The study presents a statistical analysis of the hydrological cycle in Czechia. To do so the study uses multiple gridded hydrological products, derived using remote sensing and reanalysis. First a ranking scheme regarding the performance of each product and their combination is presented. To me this is the main novelty of the study. Afterwards the best products are analyzed to provide spatially explicit estimates of the change of hydrological dynamics between  a past and a present era.

Overall, the methodology is mostly solid. My main methodological question concerns the use of GLEAM and GRUN. I am not sure why GLEAM and GRUN were used as benchmark datasets. GLEAM and GRUN are both model based. How can they be used as benchmarks for validation? They themselves carry a lot of uncertainty. For ET, the physical basis of some of the remaining datasets (e.g., ERA5 land) is much more detailed than GLEAM as they integrate a full complexity land surface scheme, rather than simplifying models (e.g., Priestley Taylor). GRUN has even less physical basis, as it is a statistical model. I would be more convinced with the analysis, if only real high-quality observations were included in benchmarking the various datasets.

Apart from that, a thorough analysis is presented, which to a large extent is consistent with previous results related to cintintental Europe. Even though the study is methodologically sound, its novelty is limited in my opinion because of (a) the data products used are all well established and have been extensively previously analyzed at regional and global scales, and (b) the limited geographical extent of the study. I find the paper better suited to journals focusing on regional studies, rather than HESS whose goal is to further advance the fundamental understanding of hydrological processes and their impacts on society and ecosystems.

A few minor comments:

Lines 17-20: Not clear what the contradiction is between the 2 statements

Line 26: define what you mean by unquantified uncertainties

Line 73: What is the meaning of the roof analogy?

Figure 1: the different shading is not clear. I suggest the authors to add in bold colors the catchment boundaries for clarity

Line 130: System instead of set of ODEs

I find the definitions of R2 and RMSE a bit redundant.

In eq 1, 2 I suggest changing the variable name of the residual term from R to something different, e.g. epsilon, to not confuse the reader as R is commonly used for runoff, and previously in the paper as the coefficient of determination

Line 211: Why were the authors surprised by the quality of ERA5-Land. Please explain further this statement? The land surface scheme of ERA5-Land (H-TESSEL) has a hydrological component, which is in compatible complexity with the remaining hydrological models of the study.

Line 225-228: Does this imply that the models do not close the water balance, or that the integration periods are not long enough, and the discrepancies are due to soil water storage dynamics?

Line 244: Change Abril to April

Figure 4: Might be better if presented as cumulative distribution functions, q-q plots or boxplots

Comments:

This article presents extensive work on comparing the performance of different datasets on the closure degree of the water budget and demonstrates the acceleration in the hydrological cycle over Czechia. Overall, the paper is well written and readable and provides direct evidence of the performance on evaluation from different datasets. However, the title of the article could probably be rephrased, as it looks like a new method for demonstrating water cycle acceleration, but the actual story of the article is more about comparing the performance of different datasets using a novel method. Here are several issues needed to be addressed or clarified, which are listed as follows.

Major comments:

1. Line 12: What does the median space pattern mean here? Why only mention spring and summer here?
2. Line 17-21: A more logical organization is needed, perhaps adding a sentence in front of “on the one hand” to introduce the relationship between the water cycle and water fluxes you have chosen here (precipitation, evapotranspiration…). The information behind "on the one hand" and "on the other hand" are not parallel associations, and these two aspects are less relevant to the focus of this article.
3. Line 36-43: The information in parentheses may be summarized in a supplementary table and moved the table to supplementary materials for detailed clarification. In addition, please add the datasets categories (which ones belong to satellites or ground-based measurements, or climate models) in the table.
4. Table 1: Add the datasets categories (which ones belong to satellites or ground-based measurements or climate models) in table 1.
5. Line 173: Are there any supporting references to this similar approach? If yes, please provide the citations.
6. Line 180: It is okay to use the medians for excluding the outliers, but can you provide a supported plot to show the distribution of values as supplementary material?
7. Line 196-197: The demonstration is on the edge, as it is not all time is overestimated and underestimated, only in some certain period.
8. Figure 4: Can you use the line plot to show the trend as this is a time series for changes in hydrological variables, while a histogram may not be very straightforward?
9. Figure 7-9: When you discuss the spatial distributions in different parts of Czechia maybe just focus on the one figure which is most representative as I see the spatial patterns are similar across Figure 7-9 and moved the rest figures to supplementary materials.

Minor comments:

1. Line 80: Nine datasets? But in Table 1 there are ten datasets, right?
2. Figure 5: Is it possible to zoom in on the y-axis limit because the boxes in the second and third rows are not clear?

Water cycle acceleration in Czechia: a water budget approach

This paper presents analyses of the water budget and water cycle for Czechia. Overall I found that the work is interesting and well written. My major comment is around the definition of the score used for the ranking of the different data set combinations and how this was derived and justified. For example the score only accounts for the anomalies and the correlations but does not consider bias in the products. This is very evident from Figure 4 where ERA5-land has substantially higher estimates of both P and ET and therefore its anomalies are similar to the other products. But presumably in some applications consistent biases may be problematic even if the anomalies are ok (e.g. water allocations or environmental flows). I think that the authors need to do far more to consider the sensitivity of the dataset ranking to the definition of the score.

Minor comments:

Figure 1: shading is difficult to interpret and I think it would be easier to use hatching or just label the rivers

Line 163: Would be interesting to do the analyses for the three main drainage basins. 

Line 170: would be good to explicitly note that you are doing the closure each year here and then Ri is the average of Rj for j in 1:60

Equation 2: this isn't actually the ranking but a score that is then used for ranking so I think all the text associated with the equation needs to be updated.

Figure 3 - we can't see most of the distributions. I don't think this is a useful presentation of the data. What are the units for the budget residual?

Figure 5 - wrong colours mentioned in caption. I am surprised by the results shown in figure 5 as there is less difference between the different models than implied by Figure 4 where ERA5 is substantially wetter and higher ET. I think you could dig further into this.