The authors have not responded to the comments of the second, more critical, reviewer. Obviously, these reviewer comments need to be carefully addressed. Unfortunately several points were unclear in the original submission, as indicated by the reviewers’ questions. These points need to be clarified to allow a full assessment of the manuscript.

As also mentioned by the reviewers, the language needs to be improved. Partly the language is not exact enough (as a small example, P3L7, not the models are downscaled but the variables which are simulated by the models).

Furthermore, any discussion of uncertainties is missing. Besides the usual uncertainties, there also is the issue of using a model beyond the conditions used for model calibration. These uncertainties cannot be ignored! (see WRR paper by Pappenberger and Beven, 2006, Ignorance is bliss: Or seven reasons not to use uncertainty analysis)

Thanks for your efforts with the revision of the manuscript, which has greatly improved. As described nicely and in more detail by reviewer #3, there are still a number of issues which need to be addressed carefully. In particular, you need to better describe the M5 model (or models?) in a way that is understandable for the (hydrological) audience. As the reference (Quinlan, 1992) is hard to access this description is really crucial. I would also like to see some better motivation on why this particular (type of) model was chosen. What are the advantages (disadvantages) compared to conceptual/bucket-type models (like SWAT, PRMS or HBV), which are otherwise often used in similar applications.

Please also get help from a native speaker, some terms just make little sense (future running flows, rear edge, ...), this is very confusing.


Best regards,
Jan Seibert