the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Combination of traditional nutrient load analysis and storm hydrograph separation unveil unexpected patterns in event-driven nutrient export dynamics in a rural headwater catchment
Abstract. First flush and dilution are major effects on solute export dynamics during precipitation events in headwater catchments but are hard to predict, even if catchment properties are well known. Normalized cumulative load (NCL) functions have been used to visualize and classify event-based discharge–load relationships, distinguishing between dilution, flushing, and linear export behavior. This study presents an enhanced version of the classical NCL function approach by combining it with hydrograph separation. Over an 18-month period, discharge and solute concentrations were monitored in an agriculturally influenced headwater catchment in the German low mountain ranges, with a focus on nitrate (NO3−) and total phosphorus, and a complementary dataset of major ions. Discharge was separated by using stable water isotope signals into event water and total discharge. Both discharge components were then analyzed for solute loads (NO3−, total phosphorus, and major ions). The results reveal significant differences in solute export dynamics between event water and total discharge, including unexpected similarities in the export patterns of nitrate and phosphorus. The proposed method also highlights a shift from predominantly linear export behavior in the total discharge (coefficient of variation = 0.13) to more pronounced first flush or dilution patterns in the event water (coefficient of variation = 0.36). These findings indicate a fundamental difference between the discharge processes governing the solute export dynamics of the catchment. While the signal of total event discharge indicates linear behavior, the separated event water exhibits strong flushing or dilution tendencies, likely linked to the activation of drainage systems and draining of NO3− legacy storages. The proposed method is straightforward to implement, yields statistically robust results for the dataset and provides new insights into solute input pathways in headwater catchments.
- Preprint
(4093 KB) - Metadata XML
-
Supplement
(677 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-3528', Anonymous Referee #1, 22 Sep 2025
- AC1: 'Reply on RC1', Lukas Ditzel, 16 Oct 2025
-
RC2: 'Comment on egusphere-2025-3528', Anonymous Referee #2, 24 Sep 2025
General Comment
The authors investigated the dynamics of nutrient export in a headwater catchment by combining hydrograph separation and nutrient load analysis. They expect this method to provide further insight into these dynamics, which are important for understanding the impact on ecosystems.
The authors monitored 15 precipitation events and took biweekly grab samples over a period of 1.5 years at the outlet of a 3.2km² headwater catchment area used mainly for agriculture. They measured stable isotopes, NO3-, Ptot, EC and major ions in the discharge. A weather station was also set up during the sampling period. This resulted in a limited dataset, yet one that was sufficient for the applied statistical analysis.
The results show large differences between total and event water export. Using k-means cluster analysis, the authors identified parameters influencing nutrient export dynamics that were not apparent from examining total discharge characteristics alone. The volatility of an event described here integrates various parameters for identifying patterns in nutrient export.
Overall, the authors gained further insights by combining two standard methods, which could be applied to other catchments. The dataset and its transfer to other catchments could be discussed in more detail. Specific comments are given below.
Detailed comments
Line 69: If applicable, please change “rural” to “agricultural area”.
Line 81: Please revise the heading, possibly change to “Sampling and measurements” to illustrate the content.
Lines 124-125: Please check that all parameters are listed correctly and use the same terms as in line 126.
Line 103: Please revise the heading, as this section also covers the clustering of events, not just hydrograph separation. Alternatively, move the second part to a separate section.
Line 196/ Table1 :The pre-event wetness in the catchment (API) appears to be a very important parameter for the subsequent analysis. Please add this to the table.
Line 215-217: Please revise the sentences to avoid repetition.
Line 225/ Fig. 5: Add a * if the clusters are significantly different.
Line 265: It is mentioned here that the red events might have taken place in seasons when no crops were grown or the fields were compacted. However, based on Fig. 4b and Table 1, the red events (3,5 and 11) took place in May, June and November so I cannot see any clear pattern. Please elaborate on this reasoning in the text, and be careful not to draw conclusions based on a very limited sample size.
Also, the events in Table 1 were not equally distributed throughout the year, which may introduce a bias towards spring and summer samples. Please discuss this briefly.
Lines 274-276: Please revise sentence.
Line 349: Could the influence of the drainage system on nutrient export be quantified? Is there any data available indicating the percentage of the catchment area with a drainage system? Similarly, this question could be extended to the crop type in the catchment over the years. In general, more information on the catchment’s specific agricultural land use would be helpful in order to understand the possible nutrient export.
Lines 375-389: Possibly add recommendations for transferring this method to other catchments and sampling campaigns.
Line 399: Do you mean general knowledge or is there a specific point, that you would like to highlight to decisionmakers?
Line 407: “given the method’s sensitivity to limited sample size”: was this tested here?
Citation: https://doi.org/10.5194/egusphere-2025-3528-RC2 - AC2: 'Reply on RC2', Lukas Ditzel, 16 Oct 2025
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 1,468 | 43 | 18 | 1,529 | 26 | 22 | 22 |
- HTML: 1,468
- PDF: 43
- XML: 18
- Total: 1,529
- Supplement: 26
- BibTeX: 22
- EndNote: 22
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
In the manuscript entitled “Combination of traditional nutrient load analysis and storm hydrograph separation unveil unexpected patterns in event-driven nutrient export dynamics in a rural headwater catchment”, Ditzel et al. analysed nitrate and total phosphorus export patterns, using normalised cumulative load (NCL) functions and hydrograph-separation into event-Q and total Q via the isotopic signature of Q and precipitation. From my perspective, the approach is promising, especially the combination of export patterns and event-water separation, and the results have the potential to reveal new insights into catchment functioning. However, I have several major points that I consider crucial to ensure the validity of the results and to enable readers to understand them and draw the correct conclusions.
Major comments:
Event water and solute concentrationsExport patterns of N and P were analysed for total Q and event-Q. However, it is not explained which concentrations have been assigned to event-Q. Assuming that pre-event Q (or everything that is not event-Q) has a constant concentration, which (if I got it right) was also the underlying assumption for event-Q estimation via the isotopic signal, every change in concentration must be attributed to the event water. This way, even a small increase in the concentration could indicate a much higher concentration in event-Q compared to pre-event Q, especially if event-Q makes only a small percentage of total Q. How was that handled in the analysis? As there is nothing explained in this regard, I assume that the same concentrations as measured in total Q were also assigned to event-Q, and I find this conceptually questionable.
In line with that, I miss a paragraph discussing the uncertainties of the event-Q calculation and what that might mean for the interpretation of results. Also, what happens if water from shallow flow path gets mobilized that has a different isotopic signature? This should at least be discussed.
Further, I strongly recommend reading von Freyberg et al. (2018) on the topic of event water calculation using stable isotopes.
1. Explanation of NCL and comparison to other approaches
While I enjoyed reading the section on how measurements were taken, other parts of the methods lack details that allow the reader to follow how exactly the methods were conducted. This is true for the event-Q calculation, but also for the NCL method. It would further help he readers to clearly distinguish what separates this method from the power law relationship between C and Q (C = aQb), or the respective L-Q relationship (L = aQb+1), for example. As C-Q and L-Q are, from my experience, more commonly applied, readers need a good reason to be convinced that the alternative method presented here is a good alternative, at least for specific cases or questions.
2. b --> 0 (dilution)
I do not agree with the definition of b à 0 for dilution patterns, as it is currently indicated in Figure 3. I might be mistaken, but why should a negative value of b not be possible? For the C-Q relationship described as C=aQb, a negative b indicates dilution. Translated into loads, it becomes L = aQb+1, which means that b < 0 implies very strong dilution, so strong that despite discharge going up, loads go down. This can only happen if the baseflow (or pre-event Q) concentration decreases as well, which might be rather unlikely, but it is not entirely impossible. Consequently, it should be b --> -∞ for dilution. I am happy to be proven wrong, but I recommend checking this carefully.
3. Start and end point of events
I could not find a description of how the start and end points of events were defined. However, I find this important, as this has the potential to severely influence the results, especially with respect to the percentage of event-Q. This needs to be clarified, and its impact on the results should be carefully checked.
4. Discussion relevance and implication
I would appreciate to hear a little more about the relevance of the topic. Why does it matter? Yes, these patterns were observed, but what does that imply? This applies to the abstract, the discussion and conclusion.
5. Data availability
I do not see a reason why data from this study should be available upon request and not uploaded to an open repository. If there is an acceptable reason for that, it should be stated in the data availability section. Otherwise, I advocate for a transparent and easily accessible provision of the data so others can replicate he presented results
The following comments are more about precision in terminology and (hopefully) easily addressed. I still add them as major points here, as I consider them important for readers to understand the results presented:
6. First flush
I got confused by the use of the term first flush and flushing behaviour in the manuscript. It appeared to me that both were used as a synonym for what in other studies is called an enrichment or accretion pattern, meaning that concentrations increase with increasing discharge. In other cases, it appeared to describe an earlier peak of concentrations as compared to discharge, which one could call first flush, or which others have described by clockwise hysteresis. I might have overread things in this regard, but the manuscript would benefit from a clear definition of what term refers to what and how these different patterns are distinctively characterised via NCL. It would also be good to clearly distinguish “first flush” from enrichment (or flushing?) behaviour, but also distinguishing it from the “first flush” that describes a disproportionally high concentration increase during the first event(s) after a drought (e.g., Winter et al., 2022).
7. Linear vs. chemostatic
In the manuscript, constant (or chemostatic) solute dynamics are described as “linear”. While I understand that this term makes sense from the perspective of the NCL approach, it is somewhat confusing to readers who are more familiar with C-Q or L-Q relationships in the form of a power law relationship. There, enrichment, chemostasis, as well as dilution are linear in the log-log space. Hence, I suggest using a different terminology.
8. Loads vs. concentrations
Throughout the manuscript, solute dynamics are often referred to as nitrate or total phosphorus, without indicating whether this is about loads or concentrations. As this makes a huge difference, also in the way results are interpreted, I recommend clarifying this throughout the manuscript.
Minor comments:
Title: I suggest either saying “The combination” or “Combining traditional…” with a tendency to the second for brevity reasons. As it is now, it reads a little odd. No point is needed at the end of a title.
Abstract
I struggle with the causal relationships in the first sentence. Is a first flush or dilution an effect of solute export dynamics? Also, see my major comment regarding the use of the term “first flush”.
L15: NO3- should be formatted to NO3- throughout the manuscript
L19: what are “discharge processes” – I suggest referring to hydrological processes here, if this should refer to transport processes and not biogeochemical ones.
Introduction
L26: “Nutrient cycle”? This sounds like a cycle of biochemical transformations. I assume this should rather be something like nutrient storage and transport within and from catchments?
L28: “nutrient and other solutes concentrations” needs adjustment: it is nutrient/solute concentrations that are measured, and nutrients are solutes as well, not either or.
L32. The point is missing.
L46: Musolff et al. and Ebeling et al. use the exponent of the power law relationship between C and Q. Not NCL, this should be distinguished.
L50-51: Include insights from von Freyberg et al. (2018)?
Methods
L67: Central Uplands, Germany.
L72: I suggest referring to the world reference base and not to the German one (i.e., brown earth). I guess it is Cambisol?
L75: Please add the year for which the mean was calculated
Fig.1: What are the black lines in the land use map? Are these tile drains or just the borders between polygons? If the latter case, they should be removed. Also, the north arrow is missing
L82: besides à except
L87: What is the quality of the calibration? I would like to see a plot comparing sensor measurements and grabs samples with a 1:1 line and R² or similar in the supplement
L93: I appreciate Fig S1. Still I would have liked to see something that gives me an idea of how noisy discharge data was and how it looked like after the correction.
L98: Where was precipitation measured and how? Can this be displayed on the map as well?
L104: How does this compare to the methods of von Freyberg et al. (2018)? How were deuterium concentrations in P estimated? As a weighted mean? Also, how was the pre-event concentration estimated? Is it a mean across several values, if so which values?
L110-112: Nice!
L115: “using” not “by using” here and elsewhere. It does not really help to know that it is a “classical method”, common is enough.
L123: what is the R base package? If it is an additional package, it should be cited. Otherwise, “computed in R (R core team…)” is enough.
L125: what is the unit of the mean rate of change?
L133: With k equal to
L139 here and elsewhere: “=” should be written out outside of an equation
L152: Terminology is not consistent. It should be either “nitrate” or “NO3-“. Further, if the minus is added, minus and plus also need to be added to all other ions (e.g. Ca2+, etc.)
L153: “Beginning with” sounds odd. Maybe just: Event water and total discharge from automated sampling were compared…
Eq.: 6: and F(X) is the concentration? Or the load? Why not say this directly? Also, Q would be the more intuitive abbreviation than X
Eq. 7: and here y is used instead of f(x), right? I recommend using the same (and ideally more intuitive) symbols in all equations.
L186-188: This statement is too general. Especially as it is underlain by the citation of two studies that only span a hand full of catchments. If any, a large sample study should be cited here. For example, across Germany, Ebeling et al., (2021) show that N tends towards enrichment patterns (increase in C with increasing Q, due to the mobilisation of diffuse sources with increasing catchment wetness) and dilution P towards dilution patterns (decrease in C with increasing Q), due to the dilution of point sources. This is, if I get it right, the opposite of your statement. Note that Ebeling et al. looked at long-term patterns from low-frequency data, and that patterns between these time scales can diverge (Winter et al., 2024). However, Winter et al. (2024) showed that the tendency towards enrichment or dilution remains the same, only less pronounced, during events.
Results & Discussion
L192: DWD 2024 à I could not find that reference in the reference section. I am not sure if it would make more sense to name this data source in the method section and remove it here?
Table 1: It would be beneficial to add total Q to the table as well and to specify if “date” refers to the starting date of an event.
L232: From the literature (von Freyberg et al., 2018) and also intuitively, I would have expected a larger event water fraction during larger storms, as during smaller events, a higher percentage of the water fills up empty storages. The manuscript would benefit from a more detailed discussion on why the results diverge from this and from comparing their results to the literature.
L249: e.g. seems to be missing in the citation, as these are just exemplary references
Figure 6 & L257: I guess it is API14?
L264-265: see my comment above (L232). Was there a seasonal difference in the events analysed? If not, I am not entirely convinced by this argument.
L283-284: Why is it likely to be mobilised, and why would it “normally” have a chemostatic export behaviour? Winter et al. is a good citation here, as it comes from a study comprising a comparably high sample size. However, the authors showed that event patterns are closer to chemostasis as compared to long-term patterns, not necessarily that all events are chemostatic.
L290-294: I assume this is largely because the event water shows a different dynamic compared to total Q?
L305: I assume Ptot and not phosphorus?
L306: Is that necessarily a dependency?
Fig.7: The labels on the right are not needed, as all information is already provided on the left. Maybe lines in the plot could be colored for the different events so that readers can see if the direction of changes remains similar for the same events.
L315: I am not convinced the difference is unexpected (see my comment to L290-294 and my major comment regarding event-Q).
L344: A deeper look into the literature would show quite a few studies where such patterns have been found (e.g., Dupas et al., 2016; Winter et al., 2021, …)
L332-334: This explanation is needed earlier in the manuscript + additional explanations (see major comments on first flush)
L351: Shouldn’t this be introduced in the method section already?
Table 2: I recommend adding the ratio (CVC/CVQ) here as well. It would enable a nice comparison to studies such as those from (Musolff et al., 2015).
L370: The drainage system is likely to have a strong impact on the results observed. A deeper discussion on this, also in comparison with other studies with and without such systems, could potentially add much value to the discussion.
L382: What is “sufficiently large”? Can this be specified?
L394-396: And this would not be possible with other methods?
References: