Decision on “a quantitative analysis to objectively appraise drought indicators and model drought impacts” by Bachmair et al.

Below are some comments on reviews and associated authors’ responses, as well as some additional comments.

- I believe the authors’ comments in response to Referee #1 are convincing and I agree with their decision to try and keep hydrological indicators in the manuscript. Additionally, I would suggest adding vertical lines in Figures 3-7 to (1) distinguish meteorological from hydrological indicators and (2) maybe also distinguish SPI from SPEI indicators (and Q from Y).

- Like referee #1, I would also appreciate to read a shorter version of section 5, even if all what is discussed there is of much interest. The challenge is therefore to make it more synthetic.

- Concerning the quality of data used, it would be worth mentioning the (relatively high in the two studied countries) underlying density of stations in the gridded E-OBS dataset.

- Authors’ comments in response to Referee #2 (W. Pozzi) appear also satisfactory.

- About the choice of NUTS1 region, it would be maybe useful to warn against a potential effect of the size range and type of regions (e.g. in Germany, BE vs BV).

- On the criticism about the reference to individual indicators, one way to make the tree approach clearer, I would suggest to explicitly show one of these decision trees as an example, possibly in the supplementary material or annexes. I may be useful for those readers not accustomed to random forest analyses.

- The authors may want to change some of the vocabulary used in the manuscript, namely the word “trigger” that led Referee #2 to comment on the relation of such a “trigger” to drought indicator threshold. I believe that the aim of the manuscript is not to define such actually drought response triggers, but to find drought indicator thresholds that would serve as proxys for these triggers.

- Additionally, and following Referee #1 (but also Referee #2), I would suggest making the conclusion section a bit stronger, maybe by incorporating some material from the lengthy discussion section.

As a conclusion, I am confident that taking the reviewers’ comments into account may clearly help in revising the manuscript for it subsequent publication in HESS.

Dear authors,

Thank you very much for submitting this much improved version of the manuscript. I can now accept it, following also recommendations from Referee #1 on this revised version and the very satisfying responses to referees’ comments.

I simply have a request for a few minor corrections that should in my opinion be addressed before publication:

1. There is still an inappropriate (following Referee #2 in the first-round review) “trigger” in the abstract L15

2. P7L11-12: Could you justify in a sentence the use of both near-natural and anthropogenically-influenced streamflow records? I point this out because this crucially may imply confusing recommendations for readers and users of streamflow records for evaluating drought impacts.

3. P8L21-22 and Figure 2: There is an inconsistency of figures for the seasonal distribution of the number of impacts in the UK. The number “other types” of events seem largely overestimated with respect to both the regional split and the total figure given in the text. Please check this.

4. P9L7: How was the month of December treated when censoring the data considering that for seasonal analysis, the winter for a given year includes the preceding December (P8L20)? Please specify this.

Best regards