The manuscript entitled “Ammonia exchange flux over a tropical dry deciduous forest in the dry season in Thailand”, written by Xu et al., presents a unique study to NH3 exchange between forest and atmosphere in the tropics. It is, to my knowledge, the first to present such data and is valuable to the general NH3 community. They systematically discuss the drivers of both NH3 concentration at different levels and the flux and suggest future studies that could answer questions raised in their own study. Their methodology, however, lacks clarity. As this could potentially affect a major part of their analysis, I recommend major revision. In addition, I have some specific comments and suggestions that could further improve their study.

Comments on methodology

1. Section 2.2: Flux calculation
   1. As the authors deployed instrument at 26 and 34 m above a forest with mean canopy height of 20 m, they are measuring inside the roughness sublayer (RSL). As they do not mention this, and refer to a paper that is not listed in the bibliography, I cannot verify whether they included this in their calculations and thus they could potentially underestimate their fluxes. (see e.g. Harman and Finnigan (2008), de Ridder (2010) or Duyzer et al. (1994))
   2. At what time interval do the authors calculate the transfer velocity D, and how does that compare to the long measurement interval of dNH3? If possible, flux measurements should be done at 30 minute intervals, as turbulence and gradients are highly variable. I understand that this was not feasible with the given set-up, however, long averages of gradients multiplied with a long average of the transfer velocity will lead to the ‘Schmidt paradox’ (Moene and van Dam, 2014), if there is not accounted for the ‘cross-term’ (see also chapter 2 of https://dx.doi.org/10.21945/RIVM-2022-0202). As I did not see a mention of this, I do not know whether the authors accounted for this. Moreover, the authors should discuss the potential error due to such a long measurement interval, including the effect of averaging over different stability regimes.
2. Section 2.3: The authors report the used meteorological instruments measure at an interval of 10 minutes. Does that mean that they sample only once every 10 minutes, or that there measurements are averaged/valid over that 10 minutes. Moreover, how do they obtain the meteorological input for flux calculation (by averaging?) and how do they obtain the meteorological conditions that they ultimately compare to the concentrations and fluxes?

Specific comments:

1. Line 43: How did NOx change during recent years?
2. Line 80: Is the range the difference over space or over time? And how heterogeneous is the surroundings? What is the standard deviation on the LAI measurements.
3. Line 99: first halve and second halve imply that the observation periods are consecutive. Therefore, I suggest to replace it with period 1 and period 2, or something alike.
4. Line 104: Matsuda et al. (2010) is not listed in the bibliography
5. Line 137: Is there any information on how these numbers compare to long term values of the site? Are they relatively high or low?
6. Line 171: Please specify what is meant by ‘a high category’
7. Figure 4 and Figure A1. the WD seems to have a diurnal cycle during a distinct amount of time, with different WD during night and daytime. I wonder what could cause such conditions and whether there are some submesoscale processes that could explain this.
8. Section 3.2: There are more factors that could control the NH3 concentration, such as boundary layer height and entrainment, see also Schulte et al., 2020 (https://doi.org/10.1016/j.atmosenv.2020.118153).
9. Figures 5 and 6: The different relation between the two different levels implicitly say something about the footprint of the two levels. It would be nice if the authors could add a discussion on this.
10. Line 226: Is there any scientific argument to exclude 1^st February? Why are the concentration on this day so different?
11. Line 254: Do the authors have any information on the amount of emission from the source area during the measurement campaign?
12. Line 276-277: please specify what you mean with ‘roughly patterned’.
13. Lines 277 – 279. I assume that the authors note on possible measurement errors due to these causes. Should be replaced to discussion on errors.
14. Line 280: A paired t-test is usually used to compare different groups. Does that mean that you are here comparing e.g. all night samples at 26m to all night samples at 34 m? And if so, why did the authors do that? A difference could be significant at group level, but still be insignificant on a single instance. I think it would be better here to judge each gradient measurement individually by comparing the uncertainty/error on the measurements, unless the authors are only interested in an average flux (but the rest of the article addresses individual measurements). Also, how does this relate to the previous sentence? I.e., if the two levels are significantly different, how would that relate to the presence of emission sources or meteorological elements? I don’t think I fully understand the intention of this sentence.
15. Lines 298-299 I don’t understand this. Figure 9 shows a quite large difference between D1 and D2, and N for all panels (including the transfer velocity)?
16. Lines 329-349: The authors extensively discuss the stomatal conductance gs, and use measurements of August 2020 and from 1996 to explain their strongest emissions during D1. However, gs is ultimately controlled by ecosystem health and meteorological conditions, which might have changed over this time, especially when comparing over such a long period or over different seasons. The authors could make a much stronger case here by calculating the gs themselves using e.g. Embserson or an A-gs model (https://doi.org/10.1016/j.envpol.2006.04.007), which they could train on the CO2/H2O fluxes (if available) or otherwise on the previous measurements of gs. If the authors decide against this, they should better discuss the validity of their comparison.
17. Lines 340-341: I assume the authors refer to different timing of emissions in Xu et al., 2023 and this study, please clarify the timings in the main text.
18. Lines 373-375: But then why the negative correlation with temperature during the second halve (especially during nighttime). Wouldn’t you expect it the other way around?
19. Line 396: what do you mean with ‘obviously’
20. Figure 10: Please consider to use constant binsizes for the two different observation periods, as this would allow for an easier comparison.
21. Lines 434 – 459: This reads more as a results and discussion section. Please consider moving it to Section 3.5

Technical comments

1. Line 15: (…) we present the first observations of NH3 exchange (…)
2. Line 44: therefor >> Therefore
3. Figure 1: Please use a white font for more clarity
4. Line 80: (…) at the beginning of the observation period (…)
5. Line 275: (…) dC, because (…)
6. Line 295: please rephrase “about four times”
7. Line 307: (…) acted as a source of NH3 (…)
8. Figure 9: The authors could consider boxplots for more clarity
9. Lines 314-319. The authors could consider to uniform the units of all previous studies (i.e. all to ug/m2/s), as that would allow for an easier comparison.
10. Line 336: please rephrase “And there was no change during this”
11. Line 364: closed >> close
12. Line 442 & 456: Wentwortth >> Wentworth
13. Line 456: (…) conditions. Notably (…)

General comment:

This study presents a description and analysis of NH₃ flux measurements conducted in a tropical deciduous forest in Thailand.  Given the novelty of NH₃ flux measurements in this ecosystem type, and the fact that most measurement campaigns have been conducted in temperate climates in Europe or North America, this study provides valuable new insights into the biosphere-atmosphere exchange occurring in tropical ecosystems. The authors discuss which exchange dynamics could be most relevant and identify the possible sources of emission fluxes. However, the presentation of the results, both in the text and in the figures, could be improved to enhance clarity and readability. In particular, some of the wording is vague or imprecise. Additionally, the manuscript would benefit from a more thorough and critical evaluation of the uncertainties associated with the NH₃ flux measurements. Addressing these issues would significantly improve the manuscript and better highlight its important contributions to understanding NH₃ biosphere-atmosphere exchange in tropical ecosystems.

Specific comments:

Line 81 – 82: The agricultural fields and residential areas are 400 m away from the observation tower. Have you performed further analysis (e.g., with the tool by Kljun et al. (2015)) on the footprint to ensure that anthropogenic activities do not interfere with your measurements?

Figure 1: Given the novelty of the flux measurements in this ecosystem type, I would consider adding one of the pictures of Figure A2 (or even merging these two figures). Moreover, the last sentence of the figure caption states that the data for the product has been filtered to include only those with a cloud cover percentage of less than 20%. I believe this refers to the satellite-derived LAI, but it is not clear in this capture. Please specify this.

Line 103: Can you be more descriptive regarding the physics behind the transfer velocity, or at least include the unit?

Line 179 – 186: I would recommend putting these statistics in a table, making it easier for the reader to compare the meteorological circumstances between the first and second phases of the study.

Figure 3: For the interpretation of the NH₃ concentration during the two observation periods, it would be useful to also add the measurement error of the concentrations to this figure as error bars or shaded uncertainty ranges. Now it is hard to judge how meaningful the patterns and differences between the measured concentrations are. Moreover, I think the readability of the figure would improve if you add gridlines. The latter also applies to Figures 4-6, 8-10.

Section 3.2: In this section, the factors influencing ammonia concentrations are discussed, with Figures 5 and 6 serving as the main supporting figures. Have you considered that many meteorological variables (e.g., temperature, wind speed, and relative humidity) are often strongly correlated, which can complicate the interpretation of individual relationships. Drawn conclusions could be misleading when potential multicollinearity is not accounted for. I would encourage the authors to explore the interrelationships between the variables and assess their potential impact on the analysis.

Figure 4: I suggest moving the two column titles “1st half” and “2nd half” to the top of the figure (i.e., above 4a and 4e), as these column titles are now positioned between the tick labels and the caption of the figure, and get “lost” between the other elements. Moreover, have you considered also adding solar radiation and LAI data to this plot?

Line 206 – 207: “[…] found that daily NH₃ concentrations have a strong relationship with the magnitude of temperature and may be affected by different processes during the daytime and nighttime”. Could you briefly elaborate on the kind of processes referred to here to make this more informative?

Figure 5: I have several remarks about this figure. First, I would consider adding the y-axis label currently in subplot (c) to subplots (a) and (e) as well, to clarify that the y-axis concerns the NH₃ concentration. Second, to justify the found correlations, I would recommend mentioning whether the regression functions here are statistically significant (e.g., by reporting the p-values, or if p-values are below 0.05). Finally, correlations have been shown for both at 36 and 24 meters in height, but the rationale for including the correlations at 24 meters is currently not explained. In other words, what was the reasoning for checking the relationships at height z₁ as well?

Line 222 – 229: This paragraph discusses how in the second half of the study campaign, the relationship between NH₃ and the concentration was inverse at night time and little during D1 and D2. You discuss that omitting the data from February 1^st changes the relationship again to a moderate correlation of 0.57 and 0.59. You mentioned earlier in the manuscript that the wind direction changes on this date. I suggest providing a (brief) explanation of why the correlations between temperature and NH₃ concentration improve so significantly when omitting this data. Additionally, justify in this paragraph why you would omit this data.

Section 3.3: I was wondering if it was a deliberate choice not to include a time series of the measured fluxes as well.  Such a plot can be a helpful way for readers to quickly understand the flux dynamics, such as day-night variations or temperature influences. Moreover, it would also be useful to display the (estimated) measurement uncertainty here per flux measurement. Finally, it would also be interesting to know whether the emission pulses increased over time in response to decreasing LAI, potentially due to enhanced leaf litter decomposition. Was there an observable increase in NH₃ emission fluxes as more leaves fell from the canopy?

Sections 3.3 and 3.4: I would recommend restructuring these two sections, if possible, into one section. The first section focuses on the possible sources of the fluxes, whereas the second section focuses on the factors controlling the NH₃ fluxes. In my view, these two subjects are too tightly interconnected to treat separately. I believe that discussing these two together will both aid in conveying your message more clearly and concisely, and in avoiding repetition.

Lines 291 – 303: I would recommend putting the statistics (i.e., the weighted means and the standard deviations) in a table to improve readability.

Line 346 – 349: Have you also considered that a decrease in LAI can also indirectly cause an increase in emissions, because (a) the forest floor becomes more exposed, and (b) the reduced LAI may diminish the canopy’s capacity to act as a deposition sink through cuticular deposition?

Line 393: “Observations and bi-directional exchange models have demonstrated that temperature is the most important factor controlling NH3 emissions (Flechard et al., 2013; Zhang et al., 2010).” Consider nuancing this statement, as indeed temperature is an important driver in the NH₃-NH₄ dissociation and Henry equilibrium – but temperature is also closely correlated with other key meteorological variables such as relative humidity and solar radiation.

Line 403 – 405: Have you considered the potential influence of leaf phenology on the NH₃ emission strength? Mattsson and Schjoerring (2003) found that the leaf senescence caused changes in the apoplastic NH₄⁺ concentrations. Incorporating this aspect could provide additional insight into the observed flux patterns.

Line 423 – 434: The median flux errors reported by Wolff et al. (2010), Ramsay et al. (2020), and Melman et al. (2025) are derived from high temporal resolution measurements, while this study relies on manual measurements. I am not fully convinced that a comparable measurement error of approximately 50% can be assumed without further explanation. Could the authors clarify whether they attempted to quantify the uncertainty of the flux, for example, through error propagation? In my opinion, the uncertainty of the fluxes needs a more critical evaluation to strengthen the credibility of the conclusion that the DDF acts as a net NH₃ source during the observation period.

Figure 20: The caption of this figure contains repetitive information in the last two sentences, which have already been mentioned in the methodology section. Moreover, I would consider removing the sentence “(x” indicates values greater than x, and “x]” indicates values less than or equal to x, as this notation is standard and generally well understood. Moreover, increasing the font size of the axis labels and the tick marks would improve the readability. Finally, a technical correction is to change the numbering of this figure from 20 to 10.

Line 434-436: You discuss the presence of outliers in the measured fluxes of D1, but it is unclear based on which criteria these are qualified as outliers. Moreover, are these outliers still taken into account in the calculation of the statistics mentioned in the paper?

Line 464: “On the other hand, we hypothesize that NH3 concentrations are controlled by meteorological elements as well as by changes in canopy structure accompanied by defoliation. Specifically, the dominant NH3 emission source may shift dynamically from leaf stomata to leaf litter in response to changes in canopy, forest floor, and meteorological conditions.”. These sentences refer to ‘meteorological elements/conditions’, which are a very general formulation. I would recommend rephrasing or combining them with the following sentences to improve clarity and avoid vague wording.

Section 4: I have a suggestion for future research: This study clearly demonstrates that the DDF ecosystem serves as an emission source during the dry season, likely due to leaf litter decomposition. However, this concerns a very seasonal process and may not be representative of NH₃ flux dynamics on an annual basis. Have you considered analyzing whether this phenomenon of NH₃ emission during the dry season is more broadly applicable to deciduous forests in tropical regions, with satellite observations (e.g., CrIS, IASI), could offer additional insights?

Figure A3: I suggest combining this graph with Figure 4, as this is important information for the NH₃ flux dynamics taking place at SERS.

Technical comments:

Line 79 – 81:  The sentence “Leaf area index (LAI) […] with a mean value of 3.1.” would benefit from clearer structure and improved grammar for better readability. Second, do I understand correctly that the LI-COR has only measured the LAI at SERS once? And could you specify what the range “2.5 to 4.2” indicates here? Is this the uncertainty, or has the LAI been observed at multiple locations around the tower?  Finally, you could also report that LAI values have been derived with the MODIS satellite, as mentioned at lines 231 – 233. This is relevant information, which should have been mentioned earlier in section 2.1.

Line 104: i.g. should be e.g.

Line 104: To maintain consistency in notation, consider including the symbol u* for friction velocity after introducing the full term, similar to how you denote displacement height (d).

Line 114: I found myself needing to return to this line when interpreting Figure 5 to recall what D1, D2, and N referred to. It may help the reader to place slightly more emphasis on the naming convention of the daytime and nighttime samples, as it is currently understated. E.g., rephrase it to: “We continuously collected two daytime samples, denoted as D1 and D2, corresponding to the periods 09:00–13:00 and 13:00–17:00, respectively.”.

Line 171: I suggest replacing the phrase “[…] in a high category [...]” with more specific wording, as it is currently vague.

Line 183: I would suggest replacing ‘variation’ with ‘relationship’.

Line 236: I suggest clarifying briefly that “[…] which is close to the observed mean value around the observation tower” is measured by the LI-COR LAI-2200 for clarity.

Line 242: “process” should be “processes”.

Line 271: “[…] during the daytime, it remained above a certain level” – I would recommend making this sentence more specific about what the “certain level” refers to.

Line 355-356: “[…] which has an even lower pH and poorer nutrient”; this sentence seems unfinished.

Line 273: Consider replacing ‘progress’ with ‘occur’ or ‘proceed’

Line 408-409: The sentence “In contrast, the change in median flux after the first increase was smaller in the second half” is vague and would benefit from clarification.

Line 423: The phrase “Although the case is limited, […]” is vague and requires more specific wording.

Line 429: replace ‘term’ with ‘terms’

Line 457: replace ‘depend’ with ‘depends’

Sources:

Kljun, N., Calanca, P., Rotach, M. W., & Schmid, H. P. (2015). A simple two-dimensional parameterisation for Flux Footprint Prediction (FFP). Geoscientific Model Development, 8(11), 3695-3713.

Mattsson, M., & Schjoerring, J. K. (2003). Senescence‐induced changes in apoplastic and bulk tissue ammonia concentrations of ryegrass leaves. New Phytologist, 160(3), 489-499.

Melman, E. A., Rutledge-Jonker, S., Frumau, K. F. A., Hensen, A., van Pul, W. A. J., Stolk, A. P., Wichink Kruit, R. J., and van Zanten, M. C.: Measurements and model results of a two-year dataset of ammonia exchange over a coniferous forest in the Netherlands, Atmos. Environ., 344, 120976, https://doi.org/10.1016/j.atmosenv.2024.120976, 2025.

Ramsay, R., Di Marco, C. F., Sörgel, M., Heal, M. R., Carbone, S., Artaxo, P., de Araùjo, A. C., Sá, M., Pöhlker, C., Lavric, J., Andreae, M. O., and Nemitz, E.: Concentrations and biosphere–atmosphere fluxes of inorganic trace gases and associated ionic aerosol counterparts over the Amazon rainforest, Atmos. Chem. Phys., 20, 15551–15584, https://doi.org/10.5194/acp20-15551-2020, 2020.

Wolff, V., Trebs, I., Ammann, C., and Meixner, F. X.: Aerodynamic gradient measurements of the NH3-HNO3-NH4NO3 triad using a wet chemical instrument: an analysis of precision requirements and flux errors, Atmos. Meas. Tech., 3, 187–208, https://doi.org/10.5194/amt-3-187-2010, 2010.