The authors are acknowledged for the answers to the two referees and are encouraged to submit a revised version of their manuscript. Please, specifically, consider the major remarks of the referees, namely:
- include omitted results to show if the method is actually robust;
- provide more evidence that “several climate mode indices and combinations of indices for ensemble member selection and conditioning of the subsampler were evaluated” as mention in Section 3.2 by the authors;
- clarify how parameters were tuned or determined, indicating when cross validation was performed or not (and explaining the reasons behind the choices made);
- provide details on the methodology, as stressed by Referee 2;
- provide enhanced figures for the results, such as these can be better evaluated.
If these requests are not (or cannot be) satisfied, please explain why in details.

Dear authors,

I would like to thank you for the efforts put into taking into account the suggestions of the reviewers. I believe your manuscript has significantly improved. The two reviewers agree that it can now be considered for publication in HESS. I also share their evaluation.

I would like to suggest you to consider the final minor remarks/technical corrections provided by one of the reviewers (see below). I also have some minor remarks from my own reading of the revised version, which I am listing here below too. I think these final suggestions will also positively contribute to your final paper.

Best regards,
Maria-Helena Ramos.
======================
My remarks:
- Page 2, line32-33: please, consider checking the expression “an improved specificity of the ensemble forecast”. It is not fully clear to me what you mean by that.
- Page 4, line 10: “loosely based on…” Can you be more precise about the differences between the original proposal and your adaptation of the method referenced? As it is, the sentence is very vague and not much informative for the reader.
- Page 4, lines 15-16: please, check the use of the English language in this part of the sentence: “... represent realistic and equally likely representations of…”
- Page 4, line 30-31: If I understand well, the year of reference (which is the year associated with the time of forecast) is used in the re-sample. This would however not be feasible operationally (as one would not have the data available for the time steps after the time of forecast). I am however not sure I understood it correctly because on page 8, line 24, you mention that “The year of hindcast was excluded…”. Could you please check these two sentences and make then clearer for the reader?
- Page 7, line 3: when you mention CHPS, are you making reference to the CHPS used by NOAA too? http://www.nws.noaa.gov/ohd/hrl/chps/ The way it is mentioned in the paper, it makes the reader think that CHPS is a specific system developed at BPA. Maybe if you could add a reference here, the ambiguity would disappear.
- Page 7, line 8: can you explicitly indicate what the period (years) is used for calibration?
- Page 8, line 5: consider changing “in the next year” to “in the following year”
- Page 8, line 6: consider changing “in three sub-basins of interest” to “in the three sub-basins of this study”
- Page 8, line 10: considering the period 1871-2013, is it a stationary period? Aren’t there trends observed during this period? How does non-stationarity impact your study and method? I think that maybe a sentence or two about the dependence (or not) on the hypothesis of stationarity of historic time series used in the methodology should be added to the paper (for instance, in the “Discussion” section).
- Figure 3 and Figure 4: I guess these results pertain to a specific sub-basin of the study area. Is that right? In that case, could you indicate that in the caption? Also what is the difference between “ENSO-MEI difference”, “ENSO-MEI signal” and “MEI difference”? You use these in the figures. If they refer to the same thing, maybe it would be better to choose one unique term to refer to the same thing. Please check terminology (see also remark below)
- Captions of Figure 1, Figure 2, and Figure 5: you use three different terminologies for the same object: “test-sites”, “test-basins” and “test location”. Could you please check throughout the paper and make sure that you use the same terminology for the same objects that you make reference to? This makes it much easier for a reader to follow the reasoning and the results.
- Page 9, line 7 and Page 11, line 1: please, consider mentioning also in the text the sub-basin to which Figure 5 and Figure 6 refer to. This information should also be added in the caption of Figure 6.
- Figure 6: consider changing the caption to “Number of…”. Also, please, check for the use of a unique term here too: “ensemble members” or “ensemble traces”. It is better to choose one and stick to it all over the text.
- Page 11, line 19: check the use of punctuation (comma) here.
- Page 11-12: explain what you call “relative reduction in RMSE” (introduced on Page 12, line 7) already in the sub-section 3.3.
- Page 12, lines 6-8: check the use of the English language. I think some words (ex., “the”) are missing here.
- Page 13, line 9: are you sure it is “… attributed to statistical uncertainty of the skill score calculation”? How did you compute that? Otherwise, please moderate it with “probably” or “might be due to”.
- Page 15, line 2: I think you mean “Sect. 3.2” instead of “Sec. 3.1”. Please, check.
- Page 15, line 10: consider changing to “… in two of the three sub-basins …”
- Page 15, line 14: “… used a separate calibration post-processing parameters per sub-basin”. I do not understand this sentence. Please check to make it clearer.
- Page 15, line 21: consider changing to “…for very small sub-samples.”
- Page 15, line 32: consider changing to “…as input do not need to be updated. Finally,…”
- Page 16, lines 2-3: Is “water demand” stationary over historic periods? Can we use “historic data” to infer water demand of today? I think you could discuss some limitations also of using historic data for today’s inferences. Water demand may have changed with societal changes (for instance, with growing concerns towards water economy). Again, I think that “stationarity” issues and implications to the methods presented would deserve a few lines in the discussion section.
- Page 15-16: Discussion section: I miss a discussion on the issues of cross-validation, i.e., taking some years out from calibration and then using them for validation. Could you add a sentence or two in the discussion section about it? In your opinion, how it may play a role and be considered in further studies?
- Finally, I think it is strange to have a paper without a “Conclusion” section. Could you add a section with the main conclusions of the study for the sake of completeness of the paper and clarity of the message and highlights of your results? Maybe it could be derived from a re-organization of the “Discussion” section.

Reviewer#2 remarks:

Summary
The authors have done a nice work addressing the issues raised in the first round or reviews. I have a final set of suggestions that should be easy to address, and will put the manuscript in good shape for publication.

Minor comments
1. P2, L2-5: The papers by Tootle, Abudu and Sagarika should not be cited here, since they do not relate to GCM-based seasonal forecasting. Instead, these should be included in L15.

2. Section 2.2: The authors may agree or not on this, but after reading this section several times, I still find difficult to understand how the resampler technique works. In my opinion, a comprehensive diagram would help a lot to clarify the method, and therefore help other readers to reproduce it.

3. Figure 1: I encourage the authors to add latitude, longitude and a scale bar.

4. Table 1: Can the authors please add in the caption the period used to compute mean flow, precipitation and runoff ratio?

5. P13, L8-9: The authors state that “the marginal loss of skill for Libby is attributed to statistical uncertainty of the skill score calculation”. Can the authors clarify what does this mean? So far, no uncertainty (using bootstrapping, for instance) has been considered in the calculation of any score, so this statement seems misleading. Perhaps delete?

6. P15, L2: “as discussed in Sect. 3.1”. Should it be Sect. 3.2?

7. P16, L18: Can the authors please clarify what they mean with “introducing a random time shift”?

Suggested minor edits
8. P1 L12: “a number of… are” -> “a number of… is” (noun is singular, not plural).
9. P1 L17: “in the Pacific Northwest” -> “in the U.S. Pacific Northwest” (same in P5, L20).
10. P3 L9: “latter” -> “later”.
11. P3 L16: “In a pre-processing” -> “In a pre-processing step”.
12. P6 L4: Add a comma after “Oregon”.
13. P15 L24: “resampler” -> “resampled”.