Crossing hydrological and geochemical modeling to understand the spatiotemporal variability of water chemistry in a headwater catchment (Strengbach, France)

Ackerer, Julien; Jeannot, Benjamin; Delay, Frederick; Weill, Sylvain; Lucas, Yann; Fritz, Bertrand; Viville, Daniel; Chabaux, François

doi:https://doi.org/10.5194/hess-24-3111-2020

Articles | Volume 24, issue 6

https://doi.org/10.5194/hess-24-3111-2020

© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/hess-24-3111-2020

© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 24, issue 6

Research article

|

17 Jun 2020

Research article |

| 17 Jun 2020

Crossing hydrological and geochemical modeling to understand the spatiotemporal variability of water chemistry in a headwater catchment (Strengbach, France)

Julien Ackerer, Benjamin Jeannot, Frederick Delay, Sylvain Weill, Yann Lucas, Bertrand Fritz, Daniel Viville, and François Chabaux

Download

Final revised paper (published on 17 Jun 2020)
Supplement to the final revised paper
Preprint (discussion started on 08 Jan 2019)
Supplement to the preprint

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'referee comments', Anonymous Referee #1, 19 Feb 2019
- AC1: 'Answer to the referee's comments', F. Chabaux, 02 Apr 2019
RC2: 'Reviewer Comments', Anonymous Referee #2, 19 Feb 2019
- AC2: 'answer to the rewiever's comment', F. Chabaux, 02 Apr 2019

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

ED: Reconsider after major revisions (further review by editor and referees) (02 Apr 2019) by Jan Seibert

AR by F. Chabaux on behalf of the Authors (03 Apr 2019) Author's response Manuscript

ED: Referee Nomination & Report Request started (04 Apr 2019) by Jan Seibert

RR by Anonymous Referee #1 (22 Jun 2019)

Suggestions for revision or reasons for rejection

General Comment: I commend the authors for their efforts to address the recommended revisions by both myself and Reviewer #2. Clearly, the authors addressed the hydrological concerns with a lot of thought and effort. I particularly appreciated the authors’ incorporation of a geochemical database and modeling parameters that are included in the supplementary material of the modified manuscript. However, after this first round of review I still have some concerns on the geochemical portion of the model and its reproducibility. In particular, the saturation states for the primary and secondary minerals are not presented in the current manuscript. Additionally, the specific clay phases precipitated are not identified (i.e kaolinite vs. smectite, etc.). Without this information, which was requested in the last round of reviews, it’s hard to justify the authors claims that only secondary clays were precipitated (and not other phases) as stated in Lines 270-273 and the representation of clay solid solution series. In general, treatment of the secondary phases still needs to be more adequately described. This is rather crucial for interpretations on C-Q relationships for Na and Si. If these issues can be addressed in the next round of revisions then I believe the manuscript will be suitable for acceptance.
Specific Comments:
Lines 50-52 (Abstract): I’m not sure “originality” is a good word to start this sentence. Also it’s still not clear what is original about this study: is it the approach? – i.e. using a hybrid hydrological + geochemical model to characterize a mountainous catchment? Or is it the hybrid model itself? – i.e. the fact that its “dimensionally” reduced and thus unique from other hybrid models?

Lines 57-60 (Introduction) C-Q relationships are not qualitative. How they are used and interpreted can be qualitative. In this regard, I’m assuming that the authors were speaking specifically about their interpretations based on simple 3 end member mixing relationships (Hornberger et al. 2001). But, C-Q relations themselves show how measured, dissolved solute concentrations evolve with measured changes in discharge. The goal is to explain these observed C-Q relations or, in other words, the observed behavior of dissolved solute concentrations with discharge in watersheds. This is the motivation for using hybrid hydrological and geochemical models – to properly constrain important parameters such as water transit times and chemical reaction times that influence the overall fluid chemistry in order to accurately predict C-Q relationships. This is what was missing in the introduction and I highly suggest that this be rewritten to emphasize this point considering that C-Q relationships represent a large portion of the results and discussion section.
Lines 114-115 (Introduction): Nit-picking here, but what is meant by a “dimensionally-reduced” approach and how does it differ from a “fully dimensionalized” approach? I’m assuming that this is referring to 2-D vs 3-D approaches, but deserves a clear explanation.

Lines 267-268 (Hydrogeochemical modeling): How is the reactive surfaces of the primary minerals tracked? What was actually tracked- changes in the specific surface area (SSA, m2/g) or the bulk surface area (BSA, m2mineral m-3porous media)? Is a shrinking sphere model used (an example can be seen in eqn.3, Navarre-Stichler et al. 2011)? What assumptions are made in the calculation? This needs to be outlined in detail, especially considering that (1) estimating the SA evolution is generally challenging and there are several approaches that can be taken, each with their own assumptions, (2) the authors conclude that “changes in the reactive surfaces of primary minerals became negligible” (line 270) and (3) one of the principal conclusions is centered on surface area estimations.
I highly recommend including in the appendix how the model calculates the evolution of the surface area and present the associated parameter values.

Lines 270-273 (Hydrogeochemical modeling): It’s unclear what is meant here by “hydroclimatic context” in regard to the precipitation of secondary minerals other than clays. It’s the saturation state with respect to these secondary minerals specifically that dictates whether they form or not. This sentence should be changed to something along the lines of: “these secondary phases were not formed based on calculated saturated states…”
Along these lines, how is the saturation state calculated in KIRMAT? A quick statement to address this would suffice. What are the calculated saturation states for both the primary minerals dissolved and clays precipitated? Also, unless I am missing something, the type of clays precipitated is not described in the paper. Was the clay mass fraction mostly kaolinite? Or some mixture with other commonly formed clay phases like smectite? This information must be provided in the manuscript so that this model is reproducible. A good place to put it would be in Table EA11. Otherwise the interpretations on the geochemical variability in the fluid phase as well as tracking clay mass fractions as solid solutions are unsubstantiated. It’s possible that these concerns are all considered in the model, but the point is that the authors do not provide any descriptions of these important chemical parameters and thus makes it hard for the reader to follow the logic of the paper.

Lines 642-643: Chemostatic behavior in this study hasn’t yet been justified. I would suggest validating chemostatic behavior using a power-law relationship to fit the C-Q relationships in log-log space and reporting the value of the exponent, “b” (Godsey et al. 2009). Generally, a b close to zero indicates chemostatic behavior in the catchment.
Figure 8: I would suggest having the data points color coded according with time in a gradient that goes from oldest (2005) to youngest (2015) with the date of each sampling point presented in the legend. This would help the reader understand the evolution of the C-Q relationships with time and help the reader follow the results (lines 647-650) for dissolved Si and Na behavior.

Hide

RR by Laurent Pfister (18 Jul 2019)

Suggestions for revision or reasons for rejection

Dear authors,
The overarching framework of this contribution is the need for developing innovative hydrogeochemical modelling approaches for the watershed scale, able to integrate both the complexity of water flow paths and the diversity of water-bedrock interaction processes. The presented work is novel and timely in the context of existing literature on this topic. The submitted work is equally well aligned with the scope of the journal.
As a way forward, the authors propose – like other recent studies – to merge hydrological and geochemical codes. Instead of solving fully dimensioned problems, they here explore an alternative avenue, consisting of a so-called dimensionally-reduced approach with modest computational needs (even when applied to an entire watershed). Their aim is to model the spatial and temporal distribution of water flow paths (or trajectories), weathering reactions and the subsequent evolution of water chemistry. The authors expect to improve their knowledge of water flow paths – including their variability – between wet and dry seasons, as a prerequisite to better assess water transit times and eventually to better understand water chemistry dynamics.
While the overall goal and proposed approach are highly valuable, this contribution could have its impact certainly increased by further highlighting and leveraging the more than three decades of research into hydrological processes in the Strengbach catchment. While mentioned throughout the manuscript, many of the historical investigations in the area of interest are not presented clearly enough as contributions to what certainly is a textbook case of an evolving perceptual model of a key experimental watershed in critical zone research. By developing (even briefly) in the introduction on what is the perceptual model of fundamental catchment functions of water (and matter) collection, storage, mixing and release in the Strengbach catchment, the authors could build a much stronger case (e.g. in the discussion and conclusion) for how their (unquestionably) important work is a major milestone towards improving the understanding of water chemistry dynamics – in the Strengbach watershed and elsewhere. A new improved perceptual model – leveraging the work presented in this contribution – could (for example) be introduced in the discussion.
Since building on previous hydrological and geochemical modelling work, the authors rather marginally develop on the modelling concepts and mostly refer to existing literature. For aspects related to model parameterization, choices are backed in most cases by references to past investigations and/or field data. However, a major problem for the hydrological modelling in the Strengbach watershed is related to the fact that no O and H isotope data in precipitation and stream water is available. Therefore, mean transit time estimations obtained through the hydrological model cannot be validated by experimental data. While certainly not perfect, a (simple) way to consolidate the outcome of the hydrological modelling could consist in comparing the MTTs obtained through the NIHM model to calculations of the hydraulic turnover of the watershed. Other (albeit more sophisticated and time-consuming) approaches rely on time-variant transit time concepts (e.g. Hrachowitz et al., 2016; https://doi.org/10.1002/wat2.1155). Surprisingly, the range of MTT values is 1.5 to 3 months in the abstract, while in section 4 the range given for MTTs is ‘approximatively 1.75 to 4 months between the strongest flood and the driest conditions.’ These MTTs also differ slightly from those given for the same watershed by Pierret et al. (2018; ‘The Strengbach Catchment: A Multidisciplinary Environmental Sentry for 30 Years’; Vadose Zone Journal): 100 to 200 days (i.e. ~3 to 6 months). While not contradicting per se, these MTT values nonetheless are not totally coherent and should be homogenized in the contribution, if not discussed at some point in the manuscript for if and why they differ from previous work.
Another important point is how the results of this contribution relate to the non-stationarity documented in Pierret et al. (2018) for various parameters in the Strengbach watershed. They have described long-term (i.e. over several decades) significant increasing and decreasing trends in pH and sulfate concentrations in precipitation, spring water and stream water in the Strengbach watershed. As an example of this non-stationaity, mean annual pH in precipitation still remains lower than at pre-industrial levels. In the Strengbach watershed, stream water pH is strongly correlated to precipitation acidity. The slope of pH vs. time for precipitations is more than twice that for stream water, suggesting that some protons are neutralized during their transfer through soils, saprolite and bedrock via exchanges and mineral weathering processes. Pierret et al. (2018) also hypothesize long-term signal fluctuations being dampened during this transfer process – translating in lower standard deviations at the watershed outlet, confirming the neutralisation reactions. Alongside other examples of past research carried out in the Strengbach watershed (e.g. on the interception process and how tree species influences chemical concentrations of atmospheric inputs reaching soils, on the reported 7-year periodicity in precipitation and outflows, on a reported increasing difference between annual inflows and outflows over the past 3 decades). Discussing the results of this contribution in the light of three decades of research in the area of interest and especially how the proposed combination of hydrological and geochemical modelling may help anticipating non-stationarity in hydrological catchment functions of water and matter collection, storage and release in the Strengbach watershed would certainly contribute to further increase the impact of the presented findings.
I hope that these suggestions – for what I consider minor changes – will help to further improve what I consider as a very interesting and valuable contribution.
Best regards,
Laurent Pfister

Hide

RR by Anonymous Referee #4 (15 Aug 2019)

Suggestions for revision or reasons for rejection

Ackerer et al. coupled the catchment hydrology model NIHM with the geochemical model KIRMAT, enabling the much-needed connection between hydrological processes and geochemical reactions. Such connection is particularly important as the hydrology and biogeochemistry fields advance to resolve pressing issues at the interface of water quantity and quality. In addition to the model development, the authors also validated the model with field data, and explored the connections between transit time distribution and reaction rates, another important missing link. While I applaud these novel aspects of the work, there are issues that need to be addressed before publication.
First, with the advanced modeling tool and particle tracking technique that quantifies the travel time, the scientific results from this work appear weak. In particular, the conclusion that “durations of water rock interactions exert a first order control on the chemical composition of waters and that the acquisition of the water chemistry can be explained by weathering processes that are spatially fairly homogeneous over the catchment.” This seems a fairly obvious conclusion that does not need a coupled watershed modeling tool. In fact, the dependence on residence time and the closely related concept of Damholker number have been discussed extensively in existing literature. See, for example, (Maher, 2010; Wen & Li, 2018) and the literature therein.
The authors also concluded that “… the chemostatic behavior of the water chemistry is a direct consequence of the strong control exerted by hydrological processes on water transit times.” This very general statement does not provide specific insights on how water transit time control water chemistry. We all know that transit time controls water chemistry. But how and via what mechanism it leads to chemostatic behavior? Transit time influences dissolution rates but that do not necessarily would have chemostatic behavior. This needs to be better explained and discussed from view point of how processes occurred. The real strength of process-based model like the one presented here is its power in linking observations to processes and mechanisms.
Second, despite of the comments from previous reviewers about earlier work, the authors are still not up to speed about literature. For example, the authors still state that (in Line 186 – 187) “To the best of our knowledge, this is the first Time that such a coupling between hydrological and hydrogeochemical modeling approaches has been attempted at the watershed scale.” This work is obviously NOT the “first time” such coupling has been done, as the previous reviewers have pointed out. Although Beisman et al (2015)’s model does not consider the surface hydrology processes and temporal dynamics and therefore is not strictly a watershed scale hydrological and biogeochemical model, RT-Flux-PIHM (Bao et al., WRR, 2017; Li et al., WRR, 2017) is a coupled hydro-biogeochemical model with the relevant surface hydrology processes. They in fact have a string of new papers out based on this model, including (Wen et al., 2019; Zhi et al., 2019). In that context, the authors really cannot claim the “first time”. Even not being the first time doing this, this paper is still valuable and publishable. It is better not to claim “the first time” when it is not. The series papers from RT-Flux-PIHM did not link water transit time with geochemical reactions, which may be the angel that the authors CAN claim as the major novelty of this work.
Third, although a major focus is on concentration discharge (CQ) literature, the authors seem not aware of the most recent CQ literature. For example, (Musolff et al., 2017) explored relationships between travel time and emergent CQ patterns, although they used a different approach for quantification of travel time. Zhi et al (2019) showed that the contrasts between shallow and deeper water composition governed CQ patterns, which in fact can explain the chemostatic behavior observed in this work, as the dissolving minerals are homogeneously distributed here (if I understand correctly). It would be meaningful and increase the readership of the paper if the authors can discuss results from this work in the context of previous topics on similar topics. Other relevant papers include, for example, (Diamond & Cohen, 2018; Herndon et al., 2018; Musolff et al., 2015).

I generally believe that more insights can be gained via more detailed analysis. For example, when and where the dissolution rates are highest in the watershed at dry and at wet times? In addition to the spatial patters of conductivity, can you show rates or concentration spatial patterns over the entire watershed?

I also think the writing of the manuscript can be improved by being more specific and concise. Some of the discussion appears lengthy and diffusive. For example, 6.1 and 6.2.

Detailed comments:
title: what is "elementary" watershed? it seems an unusual name.
Introduction: Motivation for coupling is still not strong. It can use literature review of CQ to motivate the need of coupling, as previous reviewers have pointed out.
Line 62, Ameli et al 2017 is not at the watershed scale. it is hillslope scale.
Line 85-87: please define “depth-integrated models”. Do you mean there is no resolution in the vertical direction and there is only one grid in the vertical direction?
Line 103 – 107: do you need “if” at the beginning of the sentence. Reads awkward. Can be separated into 2 sentences.
Line 114 – 115: please define “dimensionally-reduced” here and later in approaches. What specific did you do? again related to the “depth-integrated”. Are these two terms equivalent here? if so, maybe stick to one term? Do you mean you only have two grids in vertical direction with unsaturated and saturated zones?
Line 202-245: this is a rather long paragraph. I suggest separate the particle track part as a separate paragraph and with its own subtitle (starting from line 222). this would help give attention to this important section.
Line 222: are there existing references for backtracking approach? does figure 3 indicate that the backtracking only track through flowing water? what about the areas that were dry and disconnected at dry times and reconnected to the stream at a later time.
Are mineral phases homogeneously distributed? I think they are. Please be explicit.
Section 3.2: it sounds like the concentrations are calculated based on TST rate law. but the description also sounds like the calculation is based on travel time. The TST itself does not have a time component to take into account of travel time. so I am confused about how exactly the rates were calculated. Please clarify.
Figure 7, 11: these bar figures are unnecessary and not effective. Why not plot lines for conc vs time for different solutes? you can still add the measurement data for comparison. It would be nice to also include conc. vs. mean travel time, as they may reveal different trend as conc vs discharge.
518-523: “give weight to” were used for a few times. Suggest rephrasing
The discussion about geometric surface area is applicable together with lab-measured reaction constants may need further consideration. The text describing Table 3 emphasizes similar geometric surface area and BET surface area. But the geometric surface area has a large range, often by orders of magnitude. if one takes log average instead of arithmetic, the geometric surface area is in fact much lower than the BET surface area, which mean a much lower surface area is needed to reproduce the concentration data. This in fact is consistent with many previous studies showing that lower surface area needs to be used in order to directly use TST rate law at the field scale, see for example (Heidari et al., 2017; Moore et al., 2012).
657-659: “the study of concentration discharge relationships has been intensively used to assess the chemostatic behavior of waters (Godsey et al., 2009; Kim et al., 2017; Ameli et al., 2017).” Oddly phrased sentence. Please rephrase.
References:
Diamond, J. S., & Cohen, M. J. (2018). Complex patterns of catchment solute–discharge relationships for coastal plain rivers. Hydrological Processes, n/a-n/a. doi: 10.1002/hyp.11424
Heidari, P., Li, L., Jin, L., Williams, J. Z., & Brantley, S. L. (2017). A reactive transport model for Marcellus shale weathering. Geochimica et Cosmochimica Acta, 217(Supplement C), 421-440. doi: https://doi.org/10.1016/j.gca.2017.08.011
Herndon, E. M., Steinhoefel, G., Dere, A. L. D., & Sullivan, P. L. (2018). Perennial flow through convergent hillslopes explains chemodynamic solute behavior in a shale headwater catchment. Chemical Geology, 493, 413-425. doi: https://doi.org/10.1016/j.chemgeo.2018.06.019
Maher, K. (2010). The dependence of chemical weathering rates on fluid residence time. Earth and Planetary Science Letters, 294(1-2), 101-110. doi: 10.1016/j.epsl.2010.03.010
Moore, J., Lichtner, P. C., White, A. F., & Brantley, S. L. (2012). Using a reactive transport model to elucidate differences between laboratory and field dissolution rates in regolith. Geochimica et Cosmochimica Acta, 93, 235-261. doi: 10.1016/j.gca.2012.03.021
Musolff, A., Fleckenstein, J. H., Rao, P. S. C., & Jawitz, J. W. (2017). Emergent archetype patterns of coupled hydrologic and biogeochemical responses in catchments. Geophysical Research Letters, 44(9), 4143-4151. doi: 10.1002/2017GL072630
Musolff, A., Schmidt, C., Selle, B., & Fleckenstein, J. H. (2015). Catchment controls on solute export. Advances in Water Resources, 86, 133-146. doi: 10.1016/j.advwatres.2015.09.026
Wen, H., & Li, L. (2018). An upscaled rate law for mineral dissolution in heterogeneous media: The role of time and length scales. Geochimica et Cosmochimica Acta, 235, 1-20. doi: https://doi.org/10.1016/j.gca.2018.04.024
Wen, H., Perdrial, J., Bernal, S., Abbott, B. W., Dupas, R., Godsey, S. E., et al. (2019). Temperature controls production but hydrology controls export of dissolved organic carbon at the catchment scale. Hydrol. Earth Syst. Sci. Discuss., 2019, 1-35. doi: 10.5194/hess-2019-310
Zhi, W., Li, L., Dong, W., Brown, W., Kaye, J., Steefel, C., & Williams, K. H. (2019). Distinct Source Water Chemistry Shapes Contrasting Concentration-Discharge Patterns. Water Resources Research, 0(0). doi: 10.1029/2018wr024257

Hide

ED: Reconsider after major revisions (further review by editor and referees) (22 Aug 2019) by Jan Seibert

AR by F. Chabaux on behalf of the Authors (07 Oct 2019) Author's response Manuscript

ED: Referee Nomination & Report Request started (08 Oct 2019) by Jan Seibert

RR by Anonymous Referee #4 (20 Oct 2019)

RR by Anonymous Referee #1 (03 Jan 2020)

Suggestions for revision or reasons for rejection

General Summary: This study presents the application of a coupled hydrological—reactive transport model capable of characterizing monthly to annual scale hydrogeochemical variability in Strengbach, a small granitic headwater catchment in France. The model proposed is composed of both a depth-integrated and spatially-distributed NIHM hydrologic model and a Kinetic Reaction and Mass Transport KIRMAT reactive transport model. Although there are several combined hydrologic + geochemical models that have been developed in the literature, this particular model has some unique features that make it quite appealing in application to watershed scales. One feature in particular is the low-dimensional approach to the hydrologic model, which consists mainly of combining vadose and saturated zones into a single subsurface compartment that can be modeled as a simple 2-D layer. Reducing the “dimension” of the hydrologic model has significant advantages in that it reduces the complexity of solving numerically for both unsaturated and saturated flow and the computational costs. This is not the first study to take such an approach, but it is novel in that it’s one of the first studies to combine this “non-dimensionalized’ hydrologic model with a fully realized reactive transport model. This model was further validated with field data where independently constrained water transit times through the hydrological simulations were used to accurately predict observed geochemical variability.

General Comments: This last round of revisions has shown large improvements to the overall quality of the manuscript and I thank the authors for their efforts. In light of the corrections that have been made by the authors, I believe that this article is now suitable for publication granted that certain minor modifications are made beforehand. I suggest the authors try to concentrate more on what these model simulations tell us about the catchment dynamics in Strengbach and how this build on findings from other studies of this catchment. As one of the other reviewers noted, Strengbach is a well-studied catchment and, thus, articulating how the results from these novel hydrological + reactive transport models provides a (large?) step forward in our understanding of the Strengbach catchment in particular would boost the impact of this paper in my opinion. While I think it is good to explore what these findings might mean at a global scale or in comparison to other similar catchments, it shouldn’t be the centerpiece of the present conclusions; rather it should be used as motivation for conducting similar types of studies in other catchments in the future.

Additionally, I found quite a bit of spelling and grammar mistakes as I was reading through this latest version of the manuscript. There are also areas where I still believe the writing can be paired down. I strongly urge the authors to address these issues before publication.

Specific Comments:
Lines 36- 37: “explains why transit times span much narrower ranges of variation than that water discharges.” Does this statement apply to Strengbach specifically or is this a more general observation? Please provide citations for the latter.

Line 50: remove “the” in “ effects of the ongoing climatic changes…”

Lines 87- 89: How the flow is treated in a dimensionally reduced hydrologic model should be quickly added here. Even the explanation that was provided to one of my comments in the last round of revisions is suitable as shown below:
“NIHM solves subsurface flow via an integrated Richards equation …”

Lines 607-610: Is the model able to track the areas of maximal clay formation in the catchment and how this varies between wet and dry seasons? This would be very useful information to complement the predicted areas of max dissolution rates proposed by the model.

697-699: Yes, but only on a monthly to annual timescale resolution. It would be interesting to see if this model can replicate observed water chemistry in more high frequency daily or hourly geochemical datasets during major flooding events.

Line 713: Be specific in regard to the explanation of this themodynamic equilibrium, which is (I think) the primary mineral dissolution reaction. I think I found this elsewhere throughout the manuscript.

Lines 727-729: “…spatial and temporal variability in flow paths is a key process to explain C—Q relations” I would add for this particular catchment.

Lines 748-752: “…good estimate of the reactive surface within the natural environment” is a little too broad. While I agree that this geometric approximation of the minerals is most likely a better approach in the models, I don’t think you can generalize these conclusions to the “natural environment”. In this case, for this particular catchment, these estimated low surface areas along with the rate constants did provide good replications of the observed geochemical data. This might also be the case for other similar granitic headwater catchments, but not perhaps the case for volcanic or karstic systems.

Figure 8: The axes for the H4SiO4, Na+, K+, and Mg2+ should really be shown here in mmol L-1, rather than mol L-1 that way it’s easier for the reader to see the variability.

Hide

RR by Camille Bouchez (09 Jan 2020)

Suggestions for revision or reasons for rejection

This work is of great interest and provides new insights in coupling hydrological and geochemical processes at watershed scale. The authors explore the development of a dimensionally-reduced model coupled with a reactive transport model. The modelling approach is validated by field data collected in springs and piezometers at wet and dry periods, which allows for the investigation of the spatial and temporal variability of transit times and reaction rates. The authors conclude on a hydrological control on the chemostatic behaviour of the watershed, due to fairly constant transit times despite highly varying flow dynamics. The conclusions of the paper thus strongly rely on the hydrological model simulating a low variability of water transit times, from which the geochemical model logically simulate a low variability of geochemical concentrations. To my point of view, a more detailed description of the hydrological model is needed to strengthen the confidence in the results (see specific comments). I recommend moderate revisions before publication of this interesting work, mainly because information is missing in the method section and because the study would gain from a more detailed analysis of the results.
Specific comments:
Figure 1. Please add a x scale.
l. 175 refer to Figure 1
METHODS
l.179 A lot of geochemical data seem available, both for springs and piezometers, but only some of them are used to validate the models. Why are you only using some of the available data? The model would be improved with a validation on all available data that span a long period of time. And if not, the authors should justify why they use only a limited part of their dataset and how they chose the data used.
l.200 “The exchange of water between the surface and subsurface flows are addressed via the hydraulic head differences between the compartments.“ A hydraulic conductivity value of the interface between the two compartments must be also considered. Which value was used? Was it calibrated or fixed? Please clarify.
The model parametrization paragraph needs clarification and additional precisions.
- “several zones of heterogeneity” please refer to the Figure
- Please specify all parameters instead of “other parameters” l.206 for clarity
- I guess the aquifer thicknesses given Fig. 2 correspond to the values obtained after the calibration. If so, please be specific on the figure or in the legend. Also add in the legend that the grey lines show the mesh grid of the hydrological model (if it is the case).
- The hydraulic conductivity is calculated using with the Van Genutchen model (which involves parameters such as the saturated hydraulic conductivity, n and alpha), and therefore is a result of the simulations at the grid size, is it correct? I don’t think it is clearly said in the paper. As the temporal variability of the hydraulic conductivity is an important point of this study, I think a very clear explanation of how the depth-integrated hydraulic conductivity is calculated would make it easier for the reader to follow the logic. Fig.6, do the red colour (highest hydraulic conductivity) correspond to the value of the calibrated saturated hydraulic conductivity? If yes, it might be worth saying it, as it shows that the subsurface compartment is fully saturated at high flows.
- How is the calibration realized? Which algorithm was used? Could the authors add some uncertainty estimates on the calibrated parameters?
- The hydrological model over the whole catchment is calibrated only on the discharge time series located at the outlet of the catchment. Do the authors have other hydraulic data they could use to strengthen their calibration, such as piezometric heads or spring discharge rates? If not, I am worried that equifinality might not be negligible, which also points for a serious estimation of parameter uncertainties and/or sensitivity analysis (see previous comment).
The hydrogeochemical modelling strategy presented Figure 5 deserves more details. For each flow line, how much is “several”? What is the value of the “constant distance along the flow line” that is used? Do the boxes drawn Figure 5 correspond to the grid of the NIHM model or not? KIRMAT simulations were performed for different flow lines independently and then mixed at the outlet of the flow lines, if I got it correctly (Figure 5). The mixing could also occur all along the flow line, each percolated soil water mixing with the water coming from the upstream “box”. Would the results be different? What is the justification for no mixing between flow lines within the subsurface compartment?
RESULTS
l. 320 Where are the results of the water velocities?
l. 328 Please specify which characteristics of the flow lines you consider similar (geometric characteristic? length, position…), as the velocity along the flow lines for instance differ from given dates (4.2). Maybe not talking about water velocity along flow lines in this paragraph might be less confusing.
l. 341 “or parameters” Parameters should not change under transient conditions as they have been previously calibrated, right?
l. 344 Please describe the driving factors of the spatial and temporal change in the simulated hydraulic conductivity. How much is this result related to the geometry of the watershed (small thickness, steep topography…)?
l. 385 refer to Figure 7
Figure 7. Where do the uncertainty bars come from in the KIRMAT simulated concentrations? Please, clarify what you mean by “induced” in the legend. A priori it could not be induced by the hydrological model as uncertainties are not taken into account. Is it only coming from the propagation of soil solution uncertainties?
Figure 8. The CS1 geochemistry was simulated at 6 different dates, but we cannot tell which of the observation blue point corresponds to each simulated orange point on the Figure. This information is needed to assess the reliability of the modelling. Maybe using different colours, please link each simulated point to the corresponding observation point.
Figure 9. and l. 403-405. The sentence is quite vague. The authors chose to present the differences between elements in a figure, which is very interesting. But then their explanation for the differences remain very vague and not specific to any element. Could the authors expand on the geochemical mechanisms yielding to the different C-MTT relations?
PZ3 and PZ5 piezometers. You already specify paragraph 4.2 that, such as for CS1 and CS3 springs, single flow lines can be used, so you can shorten the first part of the section. Maybe this section could be merged with the previous one as the approach and the results are similar.
CS2 and CS4 springs. Same comment as above, you can probably shorten the justification of the scattered flow line distributions as it is a repetition of paragraph 4.1
DISCUSSION
Have the standard kinetic constants used here been determined on minerals all coming from the Strengbach catchment? Then would you recommend to use site-specific kinetic constants to account for local aging effects?
I am not sure if Figure 13 brings anything new. It might be more interesting to show the distribution of dissolution rates over the catchment.
The discussion on the chemostatic behaviour would gain from a more concise and straightforward argumentation. Some repetitions with the result section could be avoided. For instance, the whole paragraph describing changes in the simulated hydraulic conductivities, in water velocities and in mean transit times should not be detailed as much or details should be moved to the result section 4.2. Lines 693-699 aim to justify the modelling approach and results (same holds true for the discussion on the clay solid solution), which could be moved elsewhere for clarity. I would advise to refocus the whole paragraph only on the discussion on the potential origins for the chemostatic behaviour.

Referee Report: PDF

Hide

ED: Publish subject to revisions (further review by editor and referees) (12 Jan 2020) by Jan Seibert

AR by Lorena Grabowski on behalf of the Authors (12 Mar 2020) Author's response

ED: Referee Nomination & Report Request started (12 Mar 2020) by Jan Seibert

RR by Camille Bouchez (24 Apr 2020)

ED: Publish subject to minor revisions (review by editor) (24 Apr 2020) by Jan Seibert

AR by F. Chabaux on behalf of the Authors (01 May 2020) Author's response Manuscript

ED: Publish subject to technical corrections (15 May 2020) by Jan Seibert

AR by F. Chabaux on behalf of the Authors (19 May 2020) Author's response Manuscript