Thank you very much for your review and useful comments. We changed the manuscript to include your corrections and clarify the point regarding f_wtd.



Regarding your first two comments, we followed your suggestions and changed the sentences in the abstract (lines 10-15).

In the first version of the manuscript, it read : "This study found that the impact of groundwater and floodplains on precipitation and 2-meter air temperature biases is predominantly positive in comparison to observations, although on a very regional scale. Additionally, the model's ability to simulate terrestrial water storage and river discharges is enhanced by these processes. The improvement in land hydrology with floodplains and groundwater is attributed to their ability to increase the hydrological memory of the model."

It now reads :  "The simulated hydrological cycle is improved by representing floodplains and groundwater, thanks to an increased hydrological memory which allows to better capture the seasonal cycle of the terrestrial water storage and river discharge. Additionally, the inclusion of groundwater and floodplains reduces precipitation and 2-meter air temperature biases at the regional scale."

Yes, f_wtd is variable in time, as it depends on wtd. We clarified this point in the manuscript.

In the model’s description section, lines 114-115, we changed the sentence « with vertical upward capillarity fluxes into the superficial unsaturated soil in a fraction of each grid cell defined by the actual distribution of the sub-grid topography » into « with vertical upward capillarity fluxes into the superficial unsaturated soil allowed over a fraction of each grid cell which varies with the water table head and its position with respects to the sub-grid topography distribution ».

Also, line 148, the sentence « Subgrid fractions, f_wtd, of each grid cell that allow z_wtd to rise into the superficial soil »  was changed into « The temporal mean values of f_wtd the fraction of each grid cell over which z_wtd is allowed to rise into the superficial soil ».



We also corrected the typo line 214 ("bay" instead of "by").



Thank you again.

Thank you for your specific comments which helped us clarify the manuscript by giving more details on the points you commented on. We also thank you very much for having taken the time to make all these corrections on typoes and english mistakes. For the sake of clarity in our answer below, we copied each of your comments in bold characters before answering them.

Specific comments



1. The methods used in this study are clearly outlined overall, with a small handful of requests for added clarity to be found within this section of the peer review.



2. Will the authors please add a description of GRACE and how it is used in the methods section? It would also help to know the rationale for selecting GRACE over other comparable products. And if other comparable products exist, it would also be good to learn why an observational ensemble is not being used. To be clear, I think it’s okay to use just one observational dataset, it would just be good to know why this choice has been made.

We added information on the GRACE data and the reasons why we chose it at the beginning of section 3.2 (after line 171) : "The Gravity Recovery and Climate Experiment (GRACE) satellite mission, developed by NASA and the German Aerospace Center (DLR), measures temporal variations in the Earth's gravity field. These variations are used to estimate changes in terrestrial water storage (TWS), including surface water, soil moisture, and groundwater. Therefore, we used these GRACE data to evaluate the TWS variations simulated by our climate model (see also Data availability). This dataset consists of an ensemble of three monthly TWS estimates, from 2002 to 2014; they are derived from solutions provided by three independent processing centers: the Jet Propulsion Laboratory (JPL), the German Research Centre for Geosciences (GFZ), and the Center for Space Research (CSR). These data are uniquely suited for evaluating large-scale hydrological processes as they provide a global coverage of direct measurements which can not be obtained through other observational products."

However, this was not done in the methods section, where we chose not to include a "dataset" subsection, given that all the datasets we used are very well known, accessible and documented in the references we give in the "data availability" section.



3. Also related to the methods section, will the authors include a sentence to indicate why performing this set of experiments in offline mode is preferred?

All the simulations analyzed in the paper were run in "inline" mode, that is with the land surface model coupled with the atmospheric model. Did you mean "inline" instead of "offline" ? Or by "offline", you meant "not coupled to the ocean model"?

At any rate, we specified again in the methods section (2.2) that the simulations were run with CNRM-CM6-1, not coupled to the ocean model.

The simulations were run inline because the paper deals with the effects of groundwater and floodplains on the atmosphere. They may also have a limited impact on the ocean dynamics near the rivers mouth, but this is beyond the scope of the paper. It is very unlikely that the expected limited impact on the ocean would affect the atmosphere, and running the same experiments in a fully coupled mode (that is, coupled to the ocean model used in CNRM-CM6-1) would require a much larger number of members as the model's internal variability is much greater in this configuration, so we felt it was unecessarily expensive.



4. On line 174, skill scores are referenced. I assume this is related to the bias scores presented in Figure 3, but is it possible to state which test of skill is used and refer to the corresponding figure within the main text to make this more explicit?

"Skill scores" in this sentence actually refers to the mean biases and root mean square errors, shown on Figure 3. We clarified it in the manuscript.



5. Could figure 11 be moved to before the discussion and conclusion section?

Agreed. It is usually the journal that takes care of the final editing and positionning of the figures.  We will make sure the figures are well positioned in the proofreading file.



6. Is it possible to include a model structure diagram to show how CRNM-CM6-1 incorporates the ISBA and CTRIP models, as well as sub-grid processes? This might help other ESM modelling teams envision how floodplain and groundwater processes could be incorporated into their own models.

We did. This new figure and its descriptive caption is attached to this reply.



7. Figure 5b seems to appear before it is mentioned in the main text. Is it possible to move this figure (or the mention of it) so that this is no longer the case?

Agreed. It is usually the journal that takes care of the final editing and positionning of the figures.  We will make sure the figures are well positioned in the proofreading file.



8. Ask for confirmation of what the model uses as maximum rooting depth

It depends on vegetation types. We added a paragraph describing this point in the methods section (2.1.2) : "The soil moisture profile is computed within the rooting depth, which varies from 0.2 to 8 meters depending on the vegetation type, as detailed in Figure 2 and Table 1 of Decharme et al. 2019.These rooting depths values are derived from the ECOCLIMAP database. They reflect vegetation-specific adaptations to climatic and soil conditions. In contrast, the soil temperature profile is computed down to a depth of 12 meters to account for thermal dynamics beyond the rooting zone. To represent land cover and rooting depth heterogeneities, a tile-based approach is employed. This approach allows multiple vegetation types to coexist within a single grid cell. Distinct energy and water budgets are computed for each tile, and their relative fractional coverage within the grid cell is used to determine the grid-box-averaged water and energy budgets."



9. Please include a sentence on why the False Detection Rate test was used to test for statistical significance instead of other sign-based tests.

This was done in section 2.3 of the method section, where we state "To assess field differences significance, we use the False Detection Rate (FDR) test from Wilks (2016), with a 95% confidence level. This test, based on local t-tests, allows to reduce the false detection rate (i.e the detection of a signal which is actually not significant) in the case of auto-correlated fields, such as those analyzed in climate science. Further details on this test are given in Wilks(2016) and Colin et al. (2023)."

To address your comment, we added a sentence to this paragraph. The paragraph now reads : "To assess field difference significance, we use the False Detection Rate (FDR) test from Wilks (2016), with a 95% confidence level. It relies on t-tests performed for each grid point. But instead of comparing the p-values to a fixed value (0.05 for a 95% confidence level) as it is classically done, the FDR test consists in computing a threshold which also corresponds to a given confidence level (here 95%) but depends on the series of all P-values. This allows to reduce the false detection rate (i.e the detection of a signal which is actually not significant) in the case of auto-correlated fields, such as those analyzed in climate science. Further details on this test are given in Wilks(2016) and Colin et al. (2023)."





Technical corrections

We agree with the vast majority of your correction and modified the manuscript accordingly. Below, we only reply to the comments with which we disagree or for which we came up with a correction different from the one you suggested.



7. Page 2, line 40: Depending on the intended meaning, it may be clearer to say “potential evaporation” rather than just “potential”. 

We wrote instead that " [evapotranspiration] reaches its potential rate".

11. Page 3, line 54: Delete “these” before “water bodies”.

We changed "these water bodies" into "inland water bodies".

12. Page 3, line 57: Delete “s” after “impacts”.

13. Page 3, lines 57 and 58: Reverse the order of “enough documented” to read “documented enough”.

We changed the whole sentence from "The climate impacts of groundwater stored in large unconfined aquifers are however not yet enough documented while it acts as a lower boundary for the overlaying unsaturated soil through upward capillarity rise" to : "However, the way groundwater stored in large unconfined aquifers might impact climate worldwide has not yet been documented enough, even though these aquifers act as a lower boundary for the overlaying unsaturated soil it can affect through upward capillarity rise".

21. Page 3, line 76: Add an “s” to the end of the word “remain”.

Instead, we added a "s" to feedback.

23. Page 3, line 84; Page 8, line 186: It’s possible that the word “assess” should be used instead of “access”.

Yes. And also page 8, line 166.

26. Page 4, line 114; Page 5, line 117; Page 7, line 149: This may be a matter of style, but it’s possible that the word “upper” is more precise in English than “superficial”, which would be preferred in some other languages, such as French (example: superficiel).

By "superficial", we meant the part of the soil column of the ISBA model, based on the Richard’s equation (unsaturated soil). We understand what you mean but we are afraid "upper" might be misleading as it can refer to the whole unsaturated soil, and not just its upper part.

Line 114, we thought it was best to remove "superficial" from the sentence altogether.

Lines 115-119, we rewrote the 2 sentences to change the two occurrences of "superficial". Going from : "The second difference compared to the standard version of the CNRM-CM6-1 climate model appears in the coupling between the aquifer water table and the superficial soil. While the standard coupling imposes the water table depth to be lower or equal to the hydrological soil depth in ISBA to compute upward capillarity fluxes (Vergnes et al., 2014; Decharme et al., 2019), in our upgraded version the Richards equation is modified to allow the water table to penetrate into the superficial soil (Colin et al., 2023)." to : "The coupling between the aquifer simulated by CTRIP and the soil column simulated by ISBA has been modified compared to the standard version of the CNRM-CM6-1 climate model. While the standard coupling imposes the water table depth to be lower or equal to the hydrological soil depth in ISBA to compute upward capillarity fluxes (Vergnes et al., 2014; Decharme et al., 2019), in our upgraded version, the Richards equation is modified to allow the water table to penetrate into the soil column of ISBA (Colin et al., 2023)."

Line 149, "superfical" was changed into "ISBA soil column".

Line 152, "Unsaturated superficial soil" was changed into "above unsaturated soil".

37. Page 7, line 149: The word “globally” may not be needed.

We changed "represent globally 12 %" into "amount to 12 %".

53. Page 11, line 219: The meaning of the words “to dawn” is not clear in this context.

We rewrote this part of the sentence, which now is "the strong underestimation of the simulated precipitation limits the development of floodplains".

55. Page 11, line 227: Delete the letter “s” at the ends of the words “floodplains”, “models”, and “performances”.

OK for "floodplain" and "performance", but we do mean to talk about land-atmosphere coupled models in general, hence the plural.



69. Page 16, line 310: The term “downward surface solar heat flux” is a bit confusing.

We changed it to "surface downward solar radiation". We also changed it in the caption of figure 8.

73. Page 20, line 363: Replace the “s” at the end of “biases” with a “d”.

We don’t think this would work. We do mean "the model’s biases" as in "the biases of the model" ("les biais du modèle" in french).

75. Page 21, Figure 10 caption: Note that “2m” is written as “2-meter” some of the time.

Yes. We wrote "2m" in figures captions and in the data availability statement but felt it was better not to use this abbreviation in the text, where we prefered "2-meter". We think it is quite clear that 2m = 2-meter.