3.1 Zonal-mean perspective on transport

The evolution of the ASM tracer after its release is very similar to results shown by Ploeger et al. (2017, their Fig. 1). To avoid redundancy, we only show the AOF distributions around 10 months after the tracers are initialized (i.e. during April–June), focusing on tracers that have been transported into the global stratosphere. Figure 3 shows the average AOF from the ASM and NASM regions initialized in the 350–360 K (Fig. 3a and b) and 370–380 K (Fig. 3c and d) layers in July and August of 2010–2013 in the CLaMS-EI simulations during the following April–June. These results are based on four sets of CLaMS simulations, with the AOF set to 1 in the 350–360 or 370–380 K layer over the ASM or NASM region on each day during July–August. Air is transported to the global stratosphere from both the ASM and the NASM. However, much more monsoon air originates from the ASM region than from the NASM region, and ASM air is transported to a higher level in the TrP by the following April–June when compared to NASM air. More monsoon tracers are transported into the stratosphere from the 350–360 K layer than from the 370–380 K layer for both the ASM and NASM regions. Monsoon air initialized at 350–360 K is more prevalent in the SH stratosphere than in the NH, while monsoon air initialized at 370–380 K is more likely to remain in the NH. The distribution of ASM tracers initialized at 370–380 K (Fig. 3c) is similar to that reported by Ploeger et al. (2017, their Fig. 1d) based on a PV-barrier definition of the ASM, but the abundance of ASM air is greater in our results because our ASM domain is larger. The distribution of the monsoon tracer initialized at 340–350 K (not shown) is also skewed toward the SH, similar to the results for 350–360 K, while the distribution of the monsoon tracer from the 360–370 K layer is more symmetric between the hemispheres. In summary, during the following April–June, the monsoon tracers are more likely to end up in the SH stratosphere if they are initialized at lower levels (340–350 or 350–360 K), and more likely to end up in the NH stratosphere if they are initialized at higher levels (360–370 or 370–380 K). To avoid redundancy, we only show the results for tracers initialized in the 350–360 and 370–380 K layers in this paper.

Ploeger et al. (2017) showed that the simulated ASM tracer based on driving CLaMS with ERA-Interim correlates well with ACE-FTS observations of HCN. Here, we find that the HCN observations show strong correlations with both the ASM and NASM tracers in the TrP. This result suggests that both monsoon regions could serve as sources of enhanced HCN to the stratosphere, with the ASM region representing a relatively strong source and the NASM region representing a relatively weak source. Although the HCN mixing ratio over the NASM is lower than that over the ASM, it is much higher than that over the tropics (Randel et al., 2010, their Fig. 1). The peak HCN mixing ratios in the TrP closely overlap with the peak contributions of monsoon tracers initialized in the 350–360 K layer and are located slightly below the peak contributions of monsoon tracers initialized at 370–380 K. However, such vertical offsets should be interpreted with caution given likely overestimates of tropical upwelling in ERA-Interim (Dee et al., 2011; Wright and Fueglistaler, 2013; Ploeger et al., 2019; Tao et al., 2019).

Column integrals are calculated to further investigate the transport of the monsoon tracers. These integrals are defined as the thickness (in metres, m) of pure monsoon air assuming standard temperature and pressure for each altitude. Column integrals of monsoon air above the tropopause (SC), below the tropopause (TC), and through the entire atmosphere are shown along the bottom of each panel in Fig. 3. The column integrals of ASM and NASM tracers show similar patterns in both hemispheres. The TC, SC, and total monsoon tracer columns initialized at 350–360 K show slightly larger values in the SH. The larger columns in the SH originating from the 350–360 K layer over the boreal monsoon regions may arise from a combination of three effects: weaker confinement of air inside the monsoon regions at this layer (Garny and Randel, 2016; Vogel et al., 2019), hemispheric differences in isentropic mixing (Orbe et al., 2013; Konopka et al., 2017), and the seasonality of the Brewer–Dobson circulation (Seviour et al., 2011; Konopka et al., 2015). A portion of the monsoon tracers initialized at the 350–360 K level enters the tropics due to wave-driven isentropic transport, ascends through the tropical tropopause layer (TTL), and is then advected into the SH through the shallow branch of the Brewer–Dobson circulation (Rosenlof, 1995; Konopka et al., 2015). By contrast, monsoon tracers released in the 370–380 K layer are more tightly confined to the NH, with higher column-integrated values for both the troposphere and stratosphere in the NH.

Figure 4Same as Fig. 3 but for CLaMS-M2.

Download

Figure 4 shows the AOF during the following April–June for the ASM and NASM tracers initialized at 350–360 and 370–380 K in July–August based on CLaMS-M2 simulations. Similar to the results from CLaMS-EI, the fraction of ASM air is larger than that of NASM air in the TrP, and the ASM air reaches a higher vertical level. More monsoon tracers are transported into the stratosphere from lower levels than from higher levels, and the distributions of monsoon tracers also show hemispheric asymmetries similar to those in the CLaMS-EI case. However, despite these many similarities, the CLaMS-M2 simulations show much larger fractions of ASM and NASM air in the global stratosphere in comparison to those from CLaMS-EI. Another prominent difference concerns the vertical transport of monsoon air. The tracer peak from CLaMS-M2 is slightly below the HCN peak from satellite observations for tracers initialized at both the lower and the upper layers. This can be attributed to slower upwelling, which stems from smaller diabatic heating rates in the lower stratosphere in MERRA-2 (Ploeger et al., 2019; Tao et al., 2019). Slower upwelling leads to larger contributions from quasi-horizontal transport and smaller contributions from vertical transport relative to the CLaMS-EI simulations. Transport of tracers into the stratosphere takes a longer time in the CLaMS-M2 simulations, but a larger fraction of monsoon air reaches and remains in the stratosphere. Column-integrated monsoon tracer values from CLaMS-M2 are also much larger than those from CLaMS-EI. This difference arises in part because monsoon tracers from CLaMS-M2 reach relatively lower altitudes in the stratosphere by the following April–June compared to those from CLaMS-EI and air density is larger at lower altitudes. Differences between the SH and NH for monsoon tracers initialized in 370–380 K layer are smaller than those based on CLaMS-EI.

3.2 Two distinct transport pathways

In Sect. 3.1, we have discussed differences in transport into the stratosphere from the ASM and NASM regions, focusing on two different vertical levels of the monsoon source regions in the upper troposphere (the 350–360 and 370–380 K layers). The differences in simulated transport of monsoon tracers released at different vertical levels imply the existence of multiple transport pathways from the monsoon source regions to the TrP. Along the first pathway, termed “tropical pathway” in the following, the tracer is first horizontally advected to the tropics, where it ascends through the TTL into the TrP. Along the second pathway, termed “monsoon pathway” in the following, the tracer is first lifted across the tropopause within the monsoon anticyclone and then transported isentropically to the tropical lower stratosphere and TrP. In this section, we endeavour to disentangle these two pathways and clarify the extent of transport from the monsoon source regions to the global stratosphere by separating the contributions of vertical transport within the tropics from those of horizontal transport into the tropics. We also compare the artificial tracers from the model simulations with MLS water vapour observations and CLaMS-simulated water vapour to investigate the influence of the monsoons on stratospheric chemical composition.

Figure 5Potential temperature–time sections during July 2010 to June 2014 of zonal-mean monsoon air mass origin fraction (AOF: colour shading) between 15^∘ S and 15^∘ N released in the 350–360 K (a, b) and 370–380 K (c, d) layers within the ASM (a, c) and NASM (b, d) based on CLaMS-EI simulations. Tape-recorder signals calculated from Aura MLS water vapour (in parts per million of volume, ppmv) with the annual average from each year removed are shown as black contours for context. Grey dashed lines mark the 400 K level. Note that the tracer is set to zero everywhere on 1 July of each year.

Download

The vertical transport of ASM and NASM air into the tropical stratosphere (15^∘ S–15^∘ N) in CLaMS-EI is illustrated using the familiar “tape-recorder” form (Mote et al., 1996) in Fig. 5. The transport barrier between the tropics and subtropics is weaker during boreal summer (Chen, 1995; Haynes and Shuckburgh, 2000). Monsoon tracers released at lower levels are especially abundant in the tropical UTLS during summer, indicating that a large fraction of the monsoon air from the 350–360 K layer reaching the TrP may be advected quasi-horizontally into the tropics before ascending, in agreement with previous results based on trajectories initialized at 360 K (Garny and Randel, 2016). We also find that less of the monsoon tracer reaches the tropics when it is initialized at higher levels relative to when it is initialized at lower levels. This difference suggests that tracers initialized at higher levels within the monsoon regions primarily ascend locally before being mixed or advected into the tropics. ASM air rises slightly faster than NASM air and reaches slightly higher altitudes by the end of the simulation period. These features are consistent with the results discussed above.

MLS water vapour observations are included for additional context in Fig. 5. The annual mean of MLS water vapour is removed to highlight the annually repeating water vapour tape recorder. The simulated ASM and NASM tracers correlate well with the “wet” phase of the tape recorder (positive anomalies after removing the annual mean), which starts each year in boreal summer. The close correspondence between the monsoon air mass tracers and the “wet” phase of the water vapour tape recorder in the tropics is consistent with the idea that the ASM and NASM contribute to moistening the tropical lower stratosphere during boreal summer (e.g. Dethof et al., 1999; Bannister et al., 2004; Fu et al., 2006; James et al., 2008; Randel et al., 2012). Note that large concentrations of monsoon tracers in the tropics do not always correlate with large water vapour mixing ratios in the tropical UTLS, as water vapour content at these altitudes depends not only on the origin of the air parcel but also on its temperature history (Fueglistaler and Haynes, 2005; Nützel et al., 2019).

CLaMS-M2 simulations (not shown) have much in common with the CLaMS-EI results with respect to the transport of monsoon tracers into the tropics. However, there are also some differences. As mentioned above, the vertical transport of the ASM tracers is slightly weaker, with the tracers reaching slightly lower altitudes by the end of the simulation period. Likely linked to this slower upwelling, more monsoon air arrives in the tropics between each summer monsoon season and the following spring compared to CLaMS-EI simulations. A relatively large fraction of monsoon air is transported quasi-horizontally into the tropics before ascending to the TrP. The slower upwelling in MERRA-2 also manifests in a slight lag between the simulated monsoon tracer distribution and the MLS water vapour tape recorder signal.

Figure 6Zonal-mean tracer concentrations (colour shading) for the ASM (a, c) and NASM (b, d) tracers initialized at 350–360 K (a, b) and 370–380 K (c, d) as functions of time and latitude on the 400 K isentropic surface based on CLaMS-EI (see also Fig. 5). MLS water vapour retrievals (in parts per million of volume, ppmv) with the annual average from each year removed are shown as black contours for context. Note that the tracer is set to zero everywhere on 1 July of each year.

Download

To investigate the influence of transport from the boreal monsoon regions on the lowermost stratosphere, we examine the horizontal spread of the ASM and NASM tracers on the 400 K isentropic surface based on the CLaMS-EI simulations (Fig. 6). Tracers initialized on both levels show broadly similar features at 400 K, although the ASM contributes much more to the lower stratosphere than does the NASM. The ASM and NASM tracers initialized at 350–360 K are less confined in latitude, with much more transport to the SH than tracers initialized at 370–380 K. Tracers initialized at 370–380 K are more confined to the monsoon region and the NH, consistent with trajectory simulations initialized at 380 K (Garny and Randel, 2016). The ASM and NASM tracers coincide well with the MLS water vapour “wet” signal in the lateral tape recorder, which is again consistent with the idea that the monsoons contribute substantially to water vapour concentrations in the stratosphere. The monsoon tracer distributions based on CLaMS-M2 are more widespread than those based on CLaMS-EI, with a greater tendency to remain in the extratropical stratosphere regardless of the initialization level. Both sets of simulations produce close matches between the ASM and NASM tracer peaks and the “wet” signal in MLS water vapour, indicating that horizontal transport in CLaMS-M2 is in good agreement with that in CLaMS-EI.

Figure 7Correlations between CLaMS simulated water vapour and the ASM (a, c) and NASM (b, d) tracers in the tropical stratosphere (15^∘ S–15^∘ N; 420–500 K) from November 2010 to May 2011 with monsoon tracers released in the 350–360 K (a, b) and 370–380 K (c, d) layers based on CLaMS-EI. Contours mark young (less than 9-month-old) air mass fractions of 24 %, 33 %, and 42 % (from thin to thick dotted isolines) and 51 %, 60 %, and 69 % (from thin to thick solid isolines). The cyan and magenta arrows respectively show the monsoon pathway and tropical pathway.

Download

The qualitative correspondence between the monsoon tracers and the “wet” phases of the vertical and horizontal tape recorders suggests a strong connection between stratospheric water and transport from the monsoon regions. To further investigate the extent to which the monsoon tracers can explain variability in lower stratospheric water vapour and clarify the differences between the tropical pathway and the monsoon pathway into the TrP, we correlate the monsoon tracer amounts with CLaMS-simulated concentrations of stratospheric water vapour, which have been successfully validated and used in many previous studies (e.g. Vogel et al., 2016; Rolf et al., 2018; Tao et al., 2019, and references therein). We analyse the data from November to May, i.e. about 4–10 months after the release of the monsoon tracers (as discussed below, this is the earliest that the tracers may reach the TrP). As shown in Fig. 7, a strong positive correlation exists between the monsoon tracers and water vapour in the tropics. Larger ASM and NASM air fractions correspond to larger water vapour mixing ratios in the tropical lower stratosphere (420–500 K). The distinct differences between the upper panels (tracers released in the 350–360 K layer) and the lower panels (tracers released in the 370–380 K layer) reflect the two transport pathways from the ASM and NASM to the tropical lower stratosphere. The right branch in Fig. 7a (Fig. 7b) mimics the joint distribution of ASM (NASM) tracers from the 370–380 K layer and water vapour shown in Fig. 7c (Fig. 7d). This branch thus indicates the correlation along the monsoon transport pathway (cyan arrow), in which air ascends locally across the tropopause and then enters the tropics laterally. The left branch in Fig. 7a (Fig. 7b) is then linked to the correlation along the tropical pathway (magenta arrow), in which air from the upper troposphere above the ASM (NASM) region enters the tropics quasi-horizontally before ascending via tropical upwelling.

The tropical pathway is more common for tracers released at 350–360 K. Most of the monsoon tracers released at 370–380 K are transported to the TrP along the monsoon pathway, although the tropical pathway is not completely absent for tracers released in this layer. The monsoon pathway coincides well with the upward spiraling range highlighted by Vogel et al. (2019). The water vapour mixing ratio along the tropical pathway is lower than that along the monsoon pathway because more dehydration is caused by the lower tropopause temperature over the tropics (figure omitted). The mass fraction of air younger than 9 months is also shown in Fig. 7 (black contours) to further evaluate the vertical components of the pathways by which monsoon air reaches the tropical lower stratosphere. Young air mass fractions are considerably larger for the tropical pathway, suggesting that overall ascent rates over the tropics (along the tropical pathway) are much faster than those over the monsoon regions (along the monsoon pathway) until the next early summer. Here, a single 1-year simulation is used for each tracer because interannual variability obscures the structure in the correlation between water vapour and monsoon tracers. Results from other years are similar (not shown).

4.1 Time evolution of the monsoon tracer

Figure 8a shows CLaMS-EI time series of the ASM tracer in the three destination regions (TrP, LS-NH, and LS-SH) defined in Fig. 2. The ASM tracer initialized at 370–380 K shows the largest air fraction in the LS-NH in the middle of September, with a peak value around 22 %. The peak date is earlier than that shown by Ploeger et al. (2017, their Fig. 2) (peak around 15 %). This is because the geographical domain of the ASM tracer used here is larger than that based on the PV barrier definition used by Ploeger et al. (2017). As a consequence, air parcels that are not well confined inside the anticyclone may reach the LS-NH faster. The ASM tracers released in the 350–360 K layer show two peaks in the LS-NH. The first of these is at the beginning of September and is related to rapid isentropic poleward transport. The second occurs about 1 month later relative to the tracers initialized in the 370–380 K layer due to the need for additional (spiraling) ascent within the anticyclone (Ploeger et al., 2010; Garny and Randel, 2016; Vogel et al., 2019). During the first 3 months, fewer ASM tracers released at 350–360 K are diagnosed in the LS-NH relative to ASM tracers released at 370–380 K. However, after 3 months the ASM tracers released at 350–360 K contribute significantly more to the composition in the LS-NH and remain in the LS-NH for longer when compared to tracers released at 370–380 K. Similar transitions in the relative concentrations of tracers released at 350–360 and 370–380 K can be diagnosed in the TrP region, although the maxima are about 6 months later compared to those in the LS-NH. The ASM tracers start to reach the TrP region around November, peaking in the TrP around March at a value of ∼4 % for tracers released in the 370–380 K layer. The timing of this peak is about half a month earlier than that of ASM tracers released from the 350–360 K layer, although the latter has a larger peak value (∼7 %). Tracers released at 350–360 K are much more abundant in the LS-SH than tracers released at 370–380 K throughout the year.

The corresponding time series of air fractions from the NASM source region to the three destination regions are illustrated in Fig. 8b. These time series show some important similarities to those based on the ASM tracers. The NASM tracers initialized in both the 350–360 and 370–380 K layers reach their peak values in the LS-NH about 2 weeks earlier than the corresponding ASM tracer concentrations, likely owing to the weaker dynamical confinement of NASM air in the upper troposphere (see also Fig. 1d). The amount of NASM tracer initialized at 370–380 K that reaches the LS-NH (21 %) is similar to the amount of ASM tracer initialized at the same vertical layer that reaches the LS-NH. This similarity is also manifest in the comparable CO concentrations along 90^∘ E and 90^∘ W in the LS-NH (Fig. 1c and d). The amount of NASM tracer initialized at 350–360 K is much lower, in part because the 370–380 K layer is above the tropopause in the NASM region. However, NASM air fractions in the LS-SH and TrP are much smaller than the corresponding ASM air fractions in these regions, generally by more than a factor of 2. As for the ASM, NASM tracers initialized at 350–360 K are more abundant in the three destination regions by the end relative to NASM tracers initialized at 370–380 K.

Most tropospheric air that is transported into the global stratosphere passes through the TTL via tropical upwelling. Thus, to evaluate the relative contributions of the ASM and NASM sources to the stratosphere, we use the transport of air originating in the same vertical layers over the tropics (15^∘ S–15^∘ N) as a benchmark. By comparing tracers tracking air of tropical origin in the three destination regions (Fig. 8c) with those tracking air with monsoon origin in the same destination regions, we conclude that ASM tracers are as abundant as tropical tracers in the LS-NH. Moreover, the abundance of the NASM tracer initialized in the 370–380 K layer is also comparable to the abundances of ASM or tropical tracers in the LS-NH region. Inside the TrP, the peak ASM contribution is smaller than the peak tropical contribution in early spring. However, the abundance of ASM air originating from the 350–360 K layer in the TrP is comparable to that from the tropics at the end of the simulation period. Both values are about 3 times larger than the NASM contribution. Much more air in the LS-SH originates from the tropical upper troposphere than from the two monsoon regions.

The time series of all source air fractions discussed above calculated from CLaMS-M2 are very similar to those derived from CLaMS-EI (not shown). Even daily perturbations of such tracers driven by Rossby waves and diagnosed on individual isentropic surfaces show very strong similarities, as outlined later. ASM contributions based on CLaMS-M2 lag behind those based on CLaMS-EI by about 1 month, first reaching the TrP in December rather than November. This delay is related to the slower tropical upwelling in MERRA-2 relative to ERA-Interim as discussed above. For the same reason, CLaMS-M2 produces larger amounts of ASM tracers in the stratosphere at the end of the simulation period. The time series of NASM tracers based on CLaMS-M2 are likewise similar to those based on CLaMS-EI, again including daily variability. For the tropical source, CLaMS-M2 shows similar peak values to CLaMS-EI in the LS-SH destination region, but lower peak values in the LS-NH and TrP. However, the amount of air of tropical origin reaching the TrP is larger in CLaMS-M2 than in CLaMS-EI starting from April.

Table 1Typical masses of air contained in each box for the test experiments over the ASM, NASM, and tropical source regions in CLaMS-EI (a) and CLaMS-M2 (b). All masses are reported in kilograms (kg).

Download Print Version | Download XLSX

4.2 Transport efficiency

Our comparison of AOFs conditioned on different transport pathways is partially hampered by the fact that the source regions have different masses. To illustrate such differences in the domains, Table 1 shows the mass calculated for all considered source regions (ASM, NASM, and tropics) based on both reanalyses. To better control for the potential influence of differences in the sizes of the source domains on our quantitative estimates of transport to the stratosphere, we define the efficiency of transport (TE) using the air mass of the source and destination domains, m_source and m_dest, to normalize the mass transported from each source region to the defined stratospheric destination regions. Thus, TE is defined as $\mathrm{AOF}\times (m_{\mathrm{dest}}/m_{\mathrm{source}})$.

Figure 9Climatological time series of transport efficiency (TE) as derived from the CLaMS-EI simulations and calculated for transport from the ASM (solid line), NASM (dashed line), and tropics (dotted line) source regions to the (a) LS-NH, (b) LS-SH, and (c) TrP destination regions. TE (in per mille, ‰) is defined as $\mathrm{AOF}\times (m_{\mathrm{dest}}/m_{\mathrm{source}})$, where m_source and m_dest are the masses of the source and destination regions, respectively. Different colours of the lines represent tracers released at different levels.

Download

Figure 9a compares the TE from the three source regions (ASM, NASM, and tropics), each divided into two layers (350–360 and 370–380 K), to the LS-NH destination region. In contrast to the total budget, where tropical and monsoon contributions are very similar (Fig. 8), the ASM and NASM monsoons dominate the TE into the LS-NH, especially for sources in the 370–380 K altitude range. At the beginning, the TE is larger for the NASM than for the ASM; the larger NASM TE is in part related to the fact that the 370–380 K layer is above the tropopause over the NASM region and the NASM tracers are less confined. However, transport from the ASM becomes most efficient starting from October. Perhaps more surprising, the TE into the LS-SH is also dominated by the ASM and NASM regions for tracers released at 350–360 K after a certain date (see Fig. 9b). For the first 4–5 months, the TE is larger from the tropics than from the monsoon regions, especially for tracers released between 370 and 380 K. However, transport from the 350–360 K layers in the ASM and NASM regions reaches the LS-SH more efficiently than transport from the same layer in the tropics starting from December or January (with TE from the ASM larger than that from the NASM). The time evolution of the source-resolved TE into the TrP is shown in Fig. 9c. Here, the highest TE values are along the pathways starting in the ASM source region after some certain date, followed by those starting in the tropical and NASM source regions. The NASM TE exceeds the tropical TE by March for tracers from 350–360 K. In contrast to the total source budget (Fig. 8), transport is consistently less efficient from lower levels than from higher levels.

Patterns of TE based on CLaMS-M2 are similar to those based on CLaMS-EI (not shown), although there are a few important differences. First, TEs from the ASM and NASM source regions to the LS-SH destination region are much larger in CLaMS-M2 than in CLaMS-EI. Both values exceed those from the tropical source region by December (for tracers initialized at 350–360 K) or February (for tracers initialized at 370–380 K). Second, the TE to the TrP in CLaMS-M2 shows slightly higher values for both the ASM and NASM sources and lower values for tropical upwelling relative to CLaMS-EI. TE from the NASM to the TrP slightly exceeds that from the tropics to the TrP starting in March (April) for tracers initialized at 350–360 K (370–380 K).