|Overall, this is a very interesting and complex case study. I recognise the multi-institutional and -discipline effort, as well as the European scale of it. However, I find also major issues that I outline below. Despite I value the work in its nature, my review is critical because it identified these flaws that appeared straightforward to me.|
1) What is has been defined as “object-based” approach is NOT a new approach. Many works in literature (included some of those cited by the paper) assessed the water depth by overlaying road links and flood footprint; I would say this is a “normal procedure” of spatial analysis at urban scale. Perhaps, what is new here is the application of the method at European scale and the damage curves. So, I would revisit the paper and omit the “object-based” approach is new.
2) the resolution of the hazard domain is 100m (!). I am wondering if this is compatible with an assessment of flood impact to roads. Usually for “object-based” assessment analysis to roads in cities the resolution is ~5m (or less). This because the uncertainty of the model simulation (cm? dm? m?) should be compatible with the road dimension (two lanes road, ~6m) and the scale of the object. So perhaps, applying such a coarse flood hazard to a detailed exposure could not make sense. This point is even not discussed or considered in the paper, and I expected authors to address it properly.
3) I don’t understand, and I invite authors to explain, why Sec. 5 is about areas that are different compared to Fig. 5.
4) I understand that Sec.3.4 is about offering a sort of validation of the “new” damage curves. However, I see major criticalities in validation damage curves applied at EU level with a 7x7m reference case. In particular, because this refer to one road type only (motorway), whereas damage curves were developed for six road types (Table 1). Also, even assuming this reference case as appropriate, the model estimation ranges from €3.4m to €28.6m (!) – when the comparison value is €3.8m. I can’t accept this refence case as validation of all the curves (or even the motorway one).
Perhaps, this work is trying to achieve too much in just one paper. The “new” damage curves could be a paper on their own, if well-documented and validated.
Other detailed comments (some also not minor) are included in the following list.
L19 and in general: I find the term “direct” very misleading. This because, until specified, could be misinterpreted (direct risk? Direct assessment? Direct method?). Also, to my knowledge, “direct losses” is the correct term (rather than “direct risk”). I think a simple title “Flood risk assessment of the European road network” would be much more effective.
L23: “more precise” – is it possible to quantify it, with numbers?
L26: move “the Alps” after, i.e. “such as roads in the Alps and along the Sava river”
L38: here “direct” make sense because it is associated to losses.
L40 and below: also vulnerability of parked vehicles https://www.sciencedirect.com/science/article/pii/S0048969718345388
L47 and below: as said, this is not true. Most of urban studies use road links to assess impact of flooding to roads.
L57: I think an overview about the type of damage/losses is needed: direct/indirect, tangible/intangible. Indirect losses should be clearly defined as e.g. “physical damages” of road surface, etc
L68: “complete” – what does it mean?
L73: “object-specific attributes” – what does it means? Just…attributes?
L106: it’s not “new”
L109-110: authors need to explain before this list what CORINE and LUISA are
L112-113: is “object representation” different from “object-based”?
L133: “not include…pluvial, …coastal,, ..river, ..flash”- so what does it include?
L223: “motorways” – in italics with no reason
Table 1: worth adding the labels C1, C2, etc here
L315: “same damage curves” – specify if Huzinga’s, if this is the case
L336: “strong increase” – of which one? specify
L339-340: this point seems quite important to me, because there is the risk to compare “apple and oranges”. Is it possible to quantify the “small” contribution (L342)?
Fig. 4: I would try to use the colours in a way that it is clear what is compared between (a) and (b/c)
L376: from here, the text seems to belong to the following section
Fig. 5: I can’t see the areas mentioned in the main text, e.g. Po area
Fig. 6: I would specific on the top of a/b/c the location (Nl/Alps/Balkans); also a/b/c have a different style compared to Fig. 5. The colours of the figure are not clear. Although I recognised it is difficult to represent road type and risk, this picture is not effective at all since the colours do not convey the perception of risk (trivially: red for high risk, green for low risk) and in general are difficult to be seen
L431: specify the type of road, I assumed A3 is a highway but the specification is needed for including all the readers
L467-471: “resulted in comparable estimates” – I expect more justification, because it seem not to me
L472-479: I don’t understand this information here. If it is really informing your validation, I would create a section dedicated to validation where authors refer to the reference case(s) and this USA study (and others perhaps)
L485, 486 and 490: references miss year
L490: “perceive” – not sure academic results are about perceptions
L539-540: the reference case is local scale, isn’t it? So perhaps more cases and more road types are needed? Or validation that range from local (as the reference case) to urban to regional?