The authors present an interesting article on a topic which is well-known to most flood modelers, but what is under-researched so far. In that sense, this article is very welcome and the analyses are very systematic, well structured and well explained. I only have several minor points where the authors should comment on:

-    Mitigation strategies are in the title, but only investigated a bit in chap. 3.5; the authors should check whether that should remain in the title; should they want to leave it in the title, a bit more investigations should be done

-    L. 31f: I suggest here to add that the 2 methods are for the computation of the losses or are runoff generation approaches

-    Fig 1: please show both creeks in Fig 1, also Würzburg; I suggest to use the English terms, eg Bavaria; please check whether the Reichenberger and Guttenberger Bach can be shown more clearly

-    L. 118: comparatively dry for Bavaria, eg in Berlin / Brandenburg the precipitation is ~500 mm; 800 mm is average for Germany -> rephrase

-    Chap 2.2: please comment on even larger events eg for the Starkregengefahrenkarten which are currently produced in the federal states 90 or 100 mm in 1 hour is investigated; this can be done here or in the discussion later, limitation part; how would more extreme events impact your results ?

-    L. 150: would you expect better results with a finer mesh ? grid convergence, please comment on this

-    L. 151: mention the resolution of the DEM

-    Headline 2.4: wouldn't it be better to call this Plausibilisation of the Model or similar ? see also line 491

-    Chap. 2.5: comment here or in the discussion; you could also link Telemac2D with a drainage model; would that have advantages, possibly produce more realistic results ?

-    Chap. 3.3: I suggest to include 1, 2 or 3 cross sections and the related figures from the appendix here in the paper; the cross sections not shown and the related text then should occur in the appendix

-    Could one think about some statistical approaches to determine when which culvert is blocked to wish extent ? include that also in the manuscript

-    Can't one overall say the impact of the blockage is not that big except some local effects ? Should you agree, formulate that more clearly, especially in the conclusions and abstract; this would not mean that the paper is not good or not novel, but would provide a valuable information to readers and flood modelers

There are further minor points in a pdf attached; no need to comment on them 

The manuscript examines the issue of modeling culvert blockage during flood events. This process is significant in river reaches where culverts are numerous, often undersized, and susceptible to clogging by material transported by the flow. The study builds on the TELEMAC-2D hydrodynamic model, which already includes a routine for simulating culverts as one-dimensional elements. The authors extend this routine by introducing an additional head-loss term defined as a function of the blockage ratio (BR). This term is assumed to be activated dynamically when the ratio (TR) between the upstream water depth and the culvert’s height exceeds prescribed threshold values. The influence of these two parameters on flood dynamics is investigated through a set of hydrological–hydrodynamic scenarios developed for a small German catchment in which several culverts are present.

In principle, the study is relevant and aligned with the topics addressed by the journal. However, in my view, several aspects require further clarification before the manuscript can be considered for publication in NHESS.

General comments

Choice of study basin. A general concern relates to the authors’ selection of a reference catchment for which hydrological information is extremely limited. There are no hydrometric measurements available, either at the outlet or within the basin. Information on the channel network is also incomplete; several culverts were missing from the initial dataset, and the authors acknowledge the possible presence of additional undocumented structures (line 110). This lack of data has made model validation essentially impossible, meaning that all scenarios are assessed only relative to a reference simulation that may itself not reflect actual system behavior. The authors should provide a clear justification for adopting such a poorly documented case study.

Result interpretation. In several sections, the discussion of results is excessively detailed yet does not offer adequate interpretation, particularly where outcomes appear counterintuitive or anomalous. The authors should make a more substantial effort to synthesize the findings and, importantly, to explain and justify the behaviors observed rather than simply describing them. This issue is especially relevant to my comment no. 11 below.

Specific comments

In the following, I outline, point by point, the aspects that in my opinion require further clarification.

1. Line 98. The manuscript mentions here that “the study area has high gradients” yet no quantitative information is provided regarding the channel slopes or, more importantly, the slopes of the principal culverts (particularly the longer structures near the municipality of Reichemberg). This omission is nontrivial because the culvert modelling approach implemented in TELEMAC-2D - on which the entire study is based - excludes Type 1 flow conditions, i.e., situations in which the control section is at the inlet and the culvert slope exceeds the critical slope. The authors later refer (line 200) to Smolders et al. (2016), who state that Type 1 operation is very rare; however, this statement pertains to a completely different hydrodynamic setting, involving culverts between tidal basins with negligible slopes. The authors should therefore verify, at least for the longest and hydraulically most relevant culverts, whether their slopes can exceed the critical value for plausible discharge ranges, and hence whether Type 1 operation can indeed be disregarded a priori.
2. Figure 1 should be enhanced by clearly delineating the channel network, labelling the streams referenced in the text, and indicating the flow direction.
3. (Section 2.2 – Precipitation scenarios). The manuscript does not specify whether rain gauges are available, how many exist, or where they are located. Are they situated within the catchment boundaries or outside? This information should be provided, and the locations should be shown in Figure 1 if applicable.
4. Lines 121–125. Given the relatively small size of the basin (33 km²) and its likely short hydrological response times, it is not particularly informative to report, for example, that 60 mm of rainfall occurred over a 13-hour period on 9 July. What matters is the temporal distribution of this event, especially the peak intensities. Additional detail on the hyetograph should therefore be provided. Moreover, for Events 1, 3, and 4, the authors report total precipitation amounts (e.g., 30 mm in 3 hours) and also state that values “locally reached 37 mm in 3 hours.” The meaning of this statement is unclear. Does spatially distributed rainfall information exist for these events? If so, what is the data source - point rain gauges, radar-derived products, or another dataset? Please clarify.
5. (2.3 Mesh generation and parameters). At the beginning of this section, there is no information on the topographic datasets used. Only later (lines 153–154) is there a brief indication of the provider, but no details are given regarding DEM resolution or the acquisition technique (e.g., LiDAR or another method). Considering that the channels in the study area are relatively small, the quality and resolution of the DEM - along with any supplementary field surveys - are critical for accurately representing the terrain. Similarly, the main geometric characteristics of the culverts (width, height, length, slope, etc.) are not reported. Please provide this information in a specific Table.
6. (Section 2.4 – Model validation). Given the very limited amount of observational data, I do not consider it appropriate to refer to this step as “model validation”. The authors explicitly acknowledge that the only available information consists of a few photographs which, due to the lack of temporal metadata, do not necessarily correspond to the peak inundation extent (line 168) and which cover only selected portions of the study area. Under these conditions, true validation is not feasible. I recommend renaming this section to “Simulation of real events” and avoiding the term “validation event”.
7. Lines 172–175. The manuscript compares the rainfall events of 9 and 15 July 2021, asserting that the former was more severe because it produced 60 mm over 13 hours, whereas the latter produced “only” 25 mm in two hours (although locally up to 43 mm in two hours, as stated in line 124). As the authors will surely know, the severity of a rainfall event cannot be assessed solely using its total depth; the temporal distribution and the mean intensity over timescales comparable to the basin’s response time are far more relevant. For a basin of this size, the response time is likely on the order of a few hours, not 13. I therefore ask the authors to revise this paragraph and clearly state the criteria by which they consider the 9 July event to be more severe.
8. Line 178. The manuscript states that "a spatially distributed precipitation" was used, but it is not explained how this field was derived. Was it interpolated from point rain gauges, obtained from radar-based products, or generated through another method? Please clarify the data source and processing steps used to obtain this distributed precipitation input.
9. (Section 3.1). In my view, this entire section -including Table 3 and Figure 5 - introduces unnecessary confusion and does not substantially contribute to the manuscript. The authors themselves acknowledge that the inventory of culverts in the basin is incomplete, and there is no hydrometric information available to calibrate or validate the various configurations. Given these limitations, the analysis presented in this section does not appear to be supported by verifiable evidence. Considering the manuscript’s overall length, this paragraph could be omitted without loss of content.
10. Lines 322–327 and Figure A2. The delayed response observed at section 2 for the HN5 and HN10 events requires explanation. This delay cannot be attributed to culvert blockage, as the hydrographs show negligible differences across scenarios. Additionally, it is unclear why, for the HN30 event, the peak flow in the br_08_tr_1.2 scenario is slightly higher than in the reference case. The authors should provide a clear explanation for these behaviors. This issue is closely related to the concerns raised in my subsequent point 11 and should be addressed in that context.
11. Lines 354–392. This entire section requires a more thorough explanation, as both the comments provided and Figure 7 are unclear to me. Focusing on the HN5 case (5-year return period, therefore a relatively frequent event), Figure 7 (first panel, top left) presents the computed discharge hydrographs for the different culvert blockage scenarios. In the reference case (no blockage) and in the br_08_tr_1.5 scenario (delayed blockage), the flow at section 3 remains below 0.04 m³/s throughout the simulation period (over 9 hours). In contrast, in the other scenarios, starting around the fourth or fifth hour, the discharge increases sharply, reaching a peak above 0.4 m³/s around 6:20, before gradually decreasing, yet remaining higher than in the reference and delayed-blockage cases. It is unclear why the discharge in the presence of blocked culverts is up to ten times (!) higher than in the unclogged scenarios. Given that rainfall excess input is identical across all simulations, one would expect culvert blockage to reduce discharge through section 3 due to hydraulic constriction, with the excess volume appearing as surface flooding. Indeed, Figure 8 confirms that maximum water depths in flooded areas are greater under blocked culvert conditions than in the reference scenario. However, this observation contradicts the trend in Figure 7, where the discharge and conveyed volumes are higher in the blocked cases. The same unexpected pattern, although less pronounced, occurs for higher return periods: in every instance, the volume passing through section 3 is consistently higher when culverts are blocked than when they are unobstructed. The authors need to provide a comprehensive explanation for this behavior. Without such clarification, one might reasonably suspect that the model does not preserve the mass continuity, which would call all results into question and undermine the validity of the study. I also recommend that, at least for the HN5 case, the authors calculate the ratio between the volume passing through section 3 and the rainfall excess from the contributing sub-catchments. This would help identify the fate of the rainfall excess volume and clarify whether surface retention or other processes are being properly accounted for in the model.
12. Lines 370–372. The authors report a specific observation in these lines. How is this behavior explained? Please refer to the issues raised in point 11 regarding the unexpected discharge trends under culvert blockage scenarios.
13. Lines 376–377. Similarly, this observation requires clarification. How can the results be interpreted in light of the anomalies noted in point 11?
14. Lines 414 and 417. The sentences “Increased inundation does not necessarily increase the damage induced by the flood” and “As higher water depths are also linked to greater damage” appear contradictory. The authors should clarify the intended meaning and ensure consistency in their discussion of damage relationships.
15. Lines 427–428. The reported observation requires explanation. How do the authors account for this behavior in the model results?
16. Line 430. The meaning of “temporal trend” is unclear. Do the authors refer to changes with respect to return period? This phrase should be clarified or corrected.
17. Lines 434–436. The authors acknowledge that numerical instabilities in the model may explain anomalous behavior. This is concerning because such instabilities could potentially affect other scenarios but remain undetected. The authors should comment on the robustness of the simulations and any measures taken to verify stability.
18. Lines 477–478. The term “redistribution effects” is not clearly defined. Please provide a clear explanation.
19. Lines 520–522. The statement that “this study strongly cautions against deliberately inducing blockage to reduce discharge as a basis for mitigation planning” underscores the concerns raised in point 11. The model results in Figure 7 show an increase in flow rate under blockage scenarios, which appears inconsistent with this recommendation. The authors should reconcile these observations and clarify the implications for flood management.

Technical corrections (typing errors, etc.)

• Line 193: Replace “diameter” with “height”.
• Line 195: Replace “before” with “upstream”.
• Line 317: Remove the repeated phrase.
• Lines 356 and caption of Figure 7: Correct “cross-section 4” to “cross-section 3” (if my interpretation is correct).
• Line 357: Remove the duplicate phrase “a slight delay”.
• Line 368: Clarify whether it refers to cross-section 4 or cross-section 3.
• Line 443: Change “In general, the inundation…” to “In general, the relative increase of inundation…”
• Line 468: Change “the resulting inundation extents…” to “the relative differences of inundation extents…”
• Line 525: Replace “has demonstrated…” with a less assertive verb, such as “has shown…”

The paper presents an interesting case study which aims at assessing the influence of culverts’ blockage on flash flood inundations. The authors have used TELEMAC-2D (after implementing a specific culvert blockage module) to simulate different precipitation scenarios assuming different blockage conditions.

I think that the paper can be a relevant contribution, but some major revisions are required before it can be accepted for publication.

Main points:

• The title should be revised. First, it should explicitly reference “flash floods” instead of generic “floods”. Besides, the paper just suggests how to find the most critical culverts, rather than giving practical guidance on how to design mitigation strategies. The authors should try to devise a new title that better reflects the main focus of the paper.
• Figure 1 should be enriched with more labels, e.g. outlet section, streams and locations that are referenced in the paper. The results description is currently hard to follow for readers that are not familiar with this area, due to the many geographical references.
• The study area lacks accurate observations that allow calibrating the model. However, I acknowledge that the paper mostly presents comparative results (e.g. blocked/non-blocked culverts, different blockage ratios, etc.) to show the relative importance of the culvert assumptions, and this may make the lack of calibration less relevant. My suggestion is therefore to treat the case study as if it was a synthetic case, although inspired by a real catchment. After acknowledging this hypothesis, the authors can completely remove any mention of real events (e.g. lines 118-129 and section 2.4) and just state that the parameters (CN, roughness, etc.) and precipitation scenarios are set to reasonable values for this area. The lack of calibration can then be discussed among the limitations.
• Line 215. This sentence is the core of the proposed modification, but it is very cryptic: the residual discharge is distributed (how?) onto neighboring (of whom?) nodes. Please provide more information about this novelty, and maybe a simple sketch that clarifies the difference between the original and novel implementation. Moreover, did the authors perform any kind of validation of their novel implementation (e.g. for a simple test case)?
• Section 2.6 presents details about the blockage method using the coefficient k_e. I think the authors should provide a better definition for this coefficient (e.g. by showing how the culvert discharge is computed at least for one flow type). This could help justify why the coefficient is set to 0.5 at line 240.
• Results section. I think that this section should be re-organized and improved. Some suggestions are provided in the following points.
• The global results (outlet hydrographs and flood extent) should be discussed first, since they are connected (i.e. blockage leads to a peak flow reduction due to increased flood extent).
• As regards the inundation, I understand that the catchment size and relatively limited flood extent prevents displaying readable comparative maps. However, the authors should not only compare the total flooded areas, but also provide some quantitative metrics (e.g. ratio TP/(TP+FN+FP) or other metrics combining True/False Positives/Negatives) about the inundation overlapping (each scenario compared to the reference one). This could be useful to highlight if flooding occurs over different areas due to culvert blockages. Moreover, if the flooded area is overall similar but the max depths are different, another option could be to identify a limited number of classes of max depth intervals (e.g. 0-0.25 m; 0.25-0.5 m; 0.5-0.75 m; 0.75-1 m; >1 m) and compare histograms reporting the areas belonging to each class. This could also be useful to discuss variations among scenarios.
• Section 3.3 is very descriptive but hard to follow without being familiar with the layout and conveyance capacity of streams and culverts. Moreover, all hydrographs look very similar, except at cross-section 3, so I wonder if the authors could just focus on a couple of interesting locations and two/three precipitation scenarios and make an effort to summarize only the relevant findings.
• Overall, the results of this case study seem to imply that the “impact” of the culvert blockage is not so large at the global scale (just a few percentage points in terms of flood extent and peak discharge), even by assuming different blockage configurations. This should be stated more explicitly in the conclusions. However, locally the impact can be more relevant: the revision of section 3.3 could give the opportunity to show an example and better discuss this point.
• Lines 435-436. I was a bit astonished here. If there are numerical instabilities, the simulation results should not be included in the analysis.

Specific points:

• Table 1 can maybe be replaced by a figure.
• Figure 7. Check captions. Cross section 3 or 4?
• Line 351 and similar. I don’t understand the implication that flooding occurs when the discharge in the culvert is larger than zero. Is this culvert supposed to remain dry even during precipitation events?
• Line 358. Typo.
• Line 422. Please report here at least the absolute values of the flood extent for the reference scenario, so that the reader can grasp its magnitude without having to jump to the appendix.

This paper presents a hydrological-hydraulic modelling framework aimed at simulating the effect of culvert blockage in minor drainage networks. the main component is a modified version of TELEMAC-2D model that can handle dynamic blockage conditions. The modelling framework is applied to a drainage basin in Germany to simulate different scenarios of rainfall intensity and culvert blockage conditions.

Overall this is an interesting work. It addresses an important problem that is often underestimated both in scientific literature and in operational flood risk management. 

Main comments

1) My feeling is that the focus of the work is too narrow, resembling a technical report rather than a scientific paper. There is little reference to importance of the problem on a bigger scale, and little attention is given to the applicability of the proposed approach to other case studies. For instance, would the TELEMAC 2D work well also for larger river networks, or under more complex hydraulic conditions such as bifurcating culverts, underground confluences etc? Can it be coupled with stormwater models? These are all issues that should be discussed by the authors.

2) The reliability of the modelling framework is not sufficiently discussed in the manuscript. The authors state that discharge measurements are not available for the study area, and therefore they carried out a plausibility check using data from a reference flood event. However, the authors should discuss if and how modelled flood extent and depth agrees with observations. 

Furthermore, the plausibility of model results could be assessed by comparing the frequency of simulated flooding due to culvert blockages with reported frequency of flooding. For instance, results at cross-section 4 (figure 7) indicate that the study area is likely to be flooded at least once every 10 years, or more frequently in case of culvert blockage. Is this in agreement with reported flood frequency, or is the model overestimating the occurrence of flood events? 

3) L116 "Without these culverts, the sub-catchments upstream of the culverts are disconnected from the overall flood" and Lines 273-274: "Omitting culverts in hydrodynamic models can cause parts of the catchment to become artificially disconnected from the flood dynamics,..." these two sentences are not clear, do you mean that there is no hydrological connectivity? My understanding is that the model can simulate surface runoff, so channel continuity is not a requisite to convey water downstream. Can you please clarify and expand the description of the modelling framework on this?

Other comments

Line 48-49: "However, these logjams at bridges usually show larger impacts at larger bridges, which are present at large rivers and therefore not directly an issue for flash floods at smaller streams." I do not agree here: also bridges over minor rivers are prone to debris jams, and the  determine backwater effects and consequent overflow upstream, especially in steep areas. I would say that such events can be highly dangerous becasue of the quick onset of flood waves. 

L118: "the Reichenberger Bach lies in a comparatively dry region with a mean annual precipitation of 712mm..." comparatively dry to what? To the rest of Germany?

Figure 1: please add the two water courses in the map 

L120-125: what does NH means?

L128: is KOSTRA 2020 a dataset of measurements, or something different? Is it at national scale? Please explain 

L193-194: "The flow types can be further grouped into inlet (flow types 1 and 5) and outlet control (flow types 2, 3, 4, and 6)." If these flow types are relevant for the study you should provide at least a short description of each. Otherwise, this can be omitted

Figure 10: the graph shows many simulations and most of them look similar, consider omitting some of them

Section 3.5 I would rename it as "Simulation of mitigation measures", there are no indications for the design

L513: "backwater effects tend to be constrained by the bridge thickness." not fully clear: do you mean that the bridge decks are relatively thin in the study areas and hence cannot block substantial amount of floating debris?

L526: "The results show that certain culverts can exert an large influence on flood dynamics,..." This is perhaps an overstatement given the results, culvert blockage can increase flood depths and extent at certain locations, but the effect is not so large

L533-534: "Future research should further investigate the processes driving culvert blockage formation. Laboratory experiments can

provide insight into blockage mechanisms and thresholds, which could then be translated into dynamic numerical models." Even though this might be useful, I would first prioritize the installation of water level/discharge gauges in strategic points of the drainage network. This would allow to measure the response of the basin to different rainfall events and relate it with culvert blockages. Having real-time measurements could also allow for more timely interventions to remove blockages