the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 31 May 2024
Nitrate and Water Isotopes as Tools to Resolve Nitrate Transit Times in a Mixed Land Use Catchment
Abstract. To understand the transport and fate of nitrate in catchments and its potential hazardous impact on ecosystems, knowledge about transit times (TT) and age of nitrate is needed. To add to that knowledge, we analyzed a 5-year low-frequency dataset followed by a 3-year high-frequency data set of water and nitrate isotopic signatures from a 11.5 km2 headwater catchment with mixed land use within the Northern lowlands of the Harz mountains in Germany. For the first time, a combination of water and nitrate isotope data was used to investigate nitrate age and transport and their relation to water transit times. To do so, the numerical model tran-SAS based on Storage Age Selection (SAS) functions was extended using biogeochemical equations describing nitrate turnover processes to model nitrification and denitrification dynamics along with the age composition of discharge fluxes. The analysis revealed a temporally varying offset between nitrate and water median transit times, with a larger offset at the beginning of wet periods due to higher proportions of young nitrate that is released more quickly with increasing discharge compared to water with larger transit times. Our findings of the varying offset between water and nitrate transit times underline the importance of analyses of solute transport and transformation in the light of projected more frequent hydrological extremes (droughts and floods) under future climate conditions.
- Preprint
(1681 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on hess-2024-109', Anonymous Referee #1, 28 Jun 2024
Review of “Nitrate and Water Isotopes as Tools to Resolve Nitrate Transit Times in a Mixed Land Use Catchment” by Radtke et al.
This paper presents a joint analysis of travel times of water (derived from SAS-based transit time modelling on water isotopes) and nitrate (based on the same SAS model, but with included biogeochemical reactions). This is an important topic and a “next step” in research on combining (conservative) transport of water with non-conservative solute transport. Of course, this is a tricky task, and the study is only able to propose some solutions, but it is nonetheless an important step in the right direction.
I think the paper is a very useful contribution to the community. It is mostly well written, but clarity could be improved in a few sections. With some better explanations, this might become a key paper at some point, and I am providing some suggestions that may help with this.
Specific comments:
Abstract: the mention is of 5 years of low frequency data followed by 3 years of high frequency data. All figures only show five years of data. This is somewhat inconsistent and should be clarified.
Lines 52-63: these are very relevant processes, and it might be helpful for readers if this could be illustrated in a schematic plot.
Line 71: the authors mention here that mobilisation as nitrate “can take up to decades”. This is an important point, and I am glad it is being made. But it is not entirely clear to me how they constrain these inputs and their enormous temporal variability and uncertainty in their model (this is not really explained in the methods – at least not prominently, so I missed it if it was). I would encourage the authors to explain this more clearly in their methods.
Line 80: hypothesis (i): can the authors add an explanation of why they expect this temporal offset?
Line 107: please add units to the kf values.
Lines 107/108: Can you provide information on the depth of the storage/aquitard? This would be helpful information for some of the conclusions drawn later, regarding the contributions from shallow and deeper storage.
Line 118: I find it a bit confusing and distracting that delta15_N values are reported and presented in great detail here, after it is explained in the introduction that it’s the oxygen isotopes of nitrate that are of relevance. Consider adding a note here that only oxygen isotopes are analysed.
Section 2.2: The separate description of the water isotope and nitrate sampling is quite confusing here. It appears that these span the same periods, but it is not explained if these are actually the same samples or samples collected at totally different times during these intervals. If the latter is the case, then it would be useful to present a flow duration curve, to illustrate that similar conditions are covered by the two different sample sets. Else this would question comparability between the water and nitrate travel time analyses.
Figure 2: This figure tries to convey a lot of information, and I’m unsure it helps. I am finding it slightly confusing at least and would recommend simplifying it. I’m not sure what the incorporation of solutes into the “legend” on the left means (e.g. for “leaching” or “discharge”). Does it imply that these compounds are measured or modelled in these fluxes? It would also be helpful if it was explained which of these fluxes and concentrations are measured, and which ones are modelled. Including the relevant equations from the routing storage is likely more confusing than helpful.
Section 2.4 (Model setup): I think it is very useful that the authors are trying to provide a good explanation of the used SAS approach. However, to be truly beneficial, a bit more information needs to be added. For example, it is not explained anywhere what the backward transit time distribution is (line 165) or what the implications and definitions of time variant and time invariant SAS are (line 172). The variables T and deltat are also not explained (equations 2+3), and neither is the wetness index (line 177), or the parameters kQ1, kQ2, kL1, and kL2 (lines 180, 181).
Line 219: “as a time stamp” – this would require that every point in time has a unique isotope signal, but this isn’t the case as the isotope signal is variable over time. Instead, a better wording may be “used as a finger print” or similar.
Lines 217-219: This is a really helpful explanation, but I would recommend bringing it much earlier to limit confusion that arises before this point regarding the methods and sampling.
Lines 224-258 (and lines 505-506): it may well be true that an average value of 23.5 permil is adequate for most approaches. However, the isotope approach is mostly based on isotope fluctuations, hence employing an average value has the potential to artificially dampen the fluctuations, resulting in an overestimation of the age. I would encourage the authors to employ a random sampling approach from a normal (?) distribution of values within this range to evaluate the effect of the soil isotopic signature. At the very least, the possible range of variability in the soil gas should be placed in context with the range of variability in the isotopic signal of the leaching water, to evaluate how significant the potential damping effect of using the average value is.
Line 250: Could the authors use a different letter to define the denitrification rate to avoid confusion with the parameter k introduced in line 174?
Table 1: the entries in the parameter column should be formatted consistently, with consistent use of subscripts etc.
Figure 3: why are d15N presented when they aren’t relevant for the analysis (see above)? And what is the cause of the high variability in isotope values of N mid 2019? These seem to be the same four outlier points in Figure 4. Is it likely that these samples aren’t useful/something went wrong during the analysis/handling?
Figure 3: Please add in the caption that these are measured values.
Line 315-317: It would be helpful if an explanation was added here of how a case of denitrification would be identified based on the monthly distributions.
Line 322: please add the statistical significance (p-value) of the Pearson correlation coefficient.
Figure 6: which of the boxes indicates the “measured range of d18o-N in streamwater”? The first one that is labelled NO3? And which ones are calculated? And how? Do the authors mean modelled? Please clarify.
Line 345: change to “range in nitrate in streamflow”.
Lines 341-654: This is a lot of text to explain that denitrification likely didn’t play a major role at this site. While this is an important explanation, I wonder if it could be communicated in a more concise way.
Figure 7b: I would encourage that the authors add the R2 value to panel b (like they did in panel a), to support the statement they make in the text, i.e. that the relationship is essentially non-existent.
Line 373: “highest proportion of young water”. What does this refer to? Do the authors mean young water fractions? I found no place where they were actually calculated, even though he sinusoidal fit in Figure 8 implies that they may have been thinking along these lines. If so, then I would caution against them in this context, because they are not the right measure to establish short-term changes.
Lines 384-385, and 458-460: I’m not certain that the findings here are logical. After a very dry summer like 2018, the older water stores should have been depleted, meaning there would be expected to be MORE young water in streamflow after a precipitation event following a dry spell. There is also higher variability in streamwater isotopes in 2019, supporting shorter travel times (less damping means less mixing). For nitrate, I would expect the opposite: Nitrate gets added to the surface even while it’s dry, hence the nitrate clock starts ticking, resulting in older (compared to water) nitrate ages after droughts. The findings here are opposite to this, and the explanation isn’t completely coherent.
Technical comments:
Line 63: I think the authors mean “along its flow path” rather than “on its flow path”.
Line 65: delta symbols are not displaying correctly here.
Line 193: should be “through” rather than “leaving” the routing.
Line 284: delta symbols are not displaying correctly here.
Citation: https://doi.org/10.5194/hess-2024-109-RC1 -
AC1: 'Reply on RC1', Christina Radtke, 08 Aug 2024
Dear Reviewer, thank you very much for your valuable feedback. In the following, we would like to answer your comments:
Specific comments:
Abstract: the mention is of 5 years of low frequency data followed by 3 years of high frequency data. All figures only show five years of data. This is somewhat inconsistent and should be clarified.
Answer: We are very sorry for this mistake and we will correct it in the revised manuscript.
Lines 52-63: these are very relevant processes, and it might be helpful for readers if this could be illustrated in a schematic plot.
Answer: We will think about a schematic plot to visualize the processes. Granger & Wankel (2016) already gave a good overview of the processes and how the oxygen isotope of nitrate is incorporated during the processes in comparison to the nitrogen isotope of nitrate.
Line 71: the authors mention here that mobilisation as nitrate “can take up to decades”. This is an important point, and I am glad it is being made. But it is not entirely clear to me how they constrain these inputs and their enormous temporal variability and uncertainty in their model (this is not really explained in the methods – at least not prominently, so I missed it if it was). I would encourage the authors to explain this more clearly in their methods.
Answer: In the revised manuscript, we will add more information. In general, the model set up allow us to track the isotopic signature of nitrate since its forming in the soil until it is released to the stream. That means when for instance organic fertilizer is applied on agricultural fields, it will take some (unknown) time to form nitrate in the soil that can be taken up by plants. We can track the forming processes by using isotopic signatures that are incorporated during the nitrification (as shown in Granger & Wankel, 2016). It would be extremely challenging to quantify how long it takes from the field application to the formation of nitrate because many factors influence the transport and processes in the soil until nitrate is formed. Therefore, we decided to focus on the transport of nitrate since it has been formed in the soil. This would act as an important initial step moving to resolve the full lifetime of catchment nitrate dynamics.
Line 80: hypothesis (i): can the authors add an explanation of why they expect this temporal offset?
Answer: The water transit time is tracked with conservative tracers and water does not underly short-term changes in their molecule structure. Instead, nitrate is degraded along its flow path, therefore we assume that a temporal offset between transit times of water and of nitrate will occur due to the degraded nitrate (mostly the oldest nitrate is degraded).
Line 107: please add units to the kf values.
Answer: We will add m/s as units to the kf values.
Lines 107/108: Can you provide information on the depth of the storage/aquitard? This would be helpful information for some of the conclusions drawn later, regarding the contributions from shallow and deeper storage.
Answer: Unfortunately, we don’t have specific measurements of the depth of the storage/aquitard.
Line 118: I find it a bit confusing and distracting that delta15_N values are reported and presented in great detail here, after it is explained in the introduction that it’s the oxygen isotopes of nitrate that are of relevance. Consider adding a note here that only oxygen isotopes are analysed.
Answer: We want to consider both isotopes in our analysis. Especially, for the evaluation of possible sources, the isotopic signature of nitrogen is necessary. Therefore, we would like to present both isotopes here. For the revised manuscript, we will keep in mind to add a note regarding the usage of oxygen isotopic signatures for the modelling part.
Section 2.2: The separate description of the water isotope and nitrate sampling is quite confusing here. It appears that these span the same periods, but it is not explained if these are actually the same samples or samples collected at totally different times during these intervals. If the latter is the case, then it would be useful to present a flow duration curve, to illustrate that similar conditions are covered by the two different sample sets. Else this would question comparability between the water and nitrate travel time analyses.
Answer: We are sorry that this is so confusing. We set up one autosampler to sample water from stream water. Usually each sampled water is used to analyse water isotopic composition as well as the nitrate isotopic composition of the nitrate in that water. For the determination of water isotopes, we only need 2ml, instead for nitrate isotope we need minimum 60ml (after filtration). In addition to that we also measured nitrate concentration in stream, for that we also needed 60ml. At some days of sampling we didn’t have enough sampled water from the stream to conduct both analyses. Therefore, we have water isotopic signatures of each sample, but nitrate isotopes only from most of these samples. Just to be clear, one example: Considering 3 days yields 3 different water samples. On day 1, we sampled 250ml of water. On day 2 we sampled 150ml. On day 3 the water level was very low and the sampled water was under 40ml. With these volumes we were able to measure water isotopes, nitrate isotopes and nitrate concentrations on day 1 and 2. But on day 3, we only had enough water sample to analyse the water isotopes.
Figure 2: This figure tries to convey a lot of information, and I’m unsure it helps. I am finding it slightly confusing at least and would recommend simplifying it. I’m not sure what the incorporation of solutes into the “legend” on the left means (e.g. for “leaching” or “discharge”). Does it imply that these compounds are measured or modelled in these fluxes? It would also be helpful if it was explained which of these fluxes and concentrations are measured, and which ones are modelled. Including the relevant equations from the routing storage is likely more confusing than helpful.
Answer: With figure 2 we are showing the conceptual model and the fluxes of water and nitrate. This figure is only to help to explain the model structure. For understanding the model structure in general, it is not necessary to know which flux is measured or modelled. In this particular case, all fluxes are modelled as described in the same section starting in line 205. The measured isotopic signature (of water and nitrate) in stream are used for calibration.
Section 2.4 (Model setup): I think it is very useful that the authors are trying to provide a good explanation of the used SAS approach. However, to be truly beneficial, a bit more information needs to be added. For example, it is not explained anywhere what the backward transit time distribution is (line 165) or what the implications and definitions of time variant and time invariant SAS are (line 172). The variables T and deltat are also not explained (equations 2+3), and neither is the wetness index (line 177), or the parameters kQ1, kQ2, kL1, and kL2 (lines 180, 181).
Answer. The parameters are described in table 1, including the parameter name, parameter description and the range of values that have been used during the optimization process. For the revised manuscript we will consider to write out that T stands for transit/elapsed time and t is the timestep/iteration. The delta values are an usual mathematical expression for the change, therefore we assume that is basic knowledge that has not necessarily been mentioned.
Line 219: “as a time stamp” – this would require that every point in time has a unique isotope signal, but this isn’t the case as the isotope signal is variable over time. Instead, a better wording may be “used as a finger print” or similar.
Answer: We agree with you and will change it in the revised manuscript.
Lines 217-219: This is a really helpful explanation, but I would recommend bringing it much earlier to limit confusion that arises before this point regarding the methods and sampling.
Answer: We find it makes sense to mention the usage of oxygen isotopic signatures here and not earlier, because we present and discuss the measured nitrogen isotopic signatures as well as the oxygen isotopic signatures of nitrate.
Lines 224-258 (and lines 505-506): it may well be true that an average value of 23.5 permil is adequate for most approaches. However, the isotope approach is mostly based on isotope fluctuations, hence employing an average value has the potential to artificially dampen the fluctuations, resulting in an overestimation of the age. I would encourage the authors to employ a random sampling approach from a normal (?) distribution of values within this range to evaluate the effect of the soil isotopic signature. At the very least, the possible range of variability in the soil gas should be placed in context with the range of variability in the isotopic signal of the leaching water, to evaluate how significant the potential damping effect of using the average value is.
Answer: We acknowledge your suggestion. We intensively investigated how the isotopic signature of nitrate changes according to a varying soil isotopic signature. In natural systems, there is a changing soil isotopic signature during the year (like the sinusoidal variation of precipitation isotopic signatures). The processes behind the variation are not only limited to temperature, instead many other factors such as microbial activity, uptake by plants, assimilation and more are affecting the isotopic signature of oxygen isotopes of soil air. Therefore, it if very uncertain to use varying isotopic signatures of soil air. One could also argue that with the varying artificial isotopic signature of soil air, we would influence our results to (probably) better values. In the case of measured isotopic signatures of soil air, it would be a big benefit to improve the model based on real data. Considering these facts, we decided to rely on literature values of soil oxygen isotopic signatures that have been intensively tested in different studies (e.g. Kendall, 2009; Kool et al., 2011; Boshers et al., 2019) and yielded satisfactory results for nitrate isotopic signatures during nitrification.
Line 250: Could the authors use a different letter to define the denitrification rate to avoid confusion with the parameter k introduced in line 174?
Answer: Thank you very much for pointing that out. We will think about another parameter letter for the revised manuscript.
Table 1: the entries in the parameter column should be formatted consistently, with consistent use of subscripts etc.
Answer: Thank you for pointing that out, we will correct that in the revised manuscript.
Figure 3: why are d15N presented when they aren’t relevant for the analysis (see above)? And what is the cause of the high variability in isotope values of N mid 2019? These seem to be the same four outlier points in Figure 4. Is it likely that these samples aren’t useful/something went wrong during the analysis/handling?
Answer: The isotopic signature of nitrogen isotopes is relevant to investigate the source of nitrate (Fig. 4). Therefore, we want to keep the isotopic signatures of nitrogen in our manuscript.
In 2019, we didn’t have enough water sample to analyse the nitrate concentrations of each sample. Instead, we focused on analyses of nitrate isotopic signatures. That’s why there are some missing values of the measured nitrate concentration in 2019. The increasing nitrate isotopic signatures can be a hint for denitrification processes that occurred, but without the corresponding nitrate concentrations we are not able to validate them. With this figure we want to show that there are some missing data, that makes it hard for further analyses and resulting assumptions.
Regarding the analysis, we are sure that the measured data shown in this manuscript is correct and have been checked before further usage.Figure 3: Please add in the caption that these are measured values.
Answer: We will add this information in the revised manuscript.
Line 315-317: It would be helpful if an explanation was added here of how a case of denitrification would be identified based on the monthly distributions.
Answer: In line 331 we mention that depleted nitrate concentrations and increasing nitrate isotopic signatures are an indicator for denitrification. We first showed and presented the figure and after that we directly discuss what we find there. Adding this information already at lines 315-317 would be a repetition.
Line 322: please add the statistical significance (p-value) of the Pearson correlation coefficient.
Answer: We will add the p-value in the revised manuscript.
Figure 6: which of the boxes indicates the “measured range of d18o-N in streamwater”? The first one that is labelled NO3? And which ones are calculated? And how? Do the authors mean modelled? Please clarify.
Answer: In lines 341ff we explain what we did. For the revised manuscript we will extend the description.
What we did: The first box with the name “NO3” is the measured isotopic signature of delta 18O-NO3 in stream water. The other boxes are results of the calculation using equation 8 (nitrified nitrate isotopic signature), for that we used measured isotopic signatures of delta 18O of e.g. precipitation water and the fixed isotopic signature of soil air (23.5) to compute the range of possible values if the precipitation would have been the source for delta 18O-H2O during the nitrate formation. That means, based on measured values that we have (delta 18O-H2O) we use the equation 8 to get an idea which possible water sources were used during nitrification.Line 345: change to “range in nitrate in streamflow”.
Answer: We will change it in the revised manuscript.
Lines 341-654: This is a lot of text to explain that denitrification likely didn’t play a major role at this site. While this is an important explanation, I wonder if it could be communicated in a more concise way.
Answer: Denitrification plays a major role for the shift of isotopic signatures of nitrate. Therefore, we think it is necessary to mention and discuss why in our catchment, the denitrification does not play a major role.
Figure 7b: I would encourage that the authors add the R2 value to panel b (like they did in panel a), to support the statement they make in the text, i.e. that the relationship is essentially non-existent.
Answer: It is not the aim of dual isotope plots to evaluate how strong the correlation is, therefore there is usually no R2 mentioned in the dual isotope plot. What we wanted to see in the dual isotope plot of nitrate is the slope of the distribution of the isotopic signatures of nitrate. As described in lines 361ff we don’t see the slope of isotopic signatures that would show a denitrification, therefore it wouldn’t make sense to plot a line that could have been the denitrification line, because it could irritate readers.
Line 373: “highest proportion of young water”. What does this refer to? Do the authors mean young water fractions? I found no place where they were actually calculated, even though he sinusoidal fit in Figure 8 implies that they may have been thinking along these lines. If so, then I would caution against them in this context, because they are not the right measure to establish short-term changes.
Answer: We will correct this part in the revised manuscript. We were not aiming to describe young water fractions. With lowered median transit times one can assume higher proportions of young water compared to time steps with higher transit times. In this section we haven’t describe it so clear, therefore we will revise this part.
Lines 384-385, and 458-460: I’m not certain that the findings here are logical. After a very dry summer like 2018, the older water stores should have been depleted, meaning there would be expected to be MORE young water in streamflow after a precipitation event following a dry spell. There is also higher variability in streamwater isotopes in 2019, supporting shorter travel times (less damping means less mixing). For nitrate, I would expect the opposite: Nitrate gets added to the surface even while it’s dry, hence the nitrate clock starts ticking, resulting in older (compared to water) nitrate ages after droughts. The findings here are opposite to this, and the explanation isn’t completely coherent.
Answer: We mention in these lines that there is more old water in summertime and during/after drought conditions. As we discuss in the lines 385ff, we assume that the replenishment of water took a few months, which was also found by other studies (Smith et al., 2020; Kleine et al., 2020). That would mean, that water from previous precipitation events took a while for infiltrating in the soils. This observation can differ for different catchments. We analysed the release of young water in several catchments in another study, that is currently under review.
Considering the isotopic signature in stream water in 2019, we can see a variability of the measured isotopes between -8 and -10, while he variability is smaller in the beginning of 2019 and increases to the end of 2019, when wetter conditions occur. Considering only the measured values and the dampening, we can see that the isotopic signatures of stream water are quite damped until autumn 2019 and they start to react slowly on a precipitation event in late summer. From that graph, we can not say how transit times behave in general. But we expect that with the long dampening, there is a mixing and filling up of the storage over the first 6-8 months of 2019, which is in line with findings by Smith et al. (2020).
Considering nitrate, usually the fertilizer that is causing high loads of nitrate in surface waters is applied on the fields as organic fertilizer or mineral fertilizer. It takes some time to degrade the products and until nitrate is formed in the soils, therefore our model considers only the time since the forming of nitrate until it is released to the stream. In addition to that, some conditions (oxygen demand, temperature) must be met so that nitrate can either be formed or degraded. Moreover, with longer transit times of nitrate (non-conservative) it will be degraded (or taken up by plants) and therefore the concentration will decrease along the transport. Most likely, the older nitrate underly stronger denitrification (degradation) than the younger nitrate, therefore the older contributions of nitrate are reduced and that is what we can see at the nitrate transit time as well: the old nitrate is reduced, more young nitrate is in the system, which can be released to the stream. This yields in lowered nitrate transit times compared to water transit times.Technical comments:
Line 63: I think the authors mean “along its flow path” rather than “on its flow path”.
Answer: We will change it in the revised manuscript.
Line 65: delta symbols are not displaying correctly here.
Answer: Thanks for mentioning, we will correct that.
Line 193: should be “through” rather than “leaving” the routing.
Answer: “Through” would be a very different meaning compared what we aimed. Here, the focus is on the transit time distribution of the water that is flowing out of the system, therefore “leaving” is a suitable word at this point. Substituting it with “through” would be irritating and could be the water and its transit time distribution that is still stored in the routing storage.
Line 284: delta symbols are not displaying correctly here.
Answer: Thanks for mentioning, we will correct that.
Citation: https://doi.org/10.5194/hess-2024-109-AC1
-
AC1: 'Reply on RC1', Christina Radtke, 08 Aug 2024
-
RC2: 'Comment on hess-2024-109', Anonymous Referee #2, 11 Aug 2024
- AC2: 'Reply on RC2', Christina Radtke, 06 Sep 2024
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 345 | 102 | 44 | 491 | 20 | 12 |
- HTML: 345
- PDF: 102
- XML: 44
- Total: 491
- BibTeX: 20
- EndNote: 12
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1