Search

[1] This paper presents a comparison of NO 2 data measured with the Ozone Monitoring Instrument (OMI) on board the EOS-AURA satellite with ground-based direct-Sun Brewer measurement data. Since its deployment in July 2004, OMI has provided more than 2 years of daily high-resolution (∼13 x 24 km 2 at nadir) NO 2 vertical column density maps. We describe the retrieval, which includes an estimation of the stratospheric and tropospheric fraction of total NO 2 columns, the air mass factor (AMF) correction based on detected tropospheric NO 2 enhancements, and the generation of the gridded data product. We present a validation study of the gridded NO 2 data set using data from a Brewer MK3 double monochromator in direct-Sun mode located at NASA Goddard Space Flight Center in Greenbelt, Maryland, USA. Monthly averages of coinciding measurements correlate well (r = 0.9) but OMI data are about 25% lower than the Brewer measurement data (slope 0.75, intercept -0.38 x 10 15 molecules/cm 2 ). We present a detailed uncertainty analysis for both ground and satellite data and discuss the possible reasons for the observed differences.

[1] We present an analysis of in situ NO2 measurements from aircraft experiments between summer 2004 and spring 2006. The data are from the INTEX-A, PAVE, and INTEX-B campaigns and constitute the most comprehensive set of tropospheric NO2 profiles to date. Profile shapes from INTEX-A and PAVE are found to be qualitatively similar to annual mean profiles from the GEOS-Chem model. Using profiles from the INTEX-B campaign, we perform error-weighted linear regressions to compare the Ozone Monitoring Instrument (OMI) tropospheric NO2 columns from the near-real-time product (NRT) and standard product (SP) with the integrated in situ columns. Results indicate that the OMI SP algorithm yields NO2 amounts lower than the in situ columns by a factor of 0.86 (±0.2) and that NO2 amounts from the NRT algorithm are higher than the in situ data by a factor of 1.68 (±0.6). The correlation between the satellite and in situ data is good (r = 0.83) for both algorithms. Using averaging kernels, the influence of the algorithm's a priori profiles on the satellite retrieval is explored. Results imply that air mass factors from the a priori profiles are on average slightly larger (∼10%) than those from the measured profiles, but the differences are not significant.