Anonymous referee #3 has raised a number of important points regarding the revised version of your manuscript. In particular, the referee points to some problems with the evaluation of the model for present day simulations, where the model shows some bias in certain cases, and the lack of an assessment of the model simulations for multiple historical years.

While these are both important and valid criticisms, in my opinion they do not warrant preventing publication of the manuscript in ACP, so I have decided to accept the manuscript in its current, revised form.

Regarding a possible model bias in ozone in comparison with station data, you make a strong case that the stations in question are influenced by small scale NOx titration effects, and regarding a bias in satellite data, you show that this is consistent with previous work.

Regarding the need to simulate two independent years in order to establish how well the model can simulate trends, the reviewer has a very good point here, but in practice this would require a significant amount of extra work to produce an emission inventory for this extra year which would be consistent with the year you have already produced. In my opinion this would be significantly more work than the standard best practice for studies such as yours. Perhaps this may be something worth considering for a future study.

Model simulation of tropospheric ozone, in particular trends in response to changes in precursor emissions, remains a difficult problem. Based on recent work such as the ACCMIP project, it is clear that this is a major shortcoming of the current generation of models. Nevertheless, studies such as yours which investigate potential future responses of air quality to projected changes in emissions continue to be published, and your work is of a sufficiently high quality to be counted among these.