Dear Authors, thank you for your comprehensive replies to the 3 reviewers. I find the paper very interesting and so do all the reviewers. I think we've had a great exchange here and all the reviewers have clearly taken the time to read your manuscript thoroughly. They also, in general, are very favourable to the paper all suggesting minor comments. I don't think I have ever edited a paper that is 'minor' and yet with so many questions/clarifications from the reviewers, due diligence has definitely been shown! So I look forward to the revised manuscript, which I am sure will improve with the check and balances the reviewers have noted and your suggested replies. Then at that point I will make my final decision and do some quick checking of the revisions!, best wishes, Jim

Dear Authors,
Thank you for your changes to the manuscript and responses to 3 comprehensive assessments by the reviewers, all who were positive to your paper. I have no problem with the fact that these minor corrections have been well tackled and you have developed considerable changes to the manuscript, your due diligence is appreciated.
One point though, that I was surprised did not come out in the reviews, but I would like personally to tackle you on. Namely that there seems a general disconnect in the paper that I think needs to be better discussed. The issue, as I see it, is as follows (and I know some of these matters you could note for a number of other hydrological modelling papers):
1) KGE appears to be your benchmark for the basis of the forecast model. Clearly KGE has certain properties of evaluation to streamflow. You do not anywhere in the paper state why models based on maximizing KGE are valuable for your forecasting objectives - in fact this whole matter (objectives over experimental design) is hardly discussed anywhere. A model with different calibration statistics will clearly be different and thus be different as the underlying vehicle for the forecast analysis. Please can this be discussed in a meaningful way, so why these choices?
2) Then we move to the statistics on the output where you have table 2 showing a whole bunch of hydrological signatures used to explore a 15 dimensional space of forecast skill. Again there is a big disconnect here to any intelligent discussion of the objectives of the forecasts and why these signatures, and these signatures alone, have value. Indeed if at all why they should be considered (as it seems they are) equally weighted in the forecast assessments (do you even have an appreciation of their individual sensitivity to how good or bad the forecasts are?). Again I really think it should be explained to the reader why this table and why these metrics and do they really all have value to the types of the forecasts you want?
I am often frustrated at the 'many hydrological signatures' means a good jib has been one but without justification for their use. So we should justify this better, to explain the experimental designs we use. Again a different set of metrics or more or less of these signatures would generate potentially different results, and who is to say what best highlights 'what you need' for a given forecasting objective... And as I stated before why indeed a more comprehensive treatment of the model hindcast skill using various calibration metrics to test different periods and magnitudes of flow, were not used?
Finally I also want to clarify your use of a scale for KGE in Figure 1. Can you explain why you have set the scale to zero? KGE does not have the same properties as NS Efficiency where 0 is the mean predictor, in fact the mean predictor is -0.41 Knoben, W. J. M., J. E. Freer, and R. A. Woods (2019), Technical note: Inherent benchmark or not? Comparing Nash-Sutcliffe and Kling-Gupta efficiency scores, Hydrology and Earth System Sciences, 23(10), 4323-4331. for an explanation of this (please note I am not at all trying to get you to cite our paper). I just think you want to explain better why your figure has chosen certain limits.
Anyhow I look forward to your responses and I hope changes in the above. I will simply review this as an editor, but I think it is worth taking some time on these matters to improve the justification of your choices and experimental design, best wishes, Jim

Whilst the authors have not really fully engaged with the points I raised in terms of their experimental design I appreciate that these matters are difficult to resolve and they were not directly brought up by the reviewers, therefore I am happy to move the paper to publication.....