Asymmetry and pathways of inter-hemispheric transport in the upper troposphere and lower stratosphere

. Inter-hemispheric transport may strongly affect the trace gas composition of the atmosphere, especially in relation to anthropogenic emissions which originate mainly in the Northern Hemisphere. This study investigates the transport from the boundary surface layer of the Northern Hemispheric (NH) extratropics (30-90 ◦ N), Southern Hemispheric (SH) extratropics (30-90 ◦ S), and tropics (30 ◦ S-30 ◦ N) into the global upper troposphere and lower stratosphere (UTLS) using simulations with the Chemical Lagrangian Model of the Stratosphere (CLaMS). In particular, we diagnose inter-hemispheric transport in terms 5 of the air mass fractions (AMF), age spectra, and the mean age of air (AoA) calculated for these three source regions. We ﬁnd that the AMFs from the NH extratropics to the UTLS are about ﬁve times larger than the corresponding contributions from the SH extratropics and almost twenty times smaller than those from the tropics. The amplitude of the AMF seasonal variability originating from the NH extratropics is comparable to that from the tropics. The NH and SH extratropics age spectra show much stronger seasonality compared to the seasonality of the tropical age spectra. The transit time of NH extratropical origin 10 air to the SH extratropics is longer than vice versa. The asymmetry of the inter-hemispheric transport is mainly driven by the Asian summer monsoon (ASM). Both ASM and westerly ducts affect the cross hemispheric transport of the NH extratropical air to the SH, and it is an interplay between the ASM and westerly ducts which triggers such cross-equator transport from boreal summer to fall, mainly westerly ducts over the eastern Atlantic. inter-hemispheric inter-hemispheric

In this study, the surface is divided into 3 boxes to investigate the inter-hemispheric transport, which are the NH extratropics (30-90 • N), SH extratropics (30-90 • S), and tropics (30 • S-30 • N), respectively. We calculate age spectra and air mass fraction 90 (AMF) to study transport from the surface of the NH extratropics, SH extratropics, and tropics using the CLaMS model. CLaMS is a Lagrangian chemistry transport model (CTM) with trace gas transport driven by horizontal winds and total diabatic heating rates from reanalysis data (e.g. McKenna et al., 2002;Konopka et al., 2004;Pommrich et al., 2014).
We apply the boundary impulse (time-) evolving response (BIER) approach to calculate the age spectrum G, following Ploeger and Birner (2016). Multiple tracer pulses are released in the boundary source region Ω i , with i labeling the source 95 domain (e.g., NH extratropics, SH extratropics, tropics). The passive tracer with mixing ratio χ at location r and time t related to the mixing ratio χ 0 (t) from the boundary surface of different source regions, which defines the AMF from source regions, can be expressed as (e.g. Waugh and Hall, 2002;Ploeger et al., 2019): The age spectrum is calculated from 120 inert pulse trace gas species from three source regions, with 40 different species 100 pulsed from each region. These pulse tracers approximate a delta distribution lower boundary condition χ j 0 (Ω i , t)=δ(t − t j ) with j=1,...,40, defining tracer pulses at source times t j . The pulse tracer mixing ratios are set to one in the boundary layer of the source region for 30 days, and are set to zero in the boundary layer outside of the initialization region in every time step. The first 24 different species (j=1,...,24) with transit time less than 2 years are pulsed every month. The other 16 different species (j=25,...,40) are pulsed every sixth month (e.g., 25th species in the 30th month, 26th species in the 36th month, etc.). 105 Hence, all species have been pulsed after 10 years of model simulations, and are reset to zero in the whole atmosphere and pulsed again subsequently thereafter. Therefore, the model simulations provide a monthly resolution age spectrum for transit times shorter than two years and a semi-annual resolution age spectrum for longer transit times. The integration of the spectrum over time generally yields a value less than 1 and AoA is young-biased caused by the truncation of the simulations at 10 years. Therefore, we calculate the mean 110 AoA for each source region by normalizing the age spectrum to unit norm: Γ(r, t) = 10 0 τ G(r, t|Ω i , τ )dτ / 10 0 G(r, t|Ω i , τ )dτ The details about the model setup and the calculation of age spectra from multiple pulse tracers and the mean AoA from the age spectrum can be found in Ploeger and Birner (2016) and Ploeger et al. (2019). For this study, we carried out a simulation covering the period from 1989 to 2017 with transport driven by the meteorological data from ERA-Interim (Dee et al., 2011). 115 Due to the 10 year spin-up time for the age spectra, the model data from 1999-2017 is analyzed in the following to address the questions raised in the introduction.

Seasonality of air mass fractions
To evaluate the contributions from the source regions, we calculate the zonally averaged seasonal mean of AMF from the 120 boundary layer of the three source regions. In the following, we use the abbreviations of months (DJF, MAM, JJA, and SON) to represent different seasons. Figure 1 shows the seasonal variations in AMF originating from the NH extratropics (left), SH extratropics (middle), and tropics (right) during 1999-2017. The total sum of the AMF over all 3 source regions is ∼1 related to the limitation of the maximal transit time in our simulations. For direct comparisons, we use the same colorbar for the AMF transported from all the three source regions with different scaling factors. The global results show that the AMFs from the NH 125 extratropics to the UTLS are about five times larger than the corresponding contributions from the SH extratropics and almost twenty times smaller than those from the tropics. Although the contributions from the tropics to the UTLS are much larger than those from the NH extratropics, the annual amplitude of tropical AMFs in the UTLS is comparable to that of NH extratropical AMFs related to the small contributions from the SH extratropics.
Newly pulsed air masses (younger than 3 months) from the NH extratropics start to cross the subtropical tropopause in 130 boreal summer (JJA, Fig. 1g). Three months later, air masses from the NH extratropics are elevated to the lower stratosphere first mainly in the Asian summer monsoon (ASM) region driven by the monsoon circulation, and are then transported isentropically to the tropical lower stratosphere and NH extratropical lower stratosphere covering the latitude range from 30 • S up to the Arctic regions ( Fig. 1j). Later on, the NH extratropical air masses in the upper tropospheric and lower stratospheric tropics driven by the ASM are further transported to the tropical pipe and the whole SH in DJF and MAM ( Fig. 1a and Fig. 1d). Note that young 135 air masses pulsed during boreal winter and spring (DJF and MAM) are not transported to the subtropical stratosphere.
The seasonality in the transport patterns of AMFs originating from the SH extratropics are shifted by 6 months compared to those from the NH extratropics. Although the respective contributions of the SH extratropics (i.e. in DJF and MAM) show some similarities to those from the NH extratropics (i.e. in JJA and SON), there are few significant differences between transport from NH extratropics and SH extratropics. Crossing of the subtropical tropopause for SH extratropical origin air happens first 140 in austral autumn (MAM, Fig. 1e) rather than austral summer (DJF, Fig. 1b), and the overall impact of the SH extratropical boundary surface tracers on both the tropics and the high latitudes is significantly weaker. Most transport of SH extratropical origin tracers is inhibited by the tropopause in the subtropical SH during DJF (Fig. 1b) and MAM (Fig. 1e). Especially, the SH extratropical AMF in the SH lower stratosphere during austral autumn (MAM) is much smaller than the NH extratropical AMF in the NH lower stratosphere during boreal autumn (SON). These differences are most likely attributed to hemispheric 145 differences in the strengths of the monsoons (e.g. Orbe et al., 2016;Chen et al., 2017) and in the strength and downward extent of the polar vortices.    To further explore the seasonal variations of transport from the boundary layer, we remove the annual mean of the contributions from each source region. These seasonal anomalies of climatological zonal mean AMF are shown in Fig. 3 respectively from the NH extratropics (left), SH extratropics (middle), and tropics (right). Clearly, the seasonality of air in the global UTLS originating from the NH extratropics is comparable to those from the tropics, and they are about five times larger than the 165 corresponding anomalies from the SH extratropics being consistent with the results from Fig. 1.  by the monsoon circulation, which again suggests that the tropospheric air in the NH extratropics is mainly transported to the NH and tropical lower stratosphere via the ASM circulation. Another striking feature is the negative anomaly of the NH extratropical air mass in the layer around 320 K in the NH during boreal autumn (Fig. 3j). This signature might be associated with the combination of less convective activity in boreal autumn in the NH extratropics, or with the suppression of horizontal 175 transport from the subtropical troposphere in the layer around 320 K or, finally, with the southward movement of the Hadley cell enhancing isentropic, poleward transport from the tropics and across the still weak summer-autumn jet at levels above (see Fig. 3l).
In contrast to the NH case, the anomaly for SH extratropical air masses shows negative values almost throughout the SH extratropical lower stratosphere during austral autumn (Fig. 3e). This difference to the NH extratropical air (Fig. 3j) is mainly 180 related to the weak convection and the strong inhibition (strong zonal jet) of horizontal transport from the subtropical region in the SH during austral autumn (MAM). In the NH, the tropopause barrier is weak and upward motion over the ASM region is strong, and a substantial amount of NH extratropical origin air can be transported to the lower stratosphere driven by monsoon circulations.
3.2 Seasonality of age spectrum and age of air 185 In Sec. 3.1, we have quantified transport using the AMF, which measures the contribution from different source regions to the air composition at a given destination point. In this section, we provide a complementary view of transport in terms of the age spectrum derived from the same simulations and for the same source regions as in Sec. 3.1. Figures 1 and 3 show that strong isentropic transport across the tropopause occurs in the layer around 360 K. Hence, we consider the age spectrum at 360 K, as a reference location for the UTLS.

190
The age spectra of air from the NH extratropics, SH extratropics, and tropics are illustrated in Fig. 4. The transport seasonality is evident for the NH extratropics and SH extratropics origin air, even stronger than the seasonality for the age spectra with tropical origin, which is consistent with the results based on the AMF (Fig. 1). The first peak of the NH extratropical age spectrum during boreal summer and autumn (Fig. 4g, Fig. 4j and factor 0.1) is a very strong signature compared to the SH extratropical age spectrum during austral summer and autumn (Fig. 4b, Fig. 4e and factor 0.02), which means that much more 195 young air can be expected in the NH compared to the SH. The age spectrum of NH extratropical origin air always shows large PDF values at young transit times during boreal summer, and nearly zero during boreal winter, which suggests that the pollutants from the NH extratropics are being transported to the global UTLS primarily during boreal summer.
Age spectrum and mean AoA from the SH extratropics show a lot of similarities to those from the NH extratropics shifted by 6 months. However, NH extratropical origin young air (< 6 months) shows peak values around the ASM region ( Fig. 4g and  allows faster horizontal transport. Another important asymmetry is that, with exception of MAM, the mean AoA is always older in the SH than in the NH for all other seasons and for all source tracers. This is mainly a consequence of hemispheric differences in the wave-driven eddy mixing, being stronger in the NH throughout the year (Rosenlof, 1995;Konopka et al., 2015). Figure 5 confines the global age spectrum shown in Fig. 4 to partial age spectra at the latitude of 60 • N, which defines the 210 spectra from individual source region without normalization to 1. The age spectrum for the NH extratropical origin air (Fig. 5a) shows multiple peaks caused primarily by the interplay between Hadley and BD circulations. As the Hadley cell upwelling is shifted to the NH subtropics during boreal summer, this is the season favoring upward transport from the NH surface, and peaks in the spectrum are related to air originating at the NH surface in early summer. The youngest peak is in JJA at transit times of around 2 months as a result of an "in phase" interaction between the Hadley and the lower branch of the BD circulation.

215
The respective first peaks in the following seasons are shifted accordingly, which suggests that most of air in the NH high latitude region with origin in the NH extratropical boundary layer is emitted during boreal early summer. Although the tropical upwelling has its maximum in boreal winter and spring, it does not significantly transport the NH extratropical origin air to the high latitude lower stratosphere. This is mainly because the Hadley cell supports such transport pathway rather in summer than in winter and spring (Fig. 2). In addition, transport of the NH extratropical origin air to the high latitudes maximizes in boreal 220 autumn. Note that the second peak in JJA resulting from the NH extropical origin air is higher than the first peak. The mean AoA shows youngest value in boreal autumn (SON) and oldest value in boreal spring (MAM).
Although the structure of age spectrum of the SH extratropical origin air (Fig. 5b) also includes multiple peaks like that from NH extratropics, its total contribution is almost 10 times smaller than the respective contribution from the NH extratropics. The first peak in age spectra in each season from MAM to DJF is delayed by around 3 months accordingly, which again suggests 225 that the main contribution from SH extratropics originates in austral summer. The mean AoA from the SH extratropics is older than that from the NH extratropics during each season except in JJA.
The age spectrum of tropical origin (Fig. 5c) shows by far the highest partial contribution (10 and 100 times larger than that of the NH and SH, respectively). Unlike the age spectrum of the NH extratropics and SH extratropics origin air, the tropical age spectrum in JJA and SON has only one clear peak at transit time around 6 months. During DJF and MAM the tropical age 230 spectrum shows more a multimodal shape with primary peak at transit time around 6 months and a secondary peak delayed by a few months. The first peak might be related to the Hadley and BD circulations combining with the rapid isentropic transport.
The second peak shows similar transit time as the air originating from the SH extratropics, which suggests that the peak might be driven by recirculation within the shallow branch of the BD circulation. The age spectra along 60 • S on the 360 K isentropic surface shows similar patterns with 6 months shift and different amplitudes (not shown).

235
Finally, to get the global view of the mean transit time seasonality from the three source regions, we calculate for each season the mean AoA from the respective age spectrum using Eq. 2. The mean AoA for each source region and season with and without the annual mean removed are compared in Figure 6. In general, we arrive at a similar view on the well-known seasonality and hemispheric asymmetries of the Hadley and BD circulation (e.g. Konopka et al., 2015, and citations therein).
While the seasonality of the tropical upwelling dominates the tropical features, the strength of isentropic poleward transport  13 https://doi.org/10.5194/acp-2020-1153 Preprint. Discussion started: 26 November 2020 c Author(s) 2020. CC BY 4.0 License. and polar vortices explain the patterns at high latitudes. The highest amplitude in the AoA anomalies for NH sources can be diagnosed in the polar SH and vice versa for the SH sources in the polar NH. The amplitude of seasonality in the tropics, especially of the air originating in the tropics, is much smaller compared to that in the NH extratropics and SH extratropics.
The tropospheric air originating from the NH extratropics shows younger mean AoA during boreal summer (JJA) mainly in the NH (negative anomalies in Fig. 6g)  lowermost stratosphere with NH extratropical air (Fig. 1g), this old layer also exists in the distribution of tropical origin air over high latitude regions (Fig. 6i).
Beyond these known features, some interesting asymmetries of the cross-hemispheric transport can be diagnosed. Comparing the left and middle column of Fig. 6, we found that the age of SH extratropical origin air in the NH is younger than the age of NH extratropical origin air in the SH associated with the fast flushing of the NH lower and middle stratosphere with young air In Sect. 3, we discussed the transport from the three source regions based on zonal mean results. Clear hemispheric asymmetry features of transport were noticed in AMF and age spectra. The transit time from the SH extratropical surface to the NH is shorter than that from the NH extratropical surface to the SH. The contributions (AMF) of the NH extratropical air to the global UTLS are around 5 times larger than those from the SH associated with the stronger monsoons and weaker transport barriers 265 in the NH during boreal summer, which allow strong meridional and inter-hemispheric transport. To gain deeper insights into these hemispheric asymmetries in transport, we disentangle the transport pathways in this section using zonally resolved data.
Since most of the anthropogenic pollutants are emitted in the NH and the contributions from the NH extratropics to the SH are much lager than vice versa, the transport pathways from the NH to the SH are of our particular interest.
The monthly evolution of young air (AoA less than 3 months) from the NH extratropics along the latitude of 10 • S longitude-270 pressure cross-section is illustrated in Fig. 7. We choose 10 • S to reduce the influence of reversible transport across the equator.
The AMF from January to May is nearly zero and therefore only May is shown (Fig. 7a). First significant signatures become evident in June−July, maximize in August−September, and vanish in November. There is almost no inter-hemispheric ex-   [%] [%] Figure 8. A snapshot of the horizontal distribution of the NH extratropical origin young (< 3 months) air on the 340 K and 360 K isentropic surface during July, September, and November. Streamlines show horizontal winds.
change in the lower troposphere, probably related to stable easterlies, which are less disturbed by Kelvin waves and which effectively act as a meridional transport barrier (see winds in Fig. 3).

275
Most of the young air masses from the NH extratropics are transported to the SH between 0 and 120 • E (ASM region) from June to October (Fig. 7b-f), with a maximum around θ = 360 K, which suggests a dominant role of ASM in the interhemispheric transport in agreement with the findings from Orbe et al. (2016). The cross hemispheric transport over the Atlantic (around 20 • W) and Pacific (around 80 • W) happens at lower altitudes between 340 and 350 K compared to the crosshemispheric transport between about 0-120 • E over ASM region. To disentangle which flow properties in the NH cause this 280 pattern of inter-hemispheric exchange, a zonally revolved view of both the AMF (less than 3 months) and the zonal wind overplotted with PV contours is shown in Fig. 8 and Fig. 9, respectively. Here, monthly means (July, September, and November) of young AMF are plotted at 340 and 360 K potential temperature levels. mainly over the Pacific and the Atlantic. In July, the peak of the young air (less than 3 months) at 340 K is located over Tibetan plateau and is related to the elevated orography over Tibet which is very close to the 340 K level, so the peak is strongly affected by the released boundary tracers. A second peak is located in the subtropics of Western Pacific and can be attributed to the outflow from monsoon circulations at lower level. The ASM circulation keeps supplying the NH extratropical young air from 290 lower level ( Fig. 8a and Fig. 8c) and isolates most of the young air inside the center of the ASM anticyclone at 360 K ( Fig. 8b and Fig. 8d) during July-September. Part of the NH extratropical origin air which was entrained into the ASM anticyclone moves southward and westward along with the ASM circulation, and is then transported to the SH by eddy shedding detaching ASM air from the anticyclone and subsequently being transported into the SH (Popovic and Plumb, 2001;Orbe et al., 2016) and to the Atlantic by the easterly flow on the southern edge of the ASM anticyclone.

295
The westerly ducts can be clearly seen in the respective climatology of the zonal wind and PV shown in Fig. 9. The tongues of PV and the related anomalies of the westerly wind can be diagnosed in the NH, both over the Pacific and the Atlantic, from July to November at both potential temperature levels 340 and 360 K. Note that the westerlies in the NH become stronger from July to November. The impact of the westerly ducts on the cross-hemispheric transport can be deduced from the time evolution of the AMF at 340 K (Fig. 8 left) with some distinct signatures over the Atlantic (July−November) and slightly 300 weaker signatures over the Pacific (November).
The picture changes at θ = 360 K (Fig. 8 right). While the eddy shedding mechanism plays an important role from July to September, there is only weak transport from the NH to the tropics and to the SH through the westerly ducts during this time.
This implies the important role of the ASM circulation in the asymmetry of inter-hemispheric transport at the 360 K level.
However, starting from September, the westerly ducts start to drive the cross-hemispheric transport ( Fig. 8d and Fig. 8f). On 305 the one hand, the westerly ducts are getting stronger in boreal autumn and winter ( Fig. 9d and Fig. 9f) compared to boreal summer (Fig. 9b). On the other hand, most of the tracers transported across the equator through the westerly ducts originates from ASM regions following the evolution of the ASM anticyclone. This suggests that the westerly ducts alone would not transport a substantial amount of young air masses from the NH extratropics to the SH. It is the interplay between the ASM anticyclone and the westerly ducts which drives the inter-hemispheric transport from boreal summer to fall.

Discussion
The air contributions and age spectrum (or AoA) from different source regions to the destination regions in the atmosphere provide valuable information for understanding the effect of natural and anthropogenic emissions on the atmospheric composition and climate. However, recent studies show substantial transport uncertainties depending on the used methods, models, and meteorological reanalyses (e.g. Krol et al., 2018;Ploeger et al., 2019).