On the behaviour of the tropopause folding events over the Tibetan Plateau

Due to its harsh natural conditions, there had not been any intensive radiosonde observations over the Tibetan Plateau (TP) before the year 2008, when a regional radiosonde observation network was implemented through a Sino-Japan joint cooperation project. This paper reports, on the basis of these observations, on an analysis of the structure of upper troposphere and lower stratosphere (UTLS) and provides observations of stratosphere and troposphere exchange (STE) over the TP. Due to sparseness of high resolution radiosonde data, many previous studies assumed that there was only one thermal tropopause over the TP. Actually, the radiosonde temperature profiles in winter time over the TP often exhibit a multiple tropopause (MT). The MT occurs in winter with a high frequency over the Plateau. MT events during this time are associated with tropopause folds near the subtropical westerly jet. The MT consistently varied with the movement of the jet. The MT becomes a single tropopause with the development of the monsoon. The detailed analyses of MT characteristics are reported in this paper. Earlier analyses of global MT events (with data based on GPS radio occultation, ERA-40 data and Integrated Global Radiosonde Archive database) resulted in a climatic frequency of MT occurrences in the winter season over the Plateau is not more than 40 %. Based on high resolution data of intensive radiosonde observations, our estimations of MT occurrence over the Plateau can be as high as 80 % during Correspondence to: X. L. Chen (chen24746@itc.nl) certain winters. This reminds us to pay more attention to the MT events above the Plateau. The influence of the coarse vertical resolution and other effects on the estimation of MT occurrence frequency are also discussed. The stratospheric intruding episodes are generally associated with the presence of subtropical jet stream over the Plateau. The complex structure of dynamic tropopause folding over the Plateau have been reflected by the thermal MT events observed by radiosondes. The intrusion of air masses from the stratosphere may contribute to a higher upper tropospheric ozone concentration in winter than in summer above the plateau.


Introduction
The Tibetan Plateau (TP) has an average height of over 4000 m and an area of 2.5 million km 2 and exerts a profound thermal and dynamic influence on the Asian monsoon and global weather and climate.Due to the low air density and the strong solar radiation, the TP allows more energy from the surface, directly heating the middle troposphere.This heat can influence the upper troposphere and warms the tropopause region (Fu et al., 2006).The exchange between upper troposphere and lower stratosphere (UTLS) was found over the Tibetan Plateau (Cong et al., 2002;Fu et al., 2006;Steinwagner et al., 2007;Sprenger et al., 2003;Zhan and Li, 2008).As a huge elevated heating source, a rigorous deep convection area (Yang et al., 2004), a dynamical pumping and sucking unit (Duan et al., 2005), and westerly jet above it, the TP generates an active stratosphere and troposphere Published by Copernicus Publications on behalf of the European Geosciences Union.exchange (STE) region, forming a short-circuit and pathway for the STE (Fu et al., 2006).An improved understanding of the STE depends on our ability to quantify the UTLS structure and its variability (Stohl et al., 2003).Due to a lack of high resolution observational data, the structure of the UTLS above the Tibetan Plateau is still insufficiently understood (Gettelman et al., 2010;Hegglin et al., 2010;SPARC Report, 2010).
The occurrence of multiple tropopauses (MT) plays a central role in the structure and features of the tropopause layer which plays a crucial role in the exchange between the troposphere and stratosphere above the TP (Añel et al., 2008).Both Añel et al. (2008) (employing Integrated Global Radiosonde Archive database) and Randel et al. (2007) (based on GPS radio occultation measurements and ERA-40) have derived global statistics of MT.The present paper presents statistics of MT and STE events above the Tibetan Plateau based on high resolution sonde data for the first time.Therefore, it will contribute to MT statistics of Añel et al. (2008) and Randel et al. (2007) and helps us to understand the exchanges between the stratosphere and troposphere over the Tibetan Plateau.
Tropopause folds are the key feature and favourable structures for cross-tropopause exchange in the subtropics (Sprenger et al., 2003;Shapiro, 1980); The frequency of tropopause folds is highest in the subtropics related to the prevailing subtropical upper level jet stream (Reed, 1955;Schmidt et al., 2005;Randel et al., 2007).The Tibetan Plateau is situated at 28 • N-38 • N, a region which is significantly influenced by the subtropical jet stream as it moves northward from winter to summer.Sprenger et al. (2003) have analysed the frequency and global distribution of tropopause folds.Their results demonstrate that the global frequency of shallow folds above the Plateau is highest during winter (see figure 3 in Sprenger et al., 2003).
Meanwhile, satellite observations of ozone have shown "Ozone Mini-Hole" events and ozone valley phenomena over the Plateau (Zhou et al., 2005;Tobo et al., 2008;Bian, 2009).The mechanisms responsible for the low total ozone have been discussed.Previous studies suggest that the transport of tropospheric air with low ozone concentration across the tropopause influences the low ozone concentration in the lower stratosphere in summer (Zhou et al., 1995;Zou, 1996).Tian et al. (2008) found that the variability of the total column ozone over the TP is closely related to uplift and descent of isentropic surfaces.Tobo et al. (2008) observed ozone anomalies near the tropopause (150-70 hPa) having a large contribution to the low total ozone.Intrusions of stratospheric air masses with high ozone concentration into the troposphere were closely associated with tropopause folds (Reed, 1955;Sprenger et al., 2003;Beekmann et al., 1997).Due to the high frequency of tropopause folds (Sprenger et al., 2003), stratosphere intrusions should frequently happen over the Plateau.These downward transports of rich ozone air strongly influence the vertical ozone distribution espe-  cially at UTLS area over the Plateau.However, the variation of the stratospheric intrusions and their influence on the vertical ozone distribution need further analyses.
In order to understand the atmosphere heating and water vapour variations over the TP during different phases of the monsoon, the Sino-Japan joint cooperation project was carried out to collect intensive radiosonde observations in 2008 (Xu et al., 2008).As illustrated in Fig. 1, nine radiosonde observation sites are included in our study.Four sites (Gerze, Lasha, Nagqu, and Litang) were situated on the Plateau.The height of other stations was lower than 2000 m a.s.l.These sites will be discussed as low altitude observations.
There is a lack of high resolution information of the tropopause above the Plateau.This study will contribute to characteristics of MT events using the most advanced radiosonde data in this area.Section 2 describes observational data and the tropopause definition.The MT characteristics are presented in Sect.3. Section 4 discusses dynamic characteristics of tropopause folds accompanied by thermal MT events during the Sino-Japan joint cooperation project.Section 5 summarizes the study with conclusions and discussions.

Observation data and tropopause definition
Three sites, Gerze, Litang and Dali, were equipped with a Vaisala DigiCORA III GPS radiosonding system.The radiosondes used were Vaisala RS92 calibrated with local meteorological measurements before releasing.The sounding data contained profiles of temperature, pressure, relative humidity, horizontal wind speed and direction.Utilizing differential GPS theory, the receiver can automatically compute the height of the sensor and the wind speed.The radiosonde transmits data to the receiver every 2 s.With an average 5 m s −1 ascending speed, the obtained vertical resolution of the data is about 10 m.The other six sites employed the Chinese meteorological radiosonde system (detailed information about the radiosonde can be found at the following website: http://www.cwqx.com/).These systems can observe temperature, pressure, humidity, wind speed and direction at a vertical resolution of 100 m.All the observation data are interpolated to 20 m vertically equal-spaced levels using the splines method for the convenience of detailed analysis and comparison.
Three intensive observation periods (IOP) were carried out.Detailed information about the observation dates are shown in Table 1.The first observation period (IOP1) was aimed at winter time.The second observation period (IOP2) was conducted in the period of monsoon onset time.The third observation period (IOP3) was almost in the mature phase of monsoon.Four radiosondes were released every day at 01:00, 07:00, 13:00 and 19:00 local standard time (LT).
For computing the tropopause height, we employ the WMO lapse rate tropopause (LRT) definition (World Meteorological Organization, 1957): (a) The first tropopause is defined as the lowest level at which the lapse rate decreases to 2 K km −1 or less, provided also that the average lapse rate between this level and all higher levels within 2 km does not exceed 2 K km −1 .
(b) If above the first tropopause the average lapse rate between any level and all higher levels within 1 km exceeds 3 K km −1 , then a second tropopause is defined by the same criterion as (a).them in Table 2.The frequency of the MT events is given as a percentage with respect to the number of LRT1 events.The MT occurrences of the three Plateau stations (Gerze, Nagqu and Litang) for IOP1 are as high as from 72.5 % to 84 %.Our analyses show that the frequency and seasonal variation of MT events are high during winter.The frequency of MT (discussed with Randel) derived from the GPS occultation data and the ERA-40 data during DJF (December, January and Feburary) in Randel et al. (2007) on the Plateau was not more than 40 % in their Fig.9a and A1b.Añel et al. (2008) analysed global MT events performed on the Integrated Global Radiosonde Archive database (IGRA).As shown in their Fig. 2, the percentage of DT occurrence had less seasonal variation and the value of DJF is not more than 20 % above the Plateau.Thus, more attention should be paid to the MT events above the Plateau and the influence of the low vertical resolution of GPS and ERA-40 data on the estimation of MT events.At the Dali site, which has an elevation of 1960 m and locates out of the Plateau, the MT occurrence of IOP1 is about 12.9 %.This value is much lower than that of other stations.We also picked out the radiosondes of the same date in each IOP.The statistics suggest that when moving to the southern and low areas, the frequency of MT becomes lower.The MT   frequencies of all the Plateau stations during IOP3 are the lowest compared to the other two periods.This result supports the conclusion of MT occuring in winter time with a high frequency over the Plateau, which is similar to previous climatological studies.
Figure 3 shows the statistical distributions of the MT height during the three IOPs.The statistics for IOP1 demonstrate an overall bimodal distribution with maxima near 10 km (primarily associated with LRT1) and 17 km (firstly contributed by LRT2).According to the significant difference in the two tropopause heights, we mark them separately as "low" and "high" tropopause.LRT1 is more inclined to low tropopause during this time.The majority of the LRT2 height in IOP1 is around 17 km, which is closer to high tropopause of the equatorial area.The statistics of IOP2 and IOP3, in Fig. 3, show a single maximum (primarily LRT1) centred at 17 km, which means MT is rarely observed during the monsoon season.Khalili (1975) observed a significant negative correlation between the tropopause height and its temperature.We also analysed the relationship between the tropopause height and temperature.Both LRT1 and LRT2 heights are in opposite phases with their temperature scaled in hours as shown in Fig. 4. LRT3 does not show this character, suggesting itself as a lower stratosphere layer (Añel et al., 2007;Gettelman and Forster, 2002).On the contrary, LRT1 and LRT2 should be attributed to tropospheric stable layers.
The average heights of DT during the three periods are listed in Table 3. LRT1 of the three Plateau stations (Gerze, Nagqu and Litang) have heights around 11-13 km during IOP1.These tropopauses can be treated as low tropopauses.For all observation periods and sites, LRT2 is characterised by a height around 17 km.After the monsoon onset, the LRT1 height over the Plateau was elevated to 4-5 km, indicating the disappearance of the low tropopause.This can be explained as a result of both the Plateau's thermal forcing, which causes large-scale ascent flow and vertical convection, and the poleward movement of the jet stream.Tian et al. (2008) suggested the thermal sink (source), by forcing a descent (ascent) flow, lowers (lifts) the tropopause.The thermal dynamic effects of the Plateau can be one reason for the seasonal variation of the tropopause, but does not seem probable in explaining variations from 10-17 km in the LRT1 height during winter time.Our explanation will be presented in the next section.

The tropopause folds over the Tibetan Plateau
It is well known that the altitude variation of tropopause in the extratropics has a close relationship with the local synoptic situation.In order to analyse the meteorological situation associated with the single and double tropopauses, we use the ERA-40 reanalysis data provided by ECMWF, the European Centre for Medium-Range Weather Forecasts (Uppala et al., 2005).We take 25 February 2008, 12:00 LT (Fig. 5a) and 29 February 2008, 12:00 LT (Fig. 5b) as first examples.During these two times, DT and single tropopause were observed at Gerze. Figure 5 includes potential vorticity (PV) isolines (units: 1 PVU), zonal wind and potential      temperature derived from ERA-40 data.Inverted triangles show simultaneous tropopause heights observed at the two stations of Gerze and Nagqu.In order to exactly locate their positions of thermal tropopause, we interpolated the ERA-40 1.5 • latitudinal resolution to 0.1 • , and pressure levels lower than 400 hPa to 10 hPa, higher than 400 hPa to 25 hPa vertical resolution with the spline method.Due to the westerly wind, the soundings were usually moved to the east with a relative small north-south variation.The positions of MT observed by simultaneous radiosondes were marked by the same latitude as its station point in Fig. 5.
A method to identify the dynamical tropopause is based on potential vorticity (PV).A series of PV isolines (PV = 1-5 PVU) were plotted to indicate the approximate location of the dynamical tropopause (blue lines in Fig. 5).A principal indicator of intrusion of stratospheric air into troposphere is the occurrence of anomalous high PV values reaching down towards middle troposphere.It is because PV is generally greater in the stratosphere than in the troposphere (Holton, 2004).Figure 5a shows that the westerly jet stream is at a latitude of 30 • N above the TP.The observed maximum wind speed is higher than 60 m s −1 at 160 hPa level.The PV isolines were distorted in the vicinity of the northern edge of the subtropical jet stream.The 2 PVU isosurface reaches down to 380 hPa over the Plateau.Folds around 30 • N is identified on isentropic surfaces between 300 K and 340 K penetrating deeply into the troposphere in Fig. 5a.The dynamical tropopause (identified by 1-5 PVU) exhibit a medium folded structure (Sprenger et al., 2003)   ozone profile around 30 • N shows the intrusion caused by the tropopause fold.The ERA-40 ozone distributions around folds were found to be a coincident with that of PV (see Fig. 6). Figure 6 shows the dynamical effects of wind on 25 February 2008.The tongue of high ozone air were transported from polar stratosphere to middle troposphere above the Plateau by southern downward meridional wind on the polarward side of the jet streams as shown in Fig. 6.The time series of the same pictures as Fig. 5 et al., 2003).Because of the high elevation of the Plateau, the intrusions can easily transport stratospheric air mass to its ground.A similar meteorological situation on 29 February 2008, 12:00 LT (Fig. 5b) was further used to study single tropopauses.The second tropopause at 70 hPa observed by radiosondes in Fig. 5b should be a stratospheric layer, not real tropopause.Thus, the profile at this time is taken as a single tropopause event.The jet core (defined as the centre of wind speed higher than 50 m s −1 ) moved northward to 33 • N, on the north of Gerze and Nagqu site.The shallow folded structure (Sprenger et al., 2003) at this time retreated to the north of the Plateau.The equatorial tropopause extends over the South Plateau.The dynamical tropopause indicates no folds above the two stations.The radiosonde data show high thermal tropopauses at both the stations.The jet core positions, in both periods of Fig. 2, were compared.From 25 February 2008, 0:00 to 26 February 2008, 13:00 LT, the jet core propagates from around 32 • N to 26 • N. The tropopause folds also have coordinated displacement when the radiosondes at Gerze site observed MT events (Fig. 2a).During the single tropopause periods from 29 February 2008, 01:00 to 1 March 2008, 13:00 LT in Fig. 2b, the jet core locates at 32 • N-33 • N. The tropopause folds also move northward and weaken compared to folds observed during the period of double tropopause.The folds have less influence upon the UTLS above Gerze station.The tropical tropopause has entered the station, which reflects a high tropopause in this period.According to the analysis of these two time series, the tropopause folds follow the displacement of the subtropical jet over the Plateau.This process significantly influences the structure of UTLS above the Plateau.
Stratospheric intrusions of air usually occur with tropopause folds.A frequent descent air circulation above the Plateau has been observed during winter (e.g., Yanai et al., 1992).This flow strongly favours stratospheric intrusions into the troposphere during winter.Seasonal changes in PV in a pressure-longitude cross-section do not reveal any folds (see supplementary Fig. 7a).This suggests that meridional  folds rather than zonally aligned folds are observed in the UTLS above the Plateau.The southern downward intrusions of high ozone from the stratosphere to the Plateau troposphere in winter are caused by the meridional folds rather than zonal folds.The intrusions in winter caused by meridional folds are more frequent than in summer.This effect can be one reason for the total column ozone value above the Plateau in winter, higher than that of summer.Recently, a middle tropospheric ozone minimum in June (Liu et al., 2009), and low ozone concentration in the UTLS are observed (Tobo et al., 2008).Liu et al. (2009) explained middle tropospheric ozone minimum with the effect of Asian summer monsoon.Together with the effect of Asian summer monsoon anticyclone, less intrusions of stratospheric air in summer may contribute to the above-mentioned phenomenon of low ozone concentration in the middle and upper troposphere.The high values of tropopause heights above TP in summer contribute also to low total column ozone in summer.
The Dali station (25.71 • N, 100.18 • E) has lower frequencies of MT throughout the year (in Table 2), which can be related to the scarce propagation of the westerly jet and folds passing the latitude of the station (Fig. 7).The movement of the jet in a north-south direction is accompanied by the latitudinal movements of tropopause folds, which influence the UTLS structure above the Plateau.The variation of UTLS dynamical structures was reflected by the single or multiple tropopauses observed at the Plateau stations.The observed, two distinctive peak values in LRT1 height distribution in winter could be related to latitudinal movement of folds and jets.Both the heating of the Plateau and poleward extending of tropical tropopause (Pan et al., 2009;Castanheira et al., 2009) can contribute to the observed single high tropopause during IOP2 and IOP3, when the folds retreat northward.
Due to the close relationship between the jets and tropopause folds (Shapiro, 1980), we display the seasonal variations of the jets and tropopause folds (highlighted by 1 and 2 PVU contour lines) to investigate seasonal variation in tropopause structure.Subtropical jet is strengthened by the cooling effect of the Plateau (Ye and Gao, 1979), and is located at the south of the Plateau during winter season.The folds in February and March are the deepest.With the development of the monsoon, subtropical jet weakens and retreats to the north of the plateau, as can be seen by poleward movement of the jet during July.In accordance with the seasonal movement of the jet in a north-south direction, the tropopause folds also move from the south in winter to north in summer.The variation of tropopause folds over the Plateau has also been shown by Sprenger et al. (2003) and the influence of the northward movement of the jet stream by Ding and Wang (2006).Their study attributed summertime maximum of ozone concentration at Waliguan (36.28  2008) is different.All make a sound comparison between us and others impossible.The MT statistics over the Tibetan Plateau using ERA-40 and GPS data maybe have shortcomings to capture more tropopause.More attention should be paid to the MT events above the Plateau and the influence of the resolution on the statistics of MT.The phenomena of the recent rising trend of MT events (Castanheira et al., 2009) can cause our estimations of MT occurrence in 2008 to be higher than that of the past ten years and this partially explains our results of MT events higher than the climatic value of Randel et al. (2007) and Añel et al. (2008).Whether there are any other significant differences in MT frequencies, this still needs to be testified with higher resolution data of radiosonde.The dynamical tropopause above the Plateau in pressurelongitude cross-section and pressure-latitude cross-section has been discussed.No folds revealed in pressure-longitude cross-section above the Plateau, which is different from the pressure-latitude cross-section.The dynamical tropopause exhibits a meridional folded structure around the subtropical jet.This folded structure has been demonstrated by radiosonde observations over the Plateau.This paper further shows the close relationship between the folds in pressurelatitude cross-section (meridional folds) and the subtropical jet stream above the Plateau.The meridional folds related with the westerly jet dominate UTLS above the plateau.As Zhang et al. (2010) pointed out, the Plateau terrain has no significant effect on the morphology of folds.The westerly jet maybe the dominative influence on the morphology of the folds above the Plateau.
A persistent maximum of chemical constituents in the UTLS (Park et al., 2009) may also reduce the ozone content in column air above the Plateau.Tian et al. (2008) pointed out that the low column ozone over the TP is rather related to transport than to chemical reactions.We found intrusions of stratospheric air frequently happening in spring time, but it rarely occurs in summer time over the Plateau.Our results support that the reduced southern downward transport of stratospheric air may partially contribute to the phenomena of low ozone content in the middle troposphere and low column ozone in summer time above the Plateau.The height variation of the tropopause, discussed in our paper, can also be used to explain the low value of the total column ozone in summer above the TP.
The lack of high resolution data in the atmosphere has hampered our knowledge about tropopause characteristics so far and limiting our ability to evaluate model performance in UTLS (Hegglin et al., 2010).In addition to our dataset, more radiosonde data with high vertical resolution are indispensable to further quantify tropopause structure and variability above the Tibetan Plateau.Further data are needed for a better understanding of the high MT frequency during winter above the Tibetan Plateau.Discovering that MT frequencies are really high in periods of winter time, this work could further deepen our understanding in UTLS structure over the Plateau.The high-resolution radiosonde data are also an important source of information for studying the thermal structure of UTLS.

Fig. 3 .
Fig. 3. Distribution of multi-tropopause heights at different phases of the monsoon.The first tropopause (LRT1), second tropopause (LRT2) and third tropopause (LRT3) are displayed by blue, green and brown, respectively.Table 1 lists which sites and data are analysed in these three IOPs.

Fig. 4 .
Fig. 4. Multi-tropopause heights versus temperatures.The first tropopause (LRT1), second tropopause (LRT2) and third tropopause (LRT3) are displayed by red, green and black colors separately.Table 1 lists which sites and data are analyzed in these three IOPs.

Fig. 4 .
Fig. 4. Multi-tropopause heights versus temperatures.The first tropopause (LRT1), second tropopause (LRT2) and third tropopause (LRT3) are displayed by red, green and black, separately.Table 1 lists which sites and data are analysed in these three IOPs.

Fig. 7 .
Fig. 7. Pressure-latitude cross section of monthly average zonal wind (black contours, m/s) and potential vorticity (PV) (red lines, contours of 1-2 PVU units) at 90 。 E between 15 。 N and 45 。 N in latitude and between 1000 hPa and 20 hPa in the vertical derived from ERA-40 data of 2008.

Fig. 7 .
Fig. 7. Pressure-latitude cross-section of monthly average zonal wind (black contours, m s −1 ) and potential vorticity (PV) (red lines, contours of 1-2 PVU units) at 90 • E between 15 • N and 45 • N in latitude and between 1000 hPa and 20 hPa in the vertical derived from ERA-40 data of 2008.

Table 1 .
Intensive observation dates of the three periods in 2008.
Fig. 3. Distribution of multi-tropopause heights at different phases of the monsoon.The first tropopause (LRT1), second tropopause (LRT2) and third tropopause (LRT3) are displayed by blue, green and brown color separately.Table 1 lists which sites and data are analyzed in these three IOPs.
Table 1 lists which sites and data are analysed in these three IOPs.

Table 2 .
Frequency of multiple tropopause (MT) during the three observation periods.

Table 3 .
Average height of LRT during the three observation periods.
X means no observation data.
Table 1 lists which sites and data are analyzed in these three IOPs.
Table 1 lists which sites and data are analysed in these three IOPs.
Randel et al. (2007) Plateau) to stratospheric intrusions, which were generally associated with prevailing upper-level jet streams.The satellite and reanalysis data have furthered our knowledge about global and climatic characteristics of multiple tropopause (MT).The temporal coverage of our radiosonde dataset is limited and the analysis ofRandel et al. (2007)andAñel et al. (2008)represent a climatology on a larger time period including a larger set of data.The vertical and horizontal resolution of our data and dataset used inRandel et al. (2007)andAñel et al. ( • N, www.atmos-chem-phys.net/11/5113/2011/Atmos.Chem.Phys., 11, 5113-5122, 2011 100.90 •