The research aims to understand and improve the important challenge of communicating three-dimensional flood map uncertainty to various end-users through a series of qualitative surveys. The manuscript is well written and structured, there are several wordy tables that could be presented differently (see suggestions below). The probabilistic visualisation prototypes presented represent a significant step-forward in terms of communicating uncertainty to forecast end-users. The work would benefit from more emphasis on which uncertainties are being represented in the visualisations and how the prior knowledge of the surveyed participants is assessed and how this impacts their opinions and the conclusions drawn from the results. The research questions proposed in the introduction are reasonable, they should be re-addressed again in conclusion. Consideration of the limitations of the survey approach and applications of this approach outside of Quebec would enhance the manuscript. Once these concerns are addressed, I feel that the article would make a valuable contribution to HESS.   

More specifically:

1. Where do the uncertainties originate from? Are they based on uncertainties in the precipitation inputs to the hydrological model? Or are they uncertainties relating to model parameters/antecedent conditions/underlying data used to determine the flood maps such as the DTM? Or are they compound and include all the above? The paper would benefit from some discussion of these aspects of uncertainty in relation to the flood forecasting system used.
2. What determines that this is a large-scale survey? How does it compare to previous similar surveys in Canada or elsewhere?
3. How is the ‘limiting the confusion of decision makers’ (abstract L20) of end-users measured/known?
4. Section 4.1.1 What was the prior experience of the participant groups at using and interpreting flood maps (probabilistic or otherwise). This seems to be critically linked to the users’ preferences.
5. How were the visualisation prototypes developed, and by whom?
6. Tables 2, 3 and 4 could be presented graphically to enable readers to visualise results and aid comparison. Tables 5, 6, 7, 8, 9 and 10 should be ordered/sectioned by participant group to improve readability.
7. Are the survey findings applicable in other places/countries or should this type of survey be repeated elsewhere? Adding recommendations would be beneficial to readers.
8. What are the limitations of this interview style survey approach? Could a quantitative survey be used to draw more specific conclusions such as linking prior experience/understanding to visualisation preferences? Also, how can the probabilistic forecasts be linked to impacts and with users’ actions. The next step to this would be to link the likelihood of impact (or flow scenario from prototype 2) with appropriate actions. These points could be developed further in the discussion/conclusions.
9. Please see supplement for minor comments.

This study investigates the communication of probabilistic hydrological forecasts with different types of users based on phone survey and qualitative elaboration. They show some interesting findings, for example, users’ responses to uncertainty of forecasting results, similarities and differences in visualization preferences of different users, their curiosity in hydrological forecasting methods and so on. This study also shows us a blueprint of forecasting visualization schemes from a holistic view of water depth, inundation area, discharge and the uncertainty according to wide suggestions from the users’ end. The paper is generally well-organized and the structure is clear. Such study can improve hydrological early warning systems, thus, benefit flood risk management.

However, I have several major concerns that expect to authors to address:

1. The innovation of this study needs to be further addressed (i.e., things that has not been done by previous study). In the introduction, the authors fully reviewed previous investigates on the communication of flood risks and highlight the importance of survey on probabilistic forecasts. However, the difference from or increment to previous studies is not clearly pointed out. For example, previous studies may only investigate communication of deterministic forecasts or 1-D/2-D hydrological forecasts instead of inundation map, etc. Besides, this study only survey people living in south Québec, where floods are mainly caused by snow melt. However, the situation may be different for other regions and countries. It remains known to what degree the conclusion drawn from this study can be transferred to and referenced by other places of Canada and the world.
2. Survey should strictly take sample representativeness into account. The education background, gender and age of the participants and their living/working places may affect the results and the representativeness of samples. Thus, it will be essential to include statistics of these kind of information. For instance, a geographic distribution of the participants with flood risk map, proportion of people with/without hydrology or atmospheric education background, etc.
3. I also notice that the authors design different contents of phone survey for farmers and citizens from non-farmers or citizens (i.e., drop “the themes related to the nature of the information” for farmers and citizens) but did not explain the reason for doing this too much. I think the different treatment may cause the readers wondering whether the forecast maps should originally be designed differently for these two kinds of users (i.e., farmers and citizens & non-farmer or citizens). Since satisfying all kind of users with a single forecast map seems to be impossible. Therefore, why did not the authors design different kind of forecast maps for them at first and then do the survey?
4. The presentation is overall a bit too qualitative. Some quantitative descriptions and statistic plots are needed. For example, in L341-349, the authors can show the voting proportion of color scheme preferences with real numbers or a table or histogram. Table 7 offers too much unsorted information and words. Table 8-11 is the same without statistics and graph visualization.

Minor comments:

1. The structure of the abstract need to improve. The background occupies almost half of the abstract, leaving little space for results and main conclusions. The conclusion is the only one sentence with “several” statement (L19-20). And the significance of the study needs to be further stressed.
2. Figure 1: The legend of the blue polygons and lines is needed. Also, please add coordinates for the map.
3. As mentioned in L112, the investigation of color scale is one of the objectives of this study, however, there is no echo in the discussion or conclusion section.
4. In the abstract and Figure 2, the number of the citizens and farmers are 37 in total, however, in Section 3.1.4, the author said 33 citizens plus 5 farmers. The numbers contradict. Please check. Besides, in L201, the number 11 is confusing.
5. In Section 3.2, the authors said “except for citizens and farmers, one-to-one interviews are taken for the participants”. However, in Figure 2, the interview number and respondents differ, which is confusing.
6. The author should double check the upper and lower case of titles in the references. For example, the fifth and last reference in Page 29 use upper-case for the title, while others did not. The same problems can be found in Page 30.