|The authors have done a good job to address most of the comments. I still have a few outstanding issues that could be resolved.|
1) The authors stated in their responses that this paper was not intended to be a comprehensive review of hazards for NYC, but rather presents a new and interesting methodology for DRR. My problem with this response is that at a large portion of the revised manuscript text is focused on developing a new hazard assessment of NYC using novel techniques (Sections 1.3, 2.2, and 3.1, Figures 1 - 3) which is in conflict with the responses provided. If this hazard assessment merely a placeholder that you will refine in a future paper, then why is it worth publishing here? I think if you want the reader to understand your research aim is on the methodology and the hazard assessment is really just a placeholder for a more complete assessment in a future papers, then all of this cataloging of hazards belongs in a supplemental material. In the main manuscript you can then say something like: “We provide a cursory hazard assessment based on the datasets A, B, C, etc. in order to demonstrate our methodology. Multihazard estimates are provided in the supplemntary material.” Alternately, if the paper is focused on novel ways to assess hazard, then the authors could do more to address the shortcomings of the hazard assessment.
2) Line 371: Your response says that you do not suggest a uniform increase, but the revised text still does not offer enough information to make this clear. I’m not sure about the relevance of changes in annual rainfall totals to the multi-hazards that you’re discussing, and I think it’s a bit misleading. First, I would choose a temporal scale for precipitation analysis that is relevant to flooding (daily intensity or shorter). Second, since your study specifically focuses on mechanisms, you should disaggregate this into the rainfall mechanisms of interest. For example:
“Recent research using historical data has suggested that for the NYC region, daily precipitation extremes are increasing in the fall with less change in all other seasons (Frei et al 2015; Huang et al 2017; 2018), however, projections from downscaled GCMs provide less clear evidence for a shift in the intensity of flood causing rain events in the Northeast US (Knighton et al 2017; Schoof & Robeson, 2016).”
You do not need to use this text or these references, but saying something like this would help the reader to make the connection between climate change and the hazards you are discussing.
Frei, A., Kunkel, K. E., & Matonse, A. (2015). The seasonal nature of extreme hydrological events in the northeastern United States. Journal of Hydrometeorology, 16(5), 2065-2085.
Huang, H., Winter, J. M., & Osterberg, E. C. (2018). Mechanisms of Abrupt Extreme Precipitation Change Over the Northeastern United States. Journal of Geophysical Research: Atmospheres.
Huang, H., Winter, J. M., Osterberg, E. C., Horton, R. M., & Beckage, B. (2017). Total and extreme precipitation changes over the Northeastern United States. Journal of Hydrometeorology, 18(6), 1783-1798.
Knighton, J., Steinschneider, S., & Walter, M. T. (2017). A Vulnerability‐Based, Bottom‐up Assessment of Future Riverine Flood Risk Using a Modified Peaks‐Over‐Threshold Approach and a Physically Based Hydrologic Model. Water Resources Research, 53(12), 10043-10064.
Schoof, J. T., & Robeson, S. M. (2016). Projecting changes in regional temperature and precipitation extremes in the United States. Weather and climate extremes, 11, 28-40.
3) I believe my comment on the decision analysis literature was perhaps not entirely clear. A large component of this research is social, in that these frameworks are fundamentally based on the mapping of hazards to risks as a way to reduce the demand placed on the social component of DRR. This is fundamentally the focus of your paper, just from a new point of view. Citing them is certainly not a requirement for publication. I just think the authors might find some relevance in these papers to the work they’ve presented. I leave it to the authors to decide how to handle this. I only bring it up again because the authors response suggested that I left an ambiguous comment.
4) With respect to the representation of uncertainty in the weights I think my initial comment wasn’t entirely clear. As the authors state, experts were given 100 points to allocate between indicators. If I’m understanding this correctly, each indicator has a population of n allocated points, one from each person surveyed, which the authors then average. If you can calculate an average, you also can calculate some measure of the population variance. Simply presenting the standard deviations of the allocated points for each indicator would given readers an idea about how much agreement (or not) there was among the group surveyed. I believe this is a critical piece of information. If there is low variability it says the the social network of experts was highly connected and held similar perceptions of risk. If there is a large variance among points assigned, then it perhaps suggests divergent perceptions of vulnerability and difficulty in lending a broader legitimacy to this approach. If this data was not collected in a way that would allow calculation of population variance, then I think this is a major issue.