The authors investigated terrestrial water storage variability in the South Saskatchewan River Basin (SSRB) for the period 2002-2019 with use of GRACE/GRACE-FO data and the HydroGeoSphere (HGS) groundwater-surface water model. The results of the HGS model were used to decompose the terrestrial water storage anomalies (TWSA) derived from GRACE/GRACE-FO data into surface water, soil moisture, and groundwater. For each of the results, the time-frequency of the intra-annual water storage fluctuation was determined, and the impacts of meteorological and oceanic drivers were examined.

Dear authors, 



It is an interesting study to read, and the description of the data is detailed and well-explained. In my opinion, the article could benefit from a more explicit or separate description of the methods and more elaboration on the limitations or challenges of the study. Below are my comments.

Specific comments:

Parts of the methodology were described in the Results and Discussion section. For example, in line 205, 206 and line 257-262. Having this placed in a separate methods section might make it easier to grasp what was done in which way and why. I recommend considering a separate methods section in addition to the description of the data.

I am missing the limitations and challenges in your chosen approach. For example, in line 105-108, you share about the advantage of CGS compared to mascon solutions. However, it was not mentioned how this (does not) show in your results, nor how it influences your outcome. It would be interesting to have these included. Could you elaborate more on these?

Section 2.1 describes how is dealt with the gaps in the GRACE/GRACE-FO data records in between the two missions for the JPL mascon data. How was this handled with the CGS method? This was not clear to me. Also, figure 3 shows a gap in the data, but figure 4 has a linear interpolation for the TWSA.

Figure 2(b): Please add a scale, north arrow and legend in the sub-figure.

Line 187-190: NAO is mentioned to be included in the analysis, but the only mention of it is in section 2.4. Is it included in the analysis? If so, where? If not, consider removing this part.

Line 202: The HGS TWSA values are compared to GRACE-derived values. In figure 3, it also shows a difference between the two GRACE-derived values (JPL and CGS). Could you specify which of the two it is compared to, or if it is to a combination of both, how this is done?

Figure 7 is interesting and nice. It would be clearer if the different subfigures had a title or something else clearly indicating which data they are showing (e.g., HGS TWSA, HGS surface water).

Line 287: Using word precipitation for the sum of rainfall + snowmelt might be slightly confusing for the reader. Was there a clear reason to define it as such? If not, perhaps the writers could consider using a different phrasing, such as “water input”.

Line 290 and Line 295 mention two different numbers: 90% confidence bounds and 95% confidence intervals. Which one is the correct one? Or if both are, could you clarify this in more detail to avoid confusion?

Overall, the research is interesting and uses sophisticated modelling and data derivation techniques. However, the manuscript is a little confusing because more information needs included in the background to understand the research justification, all the methods, results, and discussion are mixed together, and the research uses only modelled data as comparison (e.g. no in situ data for validation). Therefore, I am recommending major revisions, which includes the addition of a several paragraphs in the introduction, a “Study Area Section”, an “Analysis” section in the methods, and a “limitations” section before the conclusion, in addition to other recommendations, which you can see below.

Line 39: An additional paragraph is needed here to explain water storage variability. This should explain the conceptual background of the variables used in the study (e.g. surface water, soil moisture, groundwater, etc.) and how they interact and how research shows they are likely to change under climate change conditions. *Note – this is different than paragraph 1, which basically introduces climate change. This second paragraph needs to justify why this research is important and highlight the knowledge gap. Explain to your reader why it is important to study water storage variability and what research has previously been conducted on it and how your research is filling that gap. Then end with an objective sentence that highlights the importance of your research goal.

Another new paragraph: There also needs to be some research and justification for the time-variable frequency. Explain in this paragraph: Why is it important to study the intra- and multi-year cyclicity of these variables? What does that tell us? What has previous research shown in terms of hydrologic cycling? What does research say regarding how climate change might impact these cycles? What is the knowledge gap? Why is your research studying these time-variable frequencies novel? (e.g. this should be in the paper – not simply an explanation for me, the reviewer).

Line 50: Figure 1: I would recommend moving the study area map to a new section called “Study Area”, which goes after “introduction”, roughly on line 83. In this new section, please detail the extent, hydrological and climatological averages, the forest/land use type, the population centers, the river systems, the name of the provinces, the mountain ranges, elevations, temperature and precipitation changes with elevation and latitude, and anything else relevant for the reader. Currently, the map includes cities, roadways, rivers, and provincial borders – but you don’t explain those to the reader and anyone unfamiliar with the location cannot geolocate the study.

Line 66: You did a great job explaining studies that have used GRACE. I recommend adding 1-3 sentences at line 66 justifying why GRACE is appropriate for your study.

Line 70: You need a paragraph or several sentences explaining the HydroGeoSphere model, previous research that has implemented it, and why it is the most effective model to answer your research question.

Section 2.1: This is a lot of detail about how GRACE works. Please provide a paragraph at the beginning of the section explaining the variables that you extracted from it, the spatial resolution, the temporal resolution, and the time/dates of the data.

Methods: The methods should have four sections: Study Area, Data, and Modelling, Analysis. Currently, you have the modelling mixed in with the methods, which is a little confusing. I am recommending that you separate the Data (e.g. GRACE, oceanic indices) from the Models (CLSM, HGS) and add a new section called “Analysis”, which includes an explanation of the least-squares regression, trend analysis, and time-variable frequency analysis (e.g. continuous wavelet transformation and power spectral density), and anything else currently mixed in with the results.

Line 122: Once again, it is unclear what variables were extracted/modelled form the HydroGeoSphere model. Please provide a paragraph up front explaining the variables that you extracted from it, the spatial resolution, the temporal resolution, and the time/dates of the data.

I think you should also include a table of the variables extracted from each model, the original data sources, the acronyms, and the resolutions. It would make the data sources easier to visualize. (For example, the cm equiv. water height, surface water anomaly, etc.)

Line 285: You need to add an explanation to the new “Analysis” section on how you calculated the variables to remove seasonal effects.

Precipitation equals snowmelt plus rainfall is a little confusing – this also needs to go in the methods.

I’m a little concerned that this study only compared modelled output. There are no in situ surface, snow, or groundwater data to validate the outputs. Although the models generally had similar outputs (e.g. Figure 3), what is to say that all of the models weren’t wrong? I think you need to justify this based on literature in a “model limitations” section before the conclusion. Also, I think it would be relevant to include literature that has found similar outputs to yours from in situ data that used validation to underscore the accuracy of the model outputs.