Dear Authors,
both Referees (but also Mr Kesharvarz) submitted rather negative comments on the HESSD paper, mainly due, in my opinion, to the lack of clarity/carelessness of the presentation, starting from the stating of the main objectives of the work.

I do believe that it’s very important providing more and more information that may be used in regions where it is difficult to obtain reliable and long times series of meteo-hydrological data for calibrating rainfall-runoff models is crucial and proposing procedures tailored to improve the model implementation in such regions is definitely an important topic for our journal.
As all the referees highlight, and as you already acknowledge in your replies, the main objective should be made clear and the presentation should be structured along such line.

I would suggest to remove the analysis on the areal rainfall estimates: I believe that focussing on the first one of the key-messages you list in your reply to Ref#1 and Ref#2 (and that automatically implies addressing also the second one) is already an ambitious objective and more than enough for the paper.
In fact I am afraid that adding also information and a cursory analysis on the important issue of the rainfall spatial field (that alone would need a separate detailed analysis, probably focussing in detail also on the features of the typical rainfall events that are expected in this specific part of the world if you want to add something to the vast literature on the subject) would not help to improve the clarity of the work, and you already need to add a number of details and clarifications, especially on the proposed procedure and on the model implementation, as required by the referees.

The proposed approach for replacing the use of unreliable rating curves and how you use the Strickler equation is the core of the work and should definitely be explained much better. The description of the available data but especially of the calibration/validation procedure (very complex per se: three river sections, multiple objective functions, ..) also need to be thoroughly revised and integrated with a number of important clarifications also on the choices (necessarily subjective in some cases) that you made.

In your reply letters you have already identified the main changes that are needed for the work to be reconsidered for publication and I think you will be able to substantially improve your paper so that it may be more easily followed (the referees’ negative comments are certainly also prompted by a justifiable irritation due to the lack of many important steps and details) and to the point in explaining your work’s main novelties and strong points.

Best wishes,
Elena Toth

HESS Editor

Dear Authors,
I believe you indeed have made a relevant restructuring operation for improving the paper.
Both Referees have revised your second version and we have all appreciated your detailed reply letter, where you have adequately addressed many of the raised issues, but additional modifications are required for the manuscript to be publishable.
Referee #1 is still not convinced by your explanation of the novelty of the work: I do agree that you should definitely add a more comprehensive a state-of-the-art review of the literature presenting approaches that use water levels for calibrating rainfall-runoff models, explaining what differentiate your work from them, in order to justify its publication.
A description of the main features (advantages and disadvantages) of the proposed approach in comparison to the previous literature, may also help in improving the section (4.5) on the Limitations of the work, putting it in the international literature context. And I would suggest merging limitations and recommendations in a single section, dealing with both the limits and way forward.
I look forward to reading the revised manuscript.
Elena Toth

Dear Authors,
thank you for having added the revisions we asked for in the second round of comments.
I believe that the state-of-the-art on the use of water levels instead of discharge (that we asked for) may be more inclusive, for example a number of data-driven technique has considered water levels as model output, as may be seen in many works and reviews on the use of neural networks in hydrology.
I would also revise the new paragraph at page 20/21: there is no need to cite again (as already done, and with the same words, at p. 3-4) the remote-sensing papers, but you should directly focus on the differences from the previous works that are more similar to your approach (Jian and Seibert).
I leave to the Authors to re-verify all the paper (language included).
Many thanks for having submitted to HESS your interesting work, that I believe is substantially improved during the revision phase (thanks to the very constructive comments of two expert referees).
Best regards,
Elena Toth
HESS Editor