This manuscript presents a novel probabilistic water balance data fusion approach for calibrating multi-scale hydrological datasets. The methodology is innovative, addressing the challenge of reducing uncertainties in datasets by integrating them through water balance constraints. The approach provides a framework for both basin-scale and pixel-scale applications. The application to the Hindon River Basin demonstrates practical utility, with reasonable error estimates and clear improvements in data consistency. The paper is well-written, structured, and accessible, making a substantial contribution to water resource management and hydrological modeling. However, some areas, such as the clarity of methodological details and validation against independent data, could be strengthened to enhance the robustness and reproducibility of the findings. Suggestions are as follows:

In Section 2, beginning on line 117, you describe the Hindon Basin and the separation of two irrigation seasons (Kharif and Rabi), yet it is unclear if the rotated crops use the same land or if they are in adjacent regions. It would be helpful to add a sentence or two clarifying this. 

In your results, the validation could be strengthened. Are you able to compare your estimates against any in-situ records? Reported standard errors are useful, but which component dominates the uncertainty (precip, evaporation, storage, discharge, canal imports)?  Standard errors are provided but there is no discussion of comparisons with independent ground-truth data or other datasets not used in calibration. Including such validation would enhance confidence in the results.

Discussion would benefit from a short explanation on generalization. For example, can this approach work in snow dominated or urban catchments or is it basin specific?

Figures with more than one panel (starting with Figure 2) need tags (a, b, c, etc) and the caption should refer to each panel specifically for clarity (like you did for Figure 4). In Table 2, indicate the meaning of the underlined values. Table 3, Table 4, again indicate the bold and underline importance. 

This work proposes a two-step method, i.e., from basin scale to grid scale, to produce a water budget closed datasets by introducing the Bayesian model with predefined prior distribution and posterior parameter estimation, considering the covariance between water components and the entire time series, under a specific case study in the severely irrigated Hindon River Basin, India. This work tried to introduce an innovative theoretical basis and apply it to a basin with abundant discharge records. Although the logical structure of the work is clear, the theoretical introduction and the equations are hard to follow since there is no deductive process of the equations provided and the code and data are not accessible with the provided links. This makes it hard for the readers to follow the work and estimate the robustness of the work.

Major:

The spatial scale problem is not thoroughly discussed. How were the different scales between water components in the budget closure equation handled? Is the resolution 50 km really feasible in such a small basin that is only one pixel wide and two pixel height?

L173/L884-885. About the gap-filling method, the authors made the assumption that the canal operations do not vary widely between years in which the conditions are similar. Is it possible to use the existing data to validate the assumption? I mean compare the data in the years with similar conditions to check whether that assumption is tenable.

It will be much clearer and easier to follow if a framework diagram is provided in the method section.

About the matrix variables mentioned in all equations, for example, in Eq. 5, it is better to provide the size parameters of each matrix.

For me, the relationships between equations are quite independent and the connections are weak. For example, Eq. 3-8, the input and output of each equation are vague thus hard to understand the method itself as a whole. The same issue exists for the entire theoretical part.

L545. The labels and tick marks of x and y missed in Fig. 8.

Minor:

L125 the abbreviation of Central Water Commission (CWC) should be explained near the figure instead later in L168.

L241. the symbols mpt and vpt with Eq.7-8 are different.

L255 I think “Evapotranspiration” is better than “Evaporation” throughout the paper.

L581. There is no ground-water level data? It is weird that the discharge of the canals have been paid great attention while no ground-water data available in such a heavily ground-water-based irrigated basin.