Manuscript presents an analysis of the database on questionnaires after the 2013 and 2016 floods in Germany, comprehending data on socio-demographic characteristics, hydraulic flood characteristics , financial impacts and recovery, coping strategies, previous experience and risk perception by citizens. The analysis is based on averages of the data within four distinct hazard classifications (=pathways: dike breach, river flood, surface water flood, flash flood); cross-comparisons are discussed on a qualitative basis; no quantitative correlations are performed.

Paper is in general well organized and written. Some sections may be shortened, but this is not a crucial point. Resulting evidences are sometimes quite expected (for example: warning is more critical for flash flood with respect to river floods; forecasting convecting storms or dike breaches is challenging). Other results are less obvious and potentially more informative.

Data analysis reveals some interesting points. However, I think that the global result is weak, and the take-home message not really sharp. I try to elaborate this.

The four scenarios (pathways) here depicted have, indeed, significant differences for some of the analysed characteristics. However: are such different a general property of the pathway, or are they related to the specific event? In other words: can we conclude, say, that all (most) dike-breach events have some characteristics differentiating them from all (most) surface water floods?

As said, some characteristics are quite obviously related to the definition of the pathway, and can be expected to be general (but this would be a relatively expected conclusion); others may not be obvious, but I do not know how to conclude that they have general validity. For example: “shortcomings in crisis and risk communication” are related to pluvial flooding (as suggested at line 398) or to the specific event and the local organization of the specific area?

Aim of the paper is to “reveal whether and how people affected by different flood types and pathways were prepared before the damaging event, how they were impacted in hydraulic, financial and psychological terms, and how they coped with and recovered from these impacts. The intention is to provide more insights that help establish risk management strategies tailored to different flood types and pathways.”

As we see, the aim is basically to qualitatively describe some general characteristics; as said, I fear that such scenarios may have no general value and, therefore, may not be applicable in similar hazard pathways but in different social and geographic contexts. Consequently, we may surely agree on the risk management strategies proposed for the four depicted situations, but we should link them to the global scenarios for the four situations, including all hazard + exposure + vulnerability indicators, rather than to the hazard pathways only.

Finally, I have one specific (but not irrelevant) objection: definitions of the pathways are not univocal. Distinction between surface water flood and flash flood for the 2016 event, even if somehow subjective, looks relatively robust. However, distinction between dike breach and river flood for the 2013 event is based on the indication of citizens: “all households that reported that they had been affected by a dike breach were included in this subsample”. Can we really assume that citizens have robust knowledge of the mechanics of the flood (pathway) hitting their house? Was any validation of their declarations performed?

On the whole, I do not think that the manuscript provides significant innovation and extra knowledge for the field of flood risk assessment and management. 

This is a simple but useful article that is using historical flood loss data to compare flood types. It presents observations such as that dike breaches typically have longer flood durations or that flash floods typically have higher flow velocities. Many of the conclusions are obvious but its good to have data that confirms these intuitions.  I’m not aware of another study doing the same and I think no other dataset would be more suitable for such a study. The article is very well written but I do have some minor comments.

• Are the samples independent enough to draw generalizable conclusions? Could some of the conclusions not be statistical significant because of a high spatial correlation between the samples? For example, you only have 128 flash flood samples but over how many different locations (e.g. villages) are they really collected? Some of the variables you look into could be highly spatial correlated (e.g. the same for the whole village or even the entire flood event). The most extreme example is table 5 which is somewhat correlated throughout the event (e.g. media coverage, quality weather forecast, etc.). So arguable the sample size for table 5 is just 1? I think this drawback of the study should be highlighted more throughout the results and discussion section so that the readers know which conclusions can be generalized to other areas/countries and which conclusions could be just a local coincidence.

• The definition of flash flood is a bit subjective in this paper and the conclusions will be very sensitive for this classification. It seems like some circular reasoning could be occurring. That is you seemed to have used some flood intensity information from the household surveys to label observations a flash flood and then you seem to have concluded that flash floods are typically more intense in the same survey data. Is this observation correct or did you merely validate your flash flood classification with the household survey? Could you discuss the potential consequences of your labelling technique on conclusions of the paper? Furthermore, I wonder whether there isn’t a more objective way of classifying flash floods. Maybe extreme rainfall in a terrain that isn’t flat? Have you done some literature review on this?

• The use of the term “compound flood event” is causing confusion. From just reading the title most readers would assume this is about the coincidence of coastal and fluvial flooding. This makes the paper title somewhat misleading and also confusing because the link to the rest of the title is then no longer clear. The abstract adds to this confusion as the term compound flooding is introduced in an unexpected context. The start of section 2 clarifies everything very well and I understand why the term is appropriate but I still recommend either explaining the unconventional use of the term compound flood event early in the abstract or using different terminology (e.g. why not use the word flood type in the title).

• Section 3 doesn’t explain the approach at a high level. This approach is quite simple so you can keep it short.