long-lasting precipitation events and very short ones with high intensity. In order to work out these differences, it could be valuable to classify the precipitation events.

Specific comments:

Can you please justify in section 3.1. why you used a unit hydrograph model and why you assume a constant runoff coefficient? Even if there is no ideal model, you could also use another methodology, so it would be interesting for readers to know why you chose this one, which has some weaknesses pointed out in the discussion (e.g. line 385, 392). In line 395 you state that for high intensity events, flood peaks are not well simulated. This again shows the problem that your analysis is not so well embedded in the topic of flash floods.

A sensitivity study with changed runoff coefficients, as mentioned by Prof. Dunkerley, would at least be very helpful.

Please comment on possible differences between the LAI survey period and the period of your data (e.g. land use changes). Please explain how you compressed the LAI data with a spatial resolution of 1 km² into a value for the correlation analysis. Did you use the mean value over the catchment area?

Please explain which soil moisture values you used in the correlation analysis? Did you use the value from the station situated in the respective catchment? If so, there are two stations in the HM catchment? Did you use the mean of both stations?

The lithological abbreviations (Km3, Li2, Li3) are confusing if one does not know the context (and therefore cannot assign the numberings – e.g. what is Li1?). In Fig. 1, some of the lithological units have an abbreviation, some do not. This looks very inconsistent. Would it be possible to do without the abbreviations (Km3, Li2, Li3)? If not, it would be important to at least cite the geological map from which these designations originate. In any case, I would delete the lithological abbreviations from the abstract.

240 ff.: Please mention that also the runoff coefficient is analysed.

384: What do you mean with: “the model overestimates the rising limb of the flood wave” – is it the duration of the rising limb, its slope or something else?

Fig. 10 and 11: please explain the size of the circles

504: Is „pseudo“ necessary?

Technical comments:

“et al” should be “et al.” in the whole paper (see https://www.hydrology-and-earth-system-sciences.net/submission.html#references)

15/17: delete “Km3, Li2, Li3”

16: add “(HM)” after lower catchment.

62/576: „Bronstert“ (instead of „Bronstaert“)

106: add “(TTD)” after transfer time distribution

110: delete “(TTD)” after transfer time distribution

Fig 2 left and right are interchanged (Depending on whether the picture arrangement or the text is changed, the text in line 162 may have to be corrected.)

Fig 3 light brown is not clearly visible, please use a darker colour

246, 322, 336, 347, 353: correct “VOL1H” to “VOL1h”

270: “one event” instead of “one events”

Fig. 8: The labelling of the x-axes is wrong: it should be ”median transfer time”, “peak flow lag time”, “runoff response concentration”. You could also omit these words because they are explained in the figure caption and just write the abbreviations.

Fig. 9: I cannot find an x-axis for reading the LAI-values.

Fig. 10 upper line: SWC20 should be green, RC should be black

458: „McGlynn et al. 2004“ instead of „McGlynn, McDonnell, Seibert, and Kendall 2004“

478: “Scaini et al. 2018” instead of “Scaini, et al. 2018”

500: “Hortonian flow” instead of “hortonian flow”

You often use „note that“, I would avoid this phrase (at least I would not use it so frequently), but this is a matter of taste.

This study compares storm runoff processes between two catchments with similar size based on the analysis of transfer time distributions (TTDs). The authors present a seasonality in TTDs, which had a different trend between the two catchments. Quick runoff transfers occurred under dry condition in a catchment. The authors attribute the rapid flows to marly plateaus, hydrophobic forest litter, and the absence of a riparian zone in the catchment.

              This paper deals with an important topic. I think their analysis of TTDs using a unit hydrograph model is effective for comparing the storm runoff characteristics between the neighboring catchments. Seasonality in TTDs (Figure 8) is especially interesting. However, data for discussing the causes of the seasonality and inter-catchment differences in TTDs are insufficient. More information about groundwater dynamics and topographic analysis is needed to discuss the causes of rapid flows. My major concerns are listed below, followed a list of specific comments and technical corrections.

General comments

1. Which is the novelty of this study, analytical methodology or the estimated causes of rapid runoff? If the TTD-based comparison of storm runoff characteristics is a novel approach, the authors should emphasize this content in Introduction and Discussion. If they think the causes of rapid runoff are the main findings, they should increase reliability of estimating the causes. In this manuscript, relationships between runoff mechanisms and TTDs are unclear and the factors causing rapid flows in their study sites are only speculated from the results in TTDs.
2. Although the authors focus on bedrock geology, groundwater dynamics in each geology are unclear. The rapid flow due to hydrophobic forest litter was also not observed in their study sites. Moreover, differences in riparian topography between the catchments were not presented despite mentioning riparian buffering. Due to lack of these data, they only speculate the causes of rapid flow. If they want to discuss the causes based on the data in these two catchments, more detailed presentation of groundwater flows and topographic characteristics in the catchments is needed.
3. I can’t understand why the authors compared only two catchments despite the observations in six nested catchments (Figure 1). How different were TTDs between the six catchments? I think examination of relationships between TTDs and catchment characteristics (including geology, topography, catchment size, and vegetation) using the data of six catchments can provide more valuable implications. Even though groundwater flows were not observed, the causes of rapid flow may be estimated with reliability if the comparison of six catchments is conducted.

Specific comments

Title: If the main theme of this paper is causes of flashy runoff, the causes should be examined more deeply based on other groundwater and topographic data or the comparison of TTDs between more catchments.

L14-17: Although the geology of the catchments is well described, there is little information about their topographic characteristics. I want the authors to clearly present the difference in topography between the catchments.

L27-29: These causes are only the speculation and remains hypotheses. As these hypotheses were not verified, this is inappropriate as the conclusion. It may also be possible that the quick runoff under dry condition was caused by direct precipitation on stream channel and/or rapid runoff from riparian zone. As the catchment got wetter, hillslope runoff with long transfer time may contribute to stream water, which can be a possible mechanism of longer TTDs in wet conditions.

L58-60: Whereas the authors wrote “The numerous faults and cracks support quick water transfer through the weathered bedrock and explain fast hydrological responses” in this sentence, they also wrote “Less permeable bedrock will lead to … smaller catchment mean transit times.” in L80-81. Whether the weathered bedrock can contribute to fast responses (smaller transit times) or not?

L89: If the main problem of previous research is the lack of observations in extreme events, this should be clearly presented in Background section. The event magnitude should also be emphasized in the Results and Discussions.

L98-99: I could not understand the difference in flash flood type between Central Europe and MA regions. Please describe the difference more clearly in Background.

L116-162: I could not understand which catchments had more permeable bedrock and larger storage capacity. According to Table 1, geology seems similar between the catchments: Both had the main geology of sandstone and second geology of marls. If the authors focus on the geological features, geological difference between the catchments should be explained more clearly. Information about vegetation is also required because the effects of litter are discussed.

Figure 1: Please add contour lines in the figure. Addition of the map of slope angle or topographic index is also helpful to understand the topographic features of study sites. As soil moisture was observed at the points of raingauges, “Raingauges” should be changed to a phrase such as “Raingauges and soil moisture observations”.

Table 1: What is the difference in river width and riparian area between the catchment sections? Similar area, elevation range, and slope range does not necessarily mean that the two catchments have similar topography.

L139: Does “deeply cut” mean that valley was deeper in HM section than KOE catchment? If so, this topographic characteristic should be quantitatively presented.

L212 “net rainfall amount after infiltration”: How did you determine the amount of loss (i.e., total rainfall – net rainfall)?

Figure 7: Hydrographs in Heffingen catchment were very clearly different from those in Koedange and Medernech catchments. Why was the runoff delayed in Heffingen? I think the comparison between various catchments may provide clearer insights into runoff mechanism than the comparison between only the two catchments.

Figure 8: Please add the results in runoff coefficients of each event. I also recommend the authors to conduct statistical in order to examine whether the difference in the TTD values between the two catchments was significant in each event.

L418: I want the authors to show the location of “large flat terrain” in Figure 1 based on topographic map with the spatial distribution of slope angle or topographic index. It would also be helpful if the area of this flat terrain can be shown in Table 1.

L421-422: Why does the limited permeability of underlying bedrock lead to large storage capacity? I think permeable bedrock has larger storage capacity because groundwater is stored within weathered layer or fractures in bedrock.

L490-510: Although only the effect of litter layer is discussed, discussion about evapotranspiration is also necessary for the impact of the vegetation because LAI directly affects it.

L513-514: Differences in geological substrates and landscape features between the catchments should be more clearly presented throughout the manuscript.

L516-517: There is no evidence that main runoff source in the KOE catchment was groundwater and deep soil water.

L521-523, L529-530: It seems that the authors attribute the difference in runoff characteristics to topography in slope and riparian zone rather than geology. If so, stories focusing on the topographic features may be better.

L531-532: Runoff coefficients were one of magnitude smaller in summer than in winter (L439). I think this result indicates that runoff during dry summer season had small risk of flooding even if the rapid flows occurred. Both results of runoff coefficients and TTDs should be considered to provide conclusion for flood risk management.

Technical corrections

L24: “Another catchment” would be better than “The HM section” because I could not understand this is the name of catchment when I firstly read Abstract.

L100: Does “mean summer and winter runoff” mean baseflow runoff in summer and winter?

L111-114: I think these sentences are unnecessary.

L154-156: The order of Figures 2 and 3 is reversed.

L177: Although it was written that “the rainfall amount had to exceed 10 mm”, there is an event with the rainfall amount of 9.8 mm (Table 2).

Figures 6 and 7: Please check if the date of (c) is true. Were they really different between the two figures?

L308-310: Were these values the ranges in both catchments?

Figure 10: The color of SWC20 and RC may be wrong.