Interactive comment on “ Intra-pixel variability in satellite tropospheric NO 2 column densities derived from simultaneous spaceborne and airborne observations ” by Stephen Broccardo

This paper presents results from an airborne field campaign in 2007 over the Highveld region of South Africa. The authors compare NO2 measurements from the iDOAS airborne instrument with NO2 columns from the OMI and SCIAMACHY satellite instruments taken from the www.temis.nl data archive. On the whole, the paper is very well written and organized, and easy to read. The results and conclusions are not very surprising (OMI and SCIAMACHY, with lower spatial resolution observations than iDOAS, are unable to resolve the high values in a plume), but the paper is a nice description of several case studies and a record of the campaign, and provides some interesting context for those looking at satellite measurements over the Highveld region. I would


Interactive comment
Printer-friendly version Discussion paper satellite retrievals should be validated locally.Secondly, this study has a global relevance in the sense that it gives valuable insights with respect to the horizontal variability of tropospheric NO2 columns when observed at high spatial resolution.The satellite instruments OMI and SCIAMACHY are known to have a moderate spatial resolution and it is not well known how this resolution compares to the typical scale of spatial variabilities in the urban tropospheric NO2 column field.This information is relevant for many related studies where satellite retrievals are used (e.g. in order to derive top-down emission inventories) or being validated (e.g.comparison with MAX-DOAS).
The manuscript is generally well written and addresses a relevant topic.However, to my opinion several aspects deserve more attention before publication in AMT.
1 / Four research flights are analysed and discussed in quite some detail, but the interpretation of differences between airborne and satellite retrievals could go more into depth.Based on the present manuscript, the reader might get the impression that systematic differences in tropospheric NO2 columns between satellite and iDOAS can be explained solely (or largely) by horizontal variability in the tropospheric NO2 columns on a scale that is smaller than the typical size of satellite pixels.What could be particularly relevant is to investigate further the potential impact of profile shape assumptions for NO2 and aerosols in explaining the difference between satellite and airborne measurements over the most polluted regions.Close to major point sources one may expect not only to find locally quite extreme tropospheric NO2 column abundances, but at the same locations also the NO2 profile shape may deviate considerably from other places further away from the main sources.In this context, it may be relevant to distinguish explicitly four profiles: the true profile at the spatial resolution of the aircraft measurements (P_true_air), the profile used in the airborne retrieval (P_prior_air), the true profile at the resolution of the satellite measurements (P_true_sat) and the profile used in the satellite retrieval (P_prior_sat).Differences in tropospheric NO2 column retrievals (space borne versus airborne) cannot be interpreted without taking into including these four profile shapes in the discussion: how much do the authors think

Interactive comment
Printer-friendly version Discussion paper P_true_air can deviate from P_prior_air close to the main sources (same for P_true_sat and P_prior_sat).Furthermore the AMF is not only affected by the (different) profile shapes, but also by the block-AMFs, and these are not identical for the satellite and the airborne point of view.This should be taken into account as well.
Despite the length of this comment, I would suggest to add just one or two paragraphs addressing this point and providing some first order estimates.It could for instance be enlightening to the reader if the impact of making wrong profile shape assumptions is worked out for one hypothetical scenario.For instance (it is up to the authors to deviate from this concrete suggestion): scale height for P_true_air is 0.2 km (e.g.close to strong isolated source); scale height for P_true_sat is 0.4 km (averaged over a larger region the true profile is less dominated by the local source); scale height for P_prior_air is 0.6 km (this number is used in present study); scale height for P_prior_sat is taken from profile used in DOMINOv2 product over this region.Block_AMFs should be applied for a representative SZA and surface reflectance.When combined, this information should provide the reader with a first order quantitative estimate of local AMF fluctuations near a strong plume: to what extent can this explain the discrepancy between the satellite and airborne retrieval?Or perhaps it is concluded that -when taking this effect into account -the observed discrepancy increases even further.
2 / Although aerosols are not entirely neglected in this study, they receive little attention considering the fact that for all four flights -each covering distances of hundreds of kilometers -just one fixed value is assumed for the AOT.It is quite remarkable that the uncertainty range of the AMF is derived using a look-up table that does include variability of the single scattering albedo, but not of the AOT.Over a region where the variability in NO2 is so large, it is almost unthinkable that the AOT can be approximated with a single value.To some extent the same argumentation as given above (in the vicinity of a strong pollution source the NO2 profile shapes may show considerable spatial variability) can be given here as well: in the same region the AOT may show a substantial variability (although probably less extreme than for NO2).In my opinion