| 28 Jul 2026
Can we achieve significant climate mitigation by optimising only for contrails?
Abstract. Contrail-avoidance is a widely studied measure to reduce the non-CO2 climate effect of aviation. However, some mitigation gain is likely to be compromised through increased emissions (CO2, NOx, and H2O) and their climate effect. Here, we analyse the impact of contrail-optimised flights on the overall net climate benefit. We evaluate more than 4,000 flights with large contrail impact traversing Northern Europe in 2023, each with two trajectory options: the filed trajectory submitted to the network manager and a contrail-optimised trajectory, allowing only vertical deviations and a maximum fuel penalty of 2%. Our model setup is based on the European Unions's non-CO2 monitoring, reporting and verification (MRV) framework, modelling contrails with the contrail cirrus prediction model (CoCiP) and employing the algorithmic climate change functions (aCCFs) for NOx and H2O effects. This paper highlights three points: First, while 93% of contrail-optimised flights emit more NOx (on average +2.5%), they emit it at lower, less climate-sensitive altitudes, so only 65% of flights exhibit an increased NOx climate effect (+1.1%, measured in efficacy-weighted global warming potential over 100 years, EGWP100), with considerable spatial and daily variability. Second, we find a risk of only 2% that the net climate benefit is not achieved when optimising for contrails alone, whereas using forecast weather data poses a far greater risk, causing optimisation to fail in ~15% of cases. The net climate benefit, however, is reduced from −16.5% to −11.1%, measured in EGWP100, when including NOx and H2O effects in the evaluation. Third, we find that the results are consistent to the choice of climate metric, with net climate benefit rates of −11.1%, −12.4%, and −16.2% obtained for EGWP100, the average temperature response over 100 years (ATR100), and EGWP20, respectively. We conclude that savings attributed to contrail-avoidance under a hard constraint on extra fuel and considering flights with large contrail impact only, exceed the NOx and H2O penalties. We do, however, recommend further research that accounts for model uncertainties, in order to determine whether the clear dominance of contrail over NOx (and H2O) effects identified here for contrail-optimised trajectories remains robust.
Competing interests: At least one of the (co-)authors is a member of the editorial board of Journal of Environmentally Compatible Air Transport System.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.
This manuscript by Smretschnig et al. presents a study of the impact of contrail avoidance operations on total climate impact of flights, including CO2, NOx, and H2O effects. The study is based on a simulation of the deviation of more than 4,000 flights forming warming contrails in the Borealis airspace. The full simulation is described in the companion paper Piontek et al. (2026, https://doi.org/10.5194/jecats-2026-14, in review). Several conclusions are drawn, in particular, (1) the additional NOx impact because of the deviation is not always positive, (2) the deviation is still beneficial when accounting for all species, and (3) the deviation is still beneficial when considering different climate metrics. This study is important given the current focus on contrail avoidance from multiple stakeholders.
However, I have major concerns about the methodology described in the paper. In my opinion, it currently does not allow the conclusions to be fully supported. With substantial revisions, in particular an analysis of the sensitivity of the results to the aCCs model configuration, the paper might be suitable for publication.
Therefore, I recommend major revisions before publication. Below I provide additional details about the major concerns I have. I also provide more minor comments throughout the manuscript.
Major comments
1) aCCFs model. In this study, the climate impact of NOx is estimated using the aCCFs v1.0A, described in Matthes et al. (2023). I have several concerns with this.
First, the scaling factors (Table D1) and the resulting climate effects (Figure D1) for the three versions of aCCFs substantially differ. Since this study investigates the impact of NOx on the outcome of flight rerouting, this should be the subject of a dedicated section, with a sensitivity analysis on the model version choice (rather than the statements P19L384 and in the limitations section). Given the magnitude of these differences, the conclusions could completely change depending on this choice.
Second, the rationale behind including data points at pressures below 200 hPa is unclear. As stated P9L231, “Whether extrapolation beyond this range is valid remains unknown”. It is consistent to both acknowledge this and to draw conclusions from these data points. Moreover, you also state P9L233 “This is taken into consideration in the evaluation”, but I couldn’t find how, nor any related sensitivity study. At the very least, the validity range should be clipped to 200 hPa, even if this means losing a significant part of the data, or the results with/without these data points could be shown.
Third, the preliminary result of Frömming et al. (2026), stating that “The use of O3 aCCFs outside 200-300 hPa is not recommended”, should be mentioned and acknowledged. I’m wondering whether clipping the data to this range would be sensible, but I leave it to you.
Fourth, the Matthes et al. (2023) manuscript was not accepted for publication. It should therefore not be presented as a model that was validated through peer-review (P9L219, Table 1, Table D1, P32L625, etc.), as it is compared to previous versions that were indeed approved through a peer-review process. I understand that choosing the v1.0A version is meant to stick to the MRV specifications, but currently it seems like this choice is based on previous peer-reviewed literature, which is not the case. I would recommend not using this version at all, but I leave this decision to you.
2) Backward calculation factors. With the current explanation, I can’t understand what these factors are and how they are calculated. Given their importance (Table D2), please give a much more detailed explanation of their significance, role, and calculation methodology. Why are they needed in the first place?
3) Forecast vs evaluation. Major differences in the modelling choices of contrails arise between the planning mode and the evaluation mode, for example, the maximum lifetime of contrails goes from 14h to 12h, the integration time from 5 to 10 minutes, etc. Most importantly, the efficacy value changed from 0.42 to 0.21. What is the reason behind these different assumptions? Otherwise, a single consistent set of assumptions should be used throughout.
4) CO2 AGWP values. In the study, you use AGWP20 and AGWP100 values for CO2 from Gaillot et al. (2023), while you use the Dahlmann et al. (2025) conversion factors. As stated in your RQ3 comment (P26L493), there is a significant discrepancy, but I struggle to understand why you did not use directly factors from Dahlmann et al. (2025), their Table A1. I suggest updating your climate metric analysis using these factors.
5) Purpose of Section 3.1. Section 3.1. is mainly descriptive, with no analysis of the climate effects of each species (as stated in the title) and no analysis of the origin of the differences to the references (e.g. P13L323). Its objective, including that of the comparison with global and regional references, should be stated clearly, together with the conclusions drawn from it. Since understanding the climate effect of each species is essential for the questions addressed in the following sections, I recommend reworking this section accordingly.
Minor comments
P1L2 It seems that “Navigational contrail avoidance” rather than “contrail avoidance” now predominates in the literature, I would suggest using the former.
P1L13 This statement is not clear, what does “using forecast weather data” means in this context? Upon reading the manuscript I understand it’s compared to reanalysis data, but in the abstract it should be clear.
P1L15 There’s likely a grammar problem, I struggle to understand this sentence.
P2L38 to L43 WHile I understand the value of large scale for capturing conflicts, this study does not discuss the impact on complexity, it is thus misleading to mention it. I would remove these sentences.
P3L88 As you say later on, this should read “changing the climate indicator” rather than “climate metric”.
P4L94 Is it “for instance”? You speak in the previous sentence of “ways of dealing with uncertainties” while here you state that uncertainty should not prevent action.
P4L100 You already introduced AGWP, so please replace it with “EGWP100” and “EGWP20”.
P4L113 Are the “most warming flights” the actual most warming flights (no matter if the contrail was formed in the Borealis airspace), or those that formed the most warming contrails in the Borealis airspace?
P4L124 Please add “ECMWF HRES” for preciseness.
P5L126 I guess something is missing from this sentence. Compared to the reanalysis? You mean that 80-90% of benefit is still achievable when comparing reanalysed results with reroutings planned using forecast data?
P6L137 I would not state the contrail efficacy value here, it’s not a constraint but a modelling choice.
P6L144 What are those “acceptable limits”?
P8L174 I guess it’s the other way around, the take-off mass is taken from the BADA default values and the aircraft type from the dataset.
P8L175 Did you use ERA5 with model levels or pressure levels?
P8L184 The SAc indicates if a contrail is formed, not whether or not it is persistent.
P8L185 Contrails can in some cases not survive the wake vortex downwash phase.
P8L186 ice crystals -> ice crystals properties
P8L192 “radiative forcing” rather than “radiation change”
P8L195 It’s not a conservative estimate, some contrails live longer than that. I would simply remove this statement.
Table 1. Where contrails aCCFs used in this work?
Table 2. Isn’t this Table a copy of the Dahlmann et al. (2025) Table 3? Why reproduce it here?
P11L262 Here and throughout this section: temperature, temperature response and ATR are very different things. Please use ATR consistently throughout the section.
P13L326 Why couldn’t you apply your efficacy value to their GWP100?
P15L338 So are you implying that small adjustments lead to lower NOx impact? If so, please say it clearly.
P16L350 It is not the case for the Latvia airspace, why is that?
Figure 5. I’m wondering whether a small description of each airspace would be useful. E.g., maybe there’s a correlation with the average flight altitude in this region? Or with its latitude? In fact I’m wondering why there are such important differences between these airspaces.
P18L375 Rather than 733 kgCO2e, it seems to be 669 kgCO2e according to Table 5.
P18L375 Indeed but there is an obvious effect of the amplitude of the initial NOx impact. Maybe you could indicate what are the relative changes for each species?
P18L383 It is strange to see a positive value of delta contrail impact, maybe remind that this is a post-flight analysis.
P19L386 Typo: two “using”
P19L389 This observation lacks a conclusion. What do you imply? Is accounting for weather uncertainty more important than accounting for the combined NOx and H2O effect?
Figure 8. I’m not sure about the added value of this Figure. It is almost not discussed in the text. I would recommend either to remove it, or to develop the part that discusses it and draw conclusions from the analysis.
Figure 9. Could you add the number of flights on the right hand part of the plot?
P21L407-409 I think the origin of these differences is not sufficiently explained / explored. Could you give possible reasons for such discrepancies?
P24L461 Typo: notes -> noted
P24L465 This paragraph comes out of the blue for a study that is solely discussing climate impacts. If you’d like to keep this comment, please move it in the concluding section.
P24L483 But in that case they wouldn’t be deviated, would they?
P26L487 “-0.064” rather than “+0.064”
P26L488 Maybe remind what the bounds are quantifying? I think it’s the 1 sigma uncertainty?
P26L498 Here you can explicitly say “compared to reanalysis data” or similar.
P27L520 Something’s missing in this sentence.
P27L528 The point of the Akhtar Martinez et al. (2025) study is that contrails may live longer than CoCiP simulates, thus comparing this assertion to the actual lifetime of contrails you simulated with CoCiP cannot lead to the conclusion that your results would only be slightly changed.
P27L529-531 Indeed, but what does it imply for your study?
P27L535 How is this a limitation then? I would simply state, in the methods or discussion, that because the flights you analysed fall within the soot-rich regime, the vPM activation is of no concern to you.
P29L585 I think this way of selecting deviated flights is very important and should be mentioned in the main text.