the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 11 Apr 2025
Aerodynamic gradient flux measurements of ammonia in intensively grazed grassland: temporal variations, environmental drivers, methodological challenges and uncertainties
Abstract. Understanding the factors controlling surface-atmosphere exchange of ammonia (NH3) in grazed grasslands is crucial for improving atmospheric models and addressing environmental concerns associated with reactive nitrogen. However, in-situ micrometeorological NH3 flux measurements in pastures remain scarce in the literature. This study presents high-resolution NH3 flux data collected during four spring campaigns (2021 – 2024) at an intensively managed grassland site in Northwestern France, using the aerodynamic gradient method (AGM) alongside continuous monitoring of environmental variables and agricultural management. AGM-derived NH3 fluxes exhibited distinctive patterns: (i) high variability during grazing from -113 (deposition) to +3205 (emission) ng NH3 m-2 s-1, influenced by meteorology, grazing livestock density, and vegetation and soil dynamics; (ii) strong diurnal patterns and day-to-day variability; and (iii) transient volatilisation peaks following slurry applications (up to 10235 ng NH3 m-2 s-1). Grazing-induced fluxes often persisted for up to 1–2 weeks following cattle departure. Relative random uncertainties associated with AGM flux measurements ranged from typically 15 % to 70 %, based on errors in vertical concentration gradient slopes and variables related to turbulence and stability. Additional methodological limitations and systematic uncertainties are discussed, in particular errors associated with fundamental AGM assumptions and flux footprint attribution in a rotational grazing setup. Emission factors (EF), calculated for NH3 derived from deposited cattle urine nitrogen, varied considerably between grazing events, from 1 to 23 g NH3-N cow-1 grazing d-1, reflecting the interplay between livestock management and environmental factors. This study highlights the importance of long-term, continuous, high-resolution measurements to document the large variability in grazing-induced NH3 fluxes. The findings also underscore the need for refining bi-directional exchange models that integrate physics (meteorology, turbulence), environmental biogeochemistry (the fate of excreted nitrogen in the soil), biology (dynamic vegetation processes), and pasture management (grazing intensity) in grazed grassland systems.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(2537 KB)
-
Supplement
(3261 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2537 KB) - Metadata XML
-
Supplement
(3261 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-1605', Anonymous Referee #1, 11 May 2025
-
AC2: 'Authors reply to all comments', Chris Flechard, 01 Aug 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-1605/egusphere-2025-1605-AC2-supplement.pdf
-
AC2: 'Authors reply to all comments', Chris Flechard, 01 Aug 2025
-
CC1: 'Comment on egusphere-2025-1605', Johannes Laubach, 04 Jun 2025
This is an interesting study, with a well-constructed measurement system and having collected a valuable dataset of NH3 emissions from a real-world grazing system. I would just like to make a couple of suggestions, because I feel the full potential of the dataset has not been sufficiently explored yet.
1) In my view, the authors are too focused on individual grazing events. Because conditions during these are so variable, NH3 emissions vary enormously. It is a valid point to show this, but not a very novel one, and discussion of all the factors that might explain this, without being able to provide any insights, could be shorter.
2) Neither does it make much sense to me to then derive emission factors for individual grazings. They are short-term, highly variable, and defining the end of the emission response period to the fresh N input is rather arbitrary (N not volatilised quickly may remain available and could contribute to later bouts of emissions, be they triggered by weather changes or by fresh additional inputs).
3) Instead of defining emission factors per grazing event and per cow, my suggestion is to determine total NH3 emissions and total N inputs for each measurement season, and define emission factors as the ratio of the former to the latter. (Whole years would be even better, maybe that's an idea for the future.) That should make the numbers more easily comparable with those from elsewhere, and reduce the EF's uncertainty, too. A season's sequence of grazing events would represent the combination of management practice and climatic conditions better than individual grazing events. If you find the datasets too patchy for that, maybe consider making the quality control a little less rigid?
4) The flux-gradient method requires to approximate infinitesimal concentration gradients, d_C/d_z, with finite (measured) differences Delta_C/Delta_z, and to resolve Delta_C accurately, Delta_z cannot be too small. As concentration profiles near the ground tend to be curved, this approximation is likely to introduce bias. Some of the systematic differences between height pairs that you find may be due to this. A way to reduce this would be to only compare pairs of adjacent heights with each other, avoiding the larger Delta_z. Or, a possibly much neater way to use all heights would be to fit the profile shape for each averaging cycle and then determine its slope at the height where u* is measured (to match heights used for gradient and diffusivity).
5) Please clarify if/how the differences between height pairs have been considered in the uncertainty estimates.
6) I recommend considerable shortening of the sections on recalibration of C (slopes vs comparison instrument) and on quality control in general. State briefly that these were done and provide the details as supplementary material. Most readers would like to get to the interesting results quickly!
7) Title: consider a) swapping the words "gradient" and "flux" (many know the method as "flux-gradient"), b) dropping the words following the colon, c) including "over 3 spring seasons" (as a point of difference to other studies).
8) The abstract states "2021 - 2024" but no data from 2021 are presented, unless I missed something.
9) Are you planning to use this dataset for the testing of process-based models?
Best regards
Johannes Laubach.laubachj@landcareresearch.co.nz
Citation: https://doi.org/10.5194/egusphere-2025-1605-CC1 -
AC3: 'Authors reply to all comments', Chris Flechard, 01 Aug 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-1605/egusphere-2025-1605-AC3-supplement.pdf
-
AC3: 'Authors reply to all comments', Chris Flechard, 01 Aug 2025
-
RC2: 'Comment on egusphere-2025-1605', Anonymous Referee #2, 04 Jun 2025
Review of Abdulwahab et al. (2025): Aerodynamic gradient flux measurements of ammonia in intensively grazed grassland: temporal variations, environmental drivers, methodological challenges and uncertainties
Abdulwahab and colleagues present ammonia flux measurements in grazed grassland over four consecutive seasons. They use the aerodynamic gradient method with a custom-built sampling system. Data analysis is comprehensive and covers temporal flux variability under different environmental and management-driven controls, methodological challenges and its associated uncertainties as well as estimations of cumulative fluxes and emission factors. The latter include a well-justified application of a gap-filling approach making the study highly relevant for a wide range of readership.Tackling ammonia exchange by direct flux measurements in the field under grazing management is challenging and every attempt using robust methodology is highly welcome. The authors did a great job in combining several well-established methods to be able to present a comprehensive assessment of grazing and fertilization-induced grassland NH3 exchange.
I do not have any major concerns or major criticism, but a number of (rather minor) comments and suggestions for modifications, which should be addressed before publication. Given the challenges in ammonia flux measurements and data analysis, in my opinion, this work forms a milestone in methodology and verification of NH3 emission factors for grazed grassland.
Line 126: Definition of LSU is needed. General comment: Could you briefly mention whether a correlation between FNH3 and LSU was found and elaborate a bit on herd movement, presence time in the footprint, etc.?
Line 126f: No fertilization at all in 2021, but two in 2022, is that correct?
Line 133: What does ‘hybrid’ mean in this context?
Line 135: Here and throughout the manuscript, also in some figures – the star in u_star is in the subscript, not in the superscript.
Line 170: Refer to Section 2.3.3 after ‘NH3 analyser’, otherwise people would expect more information here at this place.
Line 174f: I think the gradient lift is a well ‘thought through’ system for ammonia, particularly with regard to avoid several sampling lines. But what about highly dynamic situations? Did the concentrations not change significantly during the 200-s cycles (or those periods where data was finally taken from)? Can the authors show some raw data time series when the system was lifted to one level, stayed in a position and then lifted to the next level to illustrate/visualize the data selection and handling a bit better?
Line 191: The nominal precision of the analyzer of 0.2 ppb at 1-s integration time: Is this number from the manufacturer’s specifications or based on own tests? See also former and next comment to include some raw data examples for better illustration.
Lines 199-201: Have the authors done tests on step concentration changes? For AGM the differences between heights (gradient) matter(s), but precision of the analyzer has an impact on uncertainty.
Line 201: ‘…using half-hourly averaging’: Can the authors clarify how exactly the ‘half-hourly averaging’ is meant? For each height? Despite a lot of information, it is unclear how this is handled.
Line 207: Add ‘, Panel (c)’ after ‘Fig. 3’.
Line 215f: The whole Section 2.3.5 is a chance of reducing the length of the main part of the paper as the DELTA method is described in detail elsewhere. Just mention the exposure times in the previous section and move the rest to the supplement.
Lines 283-287: I’m confused now, mainly by the phrase ‘to assess the stationarity of NH3 concentrations over 30-minute intervals, …’. How is this exactly handled? Stationarity for each of the heights within a 30-minute window individually, i.e., just taking the respective chosen part of the 200-s interval? Otherwise, it is not clear how this is in accordance with earlier method description of the sampling scheme. Please consider rephrasing. Further, did you also check for stationarity of consecutive half hours?
Line 306f: Great approach for the correction of background flux interference!
Line 328: Check bracket imbalance at ‘(SE(Fmeas)’.
Line 368: Is the reference ‘(see Sect. S2 in the supplement)’ correct? Shouldn’t it be Figures S10-S12? Please check. Further, did the authors try to create artificial gaps in periods with high peaks to investigate whether or not the gap-filling method works?
Lines 375-376: ‘The variability in regression slopes resulted from different degrees of cleanliness and reflectivity of mirrors…’: How do you know that?
Figure 2: This figure is hard to grasp. The main message might be clear, but details are tough to extract. In general, x-y scatter plots with identical values on the axes and a 1:1 line should always come in a quadratic format so that an over- or underestimation of one variable against the other can be easily captured by the reader. Further, use color, a more readable legend and bigger symbols in Panel (a). I suggest replotting of Panel (b). The x-axis label is not readable. Try monthly boxplots or just plotting the differences against the DELTA numbers. I think that would make it visually more appealing without losing any main messages.
Line 408: Can the authors say something about how they differentiate plumes from housings vs. plumes from grazing animals?
Figure 3: Mention in the caption what qcA, etc. means (or refer to description in text). What is the reason for the high concentrations around 9th of April?
Line 431: Provide uncertainty ranges after flux numbers.
Table 2: I don’t understand how to read the line ‘LSU ha-1 (DG)’. Livestock unit per hectare is clear, but is the grazing duration in brackets given in days? If so, I can’t really recognize the number for example in Figure 3, Panel (d) for the April 2023 period. Please check.
Line 602: Is there a part of the sentence missing?
Line 694f: Is there any information on how the farmer decides when and how much fertilizer is added after grazing? Well, slurry is probably due to seasonal storage issues, but why is so much mineral N added as well?
Line 709: Here another u-star appearance, see earlier comment.
Line 718: I don’t get the logic here. Is the number of paddocks increased? If not, how can the N be more uniformly distributed? And how can NH3 emissions be mitigated under higher grazing densities?Citation: https://doi.org/10.5194/egusphere-2025-1605-RC2 -
AC1: 'Authors reply to all comments', Chris Flechard, 01 Aug 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-1605/egusphere-2025-1605-AC1-supplement.pdf
-
AC1: 'Authors reply to all comments', Chris Flechard, 01 Aug 2025
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-1605', Anonymous Referee #1, 11 May 2025
General comments
The article ‘Aerodynamic gradient flux measurements of ammonia in intensively grazed grassland: temporal variations, environmental drivers, methodological challenges and uncertainties’ is a valuable contribution to the field of ammonia research. It presents an interesting dataset on ammonia fluxes in grazed grassland and the data treatment is conducted thoroughly and described in a transparent way.
One concern, however, is the length of the manuscript which is, based on the number of lines, roughly twice as long as is common. The length of an article has a strong influence on its accessibility. I, therefore, strongly recommend the authors to reduce the length of the manuscript by removing all lines which are not absolutely necessary. I understand that this leads to tough choices since none of the sentences are not worthy to read but all together the article length is way too long. Examples of lines which could be removed in my opinion are e.g. l35 ‘Globally – productivity’, l66 ‘It -prevalent’, l132 ‘The surface -profiles’, the section describing the Delta denuder could be roughly halved etc.
My other major concern is related to the analysis part of the article, which is rather descriptive/ phenomenological and lacks depth or real new insights on explanatory factors related to emissions of grazing events. It must be possible to dig out more from such a valuable dataset. The authors explain that a multivariate regression analysis of the EF’s didn’t work out for various reasons. However, why wasn’t an multivariate regression analysis of the fluxes attempted?
In my opinion it also could be an option to divide the article in a part 1 and part 2. Part 1 consisting of the main part of the current manuscript extended with a sensitivity analysis of the choices made in deriving the fluxes (e.g the 5 sec length of the concentration time series at each height; how sensitive are the results when 10 secs are used?, the detrending procedure, the background flux correction and the gap filling procedure). Adding such a sensitivity analysis could provide valuable extra insight in the quality of the flux dataset; which is no luxury since each half hour flux value is only based on 45 sec of actually measurement data at each height. Although the data treatment appears to be carefully conducted the fact that only so few data points are used makes it hard to believe that the flux values are robust this being the last of my larger concerns); a sensitivity analysis could help to develop more trust in the flux values. Part 2 could then consist of an elaborated version of the current results section (e.g. 3.4-3.6) extended with a statistical multivariate analysis of the fluxes; perhaps combined with a modelling part.
Due to the length of the article itself I have not taken time to study systematically the supplementary material.
Specific comments
L75-l83: this paragraph is unclear to me, what is your main message?
L157: In neutral and stable conditions it is assumed… The same assumption is usually made for convective conditions as well.
L191: what is meant with the phrase ‘the nominal precision (1-sigma)’? Why (1-sigma)?
L199: In order to check whether I understand your procedure correctly: you apply linear detrending between two consecutive timeseries at the same height, hereby using the full 30-50 seconds of data? Using timeseries 1 and 2 and then 2 and 3 and then 3 and 4? Next you use the last 5 seconds of each of the 9 time series to obtain an average representative for 30 min, this is done for every height and thus obtained profile is used to determine the flux. Maybe it helps when you add words like ‘first’, ‘second ’, ‘next’ etc in the description. Or even better, add an conceptual figure on this crucial data treatment part.
L208-214: Why is this calibration necessary when only Delta NH3 are needed for the calculation of the flux? Add one sentence explaining that it is worthwhile to have reliable estimates of the absolute NH3 concentration as well.
L229: ‘near absolute mean’ What is the uncertainty (nominal accuracy) in the concentration as measured by the Delta?
Section 2.3.6: for easy reference and in order to reduce the length of the article I suggest to put the information in this section in a Table.
L256 Storage flux correction: with a time scale of 30 minutes? Based on the measurements below z_mean? Why is this discussed here? To me it appears more suitable as part of 2.5 (see also my comment on section 2.5 below)
L273-282: I suggest to shorten this section (once more to reduce the length of the article) only mentioning what you did and skipping the explanation of the flagging policy (just give a good reference for that).
L311: small or near-zero emission or deposition: why do you assume this? In figure 3 the time series of NH3 concentration shows rather high values around 9/04, 20/04 and 22/04; fluxes around this time could be substantial (however are apparently filtered out in the quality control round) so how valid is the assumption that background fluxes are small?
Eq 8 and 9: since 9 follows easily from 8 only one of the two equations needs to be presented (and when you really want to be lean both can be removed; l318-319 is clear enough about the procedure.)
Section 2.5 I miss a discussion on systematic errors here. Of course they are difficult to determine but a few can be discussed. For example the effect of a systematic error in the determination of the sampling heights (in a Delta z of 1 m an uncertainty of several centimeters already leads to a systematic bias of several percent) and displacement height. Also the (propagated) effect of an assumed but realistic systematic uncertainty in the concentration could be taken into account. Maybe the separate discussion of the theory on random uncertainties and the actual discussion in 3.5 on values of both random and systematic uncertainties doesn’t work that well. I wouldn’t mind when you combine the two and discuss them at one place.
L355: corrected -> you mean the correction for the background from section 2.4.5
L366-368 I didn’t check the supplementary material but these sentences are unclear without doing so: why are G9 and G10 different and do they need an modified gap-filling approach?
Section 3.1 How can you be sure that the results discussed in this section don’t influence the flux measurements. E.g. doesn’t the cleanliness of the mirror affect the uncertainty in the flux?
Figure 2b: to me it makes more sense to plot the results equidistant in time, just plot lines instead of bars and adjust the length of the line according to the delta time period it is represents.
Figure 3 Please add uncertainty/error bars to the NH3 concentration values in a and b.
Table 2 SE means standard error?
Figure 4 The visualization of the data in this figure doesn’t work very well. Maybe leave out the information on the quality of the fluxes (the information in Table 2 is sufficient), plot the figures 10 by 1 instead of 5 by 2 so that the time axis can be extended and present the information on the cattle differently (only when it is present and not the variation in number since the variation is most of the times not very influential; just mention the milking times in the text and not visual in the figure since you conclude it has not real influence).
L442-446 Please refrase, your main message is difficult to grasp.
L448-451 Is this a relevant observation (here)? Could perhaps skipped?
Section 3.4 A nice description of observed correlations is given but as a whole this section is a little bit disappointing because it gets stuck at this level. In section 4.4 it is stated that no statistical multivariate analysis could be applied to derive the share of EF variance explained by variables but in my opinion it is a missed opportunity that no such analysis is applied on the fluxes! Why not try such analysis for each grazing event separately and one for all cases together in order to infer variables which explain specific events and which variables are important explaining generic behavior of emission events due to grazing. Now the reader is left with huge variability in emission over de various events and no hint at an explanation.
Figure 5 Wouldn’t be worthwhile to add an figure depicting the diurnal cycle over the ten grazing events? Either normalized or not before combining, depending on what is interesting to show?
Figure 6 Why not present these results as profiles with height labeled with time instead of height and time the other way around? I find this (and other comparable plots in the article) hard to interpret. How do the soil mineral nitrogen concentrations relate to precipitation?
Figure 7 Did you make plots with all grazing events (labeled by color and aggregated in periods of ‘days since grazing’) in one figure? Doesn’t this show a pattern?
L511 Conversely- response This is rather an stand alone observation. It would be more interesting when information would be added whether this happened (albeit on a smaller scale) more often? E.g. during events G8 and G9??
Figure 8 Isn’t this a somewhat worrisome result? The measurements are done in the constant flux layer so I would hope a smaller systematic bias was found (in absolute sense). Did you check how choices made in the determination of the flux influenced this bias? Could this be used to determine the best choices regarding the flux derivation? E.g. the length of the period chosen (now 5 secs at each height)?
Table 3 How were the urinary N excreted values determined? Did I miss this somewhere or is this not explained?
Line 646-654 Since you so nicely discuss this point it is good to add that another assumption is made. K for momentum is observed to behave differently than a K for heat or moisture. It is unknown whether it is allowed to assume that a K for ammonia behaves similarly as the K for heat.
L736-743: these lines repeat information of lines l705-712; I think L705-712 can be removed. Make ‘near the surface’ more explicit; your measurement heights are already close to the surface so I guess you mean really close? A few centimeter?
L754 How was the estimated nitrogen input from grazing events determined?
L823-825 Please rewrite, this sentence is difficult the grasp by first reading.
Maybe I missed it but did you somewhere define the actual value of displacement height d? Based on figure 3 it appears to be ~ 10cm and time dependent?
Technical corrections
L50: is ‘crucial’ the right term? Don’t you mean ‘significant’?
L93: intensively grazed grassland: LSU per ha mentioned in the table are at the lower-to middle end of the numbers presented. Is it really ‘intensively’ grazed?
L95: makes possible -> enables
L174: for me the sentence is more clear when the first ‘height’ is removed.
L198: if-profile: is this not repeating the argument? Could be removed?
L201: ‘The concentration - averaging’ can be removed without loss of information.
L214: Maybe it is good to already refer to later sections where you further discuss this choice.
L452: It is also apparent that for some grazing events (e.g. …) the valid flux data capture was patchy, not necessarily – of interest (i.e. the wind was blowing from unsuitable directions). Please shorten to ‘For some grazing events (e.g. …) the valid flux data capture was patchy due to the wind blowing from unsuitable directions.
L630: ‘with’ missing in the second half?
L675 ‘horizon’ is maybe a correct (jargon) word? Or could it be replaced by layer?
Citation: https://doi.org/10.5194/egusphere-2025-1605-RC1 -
AC2: 'Authors reply to all comments', Chris Flechard, 01 Aug 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-1605/egusphere-2025-1605-AC2-supplement.pdf
-
AC2: 'Authors reply to all comments', Chris Flechard, 01 Aug 2025
-
CC1: 'Comment on egusphere-2025-1605', Johannes Laubach, 04 Jun 2025
This is an interesting study, with a well-constructed measurement system and having collected a valuable dataset of NH3 emissions from a real-world grazing system. I would just like to make a couple of suggestions, because I feel the full potential of the dataset has not been sufficiently explored yet.
1) In my view, the authors are too focused on individual grazing events. Because conditions during these are so variable, NH3 emissions vary enormously. It is a valid point to show this, but not a very novel one, and discussion of all the factors that might explain this, without being able to provide any insights, could be shorter.
2) Neither does it make much sense to me to then derive emission factors for individual grazings. They are short-term, highly variable, and defining the end of the emission response period to the fresh N input is rather arbitrary (N not volatilised quickly may remain available and could contribute to later bouts of emissions, be they triggered by weather changes or by fresh additional inputs).
3) Instead of defining emission factors per grazing event and per cow, my suggestion is to determine total NH3 emissions and total N inputs for each measurement season, and define emission factors as the ratio of the former to the latter. (Whole years would be even better, maybe that's an idea for the future.) That should make the numbers more easily comparable with those from elsewhere, and reduce the EF's uncertainty, too. A season's sequence of grazing events would represent the combination of management practice and climatic conditions better than individual grazing events. If you find the datasets too patchy for that, maybe consider making the quality control a little less rigid?
4) The flux-gradient method requires to approximate infinitesimal concentration gradients, d_C/d_z, with finite (measured) differences Delta_C/Delta_z, and to resolve Delta_C accurately, Delta_z cannot be too small. As concentration profiles near the ground tend to be curved, this approximation is likely to introduce bias. Some of the systematic differences between height pairs that you find may be due to this. A way to reduce this would be to only compare pairs of adjacent heights with each other, avoiding the larger Delta_z. Or, a possibly much neater way to use all heights would be to fit the profile shape for each averaging cycle and then determine its slope at the height where u* is measured (to match heights used for gradient and diffusivity).
5) Please clarify if/how the differences between height pairs have been considered in the uncertainty estimates.
6) I recommend considerable shortening of the sections on recalibration of C (slopes vs comparison instrument) and on quality control in general. State briefly that these were done and provide the details as supplementary material. Most readers would like to get to the interesting results quickly!
7) Title: consider a) swapping the words "gradient" and "flux" (many know the method as "flux-gradient"), b) dropping the words following the colon, c) including "over 3 spring seasons" (as a point of difference to other studies).
8) The abstract states "2021 - 2024" but no data from 2021 are presented, unless I missed something.
9) Are you planning to use this dataset for the testing of process-based models?
Best regards
Johannes Laubach.laubachj@landcareresearch.co.nz
Citation: https://doi.org/10.5194/egusphere-2025-1605-CC1 -
AC3: 'Authors reply to all comments', Chris Flechard, 01 Aug 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-1605/egusphere-2025-1605-AC3-supplement.pdf
-
AC3: 'Authors reply to all comments', Chris Flechard, 01 Aug 2025
-
RC2: 'Comment on egusphere-2025-1605', Anonymous Referee #2, 04 Jun 2025
Review of Abdulwahab et al. (2025): Aerodynamic gradient flux measurements of ammonia in intensively grazed grassland: temporal variations, environmental drivers, methodological challenges and uncertainties
Abdulwahab and colleagues present ammonia flux measurements in grazed grassland over four consecutive seasons. They use the aerodynamic gradient method with a custom-built sampling system. Data analysis is comprehensive and covers temporal flux variability under different environmental and management-driven controls, methodological challenges and its associated uncertainties as well as estimations of cumulative fluxes and emission factors. The latter include a well-justified application of a gap-filling approach making the study highly relevant for a wide range of readership.Tackling ammonia exchange by direct flux measurements in the field under grazing management is challenging and every attempt using robust methodology is highly welcome. The authors did a great job in combining several well-established methods to be able to present a comprehensive assessment of grazing and fertilization-induced grassland NH3 exchange.
I do not have any major concerns or major criticism, but a number of (rather minor) comments and suggestions for modifications, which should be addressed before publication. Given the challenges in ammonia flux measurements and data analysis, in my opinion, this work forms a milestone in methodology and verification of NH3 emission factors for grazed grassland.
Line 126: Definition of LSU is needed. General comment: Could you briefly mention whether a correlation between FNH3 and LSU was found and elaborate a bit on herd movement, presence time in the footprint, etc.?
Line 126f: No fertilization at all in 2021, but two in 2022, is that correct?
Line 133: What does ‘hybrid’ mean in this context?
Line 135: Here and throughout the manuscript, also in some figures – the star in u_star is in the subscript, not in the superscript.
Line 170: Refer to Section 2.3.3 after ‘NH3 analyser’, otherwise people would expect more information here at this place.
Line 174f: I think the gradient lift is a well ‘thought through’ system for ammonia, particularly with regard to avoid several sampling lines. But what about highly dynamic situations? Did the concentrations not change significantly during the 200-s cycles (or those periods where data was finally taken from)? Can the authors show some raw data time series when the system was lifted to one level, stayed in a position and then lifted to the next level to illustrate/visualize the data selection and handling a bit better?
Line 191: The nominal precision of the analyzer of 0.2 ppb at 1-s integration time: Is this number from the manufacturer’s specifications or based on own tests? See also former and next comment to include some raw data examples for better illustration.
Lines 199-201: Have the authors done tests on step concentration changes? For AGM the differences between heights (gradient) matter(s), but precision of the analyzer has an impact on uncertainty.
Line 201: ‘…using half-hourly averaging’: Can the authors clarify how exactly the ‘half-hourly averaging’ is meant? For each height? Despite a lot of information, it is unclear how this is handled.
Line 207: Add ‘, Panel (c)’ after ‘Fig. 3’.
Line 215f: The whole Section 2.3.5 is a chance of reducing the length of the main part of the paper as the DELTA method is described in detail elsewhere. Just mention the exposure times in the previous section and move the rest to the supplement.
Lines 283-287: I’m confused now, mainly by the phrase ‘to assess the stationarity of NH3 concentrations over 30-minute intervals, …’. How is this exactly handled? Stationarity for each of the heights within a 30-minute window individually, i.e., just taking the respective chosen part of the 200-s interval? Otherwise, it is not clear how this is in accordance with earlier method description of the sampling scheme. Please consider rephrasing. Further, did you also check for stationarity of consecutive half hours?
Line 306f: Great approach for the correction of background flux interference!
Line 328: Check bracket imbalance at ‘(SE(Fmeas)’.
Line 368: Is the reference ‘(see Sect. S2 in the supplement)’ correct? Shouldn’t it be Figures S10-S12? Please check. Further, did the authors try to create artificial gaps in periods with high peaks to investigate whether or not the gap-filling method works?
Lines 375-376: ‘The variability in regression slopes resulted from different degrees of cleanliness and reflectivity of mirrors…’: How do you know that?
Figure 2: This figure is hard to grasp. The main message might be clear, but details are tough to extract. In general, x-y scatter plots with identical values on the axes and a 1:1 line should always come in a quadratic format so that an over- or underestimation of one variable against the other can be easily captured by the reader. Further, use color, a more readable legend and bigger symbols in Panel (a). I suggest replotting of Panel (b). The x-axis label is not readable. Try monthly boxplots or just plotting the differences against the DELTA numbers. I think that would make it visually more appealing without losing any main messages.
Line 408: Can the authors say something about how they differentiate plumes from housings vs. plumes from grazing animals?
Figure 3: Mention in the caption what qcA, etc. means (or refer to description in text). What is the reason for the high concentrations around 9th of April?
Line 431: Provide uncertainty ranges after flux numbers.
Table 2: I don’t understand how to read the line ‘LSU ha-1 (DG)’. Livestock unit per hectare is clear, but is the grazing duration in brackets given in days? If so, I can’t really recognize the number for example in Figure 3, Panel (d) for the April 2023 period. Please check.
Line 602: Is there a part of the sentence missing?
Line 694f: Is there any information on how the farmer decides when and how much fertilizer is added after grazing? Well, slurry is probably due to seasonal storage issues, but why is so much mineral N added as well?
Line 709: Here another u-star appearance, see earlier comment.
Line 718: I don’t get the logic here. Is the number of paddocks increased? If not, how can the N be more uniformly distributed? And how can NH3 emissions be mitigated under higher grazing densities?Citation: https://doi.org/10.5194/egusphere-2025-1605-RC2 -
AC1: 'Authors reply to all comments', Chris Flechard, 01 Aug 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-1605/egusphere-2025-1605-AC1-supplement.pdf
-
AC1: 'Authors reply to all comments', Chris Flechard, 01 Aug 2025
Peer review completion
Post-review adjustments
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 946 | 147 | 26 | 1,119 | 54 | 18 | 33 |
- HTML: 946
- PDF: 147
- XML: 26
- Total: 1,119
- Supplement: 54
- BibTeX: 18
- EndNote: 33
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Mubaraq Olarewaju Abdulwahab
Christophe Flechard
Yannick Fauvel
Christoph Häni
Adrien Jacotot
Anne-Isabelle Graux
Nadège Edouard
Pauline Buysse
Valérie Viaud
Albrecht Neftel
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2537 KB) - Metadata XML
-
Supplement
(3261 KB) - BibTeX
- EndNote
- Final revised paper
General comments
The article ‘Aerodynamic gradient flux measurements of ammonia in intensively grazed grassland: temporal variations, environmental drivers, methodological challenges and uncertainties’ is a valuable contribution to the field of ammonia research. It presents an interesting dataset on ammonia fluxes in grazed grassland and the data treatment is conducted thoroughly and described in a transparent way.
One concern, however, is the length of the manuscript which is, based on the number of lines, roughly twice as long as is common. The length of an article has a strong influence on its accessibility. I, therefore, strongly recommend the authors to reduce the length of the manuscript by removing all lines which are not absolutely necessary. I understand that this leads to tough choices since none of the sentences are not worthy to read but all together the article length is way too long. Examples of lines which could be removed in my opinion are e.g. l35 ‘Globally – productivity’, l66 ‘It -prevalent’, l132 ‘The surface -profiles’, the section describing the Delta denuder could be roughly halved etc.
My other major concern is related to the analysis part of the article, which is rather descriptive/ phenomenological and lacks depth or real new insights on explanatory factors related to emissions of grazing events. It must be possible to dig out more from such a valuable dataset. The authors explain that a multivariate regression analysis of the EF’s didn’t work out for various reasons. However, why wasn’t an multivariate regression analysis of the fluxes attempted?
In my opinion it also could be an option to divide the article in a part 1 and part 2. Part 1 consisting of the main part of the current manuscript extended with a sensitivity analysis of the choices made in deriving the fluxes (e.g the 5 sec length of the concentration time series at each height; how sensitive are the results when 10 secs are used?, the detrending procedure, the background flux correction and the gap filling procedure). Adding such a sensitivity analysis could provide valuable extra insight in the quality of the flux dataset; which is no luxury since each half hour flux value is only based on 45 sec of actually measurement data at each height. Although the data treatment appears to be carefully conducted the fact that only so few data points are used makes it hard to believe that the flux values are robust this being the last of my larger concerns); a sensitivity analysis could help to develop more trust in the flux values. Part 2 could then consist of an elaborated version of the current results section (e.g. 3.4-3.6) extended with a statistical multivariate analysis of the fluxes; perhaps combined with a modelling part.
Due to the length of the article itself I have not taken time to study systematically the supplementary material.
Specific comments
L75-l83: this paragraph is unclear to me, what is your main message?
L157: In neutral and stable conditions it is assumed… The same assumption is usually made for convective conditions as well.
L191: what is meant with the phrase ‘the nominal precision (1-sigma)’? Why (1-sigma)?
L199: In order to check whether I understand your procedure correctly: you apply linear detrending between two consecutive timeseries at the same height, hereby using the full 30-50 seconds of data? Using timeseries 1 and 2 and then 2 and 3 and then 3 and 4? Next you use the last 5 seconds of each of the 9 time series to obtain an average representative for 30 min, this is done for every height and thus obtained profile is used to determine the flux. Maybe it helps when you add words like ‘first’, ‘second ’, ‘next’ etc in the description. Or even better, add an conceptual figure on this crucial data treatment part.
L208-214: Why is this calibration necessary when only Delta NH3 are needed for the calculation of the flux? Add one sentence explaining that it is worthwhile to have reliable estimates of the absolute NH3 concentration as well.
L229: ‘near absolute mean’ What is the uncertainty (nominal accuracy) in the concentration as measured by the Delta?
Section 2.3.6: for easy reference and in order to reduce the length of the article I suggest to put the information in this section in a Table.
L256 Storage flux correction: with a time scale of 30 minutes? Based on the measurements below z_mean? Why is this discussed here? To me it appears more suitable as part of 2.5 (see also my comment on section 2.5 below)
L273-282: I suggest to shorten this section (once more to reduce the length of the article) only mentioning what you did and skipping the explanation of the flagging policy (just give a good reference for that).
L311: small or near-zero emission or deposition: why do you assume this? In figure 3 the time series of NH3 concentration shows rather high values around 9/04, 20/04 and 22/04; fluxes around this time could be substantial (however are apparently filtered out in the quality control round) so how valid is the assumption that background fluxes are small?
Eq 8 and 9: since 9 follows easily from 8 only one of the two equations needs to be presented (and when you really want to be lean both can be removed; l318-319 is clear enough about the procedure.)
Section 2.5 I miss a discussion on systematic errors here. Of course they are difficult to determine but a few can be discussed. For example the effect of a systematic error in the determination of the sampling heights (in a Delta z of 1 m an uncertainty of several centimeters already leads to a systematic bias of several percent) and displacement height. Also the (propagated) effect of an assumed but realistic systematic uncertainty in the concentration could be taken into account. Maybe the separate discussion of the theory on random uncertainties and the actual discussion in 3.5 on values of both random and systematic uncertainties doesn’t work that well. I wouldn’t mind when you combine the two and discuss them at one place.
L355: corrected -> you mean the correction for the background from section 2.4.5
L366-368 I didn’t check the supplementary material but these sentences are unclear without doing so: why are G9 and G10 different and do they need an modified gap-filling approach?
Section 3.1 How can you be sure that the results discussed in this section don’t influence the flux measurements. E.g. doesn’t the cleanliness of the mirror affect the uncertainty in the flux?
Figure 2b: to me it makes more sense to plot the results equidistant in time, just plot lines instead of bars and adjust the length of the line according to the delta time period it is represents.
Figure 3 Please add uncertainty/error bars to the NH3 concentration values in a and b.
Table 2 SE means standard error?
Figure 4 The visualization of the data in this figure doesn’t work very well. Maybe leave out the information on the quality of the fluxes (the information in Table 2 is sufficient), plot the figures 10 by 1 instead of 5 by 2 so that the time axis can be extended and present the information on the cattle differently (only when it is present and not the variation in number since the variation is most of the times not very influential; just mention the milking times in the text and not visual in the figure since you conclude it has not real influence).
L442-446 Please refrase, your main message is difficult to grasp.
L448-451 Is this a relevant observation (here)? Could perhaps skipped?
Section 3.4 A nice description of observed correlations is given but as a whole this section is a little bit disappointing because it gets stuck at this level. In section 4.4 it is stated that no statistical multivariate analysis could be applied to derive the share of EF variance explained by variables but in my opinion it is a missed opportunity that no such analysis is applied on the fluxes! Why not try such analysis for each grazing event separately and one for all cases together in order to infer variables which explain specific events and which variables are important explaining generic behavior of emission events due to grazing. Now the reader is left with huge variability in emission over de various events and no hint at an explanation.
Figure 5 Wouldn’t be worthwhile to add an figure depicting the diurnal cycle over the ten grazing events? Either normalized or not before combining, depending on what is interesting to show?
Figure 6 Why not present these results as profiles with height labeled with time instead of height and time the other way around? I find this (and other comparable plots in the article) hard to interpret. How do the soil mineral nitrogen concentrations relate to precipitation?
Figure 7 Did you make plots with all grazing events (labeled by color and aggregated in periods of ‘days since grazing’) in one figure? Doesn’t this show a pattern?
L511 Conversely- response This is rather an stand alone observation. It would be more interesting when information would be added whether this happened (albeit on a smaller scale) more often? E.g. during events G8 and G9??
Figure 8 Isn’t this a somewhat worrisome result? The measurements are done in the constant flux layer so I would hope a smaller systematic bias was found (in absolute sense). Did you check how choices made in the determination of the flux influenced this bias? Could this be used to determine the best choices regarding the flux derivation? E.g. the length of the period chosen (now 5 secs at each height)?
Table 3 How were the urinary N excreted values determined? Did I miss this somewhere or is this not explained?
Line 646-654 Since you so nicely discuss this point it is good to add that another assumption is made. K for momentum is observed to behave differently than a K for heat or moisture. It is unknown whether it is allowed to assume that a K for ammonia behaves similarly as the K for heat.
L736-743: these lines repeat information of lines l705-712; I think L705-712 can be removed. Make ‘near the surface’ more explicit; your measurement heights are already close to the surface so I guess you mean really close? A few centimeter?
L754 How was the estimated nitrogen input from grazing events determined?
L823-825 Please rewrite, this sentence is difficult the grasp by first reading.
Maybe I missed it but did you somewhere define the actual value of displacement height d? Based on figure 3 it appears to be ~ 10cm and time dependent?
Technical corrections
L50: is ‘crucial’ the right term? Don’t you mean ‘significant’?
L93: intensively grazed grassland: LSU per ha mentioned in the table are at the lower-to middle end of the numbers presented. Is it really ‘intensively’ grazed?
L95: makes possible -> enables
L174: for me the sentence is more clear when the first ‘height’ is removed.
L198: if-profile: is this not repeating the argument? Could be removed?
L201: ‘The concentration - averaging’ can be removed without loss of information.
L214: Maybe it is good to already refer to later sections where you further discuss this choice.
L452: It is also apparent that for some grazing events (e.g. …) the valid flux data capture was patchy, not necessarily – of interest (i.e. the wind was blowing from unsuitable directions). Please shorten to ‘For some grazing events (e.g. …) the valid flux data capture was patchy due to the wind blowing from unsuitable directions.
L630: ‘with’ missing in the second half?
L675 ‘horizon’ is maybe a correct (jargon) word? Or could it be replaced by layer?