This is a really nice paper but needs a few issues of clarification in both the introduction and in the presentation of the methodology before publication in line with the comments in the MSS.

The manuscript proposes an innovative metric (named RUMI) to support the calibration of hydrological models, based on a combination of two approaches: BLUECAT and Mutual Information (MI). The authors evaluate the performance of the proposed metric through an extensive case study analysis, comparing RUMI to an established metric, Kling-Gupta Efficiency (KGE).

The topic is particularly interesting, and the proposed approach is promising due to its potential for providing more effective calibration and incorporating uncertainty evaluation through the BLUECAT component.

While I am inclined to recommend publication, I have several comments and suggestions for the authors, detailed below.

General Comments

The main issue to address concerns the paper’s structure and organization.

Although the abstract clearly conveys the manuscript’s aim and content, the subsequent sections may leave the reader disoriented.

Section-Specific Comments

Introduction:

The Introduction should be revised to focus more on the calibration problem, highlighting available metrics and their limitations. The new proposed metric should then be introduced, emphasizing its innovative aspects and added value. The two components, BLUECAT and Mutual Information, could be briefly mentioned at the end.

Section 2.1:

The information on where the GR4J model is available could be moved to the Appendix, as it is not directly relevant to the manuscript. Instead, it would be more helpful to provide additional details about the model itself. Currently, we only know that it has four parameters and two storage modules. The calibration strategy is omitted and should be described, particularly how the two metrics (RUMI and KGE) are applied in the calibration process. It is also essential to confirm that the model is suitable for daily-scale applications and all spatial scales. For example, the reader might question whether the GR4J model performs well at a daily scale for a 35 km² watershed, which could undermine the relevance of the evaluation and comparison.

Section 2.3:

The section title is not informative. A short introduction should be added to remind readers that RUMI is a combination of BLUECAT and MI.

Line 138: The text refers to a flowchart that is missing. Please include it or clarify the reference.

Section 2.4:

Both the section title and its content should be revised for clarity. This section contains mixed information on data (which might be better placed in Section 2.2), the comparison metric KGE, and the evaluation methods. These topics should be better organized and clearly separated.

Section 3 (Results):

This section could be better organized by providing a more detailed explanation of the plots. Consider using more informative and communicative visualizations.

Lines 220–230: These lines are somewhat confusing. It might be more effective to first show the validation dataset and then focus on two specific years.

Table 2: Is this table necessary? Perhaps violin plots in Figure 4 would suffice. If kept, the table should include comments on specific values (e.g., 1755 and other notable values).

Table 3: The meaning of the values presented is unclear and needs further clarification.

Figure 5: The phrase "only for illustration purposes" in the caption is confusing. Consider keeping only subplots c.1, c.2, d.1, and d.2, as they show a clear performance difference between the two methods.

Section 4

This section needs to be improved for clarity and coherence. Emphasize the significance of the results and their practical implications.

Section 5 :

The Conclusions should be rewritten. The first sentence belongs at the end, as a concluding remark. The section should summarize key findings and highlight future research directions.

Additional Suggestions

To thoroughly evaluate the proposed RUMI, consider conducting a virtual test in which parameters are assigned to the model and then calibrated using both metrics (RUMI and KGE). The comparison would then focus on the parameter estimates rather than the streamflow time series. This is a common approach that helps isolate model-independent performance and avoids limiting the conclusions to the GR4J model.