The revised manuscript has improved. The authors provided more insight into the causes for the differences between the NOx emission inversions, and how these differences lead to differences in simulations of tropospheric ozone.

The technical clarifications are much appreciated: in particular the case for using all available OMI measurements in a data assimilation scheme, rather than sampling rows that were outside the row anomaly throughout the entire OMI mission, has been well-argued (Figure R1).

The discussion of GEOS-Chem SCDs is also useful. The finding that GEOS-Chem NO2 SCD generated with the NASA scattering weights exceed the GEOS-Chem SCDs generated with their DOMINO counterpart, is important. It indicates structural differences in the presumptions about the satellite sensitivity to NO2 in the lower troposphere: this is presumed to be stronger in the NASA retrieval than in the DOMINO retrieval. This finding could and should be highlighted more in the final paper.

Remaining issues
1. The concepts of scattering weights, averaging kernels, and vertical sensitivity are used too loosely in the manuscript. In the DOAS-formalisms discussed in Palmer et al. [2001] and Eskes and Boersma [2003], and recently summarized in Chance and Martin [2017], a clear distinction is made between scattering weights and averaging kernels.

Scattering weights (w) are related to SCDs (SCD = | w(z) n(z) dz), with | the integral sign, n(z) the a priori NO2 profile. But averaging kernels (a) are related to VCDs (VCD = | a(z) n(z) dz). Their relationship is via the AMF: a(z) = w(z)/M, with M the AMF. I recommend to first clearly define, and then carefully check every use of the term ‘scattering weights’, ‘averaging kernels’, and ‘vertical sensitivity’, in the manuscript. This is important in order to prevent the wrong use of these concepts.

2. It remains unclear how the uncertainty in the AMF (the observation operator in generating the GEOS-Chem SCDs) is accounted for in the assimilations. This is important because, together with the estimated uncertainty on the model state, it determines how strongly OMI is driving the data assimilation. In other words, I suggest the authors provide the relative weight of the OMI SCD vs. the GEOS-Chem SCD in the assimilation scheme.

3. I strongly disagree with the phrase that “daily NO2 column densities from OMI are underestimated to the diurnally varying ground-based retrievals [Herman et al., 2019].” As stated in my previous review, OMI is simply measuring at 13:30 hrs, close to the diurnal minimum in NO2 columns. To then call this an “underestimate” is misleading.

4. L26: please clarify what is meant with “current hard-constraints on NOx diurnal variation”. There is no constraint from the satellite data, so in essence the prior diurnal variation is used. Please rephrase to make this clear.

Reference
Chance, K., and Martin, R. V.: Spectroscopy and Radiative Transfer of Planetary Atmospheres, Oxford University Press, 2017.