On the impact of future climate change on tropopause folds and tropospheric ozone

Abstract. Using a transient simulation for the period 1960–2100 with the state-of-the-art ECHAM5/MESSy Atmospheric Chemistry
(EMAC) global model and a tropopause fold identification algorithm, we explore the future projected changes in tropopause
folds, stratosphere-to-troposphere transport (STT) of ozone, and tropospheric ozone under the RCP6.0 scenario. Statistically
significant changes in tropopause fold frequencies from 1970–1999 to 2070–2099 are identified in both hemispheres, regionally
exceeding 3 %, and are associated with the projected changes in the position and intensity of the subtropical
jet streams. A strengthening of ozone STT is projected for the future in both hemispheres, with an induced increase
in transported stratospheric ozone tracer throughout the whole troposphere, reaching up to 10 nmol mol−1 in
the upper troposphere, 8 nmol mol−1 in the middle troposphere, and 3 nmol mol−1 near the surface. Notably, the
regions exhibiting the largest changes of ozone STT at 400 hPa coincide with those with the highest fold frequency
changes, highlighting the role of the tropopause folding mechanism in STT processes under a changing climate. For both the
eastern Mediterranean and Middle East (EMME) and Afghanistan (AFG) regions, which are known as hotspots
of fold activity and ozone STT during the summer period, the year-to-year variability of middle-tropospheric
ozone with stratospheric origin is largely explained by the short-term variations in ozone at 150 hPa and
tropopause fold frequency. Finally, ozone in the lower troposphere is projected to decrease under the RCP6.0
scenario during MAM (March, April, and May) and JJA (June, July, and August) in the Northern Hemisphere and during
DJF (December, January, and February) in the Southern Hemisphere, due to the decline of ozone precursor emissions
and the enhanced ozone loss from higher water vapour abundances, while in the rest of the troposphere ozone shows
a remarkable increase owing mainly to the STT strengthening and the stratospheric ozone recovery.


extrapolating their model results from present time to future, concluded that a 30% increase in the ozone flux by 2100 due to BDC strengthening would result in a 3% increase in surface ozone and a 6% increase in mid-tropospheric ozone. However, Morgenstern et al. (2018) using simulations from multiple CCMs showed that the surface ozone response to anthropogenic forcings from well-mixed GHGS and ODS remains uncertain, reflecting uncertainties related to STE. It is therefore crucial to conduct more studies towards this direction, in order to increase confidence in the future projected changes 5 of tropospheric ozone and its associated drivers.
This study aims to assess the impacts of future climate change under the RCP6.0 scenario on tropopause folds and tropospheric ozone, using a free-running hindcast and projection ECHAM5/MESSy (EMAC) simulation for the period 1960-2100.
To this end, a 3-D labeling algorithm is implemented to detect tropopause folds in EMAC simulation. Besides ozone, a tracer for stratospheric ozone is also employed to investigate the projected changes in STE of ozone. Section 2 presents the main 10 characteristics of the EMAC model and describes the 3-D labeling algorithm used to detect the folding events. Section 3 and 4 show the key results of the current study, and finally, Section 5 summarizes the main conclusions.

EMAC model
The ECHAM5/MESSy Atmospheric Chemistry (EMAC) global model is a numerical chemistry and climate simulation system 15 that includes sub-models describing tropospheric and middle atmosphere processes and their interactions with ocean, land and human activities (Jöckel et al., 2010). It uses the second version of the Modular Earth Submodel System (MESSy2) to link multi-institutional computer codes. The core atmospheric model is the 5 th generation circulation model (ECHAM5, Roeckner et al., 2006). The EMAC model has been extensively evaluated for gas tracers (e.g. Pozzer et al., 2007) and for aerosols (e.g. Pringle et al., 2010;Pozzer et al., 2012;Astitha et al., 2012). For the present study we use ECHAM5 version 5.3.02 and 20 MESSy version 2.51. More specifically, data from the simulation RC2-base-04 are used, which is part of the set of simulations performed within the ESCiMo project (Jöckel et al., 2016). The model horizontal resolution is T42L90MA, i.e. with a spherical truncation of T42 (corresponding to a quadratic Gaussian grid of approx. 2.8 by 2.8 degrees in latitude and longitude) with 90 vertical hybrid pressure levels up to 0.01 hPa.
The simulation covers the time frame 1960-2100 (10 years spin-up from 1950 to 1959) driven by prescribed Sea Surface 25 Temperature (SST) and Sea Ice Coverage (SIC) taken from simulations with the global climate model HadGEM2-ES (Collins et al., 2011;Martin et al., 2011) for the Coupled Model Intercomparison Project phase 5 (CMIP5). Anthropogenic emissions are incorporated as prescribed emission fluxes following the CCMI recommendations (Eyring et al., 2013). In more detail, the emissions data set consists of a combination of ACCMIP (Lamarque et al., 2010(Lamarque et al., , for the 1950(Lamarque et al., -2000 and RCP6.0 data (Fujino et al., 2006, for the 2000 and on). A detailed description of the simulation can be found in (Jöckel et al., 2016, and 30 references therein).
Along with ozone chemistry, EMAC also includes a tracer for ozone of stratospheric origin, denoted by O3s, which provides an indicator of the stratospheric contribution to tropospheric ozone. In the stratosphere, O3s is equal to ozone values, while in the troposphere it follows the transport and destruction processes of ozone. When O3s returns to the stratosphere it is reset to stratospheric values; however, since it is initialized above 100 hPa, only a very small fraction is recirculated by multiple crossings of the tropopause (Roelofs and Lelieveld, 1997).

Tropopause fold identification
In this work the algorithm developed by  and improved by Škerlak et al. (2015) has been adopted and 5 applied in order to detect tropopause folds in EMAC simulation (as in Akritidis et al. (2016)), using the 3-D fields of potential vorticity, potential temperature and specific humidity. As in several previous studies (Hoskins et al., 1985;Holton et al., 1995;Stohl et al., 2003;, the tropopause is defined as the combination of the isosurfaces of potential vorticity at ±2 PVU and potential temperature at 380 K, whichever is lower (refered as dynamical tropopause). For each grid point a tropopause fold is designated where multiple crossings of the dynamical tropopause are detected in instantaneous vertical Before the results from simulation RC2-base-04 can be used to estimate the future projected changes of fold frequencies, the capability to reproduce present-time folding frequencies must be first ckecked. Therefore the model results have been compared with the monthly fold frequencies climatology compiled by Škerlak et al. (2015). The climatology has been calculated using 20 the same identification algorithm used in this work from the ERA-interim dataset (Dee et al., 2011). Figure  data from simulation RC2-base-04 and data from ERA-Interim, but also the geographical distribution presents the same patterns (not shown). We can therefore consider that the data used in this work are comparable for present-time with state-of-the-art calculations based on the ERA-Interim dataset.

Future projected changes
To explore the future projected changes in EMAC meteorological and chemical parameters under the RCP6.0 emissions sce-

Jet streams and tropopause folds
At first, the impact on atmospheric circulation under the RCP6.0 scenario is explored. As it is depicted from Figure 2 there is a

Tropospheric ozone
Here we explore the future changes in tropospheric ozone under the RCP6.0 GHGs scenario. Figure 5  water vapour abundances and temperatures in a warmer climate enhance ozone destruction, leading to lower baseline ozone 10 levels, while there is medium confidence that in polluted regions it is expected to increase surface ozone. Clearly, temperature and humidity under a warmer climate play an important role for lower ozone in the tropical Pacific, due to the increased rate of the ozone destruction reactions (Revell et al., 2015). Contrary, in the extratropical lower stratosphere and the upper and middle troposphere ozone is projected to increase during all seasons. The largest increases in the upper and middle troposphere, of up to 12 nmol/mol, are seen in the subtropics and in the vicinity of the jet streams where tropopause folds formation and the 15 induced STT are favoured. The more pronounced increases of ozone are found in the NH/SH during MAM/SON throughout the entire free troposphere. As regards the lower stratosphere, an increase of ozone is projected outside the tropics reflecting the recovery of stratospheric ozone. In the tropical lower stratosphere, the projected decrease of ozone is presumably related to the BDC strengthening and the induced increased upwelling of tropospheric ozone-poor air into the lower stratosphere.
These patterns of tropospheric ozone increase are probably resulting from a global STE increase, linked to stratospheric ozone 20 recovery and a strengthening of BDC, as suggested by previous studies based on simulations with CCMs (Banerjee et al., 2016;Morgenstern et al., 2018). In the free troposphere, it seems that the beneficial reduction of ozone precursor emissions is canceled out by the projected increase of stratospheric ozone influx.

Stratospheric ozone tracer (O3s)
To estimate the impact of STE on tropospheric ozone, the projected changes of O3s are examined here. Same as in Figure 5, 25 Figure 6 depicts the differences of zonal-mean O3s concentrations between the FUT and REF periods. An increase of O3s occurs almost throughout the troposphere during all seasons. In the NH, the peak of O3s enhancement is found in the subtropics and in the vicinity of the NH jet stream during DJF and MAM (Fig. 6a and c), while in the SH the respective positive maxima are seen during JJA and SON ( Fig. 6b and d), similarily near the position of the SH jet stream. These increases of O3s in the NH/SH subtropical upper and middle troposphere, reveal an increase of isentropic cross-tropopause ozone transport, through 30 tropopause folds that in principal occur near the NH/SH subtropical jet streams. In general, the positive O3s patterns resemble that of tropospheric ozone (Fig. 5), indicating that the projected increase of tropospheric ozone is mainly driven by the increase in STT and the induced vertical transport of stratospheric ozone in the underlying troposphere, as it is was also reported from O3s loss due to a slight increase in OH and HO 2 and their reaction rate with ozone (due to increased temperature).

5
The spatial distribution of O3s projected changes at 400 hPa is presented in Figure 7, to identify the global hot spots of climate change impact on ozone STT. Overall, an increase of ozone with stratospheric origin is projected in the middle troposphere (400 hPa) during all seasons, reflecting the recovery of stratospheric ozone and the associated increase of ozone STE. Notably, the maxima of O3s increase coincides mainly with the respective maxima of tropopause folds frequency increase (see Fig. 4). In more detail, during DJF the peaks of future O3s increases (up to 12 nmol/mol) are found over the NH Pacific 10 Ocean (Fig. 7a), while during JJA the respective peaks (exceeding 12 nmol/mol) are mainly occured over the central Asia and the Indian Ocean (Fig. 7c). All in all, the emerging increase in ozone STE under the RCP6.0 GHGs scenario is mainly driven by the recovery of stratospheric ozone, still for regions where tropopause folds are projected to occur more often, the downward transport of ozone from the stratosphere seems to be more pronounced.

15
STT is of great importance for ozone levels and variability in the upper/middle troposphere over regions where the meteorological conditions favor the formation of tropopause folds and the downward transport (Roelofs and Lelieveld, 1997;, such as the eastern Mediterranean and the Middle East (EMME) (Li et al., 2001;Zanis et al., 2014;Akritidis et al., 2016), and the broader Afghanistan area (AFG) (Tyrlis et al., 2014;Ojha et al., 2017) during summer, and especially the July-August period. To explore the links of the tropopause folds frequency and stratospheric ozone, with the interannual vari-20 ability of middle tropospheric ozone with stratospheric origin over the EM (20-45 • E,30-40 • N) and AFG (60-80 • E, 30-40 • N) regions, the mean July-August timeseries of tropopause folds frequency, ozone at 150 hPa, and O3s at 400 and 500 hPa for the period 1960-2099 were constructed. Figure 8a presents the mean July-August fields of tropopause folds frequency during the REF period revealing a pronounced fold activity over the depicted EM and AFG regions. For the EM region (Fig. 8b), the interannual variability of mean July-August O3s at 400 hPa (500 hPa) is found to be positevely correlated at the 99% 25 significance level with the mean July-August tropopause folds frequency and ozone at 150 hPa, with values of r=0.53 (r=0.43) and r=0.56 (r=0.49), respectively. Employing a multiple linear regression analysis, folds frequency and ozone at 150 hPa are found to explain the 58% (42%) of the variance of O3s at 400 hPa (500 hPa). As regards the AFG region, the variance of the projected mean July-August O3s concentrations at 400 hPa (500 hPa) explained by folds frequency and ozone at 150 hPa is 73% (68%). The year-to-year variability of July-August O3s at 400 hPa (500 hPa) is found positevely correlated at the 99% 30 significance level with both fold frequency r=0.64 (r=0.58) and ozone at 150 hPa r=0.64 (r=0.64).
This study investigates the future projected changes in tropopause folds, ozone STT and tropospheric ozone under the RCP6.0 emissions scenario, using a transient simulation with EMAC CCM from 1960 to 2100 and a tropopause fold identification algorithm. In particular, we examined the long-term change in tropopause folds frequency and the potential links with atmospheric circulation changes. Moreover, the long-term changes in tropospheric ozone and ozone STT were also explored and • The spatial patterns of the projected changes in NH and SH subtropical jets seem to drive the respective patterns of tropopause folds frequency future changes, with a negative/positive dipole structure found over south Asia/NH Pacific Ocean during DJF and MAM. The most prominent features during JJA are a distinct increase of fold activity over the Indian Ocean exceeding 3%, and a negative/positive dipole structure centered over the greater Afghanistan region. 15 • The regions exhibiting the highest increases in tropopause folds occurrence in future are those with the more pronounced projected increases of O3s in the middle troposphere (400 hPa). The projected changes of zonal-mean O3s concentrations reveal a strengthening of ozone STT at the middle latitudes of both hemispheres during all seasons, which is more distinct at the NH during DJF and MAM (up to 6 nmol/mol down to 500 hPa), and at the SH during JJA and SON (up to 8 nmol/mol down to 500 hPa). Although the future increase in ozone STT on a global scale seems to be forced from 20 stratospheric ozone recovery and strengthening of BDC (Banerjee et al., 2016;Meul et al., 2018), regionally, the degree of increase in the downward transport of stratospheric ozone is partially driven from the long-term changes of fold activity.
• For specific regions considered as global STT hotspots, namely the summertime EMME and AFG, the projected yearto-year variability of middle tropospheric ozone with stratospheric origin seems to be largely governed from both the 25 variabilities of ozone at 150 hPa and folds frequency, as they explain 60% and 68% of the variance of mean July-August O3s concentrations at 400 hPa for EMME and AFG respectively, over the period 1960-2100.
• Ozone in the lower troposphere and near the surface decreases under the projected decline in ozone precursors emissions during MAM and JJA at the NH, and during DJF at the SH, as photochemical ozone production is more dominant during these seasons. In the middle and upper troposphere the projected strengthening of ozone STT results in a distinct increase 30 of ozone globally, that seems to cancel out the aforementioned ozone decrease due to emissions reduction.