The focus of this paper is an exploration of surface ozone levels predicted over India and adjacent regions using the WRF-Chem model, with two different chemical mechanisms, and three different emission inventories. The resulting predicted ozone fields are compared to each other, and to observations. The manuscript has received two detailed reviews, which raise a number of issues, the majority of which the authors have addressed, although with some areas where further refinement would improve the quality of the manuscript, and noted below.

The more substantive comment relates to the overall focus of the work. In particular, the referees challenge the use of emission inventories from very different years (2006, 2010, 2012), in a simulation for 2013, and which is compared with observational data acquired mostly between 2009 and 2013, but in some cases dating back to 2004 or earlier. The concern arises as there are substantial and well documented trends in ozone precursor emissions, and ozone levels, over this region. There are secondary concerns over the focus upon just O3 rather than NOx and VOC as the model metrics (although addressed to an extent in the revised version), and station specific dependencies in some locations which regional model resolution cannot address.

The results in the amended manuscript (i.e. most figures, tables) focus upon the difference in modelled ozone and ozone tendencies between the different inventories and chemical mechanisms, and exploration of the factors underpinning these. This is the appropriate focus for the manuscript, as the results cannot be readily compared with absolute ozone levels using ozone observations 15 years adrift from the inventory, and different meteorological years etc. However the paper text still focuses substantially upon the comparisons with measurements – this focus should be adjusted, to comparison between inventories and chemical mechanisms, in a revised manuscript. The authors response to this point raised by the external referees – that better data do not exist – may be correct, but should lead to the conclusion that the comparison cannot therefore usefully be performed (at least, without explicit consideration of precursor and ozone observational trends, local station factors, and NOx and VOC measurements).

Additional minor changes recommended :

Referee 1 comments

Comment 1 – focus should be upon ozone production between models, not absolute ozone levels. The (many) uncertainties in model chemistries need to be outlined in the paper, both between the two mechanism used, and overall

Comment 3 – the comparison with observations is of limited use to guide the choice of “correct” answer given the issues outlined above – better comparison would be with newer emissions data. Local surface site factors (eg roadside etc) will dominate the NOx levels and hence O3 levels – which cannot be captured in a regional model. Observations of “NOx” are probably (mostly) NOy, if heated Mo converter instruments are used for NO2 detection.

Comment 5 – please include an example of actual j values fom the two photolysis schemes to allow the reader to understand the impact – e.g. a simple comparison of jNO2 and jO(1D) for a surface point in the centre of the domain at midday.

Comment 11 – it would be useful to include the explanation for HCHO/NOy ratio in this manuscript. The work of Sillman et al refers primarily to a US context where VOC emissions (and hence VOC oxidation yields, ie HCHO formation) may be quite different to India – and referred to a specific trajectory duration / timepoint post emission - so the transition from NOx to VOC limitation may not occur at 0.28 in India, and will vary with post-emission processing duration.

Comment 13 The decision to turn off the aerosol radiation feedback is appropriate for the focus upon comparison between inventories (but not for comparison with absolute observed O3 levels), this should be more clearly brought out / explained in the manuscript

Comment 14 – Give specific values for the modelled vs observed water vapour data (ie comparison of means at surface / through profile, sonde profiles used etc).

Referee 2 comments

Comment 1 – please advance explanations for the difference in model performance (between models, not vs observations) as this is the focus of the study and paper title

Comment 2 – please expand on the impact of biomass burning effects in this work [not from literature] – ie from that aspect of the different emission inventories ?

Comment 8 – the pre 2004 data may not be appropriate to use for the reasons noted above. If they are to be retained explicit consideration of the trends in O3 is needed – however the manuscript focus should compare the models.

Abstract – needs to mention the second chemical mechanism, ie MOZART. Discussion of absolute model performance should be reduced given issues noted above.

Thank-you for the substantial changes to the manuscript. Two minor revisions are requested

-To comment clearly upon the comparison of modelled and measured NOx mixing ratios (as shown in Table S1) within the main manuscript, e.g. in sections 3.1 and 6

-Technical correction - clarify that figs S2 onwards refer to O3

The authors have addressed the final technical comments upon the revised manuscript, which is now put forward for publication in ACP.