On the use of MOZAIC-IAGOS data to assess the ability of the MACC reanalysis to reproduce the distribution of ozone and CO in the UTLS over Europe

MOZAIC-IAGOS data are used to assess the ability of the MACC reanalysis (REAN) to reproduce distributions of ozone (O3) and carbon monoxide (CO), along with vertical and inter-annual variability in the upper troposphere/lower stratosphere region (UTLS) over Europe for the period 2003–2010. A control run (CNTRL, without assimilation) is compared with the MACC reanalysis (REAN, with assimilation) to assess the impact of assimilation. On average over the period, REAN underestimates ozone by 60 ppbv in the lower stratosphere (LS), whilst CO is overestimated by 20 ppbv. In the upper troposphere (UT), ozone is overestimated by 50 ppbv, while CO is partly over or underestimated by up to 20 ppbv. As expected, assimilation generally improves model results but there are some exceptions. Assimilation leads to increased CO mixing ratios in the UT which reduce the biases of the model in this region but the difference in CO mixing ratios between LS and UT has not changed and remains underestimated after assimilation. Therefore, this leads to a significant positive bias of CO in the LS after assimilation. Assimilation improves estimates of the amplitude of the seasonal cycle for both species. Additionally, the observations clearly show a general negative trend of CO in the UT which is rather well reproduced by REAN. However, REAN misses the observed inter-annual variability in summer. The O3–CO correlation in the Ex-UTLS is rather well reproduced by the CNTRL and REAN, although REAN tends to miss the lowest CO mixing ratios for the four seasons and tends to oversample the extra-tropical transition layer (ExTL region) in spring. This evaluation stresses the importance of the model gradients for a good description of the mixing in the Ex-UTLS region, which is inherently difficult to observe from satellite instruments.