Observations of ozone-poor air in the tropical tropopause layer

Ozonesondes reaching the tropical tropopause layer (TTL) over the west Pacific have occasionally measured layers of very low ozone concentrations – less than 15 ppbv – raising the question of how prevalent such layers are and how they are formed. In this paper, we examine aircraft measurements from the Airborne Tropical Tropopause Experiment (ATTREX), the Coordinated Airborne Studies in the Tropics (CAST) and the Convective Transport of Active Species in the Tropics (CONTRAST) experiment campaigns based in Guam in January–March 2014 for evidence of very low ozone concentrations and their relation to deep convection. The study builds on results from the ozonesonde campaign conducted from Manus Island, Papua New Guinea, as part of CAST, where ozone concentrations as low as 12 ppbv were observed between 100 and 150 hPa downwind of a deep convective complex. TTL measurements from the Global Hawk unmanned aircraft show a marked contrast between the hemispheres, with mean ozone concentrations in profiles in the Southern Hemisphere between 100 and 150 hPa of between 10.7 and 15.2 ppbv. By contrast, the mean ozone concentrations in profiles in the Northern Hemisphere were always above 15.4 ppbv and normally above 20 ppbv at these altitudes. The CAST and CONTRAST aircraft sampled the atmosphere between the surface and 120 hPa, finding very low ozone concentrations only between the surface and 700 hPa; mixing ratios as low as 7 ppbv were regularly measured in the boundary layer, whereas in the free troposphere above 200 hPa concentrations were generally well in excess of 15 ppbv. These results are consistent with uplift of almost-unmixed boundary-layer air to the TTL in deep convection. An interhemispheric difference was found in the TTL ozone concentrations, with values < 15 ppbv measured extensively in the Southern Hemisphere but seldom in the Northern Hemisphere. This is consistent with a similar contrast in the lowlevel ozone between the two hemispheres found by previous measurement campaigns. Further evidence of a boundarylayer origin for the uplifted air is provided by the anticorrelation between ozone and halogenated hydrocarbons of marine origin observed by the three aircraft.


Principal WAS chemicals split by aircraft
The following three plots show panels of the six chemicals shown in figure 21 of the accompanying article, split into the three individual aircraft. Figure S1 shows the same panel as figure 21 for just the ATTREX data only, figure S2 shows the panel for just the CONTRAST data only, and figure S3 shows the panel for the CAST aircraft data only. In each figure, the average profile for the high-ozone case (>20 ppbv) is shown in red and for the low-ozone case (<20 ppbv) is shown in blue; the solid lines show the averages for just that aircraft, while the dashed lines show the averages for all the aircraft combined. The amount of CAST aircraft data is small in comparison to CONTRAST and ATTREX and so the effect on the overall averages in figure 21 is negligible. There is some overlap between ATTREX and CONTRAST: the highest altitude that the CONTRAST WAS samples were taken at was ∼150 hPa, and the lowest altitude that the ATTREX WAS samples were taken at was ∼180 hPa.
The ozone measurements taken on board the Gulfstream V aircraft in the CONTRAST campaign were of higher confidence than those taken on board the Global Hawk aircraft in the ATTREX campaign (see section 4.2 of the accompanying article for details on the uncertainties associated with the UCATS ozone measurements from the Global Hawk). However, the differences between the low-ozone cases and the high-ozone cases exist in both the CONTRAST and ATTREX data.

More WAS sample chemicals
The following plots are of chemical species measured by the whole air samplers (WAS) that were not plotted in the accompanying article. Firstly, dichloromethane (CH 2 Cl 2 ) and trichloromethane (CHCl 3 ) were measured by all three aircraft, but unlike the other six chemical species measured by all three aircraft, they both have a strong anthropogenic industrial source with relatively long lifetimes of around five months and six months respectively [Montzka et al., 2010;Carpenter et al., 2014;Khalil and Rasmussen, 1999]. Figure S4 shows the vertical profile of dichloromethane coloured by ozone concentration, with average profiles for WAS samples with ozone concentrations greater than 20 ppbv as a red line, and for WAS samples with ozone concentrations less than 20 ppbv as a blue line, in the same way as the panel plot in figure 21 of the accompanying article. Likewise the profile for trichloromethane is found in figure S5.
Isoprene, however is a naturally occurring chemical emitted in large quantities by vegetation rather than as a result of the petroleum industry, which accounts for the difference between the other hydrocarbons and isoprene.

Aromatic compounds
• benzene (C 6 H 6 ) lifetime = ∼months (figure S34) • chlorobenzene (C 6 H 5 Cl) lifetime = ∼2 weeks (figure S35) Benzene and chlorobenzene are industrial solvents, and both show enhancements in the high ozone régime compared to the low ozone régime, which is what is expected. However, in the mid-troposphere, chlorobenzene shows the opposite.

Sulfides
• carbonyl sulfide (OCS) lifetime = ∼35 years (figure S36) Like dimethyl sulfide, shown in figure 15 of the accompanying article, carbonyl sulfide, shows a slight enhancement in the low-ozone, clean régime. Figure S1: Panel of the six principal WAS chemicals using the ATTREX WAS sample data only.