|The authors did a wonderful job in addressing most of the comments in my previous review, and the manuscript is now much closer to publication. There’re just a few questions from my previous review that remain unanswered (or only partially answered). These questions are copied below with some notes added in the hope that they better explain the reviewer’s request. I wish the authors will take another look at these questions and provide full/direct answers to each (please use the examples provided in my question where appropriate to help with the explanation), and incorporate the answers to appropriate sections of the paper. In some cases, the answers may seem too obvious to the authors, but I think they will help the readers understand the paper better, especially those outside the authors’ field of expertise.|
Questions from my previous review that remain unanswered (or only partially answered):
Near line 20: “number of compound events”: the numbers are not fully meaningful without first defining what an “event” is, i.e., is an event counted as a day, a 3-day period, or a continuous period >=3 days?
(Note: Examples will always help. Suppose the event thresholds are met for all days from January 1st to 3rd, i.e., each of the 3-day window centered on Jan 1, Jan 2, and Jan 3 satisfies the event threshold), is this period counted as one event, or 3 events?)
Near line 20: “within +-1 days of the event”: Now I sort of understand what “centred” meant in the earlier sentence. In the example I gave above, does it mean the resulting value of a+b+c is assigned to day 2, and the 3-day period centered on day-2 is considered AR-related if an AR occurred on one or more days of day 1, 2, or 3? Please use the answer to make clarifications in the data section in terms of how a CE is defined, how an “event” is counted (e.g., if a CE lasted 6 continuous days, is it counted as one event, 2 events, or 6 events?), when a CE is considered to be AR-related or not AR-related, what “day of event” means, etc. Without clear and unambiguous definitions of terms, the statistics presented are hard to make sense of.
(Note: Please directly address each point in the above comment if possible, and use examples where they might help the explanation. Among these, the two key questions are how the events are counted, and when a CE (defined using a 3-day window but assigned to the middle day) and an AR (which presumably is based on 6-hourly time steps) are considered to be associated with each other. I understand some of the answers would seem too obvious to the authors, but the hope here is that the description should be unambiguous enough for an ordinary reader to understand the statistics and/or re-produce the analysis procedures.)
Table 1 and where applicable in the text: “on day of event”, “one day before or after event”: given that the precipitation is a 3-day total, and CEs are defined using a 3-day window, descriptions like these are quite ambiguous. For example, if a CE occurred during the period of January 1-3, then common sense is that “one day before event” is December 31, and “one day after event” is “January 4”. But that doesn’t seem to be what the authors intended in indicate here. Again, an unambiguous definition of terms is needed to avoid potential confusions of this kind, as also suggested earlier.
(Note: The authors did address this comment. But now that I understand better how an event is defined, a new question arises: how is “one day before event” and “one day after event” dealt with in cases where multiple events occur consecutively? Let’s use the earlier example again, i.e., assume the event thresholds are met for all days from January 1st to 3rd, i.e., each of the 3-day window centered on Jan 1, Jan 2, and Jan 3 satisfies the event threshold. Then, from the perspective of the Jan 2 event, Jan 1 and Jan 3 would be considered “one day before event” and “one day after event”. But that leads to a paradox, because both Jan 1 and Jan 3 are themselves “day of event”. Similarly, although Jan 2 is a “day of event”, it is also “one day before event” because it precedes the Jan 3 event, and is also “one day after event” because it follows the Jan 1 event. In other words, all of your statistics for “day of event”, “day before event”, and “day after event” are, inadvertently, a mixture of “day of event”, “day before event”, and “day after event”. This issue complicates the physical interpretation of the statistics and should be remedied/mitigated.