The goal of Wright et al aim is to estimate how various climate indices (NAO, ENSO, the annual cycle, …) affect the duration of transatlantic flights, and the associated economic costs. The editor has obviously decided the subject is within scope, so I make no comment on this and simply comment on the paper as is.

This paper was a pleasure to read. There are many complications, some obvious and some more subtle, that need to be dealt with before one can get down to business and just regress the indices against flight times. However, at every point the authors tackle these with great transparency and forthrightness, and include multiple sensitivity tests to pre-empt any concerns about robustness to their chosen methodology/assumptions. At the same time, great effort has been made to keep things easy to read, including (importantly!) the figures. Honestly, it is hard to find anything to criticise in this submission.

However, I have a few comments I consider minor. These mostly relate to round-trips, which were the only aspect that caused some confusion. I am recommending minor revisions and look forward to reading the revised manuscript.

Best wishes,

Kristian Strommen

MINOR COMMENTS

1. Should the anomalous duration of a round-trip equal (in expectation) the sum of westward and eastward anomalies? If not, why not? Intuitively this seems like it should be the case. L200 says the two legs of your synthetic round-trips are separated by half a day, and all the indices you look at have strong autocorrelation on such timescales, including the NAO. Thus, the value of each index should be approximately the same on both legs, and the effect on the round-trip duration should therefore be the sum of the effect on the two separate directions. Looking at Figure 5 suggests this might be true modulo some noise (caveat: I didn’t check all cases carefully).
   
   
   
   If this is the case, then it might be an opportunity to simplify the paper, since you could state that this is the case and therefore just focus presentation on east/westwards impacts. I would recommend considering this, because various questions about round-trips was basically the only thing detracting from the paper for me. It would also let you avoid the need to use a synthetic data set, the use of which raises questions about the computation of statistics. For example, what is the effective sample size of the synthetic data set, given that the same flight is used in multiple round-trips? Seems tricky to work out, but this question comes up because you sometimes find statistical significance for round-trip effects but not for either of the eastwards/westward legs, and I was left unsure if this was due to the synthetic nature of the data.
   
   
   
   If this is not the case, then can you explain why?
   
   
   
   
2. L301: “In all cases, the adjusted R^2 estimator suggests that our indices describe nearly a third of total variance” I am unsure if I’ve misunderstood something here. The values of R^2 in Figure 5 are often very small (0.03, 0.04 etc.), and the average value across all the cases considered is around 0.1. The maximum attained is 0.27, which is closer to a fourth than a third. Your sentence here therefore seems to oversell the actual results quite strongly, especially the way you say “in all cases”! You could rather say that the values range from 3 to 27% explained variance, with an average of 10%.

1. On page 4 where you introduce the NAO index, you should say that a positive NAO is also associated with higher jet speeds, not just latitudinal shifts. In fact, on daily timescales it can often be just one and not the other, as the jet latitude and jet speed indices are largely independent (https://doi.org/10.1175/JCLI-D-11-00087.1). For me, this makes it much clearer why the NAO has such a big impact on flight times, and in particular why there is such a strong directionality: if the jet is in the same spot but blowing much faster then it’s clear why eastward flights speed up and westwards flights slow down and the effect on round-trips ought to mostly cancel.

1. I was left wondering if all the indices you consider affect flight times primarily by modulating the speed of the jet. One could imagine testing this by computing Woollings’ jet speed index and seeing how much of all the effects can be explained by variations in jet speed. I think this might be beyond scope, so could be left as a question for follow-up work. Some brief discussion on this point in the Discussion at the end would, in any case, be nice, since the physical interpretation is otherwise relegated to cited papers.

Reveiw of "The influence of climate variability on transatlantic flight times" by Corwin J. Wright et al.

The authors use IAGOS data to examine the effects of different dynamical/climate aspects of the atmosphere on transatlantic crossing times. Overall I find this manuscript to be of interest to the scientific community and well written. I recommend publication assuming that the editors consider that it fits within the scope of the journal.

In general, I find that the article is well argued with appropriate supporting evidence and that the authors are up-front about the limitations of the study. The main concern I have with this paper is separating the dynamical signals given by the regressions against the various climate indices, and the extent to which the aircraft operators are flexible to respond to these changes given the many operational factors. As the authors state and show in figure 2, the North Atlantic Organised Track System tracks are optimised according to the position of the jet stream and other weather systems in order to capitalize on tail winds and avoid head winds, or avoid turbulence or icing or other weather systems. Figure 2 shows a large lateral spread in the tracks. I think that it is easy to believe that a lateral shift in the flight tracks  reflects climatic variability  and impacts crossing times. It might also be possible that the preferred flight level is also adapted to take account of favorable or unfavorable conditions but I wonder if the flight-levels are more operationally constrained than the lateral flight tracks. Therefore, my main concern comes at the end of the end of the article regarding the discussion on the regression of flight altitudes (section 7) as it is not clear to me how this analysis works given the step-wise climbs and the nature of the fixed flight-levels.

At the start of a flight when the aircraft is heavy, the flight-altitude is lower. As fuel is burned and the aircraft becomes lighter, the pilots are able to make step-wise climbs until the final maximum altitude is reached. The flight altitudes are constrained by the flight separation minima required for safety and there are well established and fixed flight-levels on which aircraft operate (eastbound -odd flight levels, westbound even flight levels).   Thouret et al. (1998) (JGR, vol 103, pages 25653- 25679) give a short description of the flight-levels.  The point on the trajectory at which the aircraft is able to make the step-wise climb depends on the engine performance (which changes with altitude) and load-factors as well as the air-traffic control regulations. Engine performance has evolved over time, and load-factors have changed over time being sensitive to demand ( e.g. more passengers in summer than in winter leading to an annual cycle) and changing after COVID or economic cycles. 

I think that more analysis is necessary in section 7,  to untangle these different factors (engine performance, load-factors, air traffic control requirements) and determine anything useful. My suggestion would be either to discuss more about flight operations in the article which would explain some of these issues to the reader, or simply remove this section on flight altitudes as currently I think that the physical link with climate indices, particulary TSI and ENSO in spring (conclusion 9.1 point 6)   is rather weak i.e. the aircraft are not flying higher because of the solar cycle (TSI) or EL Nino in spring but because of something else or an artefact in the data. 

Reveiw of "The influence of climate variability on transatlantic flight times" by Corwin J. Wright et al.

Reviewer: Mark Baldwin

This comment builds on reviews RC1 and RC2.

I think that this is a very good contribution to the literature, and that it should be published after the authors consider all the reviewer suggestions.

Comments:

1) I think that an important addition for most readers would be to add a brief discussion explaining that the RT flight time will always increase if there is an increase in the background wind speed in either direction. Compared to flying in still air, a uniform background eastward wind will always decrease the eastward flight time and increase the westward flight time, but the total RT flight time will increase. (imagine the extreme case in which the wind speed is equal to the airspeed of the aircraft—so in one direction the plane would make no progress.)

v - airspeed f flight

dv = uniform change in background wind

D = one way distance

Total time = D/(v-dv) + D/(d+dv) ≥ 2(D/v)

This would help to explain a lot of your results and may help at the end of the paper to explain differences with previous publications. Of course, actual flight paths attempt to mitigate this.

2) Limited data set. I looked up approximaely how many Europe - US flights actually took place between August 1994 and March 2024. There were approximately 7-9 million direct transatlantic flights operated between Europe and the United States. This estimate is based on analysis of available flight data, historical trends, and consideration of major aviation disruptions during this 30-year period.

The data used in this study neglects ~99.8% of the actual transatlatic flights. This makes me wonder how the results might differ if all the flights were available for analysis. You do mention this in 8.2 L461, but I think that you should state how small the fraction of flights available to analyze was compared to the number that took place.

3) The underlying question is: how does the atmosphere change with the 4 indices (plus trend) and how does this affect the optimal flight times for round trip transatlantic flights? I think that this is the fundamental question, and that the analysis of actual flight data has several issues that ou mention, plus those mentioned by RC2.

I suggest a comparison to a very simple theoretical flight. The basic idea would be to select several city pairs, and simply assume that flights were at some constant altitude and air speed. Using reanalysis (4X daily?) simply calculate the great circle flight path time in both directions at the reanalysis times. This comparison might reinforce the results you have obtained. I expect annual, ENSO, and NAO results to be verified. But trends would be very interesting. This approach could help sort out the Honolulu results that differ from previous wpublications (Discussion, ~L444). Stronger winds should increase the RT flight times.

4) Please convert the solar cycle altitude result to meters different (not just standard deviation) in the Abstract ~Line 12. ~Line 415 you do discuss this, but I am left wondering if this effect is real.

5) This suggestion is probably for a future paper, since it would be fairly involved: Use reanalysis to calculate millions of idealised RT flights, assuming all flights are optimised to minimise flight times. That way you would eliminate all the operational issues and address only the atmosphere. It would focus on how the atmosphere changes for ENSO, NAO, TSI etc.