Dear authors,

thanks for your responses to the suggestions and comments from the reviewers. I welcome you to send your revised manuscript after considering all comments. Please pay special attention to the suggestions with regards to discussions part, trying to be as complete and self-critical as possible.

Thanks,

Ann van Griensven

Dear authors,

thanks for the revised paper and the efforts you have done to consider all remarks raised by the reviewers. However, the most important element that came forward was related to the weights assigned in the multi-objective optimisation. The issue has been taken seriously by giving more explanations in the paper and also the report. However, these explanations are not convincing. Let me elaborate.

Weights are assigned in multi criteria analysis for various reasons. This can be done to give more weight on variables that are more important, that may have better or more data or so. These are 'subjective', defined by the modeller and are fine if well justified/described.

But weights may have a more fundamental role: to normalise against 'strenghts' of the signals. Eg if very different signals are combined, they may have very different units or order of magnitudes. The weights should balance out these differences in units.

I do not see this taking place in your study. Or, in other words, the importances of the different variables in the multi-objective function depend on the units. If, for one reason, you would change the units of one or more variable, you will also obtain a different result (you will not get different weights, but the overall strength/importance of one variable in the full optimisation may change).

I would accept the reasoning if the function would be the summation of wi*NRMSE but not for wi*RMSE, as this equation does not accounts for the 'units' or 'order of magnitude' of RMSE.

Please let me know if I misinterpreted your paper.