Time lags of nitrate, chloride, and tritium in streams assessed by dynamic groundwater flow tracking in a lowland landscape

Kaandorp, Vince P.; Broers, Hans Peter; van der Velde, Ype; Rozemeijer, Joachim; de Louw, Perry G. B.

doi:https://doi.org/10.5194/hess-25-3691-2021

Articles | Volume 25, issue 6

https://doi.org/10.5194/hess-25-3691-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/hess-25-3691-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 25, issue 6

Research article

|

30 Jun 2021

Research article |

| 30 Jun 2021

Time lags of nitrate, chloride, and tritium in streams assessed by dynamic groundwater flow tracking in a lowland landscape

Vince P. Kaandorp, Hans Peter Broers, Ype van der Velde, Joachim Rozemeijer, and Perry G. B. de Louw

Download

Final revised paper (published on 30 Jun 2021)
Preprint (discussion started on 05 Nov 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: 'Review of Kaandorp et al.', Anonymous Referee #1, 11 Dec 2019
- AC1: 'Author response to Reviewer 1', Vince Kaandorp, 30 Apr 2020
RC2: 'Review of "Aged streams: Time lags of nitrate, chloride and tritium assessed by Dynamic Groundwater Flow Tracking"', Anonymous Referee #2, 31 Jan 2020
- AC2: 'Author response to Reviewer 2', Vince Kaandorp, 30 Apr 2020

Peer-review completion

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

ED: Reconsider after major revisions (further review by editor and referees) (04 May 2020) by Markus Hrachowitz

AR by Vince Kaandorp on behalf of the Authors (29 Jun 2020) Author's response Manuscript

ED: Referee Nomination & Report Request started (06 Jul 2020) by Markus Hrachowitz

RR by Anonymous Referee #3 (17 Jul 2020)

Suggestions for revision or reasons for rejection

General comment
Dear authors,
this is an interesting explorative study, highly relevant for managing diffuse nitrate pollutions at catchment scale. It is one of the rare examples implementing nitrate transport at catchment scale using the lumped travel time distributions. This is fits very well the scope of HESS. While I like the general idea of the manuscript, I see several issues in the way the authors justify, implement and discuss their research. Note that I was not reviewer in the first round which is surely not easy for the authors as I may raise issues that have not been asked for earlier. I tried to not be influenced by the first round of review.
Here are my major points (that are also addressed in more detail in the specific comments below): The authors are not very clear in the introduction of the manuscript that this is mainly an explorative study not aiming at fitting the observed concentrations. Readers, same as me, may expect that "fitting". I thus would like to see the explorative character to be made clear earlier on. The authors stress the role of dynamic TTDs but mostly discuss and validate the long-term average behavior of concentrations. The intra-annual variability is discussed later on but not well embedded in the introduction and result sections. I found some justification of scenarios poor. Here I think the beauty of this exploratory approach lies in the interplay of those scenario components in a realistic and justified range to come up with kind of a ranking of factors influencing the observed input-output time lag. The discussion of runoff contributing area and groundwater contributing area not well embedded in the overall story.

Specific comments
Abstract
L17: It should be chloride and nitrate "concentrations", right?
L25: "absence of these processes" has an unclear meaning as e.g. 2) (L23-24) cannot be absent. Consider rewording.
I miss some key numbers in the abstract. Eg. on mean travel time in groundwater and in the unsat. zone. I can imagine this makes the abstract a bit stronger and easier to digest for the readers.

Introduction
L39f: Do you mean nitrate concentration dynamic in time or in space or the general concentration levels? You could also use "mean and variance" or moments of the probability distribution. Would be worth to note that here even if you mean all that aspects of concentrations.
L48f: That is unclearly formulated. Do you mean the travel time along the flow path or activation/ deactivation/ changes of flow paths itself? Dynamic in surface water may also be just a translation of the dynamic in N inputs (on longer time scales).
L54: What is "hydrologic approach" referring to?
L61: Typo "onf"
L65f: The lack of input and validation data is, from my perspective, not the main reason for the lack of application cases. Isn't it rather still a lack of knowledge on subsurface nitrate storage (organic N), removal (denitrification) and on catchment scale heterogeneity in these processes?
L70: This sentence seems to miss some words at the end.
L75: Recharge of the solute is not clearly defined. Maybe it can be just input to the subsurface?

Methods
L170: Nitrate availability may change within the year. Can you shortly comment on the assumptions/ simplification you do here when assuming an annual constant input?
L177: Can you specify what the nitrate transformation factor is? I suppose it represents an average retention (denitrification?) in the unsaturated zone.
L197f: Hmm. To convincingly show that you meet the concentration dynamic you somehow need to acknowledge the reaction site too. Following the Damkoehler concept (or the first order reaction equation) load or concentration over time will depend on both, transport and reaction times. Especially if both are in a comparable range and the Damkoehler number is around 1. I think this needs to be explained in more detail. For me this is also related to the choice and parameter variations implemented in the scenarios. From my point of view you first need to pinpoint, best as you can, the processes as they are and, based on that, explore the different components in a scenario analysis. Here you jump in with a selection of scenarios that sometimes needs more background and justification (see comment on Set 1b below).
L205ff: This is repetitive to L177.
L213: 5 years seems long for that shallow zone. I know that this is a scenario to be translatable to other catchments. Still there are ways to estimate travel times in the unsaturated zone (Harmann et al. 2011 10.1029/2010wr010194; Sousa et al. 2013 10.1016/j.jconhyd.2012.10.007). This could be used to justify that you go 5-fold or 10-fold of that or whatever. But starting with 5 years out of nowhere seems not ok for me.
L220: Is there an evidence of that specific time lag or an accumulation in soils from the soil data you reference earlier in the manuscript (LMM)? That lag time seems to be long when considering a first order reaction transferring organic immobile to inorganic mobile N (e.g. taking mineralization rate in van der Velde et al. 2010).
L232: "slightly" seems an unnecessary word.
L246: Suggestion: As based on the present land-use maps.
L257: Can you justify the use of the zero order kinetics?
L260f: Is that a general statement or one specifically for the Dutch situation? If the former is the case the statement is fine as it is.

Results
L293ff: The age classes chosen seem arbitrary. Can you comment, even if it repeats former studies, on the shape of the travel time distribution? Is this rather an exponential or a more gamma shape? Something like that may help the readers understanding your system.
L298: "stream's" capture zone?
L307f: You mention measurements here. That seems to refer to water quality measurements although you wrote earlier that this is the first time to compare modelled travel times with water quality observations.
L310ff: Just visual model comparison are not really convincing. I think you need to add a measure of the quality of the modelled concentrations as well. If not you need to carefully justify as the aim of you entire study is to validate the modeled travel times with modeled vs. observed concentrations. Moreover, why do you comment on only that sparsely on the modeled nitrate concentration in this part of the manuscript?
Fig. 4: Is there a specific reason for the inlets in panel a, d and f? At least in the results part I could not find a specific reference to that.
L320ff: See my comments above. I think this is a valid reasoning but needs to be introduced much earlier. You should than rather focus on the model exploration in the overall objectives than on the validation. Now I understand why you wrote "possibility of validation of the model-derived TTDs" in the objectives. I think you raise, at least with me, wrong expectations and need to address that more clearly and straightforwardly in the very beginning of the manuscript. Maybe the word "explorative" can already show up in the manuscript title.
Fig. 5: All these scenarios are mass conserving, right? Does the overall much lower concentration of nitrate in 5c means that there is a considerable build-up of a nitrate legacy in groundwater?
L365f: I see this as a missed chance. For me, at least a first order kinetic is more realistic (and used in comparable studies with less physical basis such as van Meter et al. 2017 10.1002/2016gb005498 or Musolff et al. 2017 10.1002/2017GL072630) and would change the shape of the output concentration time series as well while zero order just lowers the concentrations.

Discussion
L372f: Well, the "dynamic" part of TTD have not been shown in its implication so far. The dynamic part is aiming at day-to-day or seasonal shifts in the TTD but you dominantly looked at the long-term effects of TTs on the output concentrations. I would therefore not stress the strength of the dynamic TTDs here but rather refer to TTDs in general.
L377: The simple correction does not create a time delay!
L379: This has not been shown in the results above.
L382: Just as a remark: This argumentation can be a starter in in the introduction already. "We observe time lags between input and export here and elsewhere - here we use the strong physical basis of particle based dynamic TTDs to explore the controls of that time lag, e.g. biogeochemical legacy effects in the soil, TT in the unsat. zone, TT in groundwater, denitrification in soils and groundwater... We will also explore management scenarios such as reduction schemes of fertilizer inputs and land use arrangements".
L419: Why exponential? Also referring to my comment above on the overall shape of TTDs. For me this explorative analysis is surely valuable but also detaches from the study case. Surely, convolving to different function (input with TT) creates varying peaks that depends on both functions. Maybe this would be more telling if the TTD you use would be the long term average TTDs modeled in your catchment. But I do not insist on that if you earlier can show that exponential TT is a good approximation of your system (as literature is critical about that - see Kirchner et al. 2000, Nature).
L458ff: This is all justification for what you did earlier. Part of that may be shifted above.
L473f: Basically you argue that a spatial map of forward TTD can be used for a vulnerability assessment, right?
L486: This spatial settings-issue may be also formulated as the difference in forward TT of nitrate-contributing areas to forward TT of the entire catchment? I am not insisting on a reformulation but try to understand your discussion points here.
L502f: Maybe like this "area where at a certain moment in time water is leaving the subsurface domain to become discharge"? I find the naming of "groundwater contributing area" a bit misleading as it seems to indicate where groundwater contributes to the stream. What you mean seems similar to Yang et al. (2018), Fig. 9. Maybe "runoff generating area" vs. "runoff origin area"? I am not entirely convinced for myself here but wanted to raise that issue with the current naming.
Fig. 11: The dots and colors are hard to see. Maybe increase dot size (a bit on the costs of spatial resolution of course). Esp. RCA in the dry situation is hard to be captured - I mainly see red dots. In the plot it is not clear which colors in the background map refer to agriculture.
L539ff: Again. This is a valuable discussion. However, you neither touched on seasonality or C-Q relationships in the introduction nor in the observed data description nor in the modeled concentration time series. So the whole idea of intra-annual concentration variability is not well embedded in the manuscript while the inter-annual idea seems to be well addressed.
L550f: As you discuss later on maybe the active aquifer volume in this specific hydrogeological situation is a bigger factor of uncertainty.

Hide

RR by Anonymous Referee #1 (05 Aug 2020)

Suggestions for revision or reasons for rejection

General comments:

The authors responded in detail to the comments written down in the review file and I appreciate the effort they put into implementing many of the suggested changes. However, none of the 57 comments I made in the annotated manuscript were (directly) addressed which makes it hard for me to fully evaluate the revised version. I would highly recommend to take these comments also into consideration.
Therefore, I included the annotated manuscript from the original submission again.
In addition I prepared a new annotated manuscript with comments on the revised version.
Also, I copied the comments from both annotated manuscripts to the end of this file.
The manuscript has already improved since the initial submission, still language and sentence structure would benefit from another round of careful checks.
The discussion needs more clarity and precision in explaining the processes, concepts, methods and results. The structure is also not that clear in certain places.
I was not able to follow some of the conclusions because there is not enough explanation on how they were drawn. For example, I understand that there will be an immediate decrease in solutes in the stream if we assume that the catchment TTD is exponential (this is trivial). Now why exactly does this immediate reduction not occur if the MTT is very large – this does not make sense to me – there should still be an immediate decrease, just a slower one. So there must be another process involved that turns the exponential TTD into a different kind of TTD (like, e.g., an exponential piston-flow TTD). It could be explained by making the difference between catchment and groundwater TTD more prominent (an exponential GW-TTD preceded by a constant time delay in the unsaturated zone is basically creating an overall exponential-piston flow TTD).
In the end I am still wondering about the exact results and conclusions that this study provides.
And although, as I said in the beginning, the manuscript has improved since the last round of reviews, in my opinion it requires another iteration.

Comments in the annotated manuscript of the original submission

Page: 1
• I do not understand the title. There is no specific focus or relation to 'aged streams' in the paper. The term is not even mentioned or explained anywhere. So why is this part of your title?
• time series?
• This last sentence makes me wonder what the actual novelty of this study is. You should definitely explain better what you mean by 'groundwater distributes water in time and space' and 'groundwater makes it possible for different waters to mix'.

Page: 2
• Mediums are actually persons who talk to ghosts. Media are substances that water passes through.

Page: 3
• Maybe add some more details on the scenarios here?
• Saalian
• was
• Does 'the upstream' exist as a word?
• Sampling surface and groundwater quality
• What does that mean exactly?

Page: 4
• m³/m³
• Are you discussing the drawbacks of this method anywhere later in the manuscript?
• What are strong and weak sinks? Why were the particles stopped there? Where they removed? What happened to the other 50%?
• Are these subcatchment outlets?

Page: 5
• Can you estimate the TTs through this layer?
• How was this included?
• Curve sounds awkward.
• time-invariant?
• Any ideas on why that could be?

Page: 6
• Why don't you show any of these dynamic TTDs (individually not like in Figure 3)?

Page: 7
• What are possible implications?
• You applied this factor where/when? During the model run? After the model run?
• You should achieve this by changing the model setup (different aquifer thickness, porosity , etc.) not by applying a factor.
• What delay? Some more years, months, days?
• If the peak was kept in 1985, how did you achieve the delayed input function? Did you change the skewness, the variance?

Page: 8
• Does this not directly correspond to the fractions of agricultural land in the subcatchments? If not, why? Vegetation?

Page: 9
• larger
• Why 'although'? The shorter the time series, the smaller the chance to not capture certain events...

Page: 10
• repetitive
• superimposed
• How was that done exactly? Was it added at the end? Was it done in the model? If so, how?
• delete 'a'
• by
• Made it longer or thicker or both (added weight to the tail)?

Page: 11
• no apostrophe (TTDs not TTD's)
• The agricultural input to the GW? To the streams?
• Does that mean that nitrate was not removed at all for all other particles flowing only through shallower layers?

Page: 12
• Is that because you assume that nitrate is not removed under unsaturated conditions?

Page: 13
• nor

Page: 14
• ...and THUS removing this field only...
• Amount?
• Why? Maybe discuss it in more detail.

Page: 15
• the majority of the yearly variation
• Could you derive it by combining tritium and NO3 measurements (transport through the unsaturated zone would cause more removal - lower peak concentrations - than transport through the saturated zone)?
• Rephrase. The nitrate concentrations cannot be lower than the measurements, because you measured the measurements.

Page: 16
• How does this promote denitrification?
• What about changes to your denitrification calculations to improve model performance? I think this analysis falls a little short...
• Why is it not possible? You should try it to provide a more complete analysis.

Page: 17
• Can you quantify the uncertainty caused by the grid cell size?
• arguably

Page: 27
• This does not correspond to any other shape in this figure.
• Are these urban areas?
• What is this? The hydrological base?
• I would label the individual panels.

Page: 29
• Better connect the upper left corners (so you don't draw a line directly through the data).

Page: 30
• I do not see a difference in the scale of the x-axes.
• fraction of area

Comments in the annotated manuscript of the revised version

Page: 35
• I still do not understand the title. There is still no specific focus or relation to 'aged streams' in the paper. The term is not even mentioned or explained anywhere. So why is this part of your title? I would remove it.
• limited to what?
• does it depend on the travel time in the unsaturated zone or on the slow release of organic N or on the travel time in the unsaturated zone only if it includes slow release of organic N? This is not clear to me...
• not clear either
• Which processes?
• This list could use more explanation. How does the time lag change in response to the listed properties?

Page: 36
• You mean the location of delivery or the timing of delivery or the amount of delivery?

Page: 37
• ?
• agricultural?

Page: 38
• Backward TTDs, to be precise.

Page: 39
• Input curve sounds wrong. Input time series would be better.
• What do you mean by 'adding curves to the flow paths'?
• 'principally' is not equal to 'in principle'

Page: 41
• Do you mean time invariant? Also, do you mean backwards TTDs?
• What does that mean? A short explanation of the nitrate transformation factor method would be nice.
• although
• While

Page: 43
• This is not a result of your modeling study. Just a description of your input.
• Again, this is not a result of your modeling study but rather a description of the measurements you use. It would be a results section if you described how your model was or was not able to reproduce certain parts of the measured time series.
• This should be named figure 3.

Page: 45
• You mean the input of solutes to the stream?
• 'also' used twice

Page: 47
• This entire section needs some careful language/sentence structure editing. It is sometimes hard to understand what you want to express. I recommend more consistency in your terminology - always refer to the same processes, solutes, methods, concepts with the exact same name.
• higher/older/greater
• Be more specific, please.
• This is one of the sentences that are quite imprecise in their terminology. What change in what input? Which stream concentrations? What reduction rates in which input?

Page: 48
• Is that addition necessary?
• If the TTD is exponential there should never be a time lag in the arrival of the peak (because the maximum of each exponential TTD is always at the beginning). How exactly do you explain the observed time lag? It must come from something else in the model structure.
• Why do you spell out Travel Time distribution and Mean Travel Time and reintroduce abbreviations (MTTs) here?
• Why? This does not make sense to me? Please explain this behavior.
• But you are not looking at catchments with an exponential TTD. You are looking at a catchment with an exponential groundwater TTD. Adding a lag for the transport through the unsaturated zones makes the total catchment TTD become a piston flow-exponential TTD.
• This is where it becomes interesting and therefore it needs more explanation/analysis.

Page: 49
• Isn't this a different effect from the one you describe above? You say that if denitrification occurs predominantly close to the stream the time lag would increase. But is this not more a function of where the nitrate input takes place? Closer to the stream or farther away? I can imagine that the amount of damping is also influenced by these scenarios.
• This should be expressed more precisely and with more details:
• Longer MTTs in the unsaturated zone cause a larger time lag...
• Which combinations of MTT and solute input reduction cause which dynamics in the time lag?
• ...and so on...

Page: 50
• I do not understand this new definition. '...where water entering the groundwater-flow system is discharged' Is this water entering or being discharged? Maybe a figure would help.
• I do not understand your point here. How are the two contributing areas different?
• models
• Why not call it 'recharge area' if you're using this term to explain it?
• How about calling it subsurface flow contributing area and overland (or surface) flow contributing area? In the end your groundwater contributing area is also contributing to the runoff in the stream.
• nor
• Which consequences does this have on your model results? What about particles that reinfiltrate after seeping out?

Page: 53
• Hey, but above you said that the lag to the peak is extended for some scenarios. How does that fit together?
• Again, you should be more precise. It is a trivial result (and not surprising at all) that there is an immediate reduction if the catchment TTD is exponential. But were you even looking at that or were you also exploring the lag caused by the unsaturated zone?
• To me this is not a convincing argument, because you give too little information on the specific processes that cause your conclusions.

Page: 54
• I would like to see a summary on how exactly each of these parameters influences the time lag. (Not just a statement that they can influence the time lag).
• Again, if you expect an exponential TTD then you always expect an immediate reduction.
• You didn't show them in enough detail or with enough explanation.

Page: 66
• What about the blue line on panel c? Should also be mentioned in the captions.

Referee Report: PDF

Hide

ED: Reconsider after major revisions (further review by editor and referees) (31 Aug 2020) by Markus Hrachowitz

AR by Vince Kaandorp on behalf of the Authors (13 Dec 2020) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (15 Dec 2020) by Markus Hrachowitz

RR by Anonymous Referee #3 (05 Jan 2021)

Suggestions for revision or reasons for rejection

General comment

The authors obviously put efforts in revising the manuscript. I am still convinced about its value: A unique chance to bring together modelled travel times at catchment scale with observed output concentration changes of a conservative (Chloride) and a reactive (Nitrate) constituent. This touches the open issue of reasons for observed time lags for nitrate pollution and the potential buildup of legacies in soil and groundwater. Here, the travel time site can be largely fixed allowing for a detailed discussion on time lag reasons. Moreover this study can bridge between the spatially implicit travel time approach and the nitrate source location within the catchment with clear management implication. All this is somehow there in the paper but neither well introduced and referenced nor well discussed. The study builds partly on a scenario analysis that is, from my point of view, so simple that it can be left out. Adding a fixed time lag to all water ages to then state in the results and discussion that there is a fixed time lag in the model output does not make a story. Here, I strongly recommend to focus the scenarios on something with added value such as the spatial source arrangement and leave it to a verbal discussion on the observed Chloride and Nitrate behavior vs. modelled travel time that allows for the same conclusions. Or to enhance the scenarios with something more sophisticated than added time lags or multiplied travel times. Suggestions on that are given below in the detailed comments. However, this may not be needed for a good story.

Detailed comments

Introduction:
I still miss an introduction to the issue of spatial source arrangement and travel times.

L79ff: "Short" is a quite relative adjective. Short compared to what?
Is this time lag of nitrate different from the mean TT derived from the conservative tracers? If yes, this should be mentioned here. Why else this may be "interestingly" if there is nothing that deviates from other or previous knowledge. Is this a result of this study or can be referenced? This is a tricky point as you seem to need to state a result already to justify your methodology.

Methods
L142f: Can you spend one more sentence on the basis of recharge estimation? This is for most modellers a critical issue. Later in the text you mention ET and this would be the point to describe what you actually did.

L190: "seemed" sounds vague. Do they differ? There are statistical tests for that.

L210: You may give the % in brackets to make fully clear that this is a minor part of the catchment only.

L223: So there is ET from the groundwater? That is not clear in the model description above.

L226: Evaporation or evapotranspiration?

L243ff: I am still not fully convinced by that time lag. Why assuming a simple time shift? Effect will just by the addition of 5 years to all ages in the stream, right? Assuming an unsaturated TTD would make more sense for me. With the often used exponential forward TTD for unsaturated zones you don't even need an additional parameter than the mean.

L253ff: Same as above. This just adds a constant time delay to each particle. This effect is implemented in models such as ELEMENT (van Meter et al). Slow mineralization can be represented by a simple first order reaction that just need one reaction rate constant as a parameter and you will create something more meaningful than a constant time lag.

L260ff: This is not really an uncertainty assessment. Based on your modelling approach and the preceding papers: Is there any uncertainty assessment that you may use here? Just based on Darcy's law or Haitjema (1995): What would be a reasonable error in the recharge and what in the aquifer volume (depth, porosity)? This is something you may use.

L294: I am not convinced here. To what I know, a first order reaction would change the age distribution of the remaining nitrate.

Results
L330: The described behavior in the fig. 4 is hard to see. Esp. the peak times. I there a way to average the data to better see the long-term behavior within the inter-annual variation?

L334: Would it make sense to state the resulting time lag in this chapter already?

L346: The chloride input peak was around that time as well (1984). How can that fast peak be with a modelled median travel time of 4 and a mean travel time of 11 years? Nothing you need to comment on here - it just puzzles me. The difference between mean or median and first reaction to an input is obviously may be something different.

L348: Very old water may contribute tritium free water as well?

L361ff: This is largely as trivial as the scenario setup. I don't see the added value here of simulating that at all. This could be done in a verbal description only. Or consider to make the scenarios more interesting as suggested above.

Discussion
L427: This is not a good chapter name. What processes?

L428ff: This introduction to the unsaturated zone effects partly repeats what was written in further above in 2.7. Remove redundancies and rather hint to the text above.
The following text is all mainly stating the obvious which, from my point of view, is a result of the questionable scenario implementation.

L461f: This is unclearly phrased. How can a time lag occur before something?

L510ff: The exploration of this aspect of spatial source arrangement and the introduction of new results in the discussion section is not justified. It is not clear why you come up with contributing areas here and where this is leading to.

L602f: This is a crucial point that should be focused on most in this chapter. You have the unique chance to compare a conservative tracer with a reactive constituent. From the differences in the model to the observed behavior you can discuss what reasons for time delay are most likely. This cannot be done in one sentence while the rest of chapter's text seems to be too long.

L612ff: This exploration of TT-location-notion is very valuable but, again, not well introduced in the introduction.

Hide

RR by Anonymous Referee #1 (12 Jan 2021)

Suggestions for revision or reasons for rejection

General comments:
This manuscript is coming together. There has been a definite improvement since the last revision. In particular the discussion section has gained a lot in terms of structure and readability. The addition of more detailed explanations and the more in-depth discussion makes the overall message and conclusions much clearer.
Still, I also have to mention the fact that reading the introduction I was not expecting a significant improvement of the revised paper. The introduction still lacks clarity and structure. For someone reading the text for the first time it is quite confusing (this gets much harder to realize if you have dealt with the paper yourself for such a long period of time).
So please consider my questions and comments below and give the first part of the manuscript another iteration in terms of structure and language. Shorter, more concise sentences in the right order and connected logically could be the way forward.
Concerning the research I would appreciate a clearer distinction between the methods the authors are applying and the recommendations they give. For example, the authors write that they derive backwards TTDs with a hydrologic model and a particle tracking routine. But do they also use those backward TTDs to convolve the input time series (or do they use forward TTDs for that)? The differentiation is often lacking throughout the text.
For me, this manuscript is on the right track but needs some further clarification.

Specific comments:
Abstract
Line 23: Which ‘initial’ model run? You did not mention it before.
Line 26-27: What is the difference between the ‘long hydrological travel times’ in line 26 and ‘a long Mean Travel Time (MTT)’ in line 27?
Line 28: ‘high’ application areas? Also, the sentence is cut off after ‘from’.
Line 29: Please specify: Do you observe this lag only if the nitrate attenuating processes are located EXCLUSIVELY near the stream? What if nitrate attenuating processes are located everywhere – also near the stream or drainage network?

Introduction
Line 44-50: This paragraph is a good example for the language, structure and style issues that can be found throughout the introduction: First, flow paths are mentioned. In the next sentence these flow paths are referred to regarding their connection to nitrate concentrations. However, in the same sentence further controls are added (travel time and land use). In the next sentence it is explained why travel time and flow ‘route’ influence nitrate concentrations (but nothing about land use).
It would be good to bring all these arguments into a comprehensive order and group them: Why are flow paths important for nitrate concentrations? Because they control travel times. Why are travel times important for nitrate concentrations? Because they have a large influence on biogeochemical reactions. Why is land use important? And so on…
Further comments: Flow paths are not only the route that water particles travel through the subsoil (since surface flow paths also exist).
Why ‘flow route’, I thought you just introduced it as ‘flow path’?
Line 62: Which approach? You did not describe an approach.
Line 66: ‘while including the dynamic nature of catchments’ - Details please. What does that mean? You already wrote that you will ‘combine these dynamic travel time distributions with application rates…’! This is confusing and probably mixed up. Very hard to follow for the reader in any case.
Line 85: ‘and of different’?
Line 88: ‘as a way forward’? In how far is the exploration of the effect of different processes on breakthrough patterns ‘a way forward’ in model calibration?

Methods
Line 107: ‘The streambed of the stream…’?
Line 130: Better write ‘exhibit’ or ‘show’ instead of ‘include’.
Line 148: Are there ‘backward’ travel times? I know there are backward TTDs, but backward TTs are the same as forward TTs, no?
Line 159: ‘… showed TTDs with exponential shapes…’. It’s not just one TTD, it’s a set of dynamic TTDs, no?
Line 161: Introducing Figure 3 before Figure 2?
Line 176: But you are reporting that 20% of the water in the creek has TTs of less than 1 year. How can you be sure if you haven’t looked at anything shorter than 1 month? This could make up a large proportion of the outflowing water and increase your young water percentages significantly.
Line 216: I don’t understand. Combining the input with the calculated dynamic TTDs or using your model? I thought you said dynamic TTDs but then you mention the model and assigning the input to particles.
Line 219: What is the constant concentration of natural vegetation?
Line 223: What do you mean exactly by ‘within the months after their initial start’?
Line 224-226: This is unclear to me: What about evapoconcentration of these solutes? What happens when particles that are assigned with nitrate evaporate out of the system?
Line 229: How does the omission of chemical processes isolate input time series and groundwater travel time effects on stream concentrations? Travel time will always have effects on stream concentration, input time series (?) too. What do you mean? Please add more details to your explanation.
Line 230: What are ‘catchment mechanisms’?
Line 240: How did the agricultural fields receive chloride and nitrate following the input time series AND calculated backward TTDs? Did you deconvolute the stream signal?
Line 270: Suddenly Figure 6 pops up (before even Figure 2 was introduced).

Results
Line 341-342: Please provide more details on how this agrees with the findings of Kaandorp et al. (2018).
Line 350-351: And what did you make of this finding that the original model underestimates the travel times? Do you have some suggestions on how to fix the original model? Or some thoughts on why this underestimation happens in the first place?
Line 375: What are physical explanations for the occurrence of your different scenarios? It would be beneficial to explain this in your methods section when you introduce the scenarios. For example, scenario 3 could be caused by a larger volume of the unsaturated zone or by a smaller amount of precipitation flowing through that volume.
Line 394: The nitrate peak shifted in which direction to 1985?
Line 406: Why is that the case? I would assume that the longer flowpaths are the deeper flowpaths. Therefore the scenario where all nitrate that travels lower than 0.5 to 15 m is removed by a pyrite layer should have a lighter tail, no?

Discussion
Line 414: Did you compute any efficiency measures (NSE, KGE)?
Line 461-462: If the mean travel time is long enough or the input decrease is gentle enough you will find that the peak of the resulting convolved output time series is shifted (disregarding the fact that you use an exponential distribution with an initial maximum value). So it’s not that surprising after all. Oh, I just see that you describe exactly that in the next paragraph, so nevermind!
Line 480: Better use ‘longer’ instead of ‘higher’ in combination with time.
Line 480: In lower ‘peak’ concentrations – not in lower concentrations in general (or are you referring specifically to nitrate concentrations that experience degradation?).
Line 481: ‘rate’ not ‘rage’.
Line 518-520: So the GCA changes because in certain periods in certain areas incoming precipitation is only turned into ET instead of ET and runoff? Would be good to mention why the area of the GCA changes in time.
Line 540: This is the first time you mention ‘forward’ TTDs. So far you have only mentioned ‘backward’ TTDs. Why the change?
Line 571-572: This is unclear to me. Why would an overestimation of MTTs by the model cause higher tritium concentrations? Does the decay not compensate for that?
Line 621, 631-635: You are writing about the usefulness of forward TTDs here. Yet, in your figures you are showing backwards TTDs. Why don’t you also show the forward TTDs that you determined with your model? Or do you want to say that backward TTDs are just as useful (if so, how)?

Figures:
Figure 3: The MTT and MdTT in panel e) belong to which of the CDFs? Summer, winter, overall?
Figure 4: Put more information on the individual panels in the figure caption. Why not add the input curves for panels a) and b) too?
Figure 6: The legend is wrong. It says 4a) twice.

Hide

ED: Publish subject to revisions (further review by editor and referees) (13 Jan 2021) by Markus Hrachowitz

AR by Vince Kaandorp on behalf of the Authors (23 Mar 2021) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (24 Mar 2021) by Markus Hrachowitz

RR by Anonymous Referee #3 (07 Apr 2021)

Suggestions for revision or reasons for rejection

General comment
The authors clearly put effort into the revision of the manuscript. I think this is a state that allows for publishing this after some minor revisions. I don't believe in the scenario analysis the authors implemented but don't insist on changing this. I am still a bit more critical with the discussion as you may see below. But in the end, this is all satisfying and I really appreciate the management implications.

Specific comments
L83ff: You explore the potential reasons for time lags in addition to the physical groundwater transport component captured by the travel time model, right? Can you make that explicit here?
L85ff: The idea of adding information from Tritium and Cl is introduced after all the potential biogeochemical reasoning for time lags in nitrate response. For me it makes sense to introduce that the other way round - you verify/ challenge the model with conservative tracers and then discuss the additional reasons for time lags. This is also stated that way in line 95ff.
L170ff: It would be helpful if you state a mean and / or median calculated travel time here. Maybe even for the agricultural fields compared to the entire catchments. This can be taken up in the presentation of the initial water quality runs.
Line 256ff: I am not insisting here on changes anymore but would like to state for the records that I am not convinced about this scenario. To show that the addition of 5 years delay in the unsat zone will add a delay to the total travel time, no model or scenario is needed. In addition it is clear that the shallow groundwater table in this catchment does not allow for such long delays. For me this is useless as your aim should be to explain time lags observed in you catchment, not in general.
Line 277f: The former comment is also true for this scenario. Again, I state that there is no scenario needed to multiply the groundwater TTD by a factor of five. I would rather just state and discuss the time lags between modelled and observed (e.g., conservative) constituents and carefully state potential reasons. Surely, an underestimation of groundwater TTs is one of them.
Line 383ff: Maybe I missed something here: Is an unsaturated zone delay affecting tritium at all? During the unsaturated passage, there is still atmospheric exchange possible? Does the tritium-clock starts ticking when the rain hits the ground or when the water reaches the groundwater surface?
L411ff: This section would benefit, as mentioned above, from a description how computed travel times diverge between agricultural fields and the entire catchment.
Line 524-546: All these text introduced the idea of varying contributing areas in general. This belongs rather into the introduction. Only in lines 546-555 you refer to your catchment. This is quite unbalanced. Moreover, the specific discussion if and how the spatial arrangement of sources changes the output signal is discussed in line 513-522 already. So for me this section does not go significantly beyond what is already known. I suggest to start from the specific case made here and after that broaden the view to generalize a bit.
Line 614-617. In all the text on nitrate in your catchment, these are the only two sentences that try to explain why there is a time lag between nitrate and chloride. And this seems to be rather unrelated to all the scenarios. I would expect to see a speculation on all of what you know from the catchment and the scenarios, why there is a mismatch in time lag between model and conservative solute observations, between model and nitrate and by that also between observed chloride and nitrate. What do you think is best explaining this?

Hide

ED: Publish subject to minor revisions (review by editor) (09 Apr 2021) by Markus Hrachowitz

AR by Vince Kaandorp on behalf of the Authors (19 Apr 2021) Author's response Author's tracked changes Manuscript

ED: Publish as is (22 Apr 2021) by Markus Hrachowitz

AR by Vince Kaandorp on behalf of the Authors (29 Apr 2021)

Short summary

We reconstructed historical and present-day tritium, chloride, and nitrate concentrations in stream water of a catchment using land-use-based input curves and calculated travel times of groundwater. Parameters such as the unsaturated zone thickness, mean travel time, and input patterns determine time lags between inputs and in-stream concentrations. The timescale of the breakthrough of pollutants in streams is dependent on the location of pollution in a catchment.