the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 15 Jun 2026
Ammonia variability and trends from globally distributed FTIR measurements and model simulations
Abstract. Ammonia (NH3) is an important constituent in the global nitrogen cycle, present in both urban and remote environments. It is a source of reactive nitrogen and a precursor for particulate matter, thereby affecting atmospheric chemistry and radiative forcing. This work presents the seasonal and diurnal variability, along with long-term trends, of atmospheric NH3 total columns retrieved from Fourier transform infrared (FTIR) spectroscopic solar absorption measurements at 22 ground-based sites, globally distributed from 45° S to 80° N. Comparisons are made with simulations from the GEOS-Chem High Performance (GCHP) chemical transport model and the Tropospheric Chemistry Reanalysis (TCR-2) NH3 product. The mean NH3 total columns from the FTIR time series ranged from 0.12×1015 to 19.20×1015 molecules cm−2, with the smallest columns found at the Arctic and high-altitude sites, and the largest in urban areas. Significant enhancements were attributed to biomass burning, and NH3 emissions from volcanic eruptions were detected at the Izana site. The seasonal patterns are similar across most sites, with maxima mainly related to the volatilization of NH3 due to higher temperatures. The diurnal variability differs significantly and depends on the characteristics of each site and local sources. Most sites have positive trends in the total column, with a mean value (and 95 % confidence interval) for all sites of 3.82 (3.29–4.35) % for the FTIR measurements, 3.66 (3.35–3.97) % for GCHP, and 6.49 (2.00–10.98) % for TCR-2. GCHP exhibited a better general agreement with the FTIR observations than TCR-2; potential reasons for this are explored, including a sensitivity test on the emissions used.
- Preprint
(42059 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 27 Jul 2026)
- RC1: 'Comment on egusphere-2026-3138', Anonymous Referee #1, 29 Jun 2026 reply
-
RC2: 'Comment on egusphere-2026-3138', Anonymous Referee #2, 03 Jul 2026
reply
In this manuscript, the authors present a novel dataset of time series of total vertical column amounts of ammonia from 22 globally distributed locations. These data were obtained using high-resolution ground-based FTIR solar absorption spectrometers operating in the thermal infrared spectral region. The authors analyse the temporal variability of the dataset, including diurnal, seasonal, and long-term patterns. In addition, they compare the observations with two different global model datasets.
The paper is logically structured, well written, and the data are presented clearly in figures and tables, aside from a few revisions and corrections noted below. The arguments and discussions are clear, well supported, and appropriately referenced. Beyond the intrinsic value of the dataset itself, a particular strength of the manuscript is the comparison with model results, which clearly reveals substantial discrepancies and underscores the need for future model improvements. It also highlights the importance of continuing this type of high-precision total column observations, which complements and supports the less accurate space-borne measurements.
Overall, I strongly recommend publication, subject to consideration of the specific technical comments provided below.
Specific:
l.13
The trends are given in percentage values without stating the related time period.
l.17 ‘since 1960’
I don’t think that this happened exactly in this year. Perhaps better: ‘since the 1960ies’
l.50 ‘The NH3 columns vary widely around the world. For example, for Central America, the highest columns were during April and March and for South America during September and October, while Europe and North America showed higher columns during the spring season with columns around 1.5−2×1016 molec. cm−2.’
This implies that there is a strong difference when the maxima appear. However, it seems that these mostly appear in spring.
l.108 ‘The retrievals included the interfering species H2O, CO2, O3, CO, HNO3 and N2O …’
CFC-12 and SF6 have both strong spectral interference in these spectral regions. Why have those not been considered? Please discuss the retrieval error introduced by neglecting them.
l.122-163, 2.2 Simulations of NH3
Could you also add a short description of the physical and chemical processes influencing atmospheric NH3 concentrations in each model, as well as the related differences between the models.
l.171, Eq. 2
- Please define ‘j’ and its unit.
- Please explain how relative trends are calculated.
Chapter FITR-results
I agree with Referee #1 that the authors should think about providing the statistical analysis based on medians/quartiles and where this might be suited to avoid too strong influence from outliers/special situations.
l.210, ‘Figure A5 shows SO2 and NH3 enhancements during the Tajogaite eruption from 19 September 2021 to 13 December 2021 during which volcanic plume reached the Izana site.’
But the plot seems only to extend from 15 Sep to 15 Nov ?
Further, could you show some more time around that event to visually show the difference and enhancements in the NH3 columns with respect to the non-perturbed state.
l.218, ‘In addition, the monthly means are seen to increase for the most recent years for most sites’
This cannot easily be seen in those figures. Either point to a figure where this becomes clearer or try different color scales.
l.237, ‘However, at some sites such as Boulder, a maximum was observed later in the year, probably due to the fertilizer application season and agricultural practices.’
- I wonder if the Spring enhancements in many observations (l.231) are connected also to the fertilization period. Could you discuss this in the text?
- I think this sentence belongs more to the paragraph before as the major description of this paragraph is about comparison to previous publications?
Table 2, header last column
Shouldn’t this read ‘%/yr’ instead of ‘%’?
Figs. 4, B1
Please add the 1:1 line in all plots.
l.397, ‘Overall, GCHP NH3 total columns agree better with the FTIR measurements than does TCR-2 for most sites’
What do you mean with ‘most’: from Tables 4 and 5, GCHP agrees better for only 13 stations?
l.470, ‘Both models struggled to capture the diurnal variability of the measurements, mainly due to the lack of diurnal variation in the anthropogenic emissions used in the NH3 simulations.’
This statement seems to me a bit too strong. I don’t see a real proof for it in the manuscript.
Technical:
l.120 ‘ranging between 0.1 and 2.0’
Please add here the units.
Fig. 4:
‘blue’ -> ‘pink’
Eq. 4:
1/N missing
l.108, ‘simmulations’
-> ‘simulations’
l.410, ‘Zugstpitze’
-> ‘Zugspitze’
l.413, ‘Among the emissions, the contribution from anthropogenic and seabird emissions the most different sectors.’
Verb missing.
Citation: https://doi.org/10.5194/egusphere-2026-3138-RC2
This study uses ground-based FTIR measurements at 22 worldwide sites to track atmospheric ammonia (seasonality, diurnal patterns, trends). The authors also compare the measurements against two models. As a detailed measurement report, the paper is a valuable addition to the existing literature on atmospheric ammonia. The release of the dataset is a milestone. The trend and seasonal analysis complement results established from satellite observations. The diurnal characterisation is a useful contribution that is harder to obtain from satellite data. The GCHP/TCR-2 NH3 evaluations would arguably deserve a dedicated study, in which a more in-depth analysis could be made and/or attempts made to reconcile the simulations with the observations.
I can recommend publication of this paper subject to a major revision taking into account the following comments:
* Main comments
- As the authors acknowledge, NH3 enhancements originating from volcanoes are rarely reported. In view of the exceptional nature, the authors should present robust evidence of this, for instance in the form of a physical fit over the spectrum showing the residuals with and without NH3/SO2. This would exclude retrieval artefacts.
- The influence of SO2/NOx on the trends is stated a few times but never demonstrated. For example a figure of collocated trends of NH3/SO2/NOx could be shown and discussed. Table 9 contradicts partially the lifetime argument (showing very small correlations between NH3 and SO2/NOx).
- Can you report the error (budget) per site (currently only a range and average is given)?
- There are several problems with the reported statistics:
(1) a mean of absolute trends does not make sense, as these trends are calculated from data from stations whose average columns span three orders of magnitude. The mean trend is mainly representative of the high-column stations.
(2) There is similar objection for the mean of relative trends. In contrast to the absolute trends, these can be driven by outliers typically seen in stations with a low column average (e.g. Eureka and Ny Alesund in table 8 in the TCR-2 % column). I would propose dropping the all-site mean of absolute trends and providing the median + Q1-Q3 quartiles for the relative trends. In addition, stations with a short record (e.g. Hefei) should be excluded from this by applying some threshold over the number of years covered.
(3) mean NRMSE/MRD are likewise determined by outliers. For instance in Table 3, the mean of MRD is close to 0, but only because of a fortunate cancellation of several outliers above 100% with otherwise mostly negative values. I would recommend replacing these means with median and/or Q1-Q3 quartiles.
(4) The "Mean for all sites" row in Table 3 uses two different methods: the mean slope and R (not sure as the text doesn't specify it) were calculated from all the individual datapoints while the mean MRD and NRMSE are per-site averages.
- P3 reviews the literature on NH3 trends, could cite/discuss the following (if found relevant)
(1) https://doi.org/10.1021/acs.est.4c14020
(2) https://doi.org/10.1016/j.scitotenv.2024.176846
(3) https://doi.org/10.3390/atmos14071056
(4) https://doi.org/10.1016/j.envint.2024.108519
- Figures:
(1) All the "full page" figures are in-between single column and double column figures. Some of them are very squashed, so I would strongly recommend increasing their width, height or both to cover the entire page.
(2) Figures A1-A5 could be combined in a multipanel figure
* Typos and Inconsistencies
P5, Table 1: Lauder should be 169.68°E not W; For Eureka, the ° misplaced
P6, line 145: remove "very"
P6, Sect. 2.2.1–2.2.2: HTAP is cited as Janssens-Maenhout et al. (2015) for both the HTAPv3 GCHP sensitivity run and HTAPv2 TCR-2, but that reference is HTAPv2.2. The 0.1°, 2000-2018 coverage corresponds to HTAPv3 (Crippa et al., 2023), which is not cited. Please clarify which inventory/version and years were used for each simulation and correct the citation.
P13, line 301: duplicate "over Europe"
P16: The NRMSE is missing a 1/N
P18: the stated TCR-2 minimum is incorrect, it is -0.04 for Izana and not -0.02 at Reunion
P25, line 415: "despite the HTAP emissions being smaller than the CEDS emissions" contradicts what is written on line 412.
P26/29: "NH3 prevails in the particulate phase", more accurate is to say that the particulate form dominates.
P27, Table 7: The 20.9 for Los Angeles is not inside the CI
P28, Table 8: Can you double check the identical TCR-2 values of Mexico city and Altzomoni? The 2.58 should probably be -2.58, otherwise the CI are wrong
P29: The seasonal 2.66-6.92 and diurnal 0.20-6.74 numbers in the conclusion do not correspond to what is given in the paper.
P33: Tarjogaite should be Tajogaite