the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Impact of present aircraft NOx and aerosol emissions on atmospheric composition and climate: results from a model intercomparison
Abstract. Aircraft emissions of nitrogen oxides (NOx=NO+NO2), aerosols, and aerosol precursors provide a non-negligible contribution to the climate impact of air traffic, and the uncertainty on their Effective Radiative Forcing (ERF) of climate remains significant. This study presents results from a new model intercomparison of the impact of aircraft emissions involving five state-of-the-art global models including both tropospheric and stratospheric chemistry. Aircraft NOx increases ozone photochemical production in the free troposphere throughout the year and decreases ozone chemical loss in the high-latitude lowermost stratosphere during spring–early summer. The models generally agree on the spatial pattern of NOx, ozone, and hydroxyl radical (OH) responses. The NOx net ERF is systematically positive and ranges from 7.3 to 22.1 mW m-2 among the different models (14.1–22.1 mW m-2 without the least sensitive model). Estimates of the aerosol direct ERF are systematically negative and range between -6.5 and -17.8 mW m-2, with differences arising from the diversity in model aerosol parameterizations. This work shows encouraging results regarding our confidence in aviation NOx-induced ozone response because of a better model agreement. However, results also highlight areas where further modeling experiments are needed, both with more models and with dedicated sensitivity simulations to further understand the factors giving rise to the spread in model estimates of aviation emission impacts on atmospheric composition and climate.
- Preprint
(4631 KB) - Metadata XML
-
Supplement
(2752 KB) - BibTeX
- EndNote
Status: open (until 27 Oct 2025)
- RC1: 'Comment on egusphere-2025-4273', Charlie Wartnaby, 24 Sep 2025 reply
-
RC2: 'Comment on egusphere-2025-4273', Anonymous Referee #2, 10 Oct 2025
reply
Report of “Impact of present aircraft NOx and aerosol emissions on atmospheric composition and climate: results from a model intercomparison” by Cohen et al.
The manuscript presents a new model intercomparison to quantify the impact of NOx and aerosols emissions from aircraft on both atmospheric composition and climate. The atmospheric composition piece includes ozone, NOx, OH and aerosols – black carbon, sulfate and nitrate aerosols. The study includes three global chemistry transport models – MOZART3, OsloCTM3, GEOS-Chem – and two global chemistry climate models – EMAC, LMDZ-INCA. The authors found that aircraft NOx emissions are causing an increase of ozone photochemical production in the free troposphere throughout the year and a decrease of ozone chemical loss in the high-latitude lowermost stratosphere in spring/early summer. They also found that the net Effective Radiative Forcing (ERF) of NOx due to aircraft emissions is positive and vary with models between 7.3 to 22.1 mW m-2, when the aircraft aerosol direct ERF is negative and vary between -6.5 and -17.8 mW m-2.
The authors put in perspective their findings with prior studies and discuss the potential reasons of the uncertainties. They suggests ways to improve future model output to reduce those uncertainties.
The study is in the scope of ACP. It is a thorough study. The manuscript is well written and well organized with clear figures and tables.
I would support publication after the authors addressed the minor comments I listed below.
General Comments:
The manuscript would benefit from having a description of the aviation emission inventories used in this study: CEDS aviation emissions and Flightradar24, which could include the altitude of injection. Could the authors add a map of the aircraft NOx emissions?
The authors do compare their results with former studies but it is not always clear when the study brings new findings on the impact of aircraft emissions on atmospheric composition and climate. Does this study bring to the community more accurate numbers with less uncertainties or new numbers and how (e.g. the impact of nitrate aerosols, the intensity of the direct effect of aerosols from aviation)? Could the authors write a separate section on this point? The section could also discuss the choice of the time-period and its potential impact on the results.
Specific Comments:
L.257: “CEDS aviation emissions” - Could the authors comment on the aircraft emissions inventory? Does it include altitude of injection? Would it be possible to add in the Supplement, the equivalent of Figure S1 but with the aircraft NOx emissions?
L.260: “small” - Could the authors give a quantity?
L.274: “Flightradar24” – As for CEDS aviation emissions. Could the authors describe it more – because it is a key component of the study: altitude of injection, latitude/longitude? A paragraph dedicated to aircraft emission inventories as input for the models used in the study would be key.
L.301-306: Does the rescaling methodology implies that the vertical regridding for each model has been made with LMDZ-INCA grids as the reference? Could you clarify? Have the authors estimated the impact of the linear rescaling on GEOS-Chem’s results?
L.323-324: Could the authors be more explicit on the calculation of the impact by explaining the ratio: DeltaS/ENOx with DeltaS the perturbation. How this perturbation is calculated? How ENOx is calculated?
L.325-327: Where is the reference figure or table for the percentages (e.g. global NOx perturbation ranges of emitted NOx in terms of nitrogen mass).
L.327-328: “0.51 DU” – shouldn’t the unit be DU/NEU? / “0.90 DU” – shouldn’t it be 0.92 DU/NEU (number reported in Table 4)? / “9.9 TgO3” – shouldn’t it be 10 Tg/NEU (number reported in Table 4)?
L.329: Would the authors have a reference figure or table for mixing ratios?
L.330: “lower altitude of the OsloCTM3” – Could the author explain this characteristic in the method section? / “wider vertical range” - Could the width of the vertical range be added in Table 1?
L.332: ”AEDT” – Could the authors spell it out? / "between 2.8 and 11.2 Tg/(TgN.yr-1) from both offline and online models” – How these models compare with the CTMs of the study (emission inventories, chemical and physical processes)?
L.334: “9.9” – wouldn’t be 10 Tg/NEU instead, according to Table 4? / “The ozone sensitivity is lower in the current study” – Did the authors mean lower than what was found in Olsen et al. (2013) and Brasseur et al. (2016)? How ozone sensitivity is calculated? Is it different from the ozone burden sensitivity?
L.335: “aviation NOx emission is 36% greater in 2014–2018 than in 2006” – Do the authors have a reference, a figure or table to show this number for aviation NOx emission change?
L.336: “GEOS-Chem is characterized by the highest response” – May be worth repeating 11.2 Tg/NEU in parenthesis here, if I correctly understand.
L.342-343: “In order to capture as much of the response for each model as possible” – Could you clarify? Could you explain the vertical range characteristics of the model and give information in Table 1 that you can refer here.
L.344: “NOx response” - It would be worth defining NOx response, which seems to be called NOx change in Figure 1. It may be worth adding a paragraph in the methodology? Worth explaining why it is a different unit than in Table 4?
L.371-372: “As expected, Fig. 3 shows a more homogeneous ozone perturbation” – Do the authors mean more homogeneous than NOx? Could the authors explain why it is expected?
L.391: “for ozone” – Typo, “for ozone” is written twice in this sentence.
L.419: “lesser” – Could the authors specify which study gives a 16% lesser ozone response than the current study?
L.420: “weaker” - Could the authors specify which study gives a 16% lesser ozone response?
L.421: “perturbations” – Could the authors be specific and change to “perturbations of ozone”?
L.424: “studies that focus on the past decade” – Could the authors remind the reader of those studies by adding the references?
L.431-432: It would be worth reminding the reference to Table S2 here.
L.435: “by EMAC-NOx and LMDZ-INCA” – Could the authors remind the readers why not by the other models?
L.440: “tends to reduce the ozone vertical gradient” – Could the authors back up this statement with a figure in the supplement?
L.441: “Second, the chemical loss term (Fig. S2, in Supplement) increases in the UT” – The ozone loss seems to be maximum in the mid-troposphere (~ 600 hPa, Figure S2). It would be worth noting this and explain why focusing on the UT.
L.443-446: This finding is very interesting. Could the authors elaborate more on this competition between O3 and NO2 to react with OH? Any reference?
L.482: “The seasonality is the same as for ozone” – The seasonality here seems to be for sulfate aerosols. What about the seasonality for black carbon and nitrate aerosols? Are Jan and Apr-Jun as relevant as for ozone and sulfate aerosols?
L.523: “fewer” – Do the authors mean “fewer than for the gas-phase”?
L.533: “surface concentrations” – Could the authors provide a figure/map in the supplement for the surface concentrations? How surface is defined in the models?
L.639: “not modeled here” – Could the authors explain why they did not model “contrail cirrus formation”?
L.686-688: This is an important statement and seems to deserve being highlighted in the abstract.
Citation: https://doi.org/10.5194/egusphere-2025-4273-RC2
Data sets
Perturbation simulations for aircraft NOx and aerosol emissions in present day and future: multi-model data from the ACACIA EU project Y. Cohen, D. Hauglustaine, M. T. Lund, A. Skowron, S. Matthes, R. Thor, Z. Staniaszek https://doi.org/10.5281/zenodo.16949721
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 1,659 | 42 | 10 | 1,711 | 15 | 12 | 10 |
- HTML: 1,659
- PDF: 42
- XML: 10
- Total: 1,711
- Supplement: 15
- BibTeX: 12
- EndNote: 10
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General comments
- This paper provides an important addition to the existing literature concerning the modelled effects of aviation NOx pollution, which has high practical and policy relevance given the likely evolution of air transport.
- This paper does not itself add the strength of validating model predictions against observations. However, the referenced (published) companion paper (Cohen, 2025) does address how 4 of the 5 models compare against IAGOS observations. GEOS-Chem is the exception there though, which is often an interesting outlier in this paper.
- The overall climate forcing effect of aviation NOx is still not well-constrained by this study, as the authors discuss, but it usefully adds results and analysis from further models. The authors should make clear what is new in this study in terms of models, date ranges and emissions covered, etc to distinguish it from previous work.
- There is probably less existing literature on the climate forcing from aerosols resulting from aviation pollution, so that novelty should be highlighted.
Specific comments
- Abstract L28: "because of a better model agreement." better than what, previous comparisons? Or just "good model agreement"?
- Introduction L84: "the recent inventories used by CMIP6 for anthropogenic surface and aircraft emissions", but CMIP6 data only runs to the end of 2014 (or 2015?); the referenced McDuffie 2020 work extends CMIP6 for a few years as CEDS(GBD-MAPS). See also L255.
- Section 2.1 L104: "the pairs of mixing states correspond to the hydrophilic-hydrophobic dichotomy" I understand that organic particles may start hydrophobic but become hydrophilic, but is this "dichotomy" a well-known term here? I'm not an aerosol expert but don't recognise it. Others may benefit from more explanation here too.
- Section 2.2 L255: "The historical anthropogenic emissions are taken from the Community Emissions Data System inventory CEDSv2 (McDuffie et al., 2020; O’Rourke et al., 2021, regarding NOx, SO2, and BC emissions)." Having a look at McDuffie 2020 they only go to 2017, while the O'Rourke 2021 reference is for CEDS v_2021_04_21 which seems to go out to 2019. As some of the models run to 2018 or 2019, does that mean that the emissions for some species (those not in your O'Rourke list) were extrapolated or otherwise estimated to cover 2018 and 2019? A brief explanation of that would be good if so.
- Section 2.2 L260: "For these years (2015–2018)," but from Table 2 the GEOS-Chem model ran for 2019, so should it be 2015–2019 here? Or maybe GEOS-Chem used different emissions here too? (Later: more on GEOS-Chem emissions being different in section 2.3 L304, maybe worth a cross-reference)
- Section 2.2 L280: "We rescaled the EMAC-aer perturbations using the NOx emissions from CEDS for the period 2014–2018 (and the same rescaling factor for every species)," I'm not concerned that you had to do something special there, but don't really understand how that rescaling worked. E.g. for SEN100 I would expect you just zeroed out the aviation emissions in use by this model. So what is the "rescaling" of the perturbations? (Later: maybe related to section 2.3 around L294? In which case could cross-reference that section.)
- Section 3.1 L391: "Due to stratospheric intrusions, an extension of the mean ozone perturbation is visible at low latitudes, downward and equatorward." Is this really about stratospheric intrusion? If we had a global intrusion of that shape and size, surely it would be a named phenomenon... maybe it is partly. But we can already faintly see enhanced NOx in the same area in Figure 4. So maybe it is largely created there in situ. After all, many flight routes following approximate great circles will climb and descend from lower-latitude airports but then (for most of the flight) cruise at higher altitude and further north, giving the same kind of "diagonal" pattern on these latitude-altitude plots. (Later: at L479 you make such a case for aerosols, though with actual near-surface emissions there too.)
- Section 3.1 L392: it might be interesting to comment on the somewhat different distribution of OH, which is most enhanced inside the UT compared to NOx and O3 in the LS. I guess because there is much more H2O below the tropopause. (You come back to OH at L412+ but without venturing an explanation for its different distribution.) (Also return to OH around L444, could mention further discussion "below" at least here.)
- Section 3.1 L409-410: reading the text and looking at Figure 5 side by side, two of the quoted numbers seem mismatched by a couple of tenths of a ppbv (6.7 vs 6.9, 13.0 vs 13.2). The differences may not be scientifically significant but they should agree, if only to make relating the text to the plots more straightforward, and anything >0.1 difference cannot just be a difference in rounding here. Perhaps the text was written against an older version of the plots or vice versa?
- Section 3.1 L420 "substantially weaker... -38 %" weaker than what, MOZART3?
- Section 3.1 L403-L422: this is a long paragraph that starts with O3 (fig 5), switches to OH (fig 6), then backtracks to O3 (fig 5). It might be a smoother experience for the reader if you moved the OH discussion to be a new, following paragraph.
- Section 3.1 is pretty long overall, and without any lower-level headings you can't cross-reference easily between parts of it. Maybe split it up into subsections e.g. for the different species discussed?
- Section 3.1 L452 "stronger in models with lower NOx" I agree, also lower NOy more generally so could mention that too
- Section 3.1 L453 "The OH response increases with H2O background" the GEOS-Chem point is a real outlier here; I see no correlation if I include that as an equally valid datapoint. You do mention in the next sentence that it is much higher but I'm left not really knowing the cause of that. If GEOS-Chem is excluded as a special case, then arguably you _do_ have correlations with background CH4 and O3! So we need a good reason to include it, or not, when considering OH at least.
- Section 3.1 L454 "OH response... is correlated with the ozone response." I completely agree, but we can't really see that from figure 7, because there is no plot of delta(OH) against delta(O3). I plotted it myself with approximate values from figure 7 for each model and get a nice correlation (just MOZART3 a bit of an outlier from an otherwise straight line, as it happens, though I wouldn't draw any great significance from that). I suggest you add such a plot.
- Section 3.2 L480: "and due to subsidence"; I'm no expert on this but perhaps that should be couched with some uncertainty, as you haven't really quantified whether there is more BC close to the ground because that's where it was emitted or because it sinks. (In principle maybe you could by a comparison with the input emissions distribution fed into the models.) Out of interest a quick web search gave a figure of 0.5 km/month as a subsidence rate in https://doi.org/10.1126/science.aax1748 (but for the stratosphere?). (The BC lifetimes you later give around L492 might then suggest it disappears too quickly to subside much.)
- Section 3.2 L480: "The LMDZ-INCA and GEOS-Chem models show similar responses, with a maximum in April." Looks like it is higher for both in May not April, if my reading of the colours in Fig 8 is correct.
- Section 3.2 L523: Terrenoire (2022) could get a mention too in this paragraph about previous aerosol perturbation work.
- Section 3.2 L535: "are more significant" maybe "are more significant by mass" say, as the BC might be more significant (say) in health terms?
- Section 4 L556/Figure 11: I presume the plots shown are a global average, but it would be good to state this explicitly. Also these ones are not month-specific, so are they annual averages? (I guess so from the "in most other months" comment about GEOS-Chem.)
- Section 4 L587: I just want to check I've understood the units correctly in "a range of 1.22–1.26 % (TgN yr-1)-1 in methane lifetime". So for example if my no-aviation CH4 lifetime was 10 years, it would decrease at the low end by (1.22/100)x10 = 0.122 years for each Tg of N aviation emissions? No need to clarify anything unless that interpretation is incorrect. Though should make it clear it is a decrease (could make numbers negative to emphasise that, as they are in Table 5). Also Table 5 has (-)1.25% not 1.26% for EMAC-NOx, another rounding difference I guess.
- Section 4 L590: looks like an error in "GEOS-Chem sensitivity is ~190 % greater", should be ~90% (=(2.36-1.23)/1.23)x100, taking 1.23 as the approx average of the lower 3 model values)
- Section 5 L659: "and a moderate perturbation (> 40 ppt) confined to the northern midlatitudes in winter" looking back at Fig 2 I'd say that's also true only for the high-traffic areas, which isn't quite clear here
- Section 5 L667: "mean net aviation NOx ERF of 15 mW m-2" but at L592 it is "mean value of 16 mW m-2", i.e. 15 vs 16, again just small differences but they should be consistent, especially with this headline number
- Supplement S3 Figure S6: as the emissions inputs are the same apart from GEOS-Chem (explained L274), I wonder why the MOZART3 ENOx has a notably weaker northern latitude band, whereas the others have similar intensity in two latitude bands?
- Supplement S4 Table 3: I find the "Total CH4" values are ~1.54x the "CH4 direct effect" values. At L316 it says "the methane feedback factor from Sand et al. (2023)". But Sand et al have a model mean CH4 feedback factor of 1.45 not 1.54. Best check this, and also clarify in the table legend what "Total CH4" means.
- Supplement S4 Table 3: The "Total" row has your headline NOx net ERF values. I note these values include "CH4 direct effect" and not "Total CH4", otherwise they would all be lower (0-15 mW/m2 instead of 7-22 mW/m2). Again at L316 it mentions using the methane feedback factor (which I think gives the "Total CH4" numbers), but those feedback-adjusted values do not seem to be used in the headline total ERFs. For the uninitiated (including me), could you explain the use of feedback-adjusted (or not) CH4 ERF I guess in the main text?
Technical corrections
Many of this are very small grammatical points.
- Abstract L19: "and the uncertainty on their Effective Radiative Forcing (ERF) of climate" should be "uncertainty in their..." (in not on), and might be better to say "the uncertainty in their climate Effective Radiative Forcing"
- Abstract L28: "However, results also highlight" -> "However, the results also..."
- Introduction L41: "black carbon (BC) - or soot - particles" to be super-picky those should be em dashes not hyphens
- Section 2.4 L104: "mixted" presumably "mixed"
- Section 2.1.1 L158: "used to EMAC-aer" should be "used in EMAC-aer"
- Section 2.1.3 L189: "organic halogen compound" presumably "compounds"
- Section 2.2 L243: "without any aviation emission" should be "emissions"
- Section 2.3 L288: "derive 5-year averages for each month" might be clearer to say "calendar month" here, as I assume you mean you're making an annual climatology, e.g. averaging the 5 different Januarys together, etc?
- Section 2.3 L291: should be "factor of 5"
- Section 3.1 L328,L334,L337/Table 4: in the text we have "9.9 TgO3" (three times) but the table has "10.0" for GEOS-Chem, so I think the same number has been rounded in different ways with the same number of decimal places. Easier to relate the two if they are presented identically.
- Figure 1: maybe just a preprint thing but in the current PDF the quality isn't too good, looks like it has been compressed coarsely as a JPEG or something. But up to the journal what they expect really!
- Figure 1: I realise all the other stacked plots show the different models vertically going down the page, but this one shows models going horizontally across the page. It's not wrong, and I know it takes a lot of effort to change figures, but if you needed to rework it anyway for other reasons you might consider if it would be nicer to be consistent and have models going down and species going across here too.
- Section 3.1 L390: redundant "for ozone" in "For ozone, the perturbation even peaks ... for ozone"
- Section 3.1 L430: "for the three models" maybe "for all three models"?
- Section 3.1 L432: should make clear the direction of the methane lifetime change depending on the NOx background
- Section 3.2 L480: this paragraph is pretty long, maybe start a new one for SO4?
- Section 3.2 L485: "but also in the whole free troposphere. The maximum shifts..." it isn't totally clear to me whether the second sentence here is talking specifically about EMAC-aer still, or all the models. From the plots I think it is just EMAC-aer. Could clarify that e.g. by doing "but also in the whole free troposphere; (in this model?) the maximum shifts..."
- Section 3.2 L490: "can be caused by a number of factors, as the BC lifetime" intended "such as the BC lifetime"?
- Section 4 L638: "Last, " should be "Lastly, "
- Supplement L33: "in bracket" -> "in brackets"