I should note that my background is not in the kind of precipitation analyses presented in the paper but rather sediment sources and transport in glacial and paraglacial systems. This means that my review of the paper is only partial.

Overall, this paper is worth publishing in my view. It is extremely well-written and well-presented. It has an importance as whilst it builds upon similar papers from authors of this one, and with overlapping geographical foci, the topic is important as there is an ongoing debate over how the transition from glacier-melt dominated to rainfall-dominated catchments impacts suspended sediment yield from deglaciating basins. The analysis is largely very well done (I make some more minor comments below). On this basis I would hope that the paper can be published after some revision.

That said, I do think the authors need to be much more careful in how they present and interpret this work and I would recommend that they make some revisions.

1. The general positioning of the paper is stronger on the hydrology than on the glaciology. There is actually a very established literature on how glaciated catchments produce suspended sediment and the processes that drive it (e.g. the work of Darrel Swift, Angela Gurnell, David Hannah) and I think this should be brought into the paper. Note David Hannah’s work is also important for hydrograph analysis in this kind of setting. This is not least because you cannot assume that extreme precipitation and melt-driven suspended sediment production events are entirely independent. Some of the previous work by the authors recognises this and the point is made at l502. It is quite possible that your extreme rainfall events do not change erosion but rather increase sourcing of subglacial suspended sediment. We know that surcharge of subglacial drainage systems by water, whether melt, precipitation or precipitation-indeed melt, leads to very major suspended sediment yield, especially in “spring events”. These don’t just occur in spring but can occur at any time and they likely interact with other variables such as snow cover on the glacier. The discussion that then follows from around l410+ here appears to be highly speculative as we have nothing on this kind of driver but a lot on classical kinds of suspended sediment sources associated with very different kinds of environments.
2. Part of the problem here is that with two gauging station records of this kind, separating out individual and joint effects is not easy, and not least because (see below), your flux and yield data are collected from discharge which in turn will be a function of precipitation; so when you correlate precipitations variables with flux and yield data you have an inevitable correlation. The authors might have been much better to do a more event-based analysis than they have done.
3. The same problem under (1) is then reflected in the literature, how the paper is both set up and discussed. I am not sure how much time the authors have actually spent in this kind of river basin during extreme events, but I am concerned to see a number of papers developed for suspended sediment / soil erosion in very different environments being transferred uncritically into this one. Is Wischmeier and Smith (1978) really relevant to this case? Dunkerley’s excellent work from Australia? Processes are invoked to explain results (e.g. soil erosion due to overland flow) when field measurements of such processes question whether they really occur in at least some of the environments described here. Specifically, it is really rare to see classic overland flow in these kinds of catchments in these kinds of environments unless there is some interaction with frozen ground. The unconsolidated sediments you note on l510+ have exceptionally high vertical infiltration rates unless they contain buried ice (they typically contain very low clay fractions, for instance). The (albeit restricted) literature on shallow groundwater flow in deglaciating environments which actually suggests very low degrees of surface hydrological connectivity and hence overland flow generation – Tom Müller’s work is a good starting point (e.g. in HESS). So do these environments produce overland flow and soil erosion? What is the state of the soil development processes in this river basin that may lead to higher rates of moisture retention and overland flow generation? Where has soil developed? What is the link to buried ice (in the most upper parts of the catchment) which can be a crucial generator of impeded drainage and surface saturation? What evidence is there that processes matter besides basic fluvial reworking of former glacial sediments in high discharge events? There are many processes that could explain the results obtained and the review needs to be more balanced as there is a gap between the high quality analysis and the interpretation presented. Two things are required in revision; (a) a much more thoughtful and careful use of supporting literature, refocused on what we know of the geomorphology and hydrology of these kinds of Alpine landscapes, both in the introduction and the discussion; and (b) much more circumspect interpretation as there is only so much you can do to disaggregate catchment-scale yield data to identify potential sources within the river basin.
4. Point (3a) needs to be supported by better river basin characterisation. For each station we need to know contemporary % ice cover (we are just told it is 10% for whole system what is it for Vent). It would also be good to know what % of the catchments were ice covered during the Little Ice Age and how vegetation cover has developed since, given its potential impacts on erosion processes. This would help us to get a better understanding of process regimes associated with the basin.
5. The analytical approach in Table 2 is a bit weak. Your SSY data are effectively the sum of SSC*Q. Q is directly driven by precipitation. So, you would expect strong P versus SSY correlations. What is perhaps more interesting is to look at when P and SSY is not correlated – this might say much more about what is going on. Note, I also struggled to understand which datasets were being used for Table 2 – Tumpen or Vent?
6. Figure 6. Some of these relationships look like a linear fit it being done to a non-linear dataset. This is really difficult to see with these plots, especially with the marker being scaled by even duration. Non-linear relationships here might tell you a lot.

General comments

This manuscript addresses the critical and timely topic of suspended sediment dynamics in a rapidly deglaciating Alpine catchment under the influence of increasing extreme rainfall events. This work is highly relevant to the scope of Hydrology and Earth System Sciences (HESS) as it provides quantitative insights into how climatic shifts are altering hydro-geomorphic processes in high-mountain environments. The study is well-conducted and presents a thorough analysis linking extreme precipitation characteristics to suspended sediment fluxes in the Tumpen-Ötztal and Vent-Rofental catchments in Austria. The methods are robust and clearly described, and the resulting data analysis is systematic. The manuscript is well-written, and the discussion is generally comprehensive. Overall, I recommend publication of the manuscript following minor revisions. Some comments are listed below.

Specific Comments

Antecedent Catchment Conditions. The discussion (lines 484–486) highlights the role of sediment availability for interpreting the year-to-year variability in SSY. In my opinion, authors should also consider the influence of antecedent catchment conditions, specifically factors like soil moisture, which might impact both overland flow generation and soil erosivity, the sequence of events, which dictates the depletion of readily available in-channel and hillslope sediment supply, and the presence/absence of snow cover, which might modulate the rainfall-SSY relationship. An event-based analysis of few targeted events could provide insights on these factors.

Snow and Ice Melt Processes. In general, little space is given in the discussion on the impact of snow and ice melt processes on sediment transport. I suggest the authors expand this discussion to address the temporal dynamics and potential overlap with rainfall extremes. Specifically: When does the peak ice melt happen? Is it overlapping with the period characterized by the highest short-duration convective rainfall? Could this interaction explain the higher SSF_mean observed for short-duration events? Furthermore, describing a typical pattern of snow cover duration in the catchment would add context. Again, by analyzing a few events with different characteristics, as they did for 2020, the authors would be able to incorporate these key cryospheric processes more fully into the discussion.

Event classification. I have a clarifying question regarding the event classification described in Lines 224–226. Does this methodology imply that a genuine long-duration event could be identified or partially characterized as a sub-daily extreme if it contains a single, very intense sub-daily peak? I am not sure I understood this and I wonder how potential mis-classification might influence the results showed in Figure 6 as well as the discussion in 5.2.2. and 5.2.3.

Increasing frequency of extreme precipitation – Stations. The finding of an increasing trend in the frequency of extremes derived from the INCA product is central to the study's context. Have the authors checked if a similar increasing trend is observable in the precipitation station data used for INCA development? Recognizing the already thorough nature of the analysis, I suggest the authors check if a similar increasing trend is observable in some targeted stations. This would enhance the robustness of the signal by ruling out the possibility of the trend being an artifact of the gridded product or its calibration process.

Technical Corrections

Figure 2. The font size is rather small. The figure could be a bit bigger.

Line 100-101, 108-109: In the sentence “The accuracy of INCA estimates can vary, particularly in complex terrain, with an average error of 50-100% in the 15-minute precipitation grids and 1.0 to 1.5 ◦C in the temperature grids (Haiden et al., 2011).”, the 15-minutes precipitation grids confuses because in line 100-101 the authors describe the INCA datasets as “… hourly 1-km grids for all of Austria.”. Perhaps just add “… and sub-hourly …” to the sentence in line 100-101.

Line 115-116: Please, provide a brief (few words) explanation of the rainfall/snowfall separation method used in openAMUNDSEN.

Line 120-123: As I understand here, you used hourly, 1-km grids, precipitation for the period 2004-2024, and rainfall for the period 2011-2024. Correct? Please, clarify.

Line 125. Figure 1 to Figure 1b.

Line 190: Please improve clarity by changing “Detection thresholds, u, for each …” to “Detection thresholds, u, for each DURATION d and spatial scale (i.e. GRID-SCALE It or CATCHMENT-AVERAGED Pt).

Line 210: I believe I understood what you did, but could you please write this iterative merging in a clearer way?

Line 265: Since you are talking about events with SSC larger than the 90^th percentile, I find P₉₀(SSC_t) a confusing definition and would change it to SSC₉₀.

Figure 8. Font size rather too large. I suggest being more precise in the legend: from “extreme” to “extreme precipitation”, from “non-extreme” to “non-extreme precipitation”.

This study builds on years of meteorological, hydrological and turbidity/suspended sediment load (SSL) data collected by different institutions in and in the vicinity of a large partially glaciated catchment in the Austrian Central Alps for an event-based analysis of the contribution of different types of rainfall/precipitation/discharge events to suspended sediment yield. The study area Ötztal includes the smaller and more glaciated subcatchment Vent-Rofental. For the definition of precipitation events, the authors develop an interesting multi-scale approach, both spatially (catchment-scale vs. “grid” scale) and temporally (event duration). From the latter approach, they derive, for every duration class, a threshold at the respective 80% exceedance probability above which heavy precipitation events are delineated and characterized by a number of parameters. The authors then investigate trends in event number and characteristics, and the contribution of different types of events to the annual sediment yield. Major findings include an increase in the frequency of heavy precipitation and contrasting trends of annual suspended sediment yield that are attributed to different timing of ‘peak sediment’ and different reaction to heavy precipitation in the more glaciated part of the valley.

Regarding the study region and the focus on (changes in) suspended sediment load, the study connects with earlier publications by the same working group (Schmidt et al), but represents an original contribution. We appreciate the methodological development for the event-based analysis (and uncertainty assessment) of the INCA, discharge and SSC data that could be transferred to other study areas where comparative spatially distributed precipitation/rainfall data are available.

The manuscript is very well written and contains very rich figures, some of which might take readers some time to understand all the details contained in the diagrams. We think that the study is highly interesting for the geomorphological and hydrological scientific community and can be published pending moderate revisions being made.

General Comments

• (over)use of the term ‘extreme’. We suggest to replace this by “heavy precipitation event” whereever applicable (see e.g. L169 – the 10 mm/h judiciously mentioned there as “heavy” are for sure not extreme; L193: catalogue of heavy precipitation events instead of “precipitation extremes”). We acknowledge that extreme value statistics play a role in the analysis and that the results are used as a detection threshold for such events. However, not every single one of the events listed in the final catalogue are truly extreme in light of the whole dataset and the definition of extreme as very rare (an event with a return period of 1,25 years is for sure not to be termed extreme) and of the M sample as annual maxima that don't really need to be 'extreme' in a sense different from just being the highest in one year.
• “grid scale” vs. “catchment scale”: While ‘catchment scale’ can be understood intuitively, we suggest to rename “grid scale” (a grid can represent a whole catchment as well; it consists of grid cells) in order to better represent the ‘grid cell’ or even more simply ‘local’ scale.
• Units seem to be formatted differently than normal text. If that is intentional and typesetting rules requires that => OK
• Figure captions: We do appreciate the very rich figures, but in some cases we’d suggest to facilitate the readers’ orientation and intuitive understanding by including more direct relationships between axis title/label colour and diagram content, legend etc (see specific comments) instead of having lots of additional information in the figure captions.

Specific comments:

• L 29: While this might be the finding of the cited studies, the reader might stumble across that statement because they’ve just read in the abstract that one major finding of this study is a decline in annual sediment load at Vent-Rofental! We suggest to tune that down to “some studies have found a measurable increase…”
• L51f: The existence of high(er) amounts of unconsolidated sediments and sparse vegetation cover “downstream of glaciers” is only part of the paraglacial morphodynamics theory. According to the latter, it’s not only the existence and amount of sediments, but also the topographic and lithologic characteristics of hillslopes exposed by deglaciation. So, first, “downstream of glaciers” does not define well the proglacial areas (that include specifically important hillslopes) that are affected by paraglacial dynamics – we suggest to replace “downstream of glaciers” with “proglacial areas exposed by deglaciation”. Second, one characteristic of paraglacially enhanced erosion/reworking/transfer of sediments is that it is “system-internal”, not requiring increased external forcing. Hence, an increase in heavy precipitation would be anticipated to further enhance and maybe accelerate paraglacial dynamics.
• L58f: We feel that SSL/SSC peaks and especially hydrological events that are not driven by (heavy) precipitation (which you do address in the study!), need to be mentioned here already.
• L68ff: Start with a sentence that makes it clear(er) that you have a nested catchment approach, i.e. that you investigate the whole catchment (with characteristics such as glaciated proportion) AND the Rofental subcatchment (with much more glaciated area). The two are easily defined as the contributing areas of the two gauging stations, so reference to the nested approach also introduces 2.1 well.
• L70-72: A bit confusing: (a) 10% of the (whole?) catchment are currently glaciated => need to give values for the whole vs. the sub-catchment. (b) “Glacier volume is projected to 4-20% by 2100”: 4-20% refers to what initial volume? Present-day? “pre-industrial”?
• L80-86 refers more to the catchment (and the subcatchment); consider moving this to the introductory “study area” paragraph, while 2.1 would then focus on the measurements conducted at the gauging stations.
• Fig1: The Rofental catchment boundary is hardly visible in the big map (c). Better visibility would support the ‘nested catchment’ approach that we ask to make explicit in the text. Using a different colour than gray in (b) and (c) would probably do the job in both maps.
• Fig1 caption: “The topography of Ötztal is steep” => consider adding a table that gives e.g. average steepness and other characteristics (such as percent area glaciated see previous comment)
• Fig2: We suggest to make the Fig as large as possible (text width); it is quite dense (which is fine, other figures are even more so) and therefore maximum size is needed to help the reader.
• L92: Use “suspended sediment concentration” at the first mention of SSC (followed by a comma and SSCt). Add information on the ‘missing step’ from the originally measured turbitidy to SSC. The single “t” in line 93 can be removed, the unit is just “15min-1”).
• L95 …are hourly precipitation grids” consider adding “at XX km resolution” (same: abstract, L9) and the number of weather stations to “from weather stations”
• 2.1: We appreciate the detailed explanation of INCA and the own validation approach of this paper. Consider moving the “quality check” (L112ff) to the methods chapter (3.1)
• 1, uncertainty analysis, L145ff: Does the computation of deviations between INCA and OBS include “0 precipitation” hours? Further down a threshold is introduced to distinguish wet from dry days, which is good - but that does not refer to the amount of rain in INCA vs Station data. We suppose ME and RMSE are likely biased by including a large number of dry hours where the deviation between INCA and station are 0. This is either a point to discuss in the discussion section or preferably to change (use only hours with nonzero precipitation in the station record)
• L177: Consider referring to Fig in Appendix
• L181f: Consider adding one more sentence so that the reader does not have to look up Ulrich et al. The GEV distribution family has three parameters (location, scale, shape), - how many (which) parameters are being fit in the dGEV and why can the number of parameters be reduced?
• L190: Here, you name the “0.8 exceedance probability quantile”, while in the caption of Fig A2, you write “0.2 non-exceedance probability”. Pls homogenise.
• L187ff: This makes one think immediately of runoff/discharge events triggered by snow (and glacier) melt without precipitation. Yes, snowfall in winter is not relevant to your study, but snowmelt in spring is.
• L194f: The whole “peak detection” approach is explained in the caption of Fig. 3: While that enhances the immediate understanding of Fig3, it’s sort of missing in the text. If you like to keep it as is, consider writing “…tpeak (Fig. 3, explained in detail in figure caption)”
• Eq5/6, L198ff: We appreciate your approach (starting from peak, searching for first and last threshold exceedance before and after the peak, respectively) as an alternative to the “peak over threshold” declustering approach (that starts from a threshold exceedance) used in the corresponding extreme value statistics. However, what happens if the threshold is not exceeded for just one 15 min data point before exceeding it again? In some declustering approaches, a user-specified parameter would prevent two events from being separated by just one (especially that short) ‘break’ in a time period otherwise exceeding the threshold. We feel this could be explored and, also if not implemented, at least discussed later.
• L215: Consider giving an example of what kind of pattern would be judged as a data artefact/mistaken detection
• Headline 3.3 : Would “Classification and characterisation of heavy precipitation events” be better? We felt yes because you do not only characterise the events but also categorise/classify them
• L235: Not clear to what the “search window” refers, pls specify. It is not possible to imagine what constitutes a hydrological event (as opposed to the precipitation event whose delineation is described in detail). This has implications for the following,specifically for the hydrological events that are not triggered by (heavy) precipitation.
• Eq 7,8,9: add units for SSF and SSC
• L255f: The section heading specifies “precipitation-driven events”, but how does the approach detailed here deal with hydrological events that are not triggered by precipitation but by snow and/or glacier melt? Fig 2 clearly shows that spring snow-melt is relevant specifically at Tumpen. Pls address this in more depth in the methods and/or discussion section.
• L260: typo: Theil-Sen slope
• 6, L 262ff: This is about “sediment discharge events”. How are these detected? Do they always coincide with hydrological events? That is, are “sediment discharge events with high SSC / SSC spikes” exclusively a subset of hydrological events? We can’t know, but couldn’t it be that, for some reason, a sediment spike takes place in a minor hydrological event that does not belong to the sample/catalogue of hydrological events? E.g. a very localised rainfall event that does not lead to highly increased discharge but to a substantial input of suspended sediment? Or the sudden mobilisation of subglacial sediment without a very conspicuous hydrological event recorded at a gauge kilometres away? Is it possible that a “peak SSC event” is longer than the hydrological event(s) it is attributed to?
• 1, L269ff: See comment @ methods: Precipitation=0 included in uncertainty assessment?
• L277: INCA under-predicts => suggest to use “underestimates” instead of “predicts”
• L285: This is an example of “extreme” that we’d like you to reconsider
• Fig 4 caption/legend:
  • Dot size indicates the duration of events, but there’s no legend item. Or is dot size meant to be only a qualitative / ordinal measure of duration?
  • in addition to the event with the highes Ptot and Imax. Event 2020-k…: These three events are not shown in the figure – which is a bit confusing. Information on single events not shown should be in the text.
  • Average precipitation area: Is a relative measure (related to the total catchment size)? Shouldn’t it be called “relative precipitation area” instead? The same applies to section 3.3 L220f.
  • Especially in (c), the Vent/ROfental catchment boundary is barely visible (see earlier comment), consider choosing a different colour.
• Fig5a
  • FB is hard to see (dark purple on dark grey), consider choosing a different colour
  • Add a pink left y axis label for RMSE
  • Consider making the precipitation y axis labels blue in order to ease relation to the blue line (like RMSE and also for number of events).
  • We suggest to specifiy “Precipitation” axis title instead of just explaining in caption => “May-Oct precipitation [mm]”
• Fig5b
  • Dot size should have a legend item
  • Include grey dots/circles in the legend (otherwise, in order to understand the figure, the reader would have to read the whole caption first)
• Tab 2: add number of events (n) and type of events (all?). Moreover: Does r refer to Pearson’s r? If so: Pearson’s r only quantifies linear correlations, Spearman’s r would quantify all kinds of (monotonous) correlations.
• L300ff: We accept that your MK test was significant, and that RMSE does not greatly vary between the two major parts of your data characterised by different (INCA vs. pre-INCA) data available. However, can you exclude that the MK trend rather represents a difference between the two parts that is (partially) not due to climate change but to a change in how the data were acquired?
• L324: The difference is particularly pronounced for RFmax and SSY: We think RFmax has been confused with RFtot – here, the difference in r between SSY and SSF is 0.51 vs. 0.27, while it is much more similar for RFMax and SSY/SSF…
• Fig6: Consider adding a legend for point/circle sizes, and add “n” to the boxplots (has been done with other boxplots, should be the same here). Fig6 caption: “Labelled events” Only three out of five labelled events are mentioned in the text (2019g, 2020k, 2022d), and events are mentioned in the text that are not contained neither in Fig4 nor here (2020j,2020n, see page 24).
• Fig7: Try to separate column a/c better from column b/d
• L348: Remove duplicate “only about 10%”
• Fig8: We suggest to position the two plots horizontally (would make the figure fit better with page)
• L375: Pls specify what “geomorphological variables are sensitive to small-scale changes in rainfall spatial structure” means
• Lines 376ff somewhat repeat lines 370ff
• L417ff: We suggest to include the findings of a study that conducted sprinkling experiments on steep moraines; these could be more representative of (parts of) your study area: Maier, F., Lustenberger, F., & van Meerveld, I. (2023). Assessment of plot-scale sediment transport on young moraines in the Swiss Alps using a fluorescent sand tracer. Hydrology and Earth System Sciences, 27(24), 4609–4635. https://doi.org/10.5194/hess-27-4609-2023
• L432: add where the debris flows reported by the two papers took place (Horlach valley, tributary to Oetztal)
• L468: See comment regarding the MK test results (L300) in light of the higher RMSE in the first years. In light of this, the “robust” in L525 of the conclusion needs to be reconsidered.
• Fig A2: Title of (b) should read “Local-scale (a placeholder for an alternative to “grid scale”) maximum precipitation It, just like in the caption. Moreover, the blue threshold line (and others) use the “20% non-exceedance probability” unlike the “80% exceedance probability” terminology used elsewhere. Pls homogenise.
• Fig A3 – Legend for “Station in Ötztal”: Change fill colour to white, because the Ötztal stations have different colours and a thick black outline. Moreover, grey dots have been used elsewhere. Moreover, consider adding a legend for circle sizes
• Fig A4: Legend for point/circle sizes missing; add n to the boxplots. Similiarly to Fig6, you could add number of events and time period.