spheric models to reproduce observed atmospheric perturbations generated by SPEs, particularly with respect to NO y and ozone changes. We have further assessed the meteorological conditions and their implications for the chemical response to the SPE in both the models and observations by comparing temperature and tracer (CH 4 and CO) fields.Simulated SPE-induced ozone losses agree on average within 5 % with the observations. Simulated NO y enhancements around 1 hPa, however, are typically 30 % higher than indicated by the observations which are likely to be related to deficiencies in the used ionization rates, though other error sources related to the models' atmospheric background state and/or transport schemes cannot be excluded. The analysis of the observed and modeled NO y partitioning in the aftermath of the SPE has demonstrated the need to implement additional ion chemistry (HNO 3 formation via ion-ion recombination and water cluster ions) into the chemical schemes. An overestimation of observed H 2 O 2 enhancements by all models hints at an underestimation of the OH/HO 2 ratio in the upper polar stratosphere during the SPE. The analysis of chlorine species perturbations has shown that the encountered Published by Copernicus Publications on behalf of the European Geosciences Union. 9090 B. Funke et al.: HEPPA intercomparison study differences between models and observations, particularly the underestimation of observed ClONO 2 enhancements, are related to a smaller availability of ClO in the polar night region already before the SPE. In general, the intercomparison has demonstrated that differences in the meteorology and/or initial state of the atmosphere in the simulations cause a relevant variability of the model results, even on a short timescale of only a few days.

Abstract. Global distributions of profiles of sulphur hexafluoride (SF6) have been retrieved from limb emission spectra recorded by the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) on Envisat covering the period September 2002 to March 2004. Individual SF6 profiles have a precision of 0.5 pptv below 25 km altitude and a vertical resolution of 4–6 km up to 35 km altitude. These data have been validated versus in situ observations obtained during balloon flights of a cryogenic whole-air sampler. For the tropical troposphere a trend of 0.230±0.008 pptv/yr has been derived from the MIPAS data, which is in excellent agreement with the trend from ground-based flask and in situ measurements from the National Oceanic and Atmospheric Administration Earth System Research Laboratory, Global Monitoring Division. For the data set currently available, based on at least three days of data per month, monthly 5° latitude mean values have a 1σ standard error of 1%. From the global SF6 distributions, global daily and monthly distributions of the apparent mean age of air are inferred by application of the tropical tropospheric trend derived from MIPAS data. The inferred mean ages are provided for the full globe up to 90° N/S, and have a 1σ standard error of 0.25 yr. They range between 0 (near the tropical tropopause) and 7 years (except for situations of mesospheric intrusions) and agree well with earlier observations. The seasonal variation of the mean age of stratospheric air indicates episodes of severe intrusion of mesospheric air during each Northern and Southern polar winter observed, long-lasting remnants of old, subsided polar winter air over the spring and summer poles, and a rather short period of mixing with midlatitude air and/or upward transport during fall in October/November (NH) and April/May (SH), respectively, with small latitudinal gradients, immediately before the new polar vortex starts to form. The mean age distributions further confirm that SF6 is destroyed in the mesosphere to a considerable degree. Model calculations with the Karlsruhe simulation model of the middle atmosphere (KASIMA) chemical transport model agree well with observed global distributions of the mean age only if the SF6 sink reactions in the mesosphere are included in the model.

Good agreement with observations is found for inert SF6 transport. However, when mesospheric loss is included in the model, significant deviations are found for polar winter observations above 25 kin. Chemical loss by electron attachment without reactions yielding SF6 again is not compatible with observations. The atmospheric lifetime of SF6 spans 400 to 10,000 years, depending on the assumed loss mechanism and the value for the electron density in the stratosphere.

Abstract. During several balloon flights inside the Arctic polar vortex in early 2003, unusual trace gas distributions were observed, which indicate a strong influence of mesospheric air in the stratosphere. The tuneable diode laser (TDL) instrument SPIRALE (Spectroscopie Infra-Rouge par Absorption de Lasers Embarqués) measured unusually high CO values (up to 600 ppb) on 27 January at about 30 km altitude. The cryosampler BONBON sampled air masses with very high molecular Hydrogen, extremely low SF 6 and enhanced CO values on 6 March at about 25 km altitude. Finally, the MIPAS (Michelson Interferometer for Passive Atmospheric Sounding) Fourier Transform Infra-Red (FTIR) spectrometer showed NO y values which are significantly higher than NO y * (the NO y derived from a correlation between N 2 O and NO y under undisturbed conditions), on 21 and 22 March in a layer centred at 22 km altitude. Thus, the mesospheric air seems to have been present in a layer descending from about 30 km in late January to 25 km altitude in early March and about 22 km altitude on 20 March. We present corroborating evidence from a model study using the KASIMA (KArlsruhe SImulation model of the Middle Atmosphere) model that also shows a layer of mesospheric air, which descended into the stratosphere in November and early December 2002, before the minor warming which occurred in late December 2002 lead to a descent of upper stratospheric air, cutting off a layer in which mesospheric air is present. This layer then descended inside the vortex over the course of the winter. The same feature is found in trajectory calculations, based on a large number of trajectories started in the vicinity of the observations on 6 March. Based on the difference between the mean age derived from SF 6 (which has an irreversible mesoCorrespondence to: A. Engel (an.engel@meteor.uni-frankfurt.de) spheric loss) and from CO 2 (whose mesospheric loss is much smaller and reversible) we estimate that the fraction of mesospheric air in the layer observed on 6 March, must have been somewhere between 35% and 100%.

Abstract. We compare simulations from three high-top (with upper lid above 120 km) and five medium-top (with upper lid around 80 km) atmospheric models with observations of odd nitrogen (NO x = NO + NO 2 ), temperature, and carbon monoxide from seven satellite instruments (ACE-FTS on SciSat, GOMOS, MIPAS, and SCIAMACHY on Envisat, MLS on Aura, SABER on TIMED, and SMR on Odin) dur- Larger discrepancies of a few model simulations could be traced back either to the impact of the models' gravity wave drag scheme on the polar wintertime meridional circulation or to a combination of prescribed NO x mixing ratio at the uppermost model layer and low vertical resolution. In March-April, after the ES event, however, modelled mesospheric and stratospheric NO x distributions deviate significantly from the observations. The too-fast and early downward propagation of the NO x tongue, encountered in most simulations, coincides with a temperature high bias in the lower mesosphere (0.2-0.05 hPa), likely caused by an overestimation of descent velocities. In contrast, uppermesospheric temperatures (at 0.05-0.001 hPa) are generally underestimated by the high-top models after the onset of the ES event, being indicative for too-slow descent and hence too-low NO x fluxes. As a consequence, the magnitude of the simulated NO x tongue is generally underestimated by these models. Descending NO x amounts simulated with mediumtop models are on average closer to the observations but show a large spread of up to several hundred percent. This is primarily attributed to the different vertical model domains in which the NO x upper boundary condition is applied. In general, the intercomparison demonstrates the ability of stateof-the-art atmospheric models to reproduce the EPP indirect effect in dynamically and geomagnetically quiescent NH winter conditions. The encountered differences between observed and simulated NO x , CO, and temperature distributions during the perturbed phase of the 2009 NH winter, however, emphasize the need for model improvements in the dynamical representation of elevated stratopause events in order to allow for a better description of the EPP indirect effect under these particular conditions.