Overall I think that this is an interesting and well-written paper which is very suited to ACP.

The paper describes a new method for estimating age-of-air which offers reduced uncertainty compared to using a single trace gas such as SF6. The method is described and illustrated in some detail.

I think the paper will be suitable for publication after addressing the points below. My main comment is that paper uses modelled tracers to illustrate the method and I am not (at present) convinced that all sources of uncertainty have been accounted for. I.e. what about possible systematic errors in the model’s ability to reproduce one or more of the individual trace gases used due to errors in kinetic parameters.

Main Points

1) A major aim of the paper is to argue that the new method will reduce uncertainty. However, quantification of the errors (e.g. uncertainty below 0.3 yrs as given in the abstract) is based on a model where it is possible to get the best possible agreement between the AoA from different methods. I think that the quoted uncertainty is mainly based on atmospheric variability. What about uncertainty and bias in the modelled tracers due to e.g. photochemical data, model-calculated loss rates. Related to that, how well does the model reproduce the individual tracer profiles (and tracer-tracer correlations) from the observations such as GLORIA. Please discuss these points.

2) The model simulation used here uses a HALOE-based mid stratosphere upper boundary condition for the (non-zero) tracers which will (a) impose a circulation speed on the tracers which may be inconsistent with the model dynamics (e.g. as pointed out for ERA5) and (b) would limit the use of this method to a period for which such boundary data could be derived. Please discuss the impact of these points.

Minor Points

3) Figure 1 caption and elsewhere. ‘greenish’. This is not a colour. Please choose an actual colour and use that for lines and descriptions.

4) Page 5. Line 109. The range 0 K to 70 K. Please explain more clearly the parameter that you are using to determine the boundary layer up to 1.5 km.

5) Page 6. These loss reactions will, to varying degrees, depend on temperature (including T-dependent cross sections) and the overhead column ozone. How is the OH field calculated? What about loss rates in the troposphere (e.g. tropospheric OH). Please explain if and how these variations are taken into account, and the implications if they are not (e.g. increased uncertainty if these factors affect different tracers in different ways).

6) Pages 6 and 7. Please clarify the length of the model simulation. It must extend to 2022 for the GLORIA flights but, for example, HALOE cannot be providing the upper boundary conditions through to this date.

7) Page 10. Table 2. It would be good to include some estimates of the stratospheric lifetimes of CH4 (much longer than overall total) and N2O (very similar to overall total) from other sources.

8) Page 16. Line 347. It is confusing to say ‘boreal winter season’ and then point to results at high southern latitudes (which is in the summer). Please rephrase.

9) Page 17. Figure 7 caption. Explain what definition is being used for the tropopause (T, lapse rate, PV etc).

Typos

10) Page 2 Line 48. ‘posses’ -> possesses

11) Page 4. Line 78. ‘the’ -> they.

12) Page 6. Line 123. Small s for sulfur.

13) Page 10. Line 232. Need space after ).

14) Page 13. Line 289. The lack of meridional gradients will be partly due to the much decreased emissions by 2011. This should be pointed out.

15) Page 12. Line 288. Lifetime (no space).

16) Page 13. Figure 4 caption. ’25 km in southern hemisphere’ (not ‘on’). Small p in ‘part’.

17) Page 13. Line 293. Add ‘section’ for section 2.2.

18) Page 14. Line 314. Change ‘pretending’ to something like ‘assuming’.

19) Page 15. Line 340. Change ‘high’ to ‘large’.

20) Page 22. Line 432. ‘descend’ -> descent.