It is known that the Asian monsoon anticyclone has a strong horizontal transport barrier at about 380 K e.g.; however, this transport barrier is not well defined at higher levels of potential temperature. The weaker transport barrier at higher levels has consequences for the vertical transport at the top of the anticyclone. Before the transport at the top is discussed we show the horizontal distribution of different emission tracers at 360 K and then their subsequent transport to the top of the anticyclone up to 460 K. showed that the emission tracer for India/China is a good proxy for the location and shape of the Asian monsoon anticyclone using pattern correlations with PV, and MLS O3 and CO satellite measurements between 360 and 400 K. Therefore here we use the India/China tracer as a proxy for the location of the anticyclone.

To analyse the transport pathways at the top of the Asian monsoon anticyclone during the monsoon season 2008, we use only the tracers of air mass origin for the time pulse for summer 2008 (started on 1 May 2008, running through the end of October 2008). The geographic position and shape of the Asian monsoon anticyclone show strong day-to-day variability e.g.. In this paper, we focus on 18 August 2008 during the monsoon season 2008 as a case study. On that day, the anticyclone has two modes: the western mode located over the Middle East and the eastern Mediterranean Basin and the eastern mode over India and western China as shown in Fig. a. Further, a smaller anticyclone (eddy shedding event) is found over the northwestern Pacific. We selected 18 August 2008 for this study because, first, this day is dynamically very interesting and, second, on this day there is very good data coverage of the MIPAS HCFC-22 measurements.

To analyse the transport of young air masses to the top of the Asian monsoon anticyclone, we distinguish between air masses that experienced strong upward transport mainly inside (India/China) and mainly outside (tropical adjacent regions) of the Asian monsoon anticyclone. Figure a, b show the horizontal distribution of the fraction of the emission tracer for India/China (Fig. a) and for tropical adjacent regions (Fig. b) at 360 K potential temperature. It is evident in Fig. a and b that at 360 K the CLaMS model simulates very strong horizontal tracer gradients between the tracer for India/China and that for the tropical adjacent regions at the edge of the anticyclone. High fractions of air from India/China up to 90 % and low fractions below 10 % from the tropical adjacent regions are found in the core of the Asian monsoon anticyclone at 360 K potential temperature. Highest fractions from the tropical adjacent regions of about 40 % are found in a belt around the edge of the anticyclone. Towards the north this belt is separated by the subtropical jet from the northern lower stratosphere visible as a very sharp gradient. To the south, air masses from the tropical adjacent regions do not show a strong gradient with air masses within the tropics and therefore are not separated by a strong transport barrier from the TTL at a level of potential temperature of 360 K e.g..

Figure c, d shows the longitude–theta cross sections at 25∘ N of the fraction from India/China (Fig. c) and from the tropical adjacent regions (Fig. d) on 18 August 2008. We would like to emphasise the horizontal transport of air masses with high contributions from India/China (40 %–90 %) from the eastern part of the anticyclone to both the western part and into the eddy over the western Pacific between ∼340 and ∼380 K. Correspondingly, low fractions from the tropical adjacent regions (0 %–30 %) are simulated in these regions. The horizontal transport of air masses from the eastern to the western mode of the anticyclone indicated by the India/China tracer is consistent with simulations of CO using the Whole-Atmosphere Community Climate Model (WACCM4-SD) .

Figure  shows the latitude–theta cross sections in the eastern mode of the anticyclone at 90∘ E (Fig. e, f) and in the western mode of the anticyclone at 30∘ E (Fig. g, h) on 18 August 2008. Below 360 K, high fractions of air from India/China up to 90 % and low fractions from the tropical adjacent regions lower than 5 % are found in the eastern mode of the anticyclone (Fig. e, f). In the western mode there is still a large contribution from the India/China tracer between 20 % and 60 % and lower fractions of about 10 %–40 % from the tropical adjacent regions (Fig. g, h inside the thick white line). Below the western mode, in the tropics below ∼330 K at around 10∘ N fractions from the tropical adjacent regions (in that case from northern Africa) are up to 90 % caused by local upward transport (Fig. h).

Further, a strong vertical gradient of the India/China tracer is found at about 360 K. This level is below the thermal tropopause, which is located at around 380 K (∼100 hPa) over the eastern mode of the Asian monsoon anticyclone at 90∘ E (Fig. e). At 90∘ E, there is a layer of young air masses with enhanced fractions from India/China (up to ∼20 %) above the thermal tropopause up to about 420 K potential temperature. An obvious explanation for this model result would be that the tropopause over the Asian monsoon is not a strict vertical transport barrier and weak vertical cross-tropopause transport occurred, in contrast to the extratropics (air masses at the polar side of the subtropical jet) where the tropopause acts as a chemical transport boundary.

Below 360 K in the region with high values of the India/China tracer, the fractions from the tropical adjacent regions are below 10 %; however, above 360 K around the tropopause the fractions are much higher, up to about 30 %, and up to about 15 % around 420 K (see Fig. d, f). Thus, at 420 K the contributions of young air masses are about 20 % from India/China and 15 % from the tropical adjacent regions. From this result the question arises, “how can air masses from the outer edge of the anticyclone be transported from 360 K into the lower stratosphere above?” A straight vertical cross-tropopause transport cannot be the explanation because inside the anticyclone the fraction from the tropical adjacent regions is much lower.

Note that in Fig.  the same data range is used for all colour bars for a better comparability between the horizontal and different vertical cross sections. Therefore some features in the horizontal cross section at 360 K are not too prominent, for example the thin filament at around 50∘ E between 40 and 60∘ N in Fig. a (see Fig.  in the next section).

To demonstrate that the spatial distribution of tracers of air mass origin found in the CLaMS model in the region of the Asian monsoon anticyclone is consistent with observations of chemical tracers, we analyse HCFC-22 measurements of the MIPAS instrument . As described in Sect. , HCFC-22 is emitted in locally restricted regions in continental eastern Asia, in particular in China, and in the Middle East.

Figures  and a show the horizontal distribution for the India/China tracer and of HCFC-22 measurements synoptically interpolated to 18 August 2008 12:00 UTC at 380 K potential temperature (for details see Sect. ); also shown are the longitude–theta cross section at 25∘ N (Fig. b) and the latitude–theta cross sections at 90∘ E ±10∘ (Fig. c) and at 30∘ E ±10∘ (Fig. d). The contour line of 20 % of the India/China tracer is marked on the cross sections for better comparison with the CLaMS results shown in Fig. .

The horizontal spatial distributions of the India/China tracer (Fig. ) and HCFC-22 at 380 K (Fig. a) show good overall agreement. In particular the strong gradient at the northern flank of the Asian monsoon anticyclone is evident both in the model and in HCFC-22 observations. Enhanced HCFC-22 values up to 240 pptv compared to the stratospheric background (of around 180  to 200 pptv at 16 km altitude in JAS derived from MIPAS measurements ) are found in both the eastern and western mode of the anticyclone, in the smaller eddy at the northeastern flank of the anticyclone, and in the thin filament at around 50∘ E between 40 and 60∘ N. Further, as in the CLaMS model in the observations there is no sharp gradient at the southern flank of the anticyclone, which separates anticyclonic air from the surrounding tropics.

The vertical HCFC-22 distributions (Fig. c, d) within the eastern and western modes of the anticyclone are broadly consistent with the vertical distribution of the India/China tracer (see Fig. e, g). The highest mixing ratios of HCFC-22 are found within the anticyclone below the thermal tropopause. However, also at the top of the anticyclone above the tropopause enhanced HCFC-22 values are measured in agreement with the CLaMS tracer for India/China. Thus, measurements of HCFC-22 are consistent with our model result that young air masses from the region of the Asian monsoon are transported to the top of the anticyclone above the tropopause.

In addition, the vertical HCFC-22 distribution (Fig. d) for the western mode shows a very steep gradient in the upper troposphere between 350 and 360 K in agreement with the spatial distribution of the India/China tracer (see Fig. g). Thus below 350–360 K smaller mixing ratios of HCFC-22 are measured than above, indicating that below the western mode of the anticyclone there exists no upward transport from boundary sources for HCFC-22. The enhanced values of HCFC-22 within the western mode and below the thermal tropopause along the longitude–theta cross sections at 25∘ N (Fig. b) confirm the horizontal westward transport within the Asian monsoon anticyclone as found for the India/China tracer in the CLaMS simulations (see Sect.  and Fig. c), which is consistent with CO simulations by .

In Sect. , it is shown that enhanced fractions of both tracers for India/China and tropical adjacent regions are found above the thermal tropopause at the top of the Asian monsoon anticyclone. Figure  shows the horizontal distribution of the fraction of air originating in India/China (left) and in tropical adjacent regions (right) at different levels of potential temperature between 380 and 460 K.

Young air masses (age <6 months) from both India/China and tropical adjacent regions are found up to ∼460 K. It was shown earlier that the horizontal distribution of the India/China tracer is a good proxy for the location of the anticyclone . The comparison between the horizontal distribution of the tropical adjacent regions and the India/China tracer strongly differs depending on the level of potential temperature from a nearly disjoint distribution at 360 K (see Fig. ) to a more coincident distribution from 400 to 460 K (see Fig. ).

At 380 K, the highest fractions from tropical adjacent regions are found at the edge of the anticyclone, while at 400 K they are found within the anticyclone. Above 400 K both the India/China and the tropical adjacent region tracers show a similar horizontal distribution. We emphasise that at these levels of potential temperature the tracer distributions have the shape of rotating filaments in contrast to the more compact distribution at lower levels. The variation in the distribution of the tracer for the tropical adjacent regions with altitude is an indication that the upward transport of young air masses at the top of the anticyclone occurred more towards the edge and less inside the anticyclone itself.

To analyse the transport pathways to the top of the anticyclone in more detail, 40-day backward trajectories are calculated starting in the western (20–50∘ N, 0–70∘ E) and eastern (20–50∘ N, 70–140∘ E) modes of the anticyclone. The trajectories are started at the position of the air parcels from the three-dimensional CLaMS simulation at different levels of potential temperature (Θ=380, 400, 420, 440 K ±0.25 K) on 18 August 2018. Note that the air parcels in the three-dimensional CLaMS simulation are distributed on an irregular grid. To take into account the distribution of the boundary emission tracer at the top of the Asian monsoon anticyclone, only air parcels are selected with contributions of young air masses (age <6 months, summer 2008) larger than 70 % (380 K), 50 % (400 K), 20 % (420 K), and 5 % (440 K) (not all levels of potential temperature are presented here). The percentages are chosen in a way to obtain a number of trajectories (less than 30) that can be reasonably visualised. The results of the 40-day backward trajectories are similar at different levels of potential temperature; therefore we show a selection of trajectories to demonstrate the main transport pathway to the top of the Asian monsoon. A larger set of 20-day backward trajectories analysed statistically will be discussed below in Sect. .

Figure  shows trajectories that started in the eastern and western part of the Asian monsoon anticyclone around the thermal tropopause at 380 K on 18 August 2018. Air masses are uplifted to approximately 360 K very rapidly by various convective events occurring at different times and locations. Our 40-day backward trajectories show that preferred regions for fast uplift are continental Asia (mainly the region of the south slope of Himalayas and the Tibetan Plateau) and the western Pacific (not shown here). A lower fraction of trajectories originates in the free troposphere. The trajectories in Fig.  demonstrating convection below 380 K are only a snapshot for 18 August 2018. There are several previous studies e.g. quantifying the contribution of different source regions to the composition of the Asian monsoon anticyclone during the course of the monsoon season (see discussion in Sect. ). The backward trajectories demonstrate that above 360 K potential temperature the air masses circulate around the anticyclone in a large-scale upward spiral extending from northern Africa to the western Pacific. Here in the upward spiralling range, the vertical transport is much slower than in the convective range.

Figure  shows trajectories started in the western and eastern parts of the Asian monsoon anticyclone above the thermal tropopause at 400 K and at 440 K. The slow upward transport up to ∼1 to 1.5 K per day in a large-scale upward spiral is evident in both the western and eastern parts of the anticyclone. The 40-day backward trajectories demonstrate that at the top of the anticyclone the upward transport of air masses occurs along a large-scale upward spiral and therefore no straight vertical transport from the upper tropopause into the lower stratosphere takes place. The higher the air masses are above the thermal tropopause, the larger the contribution of trajectories is from outside the Asian monsoon anticyclone coming into the upward spiralling flow above 360 K. Trajectories at other levels of potential temperature (not shown) both in the western and in eastern parts of the anticyclone have a similar behaviour as shown in Figs.  and and therefore confirm the presented results.

In general, trajectory calculations have limitations due to trajectory dispersion by errors through interpolation of the wind data to the position of the air parcel at a specific time. Over the timescales in question, mixing can also be relevant e.g.. These errors can accumulate depending on the trajectory length over the course of the simulation. However, the frequently employed trajectory length to study transport processes in the Asian monsoon region ranges from a couple of weeks to a few months e.g.. In our trajectory analysis, the focus is to demonstrate the large-scale transport pathways of the air parcels at the top of the anticyclone; therefore small changes of the trajectory position will not affect our findings.

In the previous section, the transport pathways for a restricted number of trajectories in the region of the Asian monsoon anticyclone were discussed. Here, for a broader view 20-day backward trajectories between 0–160∘ E and 10–60∘ N are presented including the entire region of the Asian monsoon anticyclone. In this longitude–latitude region (0–160∘ E and 10–60∘ N), backward trajectories are calculated on a 1.0∘×0.5∘ longitude–latitude grid at 360, 380, 400, 420, and 440 K starting on 18 August 2008. Each point in Fig.  indicates the location of the start position of a 20-day backward trajectory colour-coded by the change in potential temperature (ΔΘ) during the last 20 days.

At 360 K, air parcels that experienced very strong upward transport by up to ∼60 K within the last 20 days were found inside the western and eastern modes of the anticyclone, within the eddy over the Pacific, and within the tropics south of the anticyclone. The patterns of ΔΘ at 360 K within the anticyclone and in the tropics are very patchy, reflecting that the strong upward transport in this region is caused by single convective events.

Above 360 K, air parcels that experienced strong upward transport larger than 20–30 K within 20 days (corresponding to a mean value of 1–1.5 K per day) are largely found in the region of the anticyclone. This rate of upwelling is much slower compared to convective upwelling shown at 360 K. Air parcels that experienced strong upward transport are mainly grouped in curved elongated filaments, reflecting a rotating movement of the air parcels at the top of the anticyclone. Often air parcels with strong ΔΘ above 360 K are located more at the edge of the eastern and western modes of the anticyclone and at the edge of the eastward-migrating eddy at the eastern flank of the anticyclone. Thus the upward transport in the region of the anticyclone is not homogeneously distributed over the entire anticyclone on a certain day. In previous studies e.g. climatological mean values (over the monsoon season of several years) are presented which can not be used to analyse the inhomogeneity of the upward transport in the region of the anticyclone at a certain point in time. This inhomogeneity is consistent with results presented above in Sect.  demonstrating that for single selected trajectories the transport at the top of the Asian monsoon anticyclone is a slow upward transport of about 1–1.5 K per day in a large-scale spiral above the anticyclone caused by diabatic heating. In the backward trajectory calculations mixing processes are not included; however, the results of the trajectory calculations are consistent with patterns found in the three-dimensional CLaMS simulation including mixing as discussed in Sect. , demonstrating that young air masses above 400 K are found at the edge of the anticyclone. Above 400 K, air masses in the tropics also experienced upward transport, but the vertical uplift is in general lower than 20 K within 20 days (i.e. lower than 1 K per day).

In Appendix , results of global 20-day backward trajectories demonstrate that during the monsoon season above 360 K an uplift of air parcels of about 1–1.5 K per day occurred only in the region of the Asian monsoon anticyclone compared to the rest of the tropics. Even in the tropics the uplift is in general slower at these levels of potential temperature. Further, the seasonal variability of this upwelling above the Asian monsoon anticyclone from monsoon onset until post-monsoon 2008 is discussed in Appendix .

Figure  shows latitude–theta cross sections at 90∘ E for the fraction from India/China from the start of the simulation on 1 May 2007 until 18 August 2008, which is a sum of the contributions of each of the three time pulses summer 2007, winter 2007–2008, and summer 2008, each set for a time period of 6 months. A signal with enhanced fractions from India/China (up to 10 %) is found at around 550 K within the tropics which is from the summer 2007 pulse. This shows that air masses from boundary layer regions in India/China are transported into the middle stratosphere within the tropical pipe within a time period of 1 year. Fractions of air from the India/China tracer for the time pulse for winter 2007–2008 are below 2.4 %, indicating that during winter in the absence of the Asian monsoon anticyclone the transport of boundary layer emissions from India/China into the stratosphere is insignificantly weak.

Figure  shows the same images as Fig. , but for the fractions from tropical adjacent regions. Because of the summer 2008 pulse, high fractions from TAR are found at the edge of or outside the Asian monsoon anticyclone below 360 K. At altitudes up to ∼460 K, enhanced fractions are found over the region of the Asian monsoon, but also above the entire tropics. For the summer 2007 pulse, an enhanced signal from TAR (up to 25 %) is found at around 550 K within the tropics similar to that for the India/China tracer. This model result shows that air masses from outside the Asian monsoon anticyclone (summer 2007 pulse) are also transported into the middle stratosphere within the tropical pipe within 1 year. The fraction of air from TAR (up to 25 %) is even larger than that from India/China (up to 10 %). In contrast to the India/China tracer, the TAR tracer shows that also emissions mainly from Southeast Asia and the tropical Pacific (for details see Sect. ) released during winter (winter 2007–2008 pulse) are transported into the lower stratosphere via the tropical pipe.

Figure  shows that the upward spiralling transport above 360 K occurs in a field of radiative heating above the Asian monsoon, in particular above the western mode of the anticyclone. Thus, the combination of the anticyclonic flow and the uplift by radiative heating results in an upward spiralling transport like a “spiral staircase”. Positive radiative heating rates are found in Era-Interim reanalyses on top of the Asian monsoon anticyclone as well as over the subtropics in the Southern Hemisphere (see Fig. ). In these regions, positive radiative heating rates were also reported in a study by , using a free-running climate model. They argue that the positive radiative heating rates are the response to very low temperatures around the tropopause, which in turn are the response to convective heating near the Equator e.g.. It is known that the radiative heating rates in the tropical UTLS are different in current reanalysis models e.g. and are most likely overestimated in ERA-Interim e.g.. Therefore, the rates of diabatic heating in the upward spiralling range found in our study are most likely somewhat too high; however, slow upward transport in the UTLS in the region of the Asian monsoon anticyclone associated with positive heating has been addressed previously e.g.. The focus of our study is to demonstrate that in the upward spiralling range (above 360 K) a slow upward transport is found over the region of entire anticyclone (west and east mode) with diabatic heating rates of up to 1–1.5 K inferred from ERA-Interim. Our 40-day backward trajectory calculations (see Figs.  and ) demonstrate that a diabatic heating above 360 K is found at both the western and eastern side of the anticyclone during August 2008.

Figure  shows MIPAS HCFC-22 measurements at the same latitude–theta cross sections as Figs.  and . The transport pathway of HCFC-22 within the tropical pipe is evident. The transport of HCFC-22 is similar to that of the emission tracer for tropical adjacent regions, in contrast to the signal for the India/China tracer, which results from a combination of just two signals, one from the summer 2007 pulse and another from the summer 2008 pulse. It was shown in previous studies that HCFC-22 is enhanced in the region of the Asian monsoon anticyclone. There are also HCFC-22 source regions outside the Asian monsoon region in continental eastern Asia , and therefore the upward transport of HCFC-22 in the troposphere can also occur during winter. In boreal winter, efficient transport into the stratosphere is found over the west Pacific and Maritime Continent, caused by strong convection and in addition by the ascending branch of the Walker circulation located over the Maritime Continent e.g.. In addition, stronger heating rates (vertical velocities) are found in the TTL during winter compared to summer e.g..

Figures , , and further show that in addition to the upward transport into the tropical pipe, transport of air masses out of the region of the Asian monsoon occurs into the northern lower stratosphere and to a much lower extent into the Southern Hemisphere. Details of the transport into the northern lower stratosphere are further discussed in Sect. .

On 18 August 2008, enhanced signals for the tracers from India/China and tropical adjacent regions from the summer 2007 pulse are found at around 550 K in the tropics between 30∘ S and 30∘ N (see Figs.  and ). To analyse in more detail the transport times from the model boundary layer into the tropical pipe, the contributions of the three different time pulses from different emission tracers are calculated within the tropical pipe at 550 K potential temperature. We use the following approach: for each day between 1 May 2007 and 31 October 2008, a mean value for each emission tracer of all CLaMS air parcels is calculated between 30∘ S and 30∘ N at 550 K (±0.5 K). This is done for the sum of all three time pulses summer 2007, winter 2007–2008, and summer 2008 and for each pulse individually.

Figure a shows the contribution of the three different time pulses (ΩS07, ΩW07, ΩS08) for the entire Earth's surface between 1 October 2007 until the end of the simulation period (31 October 2008). The time period between 1 May and 1 October 2007 is not shown because here the contribution of all emission tracers is zero, caused by the fact that the air masses need a certain transport time to reach the level of 550 K potential temperature within the tropical pipe. Air masses from the summer 2007 pulse reach this level first in November–December 2007; between January and May there is a steep increase in air masses from the summer 2007 pulse with a maximum in June 2008. Air masses from the winter 2007 pulse reach the 550 K level of potential temperature during May 2008 and increase until the end of the simulation period. No contributions from the summer 2008 pulse are found at 550 K, since the transport times are too short to reach the tropical pipe within the simulation period.

The largest contributions to the summer 2007 pulse are from the tropical Pacific, India/China, Southeast Asia, and northern Africa (Fig. a). The tropical Pacific and Southeast Asia are the main contributors to the winter 2007 pulse. Here, the contribution of India/China and northern Africa is insignificant, demonstrating that during the absence of the monsoon anticyclone in winter no vertical transport into the tropical pipe occurred.

Figure b shows that on the 550 K level, the contributions of all emission tracers (age <18 months) to the tropical pipe is up to 55 % by the end of October 2008. The contributions from Southeast Asia and the tropical Pacific are 16 % and 15 %, respectively. The contribution from India/China, which is mostly monsoon air, is much lower, around 6 %. Smaller fractions are estimated for the tropical Indian Ocean (4 %), northern Africa (3 %), and South America (4 %). The contributions of all other regions of the Earth's surface are each smaller than 3 % and are summarised in the tracer referred to as residual surface (8 %), shown in grey.

The black and the white lines in Fig. b mark the contribution of the summer 2007 and winter 2007–2008 pulses for each emission tracer, respectively (as shown in Fig. a). For the tracer from India/China as well as for northern Africa, the black and the white lines are overlapping, implying that there is no contribution from India/China or northern Africa from the winter 2007–2008 pulse in the tropical pipe (as shown above). Further, our findings show that air masses from India/China, thus mainly from the Asian monsoon anticyclone see, contribute to a smaller fraction of the composition of air within the tropical pipe at 550 K; the major part is from Southeast Asia and the tropical Pacific. We would like to point out that the CLaMS emission tracer for Southeast Asia includes both the land masses of Southeast Asia and parts of the western Pacific including the Maritime Continent, which is also known to have intense deep convection during summer e.g..