Oxygen anomaly in near surface carbon dioxide reveals deep stratospheric intrusion.

Stratosphere-troposphere exchange could be enhanced by tropopause folding, linked to variability in the subtropical jet stream. Relevant to tropospheric biogeochemistry is irreversible transport from the stratosphere, associated with deep intrusions. Here, oxygen anomalies in near surface air CO2 are used to study the irreversible transport from the stratosphere, where the triple oxygen isotopes of CO2 are distinct from those originating from the Earth's surface. We show that the oxygen anomaly in CO2 is observable at sea level and the magnitude of the signal increases during the course of our sampling period (September 2013-February 2014), concordant with the strengthening of the subtropical jet system and the East Asia winter monsoon. The trend of the anomaly is found to be 0.1‰/month (R(2) = 0.6) during the jet development period in October. Implications for utilizing the oxygen anomaly in CO2 for CO2 biogeochemical cycle study and stratospheric intrusion flux at the surface are discussed.

Transport of mass and chemical species via the large-scale Brewer-Dobson circulation and synoptic/small-scale mixing from the stratosphere to the troposphere and vice versa1234567 has a significant impact on the oxidation capacity of the troposphere8910 and the radiation budget in the stratosphere11. In the Pacific and Atlantic regions, the stratosphere-troposphere exchange occurs predominantly over storm tracks during winter, spring, and fall1213141516171819; Taiwan is located in such a region. In summer the exchange maximizes its amplitude over the central Asia, supported by various observations made at the Waliguan Observatory (36°17´ N, 100°54´ E, 3816 m, China; NOAA ESRL code: WLG) located on the Tibetan Plateau20212223242526. In the mid-latitudes of East Asia, ozonesonde and surface observations show a distinct spring maximum (e.g., see ref. 27 and references contained therein). This finding is consistent with the fact that the stratospheric intrusions bringing O3 into the troposphere reach their seasonal maximum in the summer2728. This is primarily due to unusually strong winds associated with the polar and subtropical jet streams over the east Asian coast and the persistence of cyclogenesis over the western Pacific, resulting in frequent tropopause folding and thus significant intrusion of O3 into the troposphere102529303132. The air remaining in the fold of the upper troposphere moves towards the jet exit regions and re-enters the lower stratosphere. However, stretching stratospheric intrusions to smaller scales, some of which go deeper into the troposphere, leads to irreversible transport1103334353637. Such deep intrusions of stratospheric air down to the lower troposphere or even to the surface are relevant to tropospheric chemistry8910. Ozone, however, is reactive and with numerous sources at the ground level, which makes the use of ozone for the irreversible branch rather ambiguous. In this work, we analyze the so-called “oxygen anomaly” in CO2, a species that has distinct anomaly originating in the middle atmosphere, to study how the changing meteorology affects the cross-tropopause transport. The anomaly in CO2 is potentially also a powerful tracer for improving our understanding of the carbon cycle38. One advantage of using CO2 is that this species is inert in the free troposphere; the isotopic composition of CO2 can be modified at the surface and in the middle and upper atmospheres only.

In typical biogenic/atmospheric processes, the partitioning between species’ oxygen-containing isotopologues follows a mass-dependent line, i.e.,

The factor λ is taken to be 0.516 (λ0) and may vary between 0.500 and 0.529 (ref. 39). The λ0 is chosen following the fractionation that occurs in transpiration at the globally averaged relative humidity of 75% (ref. 40). It has been discovered that some atmospheric species follow a very different relation. For example, δ17O(O3) ≈ δ18O(O3) (ref. 41, 42, 43) and δ17O(CO2) ≈ 1.7 × δ18O(CO2) in the stratosphere (reference to tropospheric CO2 ; ref. 44, 45, 46, 47, 48). The oxygen isotope distribution in CO2 is largely affected by O2-O3-CO2 photochemistry in the middle atmosphere, via the reaction O(1D) + CO2 where O(1D) is formed by dissociation of O3 (ref. 45,46,48, 49, 50). As O3 and CO2 are strongly coupled in the stratosphere and isotopically anomalous CO2 can be produced in the middle atmosphere only, one may obtain a better constraint for stratospheric O3 at the surface by measuring the isotopic composition of CO2. Symbol Δ is frequently used to quantify the deviation from the mass-dependent fractionation line, and is defined by

where δ-values are expressed relative to V-SMOW.

Methods

CO2-O2 oxygen isotope exchange method developed previously51 was followed with slight modification (see Figure S1) to measure the Δ17O of CO2 samples. The exchange was carried out in a reaction tube (made of quartz, 60 cm in length and 6.5 mm in diameter) with a cold finger and positioned horizontally inside a cylindrical heater. The heating zone is about 15 cm. Isotopic analyses were done using a FINNIGAN MAT 253 mass spectrometer in dual inlet mode. The analytical precision obtained for Δ17O values of CO2 is 0.008‰ (1-σ standard deviation and hereafter, unless otherwise stated; see Table S1). The precision is also verified by analysis of duplicate samples with difference between duplicates less than 0.01‰. To establish the accuracy of the present method, we follow a typical method52 to convert isotopically known O2 to CO2 and assume the conservation of Δ17O. Good accuracy is demonstrated in Table S2 and Figure S2.

Concentration of CO2 is measured with a LI-COR infrared gas analyzer (model 840 A, LI-COR, USA) at 4 Hz, smoothed with 20-s moving average. The reproducibility is better than 1 ppmv. The analyzer is calibrated against a compressed air cylinder, with calibrated concentration of 387.7 ppmv. This working standard is calibrated using a commercial Picarro analyzer (model G1301, Picarro, USA) by a series of NOAA/GMD certified tertiary standards with CO2 mixing ratios of 369.9, 392.0, 409.2, and 516.3 ppmv. The precision (1-σ) is better than 0.2 ppmv.

Air sampling

Air samples were collected between September 2013 and February 2014 in cleaned pre-conditioned 1-liter pyrex bottles. The cleaning was done by passing of dry high purity nitrogen overnight. Sampling bottles used for concentration measurements (~350-ml bottle) and bottles used for isotope analyses were connected in series. The sampling was carried out at Academia Sinica campus (abbreviated AS; 121°36'51'' E, 25°02'27'' N; ~10 m above the ground level or 60 m above sea level) in Taipei, Taiwan and the campus of National Taiwan University (NTU; 121°32'21'' E, 25°00'53'' N; ~10 m above the ground level or 20 m above sea level; ~10 km southwest of Academia Sinica). Sampling was done after flushing the bottles for 5 minutes by pumping air at a flow rate of ~2 liter per min. Moisture was removed during sampling by using magnesium perchlorate, to minimize subsequent isotope exchange between CO2 and water; the use of magnesium perchlorate reduced moisture content from the ambient value of 70–90% to less than 1% relative humidity, checked using the LI-COR infrared gas analyzer (model 840 A, LI-COR, USA). To get 2-liter equivalent air, we compressed the gas in the bottle to 2 bar. This allows us to get sufficient CO2 for isotope analysis (~30 μmole). In addition to major gases like N2, O2, and Ar, the flask air samples with CO2 also contain traces of water vapor and other gases that could potentially interfere with the CO2 isotope analysis. Water vapor and a few other condensable gases were removed cryogenically while pumping away the major gases using a glass vacuum system with five traps (a slight modification of ref. 53). Two traps were used at dry ice temperature (−77 oC) for removing water and volatile organics while the remaining three were used for CO2 collection at liquid nitrogen temperature (−196 oC). The flow rate was maintained at 100 ml/min during the pumping at a pressure of about 10 to 15 torr. The above process was checked by several control experiments to ensure that there is no escape of CO2 and attendant isotope fractionation.

Results

In general, about 3 samples per day were collected and analyzed, summing up to a total of 81 samples. This is the largest set of data after Thiemens et al.54 decadal record. In this paper, we focus on the changes in monthly scale and the data are averaged diurnally. The diurnally averaging is applied to minimize diurnal variation due to photosynthesis and respiration. Table 1 summarizes the results from the mission. On average, the concentration ([CO2]) is 411.8 ± 9.8 ppmv, δ13C −8.91 ± 0.56‰ (V-PDB), δ18O 40.60 ± 0.52‰ (V-SMOW), and Δ17O 0.329 ± 0.037‰ (1-σ standard deviation to represent the scatter of the data). Identification of the sources responsible for the changes of CO2 level can be done from the so-called Keeling plot (Figure S3). The intercept for δ13C is −27‰, a value that is consistent with respiration from C3 plants (major type of plant in the region), though the signature may not be distinguishable from fossil fuel burning55.

Table 1

Summary of CO2 data collected at the campus of Academia Sinica and National Taiwan University. Values of δ13C and δ18O are referenced to V-PDB and V-SMOW, respectively. The error bar represents 1-σ standard deviation (scatter) of diurnal data. Isobaric interferences of N2O to δ13C and δ18O have been corrected.

Sampling date (number of samples)	[CO2] (ppmv)	δ13C (‰)	δ18O (‰)	Δ17O (‰)
Academia Sinica
2013/09/24 (3)	397.9 ± 9.3	−8.35 ± 0.45	41.04 ± 0.30	0.291 ± 0.033
2013/09/25 (2)	400.5 ± 2.9	−8.50 ± 0.16	40.54 ± 0.59	0.340 ± 0.017
2013/10/07 (3)	421.5 ± 35.9	−9.32 ± 1.53	40.29 ± 0.95	0.244 ± 0.025
2013/10/08 (3)	427.4 ± 4.6	−9.61 ± 0.19	39.97 ± 0.32	0.277 ± 0.028
2013/10/16 (3)	409.4 ± 9.5	−8.89 ± 0.41	40.40 ± 0.15	0.287 ± 0.016
2013/10/17 (5)	409.6 ± 14.4	−8.63 ± 0.67	40.20 ± 0.40	0.321 ± 0.011
2013/10/25 (3)	412.4 ± 2.4	−8.86 ± 0.12	40.33 ± 0.16	0.314 ± 0.021
2013/10/26 (2)	420.6 ± 14.2	−9.24 ± 0.55	40.33 ± 0.10	0.310 ± 0.004
a2013/10/30 (2)	394.6	−7.15 ± 0.27	40.46 ± 0.07	0.392 ± 0.033
2013/10/31 (3)	401.0 ± 4.4	−8.48 ± 0.19	41.14 ± 0.11	0.312 ± 0.028
2013/11/04 (3)	410.6 ± 4.5	−8.81 ± 0.19	40.49 ± 0.12	0.327 ± 0.019
2013/11/09 (2)	415.2 ± 14.2	−8.89 ± 0.78	40.69 ± 0.55	0.345 ± 0.015
2013/11/19 (3)	417.5 ± 2.6	−8.79 ± 0.11	40.54 ± 0.04	0.355 ± 0.017
2013/11/26 (3)	408.1 ± 2.2	−8.87 ± 0.64	40.62 ± 0.31	0.313 ± 0.023
2014/01/27 (2)	402.5 ± 2.3	−8.61 ± 0.08	41.27 ± 0.06	0.387 ± 0.004
2014/02/03 (5)	416.3 ± 16.5	−9.17 ± 0.64	41.15 ± 0.44	0.352 ± 0.035
2014/02/17 (3)	430.5 ± 19.5	−9.65 ± 0.81	40.99 ± 0.57	0.328 ± 0.033
2014/02/19 (2)	421.0 ± 4.2	−9.25 ± 0.19	40.49 ± 0.14	0.351 ± 0.032
2014/02/22 (2)	401.5 ± 0.7	−8.40 ± 0.06	41.48 ± 0.02	0.374 ± 0.036
2014/02/20 (2)	413.9 ± 4.7	−8.91 ± 0.15	40.80 ± 0.20	0.328 ± 0.038
2014/02/24 (1)	406.6	−8.63	41.56	0.397
National Taiwan University
2013/11/14 (3)	394.0 ± 59.3	−8.64 ± 1.13	40.18 ± 1.62	0.362 ± 0.011
2013/11/15 (3)	425.2 ± 11.6	−9.41 ± 0.44	39.01 ± 0.64	0.377 ± 0.039
2013/11/16 (2)	410.3 ± 2.7	−8.74 ± 0.06	40.13 ± 0.16	0.324 ± 0.000
2013/11/24 (3)	413.0 ± 56.3	−9.25 ± 0.56	40.81 ± 0.92	0.281 ± 0.057
2013/11/28 (3)	418.9 ± 3.6	−9.36 ± 0.22	40.16 ± 0.11	0.336 ± 0.056
2013/12/01 (3)	409.9 ± 2.4	−8.89 ± 0.08	40.64 ± 0.07	0.320 ± 0.027
2013/12/07 (3)	407.0 ± 4.0	−8.59 ± 0.05	41.26 ± 0.12	0.325 ± 0.020
a2014/01/07 (2)	N/A	−10.31 ± 0.58	40.31 ± 0.56	0.332 ± 0.001
2014/01/20 (3)	427.5 ± −9.5	−9.53 ± 0.39	40.50 ± 0.21	0.364 ± 0.037

aAt least one concentration measurement is missing and so standard deviation is not available.

Figure 1 shows the three-isotope plot of oxygen in CO2 collected in the region, comparing to that at La Jolla54. Overall, our values agree with Thiemens et al.’s. The linear least-square fitting to our data yields a slope of 0.525 ± 0.013 and the value is 0.503 ± 0.008 for Thiemens et al.’s. Figure 2 compares the oxygen anomaly with the 200-mbar zonal wind, a proxy for the strength of the subtropical jet. Before Oct, the zonal wind is small and fluctuates around zero. As of then, the westerly is established and the Δ17O follows. The average Δ17O values for Sep-Oct (2013), Nov-Dec (2013), and Jan-Feb (2014) are 0.309, 0.325, and 0.357‰, respectively. During the jet development period in Oct, the Δ17O trend is 0.0035‰/day (R2 = 0.59; Fig. 2). Afterwards, further strengthening of the jet does not enhance Δ17O and the trend reduces to 0.0004‰/day (R2 = 0.22), but the short-term enhancement is apparent (see below).

Figure 1

Triple isotope plot of oxygen for CO2.

The recently published tropospheric data41 is also shown for comparison. Values are referenced to V-SMOW.

Figure 2

Time series of Δ17O obtained during the study period.

To show how meteorology and large scale transport affect the Δ17O measured at the surface, surface air temperature (taken from the Central Weather Bureau, Taiwan; station code: 466920) and 200-mbar zonal wind (taken from ECMWF Interim reanalysis) are over plotted. Note that the scale of temperature is reversed, for better comparison with the zonal wind. Linear least-square fits to Δ17O are shown by the dotted lines (see text).

Discussion and summary

Stratospheric intrusions in East Asia occurs in close association with the presence of the subtropical jet stream25262728293031. The jet is situated at ~40 °N in summer and moves southward to Taiwan at ~25 °N in winter. Summer and winter monsoons are two major climate systems responsible for the seasonal changes in Taiwan. The air mass originating from the Asian continent flows through the Pacific Ocean to the island in fall, winter, and spring. Convective activities occurring during the northeast monsoon and accompanied by the passage of mid-latitude cold fronts are largely responsible for the changing meteorology in fall. Cold surges with an abrupt change in temperature are associated with a strong northeasterly wind in winter, followed by cold fronts in spring. Such changing meteorology is also reflected in the subtropical jet stream. The correlation is clearly seen from Fig. 2 that the 200-mbar zonal wind (a proxy for subtropical jet) follows the surface air temperature; in general, the strengthening of winter monsoon (indicated by temperature decrease) is closely associated with the elevated zonal wind speed. This variable meteorology that affects the transport and mixing at all scales can potentially enhance vertical transport into the troposphere56 and sometimes also into the upper troposphere and lower stratosphere, thus leading to an enhancement of cross-tropopause exchange resulting in elevated Δ17O in surface CO2. Below we focus our discussion on the downwelling branch of transport to the lower troposphere.

To support the stratospheric origin of anomalous CO2, we analyze ECMWF Interim O3 data and the results are presented in Fig. 3. We see that the level of O3 at ~200 mbar increases from Sep, 2013 through Feb, 2014. Moreover, stratospheric air moves clearly towards our sampling site (shown by arrows). To further demonstrate the correlation of the intrusion of stratospheric air and the surface CO2 oxygen anomaly, two events in 2014 are selected: Jan 07-27 and Feb 17-24. Δ17O increases with time, concordant with the elevated zonal wind (Fig. 2; with lag of a few days); the Δ17O value changes from 0.332 to 0.387‰ for the former case and from 0.328 to 0.397‰ for the latter. During this time, a large stratospheric intrusion is seen on Jan 22 and Feb 21. The strength of this intrusion is much stronger than that on Jan 07 and Feb 17, respectively (see Fig. 3). The trend of Δ17O is calculated to be 0.0098‰/day (R2 = 0.79) for the latter case and is a factor of ~3 higher than the former (0.0027‰/day; R2 = 0.99) and the trend in Oct (0.0035‰/day; R2 = 0.59).

Figure 3

Profiles of ozone volume mixing ratios (ppbv; taken from ECMWF Interim reanalysis) at altitudes from 50 to 1000 mbar pressure level and latitudes from 0 to 50 °N at longitude 121.5 °E.

Data are either monthly (top six panels) or diurnally (bottom four panels) averaged. Arrows indicate the movement of stratospheric air toward our sampling site shown by the vertical dashed line.

The isotopic composition of CO2 in the atmosphere is an integrated signal of atmospheric and biogeochemical processes. In the atmosphere, the primary mechanism that modifies the isotopic composition of CO2 is the exchange reaction with O(1D) in the stratosphere. The stratospheric source of CO2 is enhanced in δ18O and Δ17O and has a seasonal cycle that is different from that originating from the surface2157. For example, at the Waliguan observatory, biogeochemical models21 predict maximum effects due to respiration in March-April (maximum in δ18O) and August (minimum in δ18O) and due to assimilation in ~March (minimum in δ18O) and July-August (maximum in δ18O), while the Brewer-Dobson circulation has maximum strength in ~March-July36. As a consequence of the interaction between these processes, the maximum in δ18O may occur in June57. This does not mean the seasonal cycle of δ18O is solely caused by the cross-tropopause exchange [cf. ref. 58]. Instead, in addition to natural biogeochemical cycle that results in maximal δ18O in ~April, elevated δ18O from the stratosphere is to modify the seasonal cycle to move the peak from April to June57. The presence of frequent deep intrusions over Tibet was shown recently28. However, to fully resolve the source of summertime O3, tracers like Δ17O that are seriously affected by stratospheric processes are essential. In this work, the size of Δ17O elevated during the subtropical jet strengthening period is up to ~0.1‰ (trend of 0.0035‰/day over one-month in Oct), a value that is expected by bringing air with 1‰ (referenced to the mean anomaly of tropospheric CO2) anomaly4559 from ~100 mbar to 1000 mbar. Attempts to utilize Δ17O for stratospheric and surface flux estimates are made below.

Given that CO2 is chemically inert in the troposphere, assuming steady state, we have

where Fsur and Fstr are the fluxes from the surface and stratosphere, respectively. Δ17Osur and Δ17Ostr are the corresponding oxygen anomalies. Fsur includes the CO2 fluxes associated with photosynthesis, respiration, soil invasion, and oceanic processes. Assuming the globally averaged surface emitted CO2 is in isotopic equilibrium with water at 25 °C, δ18O = 41‰ and Δ17Osur = (0.523–0.516) × ln(1 + δ18O) = 0.281‰, where 0.523 is the equilibrium constant of water and CO2 (ref. 60) and 0.516 is our adopted slope (following equation 2). If we take 0.522 equilibrium value61, Δ17Osur reduces to 0.241‰, providing a likely explanation to the obtained low Δ17O on Oct 07. The low value can also be of anthropogenic origin, as combustion produces Δ17O as low as about −0.2‰ (ref. 62), and this is supported by the elevated [CO2] and reduced δ13C on that day (see Table 1).

Taking daytime photosynthetic flux of 1015 molecules cm−2 s−1 from a direct flux measurement for CO2 in a subtropical forest63 and following the same assumption as Hoag et al.38 for C3 plants, we have Fsur = 3 × 1015 molecules cm−2 s−1. (Δ17Ostr - Δ17O) is 0.5–1‰ (ref. 45,59). Fstr can then be evaluated from equation (3). Figure 4 shows the estimated flux from the stratosphere, providing a way to assess the vertical transport in transport models in the stratosphere, troposphere, and boundary mixed layer. We note that Fsur remains poorly understood. Hence similarly, if one can get an improved understanding for Fstr from, for example, extensive mid-tropospheric measurements58, Fsur can be better determined. We expect the utilization of multiple tracers (such as N2O) obtained by the CARIBIC project58 along with Δ17O in CO2 and a global model454664 could place a strong constraint on the strength of the cross-tropopause exchange, in particular with the use of the correlation of Δ17O and [N2O] in the upper troposphere. The uniqueness of Δ17O and [N2O] lies on their chemical properties in the atmosphere: both are inert in the free troposphere but significantly altered in the stratosphere45475964.

Figure 4

Inferred flux from the stratosphere obtained by assuming Fsur = 3 × 1015 molecules cm−2 s−1, Δ17Osur = 0.241‰, and Δ17Ostr - Δ17O = 0.5‰ (see text).

In short, stratosphere-troposphere exchange carries stratospheric air to the troposphere. The air from the stratosphere has oxygen isotope signature of CO2 distinct from that originating from the surface. The interaction between the subtropical jet and winter monsoon systems could enhance the vertical mixing and cross-tropopause exchange, supported by the observed Δ17O in the near surface air CO2. The detection of Δ17O trend is clearly demonstrated. The magnitude of the trend is found to be correlated with the strengths of the subtropical jet and winter monsoon. This trend is, on average, 0.0035‰/day during the jet development period in Oct, and can be as much as 0.0098‰/day that we observe in Feb. The observed anomalous CO2 at the surface potentially provides an additional constraint to refine our view of carbon cycle involving CO2 and also provides a strong constraint on the transport of the stratospheric flux to the surface. This is the largest dataset after Thiemens et al.54 and the first attempt to monitor Δ17O at such a high sampling frequency.

Additional Information

How to cite this article: Liang, M.-C. and Mahata, S. Oxygen anomaly in near surface carbon dioxide reveals deep stratospheric intrusion. Sci. Rep. 5, 11352; doi: 10.1038/srep11352 (2015).