The observational data of the vertical temperature distribution and column ozone, obtained from 10 main stations in the Northern Hemisphere, are analyzed in order to explore the tropopause variations in conjunction with the dynamical variability in column ozone. From the analysis presented, it is evident that the summer distribution of the frequency of occurrence of the tropopause over Greece, apart from its main maximum (around 12 km), is also characterized by a secondary one around 16 km. It is proposed that this elevated maximum possibly originates from the height variation of the tropopause from 12 to 16 km depending on whether the Athens station is located below the cyclonic shear side or below the anticyclonic shear side of the subtropical jet stream. It is also suggested that the transport in the upper troposphere and lower stratosphere that originated in the equatorial region forces the appearance of the multiple tropopauses above Greece. Furthermore, the observational analysis of the vertical ozone distribution above Greece shows that the upward movement of the ozone profile is accompanied by an increase in the annually averaged tropopause height, which leads to an excessive column ozone trend around 0.5%–1.0% decade−1. Additionally, the linear regression analyses of the deseasonalized monthly mean column ozone and tropopause height indicate that the tropopause variations might be responsible for about a quarter of the observed total ozone content (TOC) trend over Greece, the same magnitude of midlatitude ozone depletion that 2D dynamical and chemical models cannot reproduce. This part of the trend is only due to the variations in the upper-troposphere/lower-stratosphere region and not attributable to all dynamical changes and forcing on TOC. Finally, the inverse relationship between column ozone and tropopause height at various geographical sites shows a longitudinal and latitudinal variability, with the strongest signal observed in the eastern midlatitudes of the Northern Hemisphere. At these geographical sites, changes in both the column ozone and lower-stratospheric temperature are roughly 10 Dobson unit (DU) K−1.
Nowadays, great interest in the unprecedented events of the major, sudden stratospheric warming and ozone hole split over Antarctica on 25 September 2002 motivates a necessity to reexamine the current understanding on the dynamics, chemistry, and climate impacts that are associated with tropospheric–stratospheric processes (Varotsos 2002, 2003, 2004).
It has long been known that the short-term fluctuations in total ozone content (TOC) originate from the synoptic-scale baroclinic disturbances in the upper troposphere and lowermost stratosphere (below about 20 km). These disturbances affect ozone through the depth of the lowermost stratosphere (Spänkuch and Schulz 1995; Ebel et al. 1997; Chandra et al. 2003; Singh et al. 2002). Therefore, to some extent, TOC is a tracer for large-scale meteorological processes in the upper troposphere and lowermost stratosphere (Holton et al. 1995). In particular, TOC tends to be highest on the cyclonic side of upper-level jet streams and in the region of isolated cyclonic vortices (cut-off lows).
The observed connection between TOC and tropopause height (TH) is valid not only on short time scales but also on long time scales. Longer-term changes are associated with planetary-scale circulation anomalies, such as the Northern Hemisphere annular mode (NAM), which elevate the tropopause at certain longitudes and lower it at others. However, the effect of tropospheric circulation on TOC, and its relation to TH, is not well understood at present (Thuburn and Craigh 1997; Krzyscin et al. 2001; WMO 2003).
Changes in TH are related to changes in tropospheric and lower-stratospheric temperatures (Hoinka 1998). An increase in the former and a decrease in the latter—as expected, for example, with increases in CO2—both lead to a higher tropopause. In each case, a 1°C change leads to an increase in the TH by about 160 m. The observed lower-stratospheric cooling attributable to ozone decreases would be expected to raise the TH on radiative grounds alone (an effect that has not been quantified yet). In contrast, sulfate aerosols are calculated to cool the troposphere and therefore lower the tropopause. Other natural and anthropogenic forcings will also act to change TH (Hoskins 2003). For instance, the recent analysis of Santer et al. (2003) suggests that over the past 20 yr, the observed globally averaged TH has increased by ∼200 m.
Notwithstanding these concerns, several studies have attempted to estimate the dynamical contribution to TOC trends by statistical regression against TH or tropospheric circulation indices that are correlated with it. There appears to be a consensus between the different statistical studies that for decadal time scales, a significant fraction (20%–40%) of the observed NH midlatitude TOC change is associated with changes in tropospheric circulation. Thus, it appears that although the ozone decreases may have led to changes in tropospheric circulation (including TH) that are qualitatively consistent with the observed changes, they are of insufficient magnitude, implying that other factors must contribute. Therefore, while there is a certain self-consistency between the various dynamical trends, without a clear mechanism connecting the NAM, TOC, and TH, any such connection in the context of long-term trends remains speculative. This question will probably not be answerable until the origin of the changes in the individual dynamical quantities has been identified (WMO 2003). Furthermore, the interrelationships between ozone and circulation parameters in the lowermost stratosphere are poorly quantified at present, and estimates of circulation effects on decadal changes in ozone are highly uncertain.
This paper is an attempt to investigate the problems described above by analyzing the observational data of TOC and TH at 10 stations distributed in the NH. Special attention is given to the area of Greece and in particular to Athens, Greece (37.9°N, 23.7°E), because this location has been suggested in the recent literature as representative of the TOC variations, not only of the Mediterranean region, but also, of the entire midlatitude zone of the Northern Hemisphere (Chandra et al. 1996; Varotsos et al. 2000; Efstathiou et al. 2003). This subject of the possibility of using TH variations as an indicator of dynamical variability in TOC is very important because of the widely discussed problem of the time required for ozone recovery (Hoskins 2003; Santer et al. 2003; Varotsos 2002, 2003, 2004). At the beginning, the characteristic features of the first two thermal tropopauses are examined alongside the column ozone over Greece. Consequently, this examination is expanded to the other geographical sites in order to explore the spatial distribution of the relationship between the column ozone and the tropopause properties.
The data used in this study are daily vertical temperature profiles and daily TOC observations, obtained at Athens for the period January 1984–March 2002. The profiles of temperature have been measured by high-resolution radiosonde ascents at Athens (Varotsos et al. 1994; Retalis et al. 1997). A portion of them has been carried out from the location of the Athens University coupled to ozonesondes and the rest from the site of the Hellenic Meteorological Service (the distance is 10 km from Athens University approximately). In most of the cases, radiosondes were launched twice a day (0000 and 1200 UTC). For all soundings, the thermal tropopause was defined according to the standard World Meteorological Organization (WMO) criterion (UNEP/WMO 1957) as the lowest level in the free troposphere where the temperature lapse rate becomes smaller than 2 K km−1 for a layer of at least 2-km thickness. The recorded THs when the radiosonde station was below the vertical tropopause break region of the subtropical jet core were filtered out because this WMO criterion is not effective in these cases.
Additionally, the available archives of radiosoundings performed at Thessaloniki, Greece (40.4°N, 23.0°E), and Heraklion, Greece (35.2°N, 25.7°E), were also used in order to occasionally confirm the obtained results from the Athens observations. Due to the fact that the total number of radiosonde ascents over Thessaloniki and Heraklion is only 3305 and 2481, respectively, the figures shown in this paper come from Athens data.
As far as the daily TOC observations are concerned, for the period January 1984–September 1991, these were obtained from the Total Ozone Mapping Spectrophotometer (TOMS), flown on the satellite Nimbus-7. TOC measurements for the period October 1991–March 2002 were obtained from the ground-based spectrophotometer Dobson No. 118, operating at Athens University since 1989 (Varotsos et al. 2000). TOC is expressed in terms of the equivalent thickness of the ozone layer at standard temperature and pressure; a typical global mean value of 3 mm, or 300 Dobson units (DU; 1 DU = 10−5 m). In this case, where the local TOC observation was not available in the second period, data from TOMS were used. The compatibility of TOMS and Dobson spectrophotometers over Athens has been taken into account by considering that the mean and standard deviation of the monthly percentage difference between Dobson and TOMS amounts to −0.91% ± 1.41% (Varotsos et al. 2000).
Furthermore, the ozone vertical profiles have been measured by ozonesondes at Athens University during the period December 1991–April 2001. The ozonesondes flown on free balloons were electrochemical concentration cells (ECC; Scientific Pump) and were well calibrated at the ground before launch with the use of the Ozonesonde Calibration Unit of Vaisala (Varotsos et al. 1994, 2001).
Finally, observations of TOC and tropopause properties obtained at various sites with continuous records of data were also used in order to identify spatial discrepancies in the long-term relationship between the two variables. The list of these stations with their geographical coordinates is shown in Table 1.
3. Occurrence frequency of tropopause height
a. The first thermal tropopause
Radiosonde profiles obtained at Athens, Greece, were grouped in classes according to the TH by averaging all ascents with THs between 7.5 and 8.5 km, 8.5 and 9.5 km, and so on. In order to avoid interference from photochemical changes (which follow the annual cycle induced from solar radiation), averaging was performed separately for periods around summer [May–June–July (MJJ)] and around winter [November–December–January (NDJ)]. During the period 1984–2002, 12 844 radiosondes were launched. From them, 3188 fall in the summer (MJJ) period and 3170 in the winter (NDJ) period. The histograms of Fig. 1 illustrate the occurrence frequency of various THs for the MJJ and NDJ periods. Occurrence frequency was considered as the ratio of the number of cases in a given height class to the total number of soundings in that period. As it is shown in Fig. 1, the NDJ distribution appears to have a single maximum around 11.3 km, while the MJJ distribution appears to have one main maximum around 11.8 km and a secondary one around 16.3 km (values are calculated from the original dataset). It is worth noting here that Steinbrecht et al. (1998), applying similar analysis on Hohenpeissenberg, Germany (47.8°N, 11.0°E), radiosonde data, found no secondary maximum in the MJJ distribution (a potential reason is given later on in this section).
Another noteworthy detail in Fig. 1 is that the NDJ distribution is wider than the main distribution for the MJJ period, which is slightly shifted toward higher altitudes compared to that of NDJ (the median for the main maximum of the MJJ distribution is 12.8 km, while for the NDJ distribution it is 11.6 km). This shift is much smaller than the width of the NDJ and MJJ distributions. It should be noted that the distributions shown in Fig. 1 are not symmetric and they do not change if radiosoundings performed at 0000 and 1200 UTC are plotted separately.
To further explore the secondary maximum in the MJJ distribution, it is interesting to compare the tropopause data stored in the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis database (model data) with those observed over Athens. Therefore, the above-mentioned analysis is also repeated by using NCEP–NCAR reanalysis data provided by the National Oceanic and Atmospheric Administration–Cooperative Institute for Research in Environmental Science (NOAA–CIRES) Climate Diagnostics Center (available online at http://www.cdc.noaa.gov/;).
Figure 2 illustrates the temperature deviations and trends of the tropopause above Athens as it is derived from in situ and reanalysis data. In other words, it is a comparison of the “real” tropopause observed at Athens with that from the reanalysis database. From Fig. 2 it is evident that there exists a very good agreement between the reanalysis and in situ data. This is probably due to the fact that in the case of temperature, the reanalysis values are strongly influenced by observations (radiosondes and satellite) rather than the model.
Nevertheless, the histogram illustrating the occurrence frequency of various THs for the MJJ period, derived from reanalysis data (not shown), does not demonstrate the secondary maximum depicted in Fig. 1. Probably, this fact denotes that the secondary maximum in MJJ distribution reflects day-to-day variations of the local weather conditions prevailing at Athens that cannot be reproduced by the NCEP–NCAR reanalysis database (see below).
Perhaps the secondary maximum in summer is evidence of the tropical air coming over Athens. In fact, as it is shown in Fig. 3, the TOC values for these cases appear to average a little over 300 DU, which is a little higher (30 DU) than would be expected for tropical air (due to convection). In this regard, a further investigation of the day-to-day variations of the TH was also attempted by considering the relative location of the Athens station with respect to the subtropical jet axis, which normally migrates at about 35°–40°N in summer at the longitude of Greece. The latter is retrieved, constructing two groups of the summer THs (low and high pressure), by utilizing the relevant meteorological charts provided by the European Weather Bulletin. An inspection of the two groups shows that the TH is approximately 12 km when the Athens site is located below the cyclonic shear side of the subtropical jet, while it reaches approximately 16 km when the Athens site is located below the anticyclonic shear side of the subtropical jet. The fact that the subtropical jet does not migrate up to the Hohenpeissenberg location might be a plausible explanation for the above-mentioned absence of the secondary maximum over that site.
b. The second thermal tropopause
The aforementioned radiosonde ascents performed at Athens are subsequently examined by focusing on the appearance of the second thermal tropopause. Figure 4 illustrates the seasonal variability of the first and second thermal tropopause above Athens. The interesting point in Fig. 4 is that the seasonal variability of the second thermal tropopause does not follow the well-known pattern of the annual cycle of the first thermal tropopause, denoting that their origin is probably different.
Figure 5 depicts the frequency of occurrence of the second thermal tropopause, demonstrating that the second thermal tropopause is more frequent during the recent time period. An investigation of the seasonal variability in the frequency of occurrence of the second thermal tropopause showed that it becomes more frequent as time goes on, in all seasons.
Figure 6 illustrates the temporal march of temperature of both thermal tropopauses observed at Athens. The most interesting feature in Fig. 6 is that temperatures of both tropopauses change approximately in phase, following a statistically significant negative trend (2σ), which equivalently means that the height of both tropopauses is increasing (with a higher rate for the second tropopause). It would be worthwhile to mention that the investigation of the validity of the latter on a seasonal basis showed that the negative trend is consistently observed in all seasons. It is also worth noting that the contribution to the TOC of the ozone content in the 12– 16-km layer is nonnegligible.
As mentioned in the previous section, it seems that the variations of the subtropical jet stream are related to the appearance of one single elevated tropopause, and the transport in the upper troposphere and lower stratosphere that originated in the equatorial region forces the appearance of the multiple tropopause.
In addition, Figs. 5 and 6 together reveal that the second thermal tropopause over Greece appears more frequently during recent years, reaching progressively higher altitudes and hence lower temperatures. This increase in TH may primarily be a result of both stratospheric cooling (due to the stratospheric ozone depletion) and tropospheric warming (due to the enhancement of the atmospheric greenhouse effect). However, the relative importance of these two factors is still uncertain. Santer et al. (2003) emphasized that anthropogenically driven tropospheric warming is an important factor in explaining modeled changes in TH. In particular, they suggested that over the period 1979–99, roughly 30% of the increase in TH is explained by the warming of the troposphere, which is induced by greenhouse gases (GHGs). In parallel, some natural forcings, like the large explosive volcanic eruptions, warm the lower stratosphere and cool the troposphere, leading to a decrease of the TH (Kondratyev and Varotsos 2000; Hoskins 2003).
Finally, the results mentioned above were also confirmed by performing similar analyses to the radiosonde observations obtained over Thessaloniki and Heraklion. Therefore, the concluding remarks discussed above concern the entire area of Greece.
4. Association between tropopause properties and TOC
As it was mentioned in the introduction, the casuality of the observed relationship between TOC and tropopause properties is not well understood. In the following, this long-term relationship is investigated by first studying the Athens observations (as a representative site for the TOC variations of the entire midlatitude zone of the Northern Hemisphere) and then by studying the observations over other geographical sites.
a. The Athens case
Significant differences are observed between the values of TOC in the MJJ and NDJ periods over Athens. In the MJJ period, TOC for the 8.5–9.5-km class is 370 DU, but only 334 DU for the 11.5–12.5-km class. In the NDJ period, TOC is 335 DU for the 8.5–9.5-km class but only 298 DU for the 11.5–12.5-km class. In both periods, the change in TH is accompanied by a decrease in TOC values of about 35 DU.
The inverse relationship between TOC and TH is presented in Fig. 7. Each data point on this graph illustrates average TH and TOC for one TH class. As can be clearly seen in Fig. 7, any TOC decrease is always accompanied by an increase in TH, with a rate of −8.5 DU km−1 in summer, and −11.2 DU km−1 in winter. The correlation coefficient between TOC and TH is −0.82 for the MJJ period and −0.66 for the NDJ period (both statistically significant at the 99% confidence level). In addition, for the same TH, TOC is higher in MJJ than that in NDJ due to transport mechanisms related to seasonal changes in the Brewer–Dobson circulations.
Similar results to those mentioned above were obtained at Thessaloniki and Heraklion. In conclusion, a vertical displacement of the tropopause over Greece by 1 km is associated with an ozone anomaly of 10 DU, approximately. In this respect, Hoinka et al. (1996) reported that a vertical displacement of the tropopause over Germany by 1 km is associated with an ozone anomaly of 13 DU.
In the following, the annual variation of ozone in the troposphere and lower-stratosphere region has been studied using all the available ozonesonde data collected between 1991 and 2001 at the Athens station.
The climatology of ozone deduced from the Athens ozonesonde data exhibits a prominent annual cycle in the upper-troposphere and lower-stratosphere region (Fig. 8). The observed change in the phase of the annual cycle from late spring–summer at 500 hPa to spring at 200 hPa and to late winter–spring at 100 hPa shows the change of the ozone control from photochemical to dynamical. The latter is in close agreement with the relevant findings of Rao et al. (2003), which were deduced from ozonesonde data collected between 1994 and 2001 at several northern European stations. In addition, Fig. 8 shows that an upward movement of the ozone profile is accompanied by an increase in the annually averaged TH. This observed shift in ozone profile in the lower stratosphere leads to a reduction of total ozone. This part of the TOC trend (0.5%–1.0% decade−1) is clearly associated with the upward movement of the ozone profile, and thus it should not be considered as a direct result of anthropogenic chemical ozone depletion. Krzyscin et al. (1998) and Forster and Tourpali (2001) have come to similar conclusions by performing observational analysis at locations in the Northern Hemisphere since 1979.
Furthermore, an alternative way to understand the dynamics of ozone in the upper-troposphere and lower-stratosphere region is to investigate the relative position of the first tropopause and ozonepause. In this respect, Fig. 8 shows that in winter the ozonepause lies below the first thermal tropopause (ozone as it penetrates into the tropopause downward) and the second thermal tropopause (shown in Fig. 4, near 100 hPa). The latter is consistent with the generally approved ozonepause pattern, according to which in winter, in polar and moderate latitudes, the ozonepause most frequently lies below the tropopause, and in equatorial latitudes, in both the summer and winter, the ozonepause generally lies above the tropopause.
From the observational results mentioned above, it appears that the most efficient mechanism to modify the TOC amounts is the vertical displacement of the tropopause. This is in agreement with the general belief that the tropospheric circulation affects ozone distribution, with the direct effects confined to the lowermost stratosphere (Canziani et al. 2002; WMO 2003). It is actually an influence of the lowermost stratosphere from below, which is also sensitive to the Brewer–Dobson circulation from above. In this respect, Krzyscin (2002) suggested that during minihole events, a 60% ozone reduction in the lower stratosphere and an approximately 50-hPa upward shift of the thermal tropopause is observed (James et al. 1997). Therefore, increases in both TH and the frequency of minihole events induce decreases in ozone over the northern midlatitudes. However, these induced decreases in ozone must not be regarded as addative, because the two causal events are not independent between them (WMO 2003).
b. The other cases
The same data analysis performed at the Athens station was also repeated for the Hohenpeissenberg station during 1967–2002 and showed that any TOC decrease is always accompanied by an increase in TH, at a rate of −11.6 DU km−1 in summer and −18.9 DU km−1 in winter. The correlation coefficient between TOC and TH is −0.72 for the MJJ period and −0.63 for the NDJ period (95% confidence level). The analysis applied on the radiosonde data obtained at Belsk, Poland (51.8°N, 20.8°E), showed a correlation coefficient between TOC and TH of −0.67 for the MJJ period and −0.60 for the NDJ period (both significant at the 95% confidence level).
Analysis similar to that described above was also made for the case of Cairo, Egypt (30.0°N, 31.3°E), during 1968–2002, where the correlation coefficient between TOC and tropopause temperature was found to be −0.43 (statistically significant at the 95% confidence level). Applying the same analysis for Sapporo, Japan (43.0°N, 141.3°E), during 1976–2002, it was found that this correlation coefficient becomes −0.70 (statistically significant at the 99% confidence level). Finally, the case of Bismarck, North Dakota (46.7°N, 100.8°W), during 1962–2002 gave a correlation coefficient of −0.3. It should be noted that some other stations in the United States (Boulder, Colorado, and Wallops Island, Virginia) also gave statistically insignificant correlation coefficients, like that of Bismarck.
From the puzzle of the aforementioned correlation coefficients between TOC and TH at various geographical sites, it is clearly evident that there exists a longitudinal and latitudinal variability. The most characteristic feature is that the strongest signal of the inverse relationship between TOC and TH appears to be present in the eastern midlatitudes of the Northern Hemisphere.
To verify, however, that there really are such large longitudinal and latitudinal differences in the ozone and temperature fields in the upper-troposphere and lower-stratosphere region, the NCEP–NCAR reanalysis tropopause and 50-hPa data for the aforementioned locations were examined along with the corresponding TOC time series deduced from TOMS observations. The analysis based on TOMS and reanalysis tropopause data showed that the aforementioned large latitudinal/longitudinal differences in the ozone/tropopause behavior are more or less real. Therefore, this analysis confirms the above-mentioned observational result that the strongest anticorrelation between TOC and TH occurs in the eastern midlatitudes of the Northern Hemisphere. However, the analysis based on TOC and 50-hPa temperature at the aforementioned locations showed that a more precise relationship between these two variables holds up; notably, a 10-DU change in TOC corresponds to a roughly 1-K change of 50-hPa temperature. This relation was discussed in many papers (e.g., Reinsel et al. 1981), and it is very important in the quantification of natural variations in total ozone.
5. Trends in TOC and tropopause properties
a. The case of Athens
To detect the dynamical contribution to the observed TOC decrease above Athens, the association between long-term trends in TOC and TH is consequently examined.
To subtract the annual variation from both time series in order to get the deseasonalized ones, a smoothing of a 13-month running mean has been applied on the raw data. Figure 9 illustrates the deseasonalized monthly means of TOC and TH, as well as their corresponding linear trends at Athens, during the period 1984–2002. As expected, high TOC is associated with a low tropopause, and vice versa. The correlation coefficient between single monthly means of TOC and the tropopause was found to be −0.5, which is significant at the 99% confidence level. The linear regression analysis of the deseasonalized monthly mean TOC indicates a long-term decrease of −7.54 ± 0.99 DU decade−1 (2σ).
It should be clarified at this point that the estimated uncertainty denotes the 95% confidence limits of the slope obtained by standard linear regression analysis. The true uncertainty is most likely higher than estimated in the present analysis due to the autocorrelation in the data. However, even increasing the level of autocorrelation in the time series (being a result of the smoothing), thus reducing the degrees of freedom in the linear regression test by a factor of 5, only increases the estimated uncertainty by 3% (Steinbrecht et al. 1998).
Since TOC is highly variable by various oscillations like the quasi-biennial oscillation and the North Atlantic Oscillation (Appenzeller et al. 2000), it would be worthwhile to mention that the application of a more sophisticated trend analysis, which includes effects of the Pinatubo Volcano, the solar cycle, the quasi-biennial oscillation, etc., on TOC over Athens, gives approximately the same results for the major long-term changes as does the linear regression analysis (Callis et al. 1997; Varotsos et al. 2000; Varotsos 2004). The latter was also observed for the Hohenpeissenberg time series (Steinbrecht et al. 1998). There is no doubt, however, that the use of more sophisticated trend models should result in lower error in the trend estimates.
Another interesting point in Fig. 9 is the long-term increase by about 167 ± 30 m decade−1 (2σ) of the TH at Athens for the period under consideration. However, if the −9.7 DU km−1 correlation (linear fitting) between TOC and TH (considering both summer and winter periods), is representative on a long time scale (shown in Fig. 7), then the observed 167 m decade−1 increase in TH (shown in Fig. 9) should be connected to a TOC decrease of 1.64 DU decade−1. Thus, the observed increase in TH could explain about 22% of the observed decrease in TOC (−7.5 DU decade−1) shown in Fig. 9.
It would also be worthwhile to mention that the 22% decrease is of the same magnitude of the midlatitude ozone depletion as the quarter of the ozone decrease that relevant models have difficulty accounting for (Steinbrecht et al. 1998). For example, 2D model estimations by Jackman et al. (1996) give TOC trends of −3.5% decade−1 for the northern midlatitudes in spring, whereas observed trends are −5% decade−1 (WMO 1995). Therefore, about 22% of the trend in TOC over Greece during 1984–2002 could be attributed to TH changes. It should be noted, however, that this part of the trend is only due to the variations in the upper-troposphere–lower-stratosphere region and is not attributable to all dynamical changes and forcing on TOC. The results mentioned above were also confirmed by using the observations obtained at Thessaloniki and Heraklion.
Because of the given strong correlation between the various meteorological indices, a similar outcome to this latter result is obtained by using NH annular mode–like or 330-K potential vorticity and meridional wind shear (from NCEP–NCAR reanalyses) dynamical proxies in statistical trend analyses of TOC during the winter/ spring period (see also Steinbrecht et al. 2001; WMO 2003).
For comparison reasons, the deseasonalized monthly means of TOC and tropopause temperature (that are calculated by using the NCEP–NCAR reanalysis data), as well as their corresponding linear trends during the period 1984–2002 at Athens are presented in Fig. 10. It is clear that the depicted trends in Fig. 10 (deduced from reanalysis data) are in close agreement with those shown in Fig. 9 (deduced from real observations).
The deseasonalized monthly means of 500-mb temperature and TOC over Athens during the period 1979– 2002 show a correlation coefficient of −0.52 (significant at the 99% confidence level) and a regression coefficient of −2.7 ± 0.4 DU K−1 (2σ). In addition, the linear regression gives an increase for the 500-mb temperature over Athens of +0.27 ± 0.04 K decade−1 (2σ). This result is in qualitative agreement with the trend reported for tropospheric zonal mean temperatures at 40°–50°N by Labitzke and van Loon (1995; ≈0.2 K decade−1 annual mean; ≈0.4 K decade−1 in winter). The latter, with the information obtained from Figs. 7 and 9, supports the consideration that the changing TH plays a crucial role in the long-term TOC decrease. The increasing TH and the decreasing tropopause temperature are qualitatively associated with the decreasing trend of ozone in the stratosphere. It should be noted that the midtropospheric temperature is used as a proxy for the temperature at the earth's surface. Also, GHGs warm the earth's surface but cool the stratosphere radiatively and therefore affect ozone depletion. In addition, due to the ozone depletion, less heat would be absorbed in the stratosphere while, due to the increase in GHGs content, more heat would be trapped in the atmosphere, and thus these effects may change the thermal structure of the tropopause region.
Furthermore, tropospheric warming pushes the tropopause up, and then the excess ozone into the stratosphere is destroyed photochemically. The latter confirms that a close connection between the increase of TH, tropospheric warming, and TOC depletion exists (Kondratyev 1998; Steinbrecht et al. 1998; Varotsos 2004).
b. The other cases
It is necessary to study radiosonde observations obtained at other geographical sites in order to elucidate any differences between TH and TOC long-term trends above those sites. Therefore, similar analysis to that of Athens (shown in Fig. 9) is attempted on the observations obtained at Hohenpeissenberg during 1967–2002 and at Belsk during 1974–96. For the Polish location, TOC data were obtained over Belsk, while the radiosondings were collected from the nearby aerological station Legionowo, Poland (WMO station number 12374). The results are presented in Figs. 11 and 12, respectively, where it is shown that the long-term increase in TH over Hohenpeissenberg and Belsk is of the same order of magnitude as that observed at Athens (Fig. 9).
The relevant analysis of radiosonde observations obtained during 1966–2000 at New Delhi, India (28°N, 77°E), and during 1973–98 at Thiruvananthapuram, India (8°N, 76°E), showed that the long-term trend of TH is increasing in the range of +0.6 to +1.0% decade−1, while the long-term trend of the tropopause temperature is decreasing in the range of −0.5 to −0.9% decade−1. These results confirm the earlier findings of Chakrabarty et al. (2001).
It should be emphasized that the results mentioned above could be combined with the recent findings of the tropopause properties at high latitudes of the Northern Hemisphere. For instance, according to Highwood et al. (2000) the multidecade radiosonde dataset (1965– 90) over the Arctic region shows that the wintertime tropopause pressure has decreased by approximately 14 mb decade−1, while that in other seasons has decreased by around 5 mb decade−1. The tropopause temperature only shows a significant trend during winter, having decreased by 1.6 K decade−1 since 1965. Also, Zou et al. (2002) reported on the ozone soundings on board the Chinese icebreaker Xuelong (at 75°N, 160°W), suggesting that the atmospheric ozone amount experienced a high–low–high variation with a low–high–low tropopause altitude.
c. Total tropopause ozone coupling and climate
The final question is whether the circulation changes might be the result of GHG-induced TH variation and further climate change. On the basis of radiative balance, Santer et al. (2003) showed a mean midlatitude tropopause pressure decrease of 3 hPa decade−1 resulting from changes in GHGs; this corresponds to a midlatitude TH increase of about 120 m decade−1, which is roughly consistent with observed changes that are discussed in the present study. Therefore, if the past dynamical changes actually represent the dynamical response to GHG-induced climate change, then these changes will increase in magnitude in the future (Kondratyev and Varotsos 2000; Santer et al. 2003). This would decrease future ozone levels and delay ozone recovery.
In this context, Gauss et al. (2003) analyzed the results of 11 chemical transport models and used them as input for radiative forcing calculations, considering that radiative forcing due to changes in ozone is expected for the twenty-first century. They addressed future ozone recovery in the lower stratosphere and its impact on radiative forcing by applying two models that calculate both tropospheric and stratospheric changes. Their results suggested an increase in global-mean tropospheric ozone between 11.4 and 20.5 DU for the twenty-first century, which corresponds to a positive radiative forcing ranging from 0.40 to 0.78 W m−2 on a global and annual average. They also found that the lower stratosphere contributes an additional 7.5–9.3 DU to the calculated increase in the ozone column, increasing radiative forcing by 0.15–0.17 W m−2. As to the ability to scale radiative forcing to global-mean ozone column change, they suggested that despite the large variations between the 11 participating models, the calculated range for normalized radiative forcing is within 25%.
From the discussion above, the following results can be drawn:
The long-term variability of TH over Athens shows that the winter distribution of the frequency of occurrence of the tropopause appears to have a single maximum around 11.3 km, while the corresponding summer distribution appears to have, apart from the main maximum (around 11.8 km), another one around 16.3 km. It is shown that the secondary (elevated) maximum originates by the following mechanism: when the Athens site is located below the cyclonic shear side of the subtropical jet, then the TH is approximately 12 km, while it reaches to around 16 km when the Athens site is located below the anticyclonic shear side of the subtropical jet.
The observed second thermal tropopause over Greece is characterized by a temporal increase in the frequency of occurrence in all seasons. It seems that the transport in the upper troposphere and lower stratosphere that originated in the equatorial region forces the appearance of the multiple tropopauses.
The temporal evolution of the temperatures of both thermal tropopauses in all seasons shows that they change approximately in phase, following a statistically significant negative trend. This equivalently means that the heights of both thermal tropopauses in all seasons are significantly increasing (with a higher rate for the second tropopause). This increase in TH may primarily be attributed to natural and anthropogenic forcings that result in both stratospheric cooling (due to the stratospheric ozone depletion) and tropospheric warming (due to the enhancement of the atmospheric greenhouse effect).
A vertical displacement of the tropopause over Greece by 1 km is associated with an ozone anomaly of 10 Dobson units. The observational analysis of the vertical ozone distribution, as it is derived from ozonesonde ascents performed at Athens during 1991–2001, shows that there has been an upward movement of the ozone profile accompanied by an increase in the annually averaged TH. This observed shift of the ozone profile in the lower stratosphere has led to an excessive ozone depletion that results in a TOC trend around 0.5%–1.0% decade−1, which should not be considered as a direct result of anthropogenic chemical ozone depletion.
The inverse relationship between TOC and TH at various geographical sites shows a longitudinal and latitudinal variability with the strongest signal in the eastern midlatitudes of the Northern Hemisphere.
The linear regression analyses of the deseasonalized monthly mean TOC and TH at Athens for the period 1984–2002 indicate that about 22% of the trend in TOC over Greece could be attributed to TH changes, or in other words, tropopause variations might be responsible for about a quarter of the observed TOC trend over Greece. This part of the trend is only due to the variations in the upper-troposphere/lower-stratosphere region and not attributable to all dynamical changes and forcing on TOC.
The observed changes in TH are consistent with the recently derived modeled ones, which represent the dynamical response to GHG-induced climate change, and would therefore decrease future ozone levels and delay ozone recovery.
The authors would like to express their gratitude to the reviewers for their fruitful comments and suggestions that much improved this work. Also, the provision of radiosonde data from Dr. Janusz W. Krzyscin (Institute of Geophysics, Polish Academy of Sciences) and Mr. Ulf Kohler (Meteorological Observatory Hohenpeissenberg) is greatly appreciated.
Corresponding author address: Dr. Costas Varotsos, Dept. of Applied Physics, University of Athens, Bldg. Phys. 5, Panepistimiopolis, GR-157 84 Athens, Greece. Email: email@example.com