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 (R2 = 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.

( ) ln 1 O ln 1 O 1 17 18 The factor λ is taken to be 0.516 (λ 0 ) and may vary between 0.500 and 0.529 (ref. 39). The λ 0 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, δ 17 [41][42][43] and δ 17 O(CO 2 ) ≈ 1.7 × δ 18 O(CO 2 ) in the stratosphere (reference to tropospheric CO 2 ; ref. [44][45][46][47][48]. The oxygen isotope distribution in CO 2 is largely affected by O 2 -O 3 -CO 2 photochemistry in the middle atmosphere, via the reaction O( 1 D) + CO 2 where O( 1 D) is formed by dissociation of O 3 (ref. 45,46,[48][49][50]. As O 3 and CO 2 are strongly coupled in the stratosphere and isotopically anomalous CO 2 can be produced in the middle atmosphere only, one may obtain a better constraint for stratospheric O 3 at the surface by measuring the isotopic composition of CO 2 . 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
CO 2 -O 2 oxygen isotope exchange method developed previously 51 was followed with slight modification (see Figure S1) to measure the Δ 17 O of CO 2 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 Δ 17 O values of CO 2 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 method 52 to convert isotopically known O 2 to CO 2 and assume the conservation of Δ 17 O. Good accuracy is demonstrated in Table S2 and Figure S2. Concentration of CO 2 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 CO 2 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 CO 2 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 CO 2 for isotope analysis (~30 μ mole). In addition to major gases like N 2 , O 2 , and Ar, the flask air samples with CO 2 also contain traces of water vapor and other gases that could potentially interfere with the CO 2 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 o C) for removing water and volatile organics while the remaining three were used for CO 2 collection at liquid nitrogen temperature (−196 o C). 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 CO 2 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. the mission. On average, the concentration ([CO 2 ]) is 411.8 ± 9.8 ppmv, δ 13 C − 8.91 ± 0.56‰ (V-PDB), δ 18 O 40.60 ± 0.52‰ (V-SMOW), and Δ 17 O 0.329 ± 0.037‰ (1-σ standard deviation to represent the scatter of the data). Identification of the sources responsible for the changes of CO 2 level can be done from the so-called Keeling plot ( Figure S3). The intercept for δ 13 C is − 27‰, a value that is consistent with respiration from C 3 plants (major type of plant in the region), though the signature may not be distinguishable from fossil fuel burning 55 . Figure 1 shows the three-isotope plot of oxygen in CO 2 collected in the region, comparing to that at La Jolla 54   0.0035‰/day (R 2 = 0.59; Fig. 2). Afterwards, further strengthening of the jet does not enhance Δ 17 O and the trend reduces to 0.0004‰/day (R 2 = 0.22), but the short-term enhancement is apparent (see below).

Discussion and summary
Stratospheric intrusions in East Asia occurs in close association with the presence of the subtropical jet stream [25][26][27][28][29][30][31] . 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 troposphere 56 and sometimes also into the upper troposphere and lower stratosphere, thus leading to an enhancement of cross-tropopause exchange resulting in elevated Δ 17 O in surface CO 2 . Below we focus our discussion on the downwelling branch of transport to the lower troposphere.  To support the stratospheric origin of anomalous CO 2 , we analyze ECMWF Interim O 3 data and the results are presented in Fig. 3. We see that the level of O 3 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 CO 2 oxygen anomaly, two events in 2014 are selected: Jan 07-27 and Feb 17-24. Δ 17 O increases with time, concordant with the elevated zonal wind ( Fig. 2; with lag of a few days); the Δ 17 O 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 Δ 17 O is calculated to be 0.0098‰/day (R 2 = 0.79) for the latter case and is a factor of ~3 higher than the former (0.0027‰/day; R 2 = 0.99) and the trend in Oct (0.0035‰/day; R 2 = 0.59). The isotopic composition of CO 2 in the atmosphere is an integrated signal of atmospheric and biogeochemical processes. In the atmosphere, the primary mechanism that modifies the isotopic composition of CO 2 is the exchange reaction with O( 1 D) in the stratosphere. The stratospheric source of CO 2 is enhanced in δ 18 O and Δ 17 O and has a seasonal cycle that is different from that originating from the surface 21,57 . For example, at the Waliguan observatory, biogeochemical models 21 predict maximum effects due to respiration in March-April (maximum in δ 18 O) and August (minimum in δ 18 O) and due to assimilation in ~March (minimum in δ 18 O) and July-August (maximum in δ 18 O), while the Brewer-Dobson circulation has maximum strength in ~March-July 36 . As a consequence of the interaction between these processes, the maximum in δ 18 O may occur in June 57 . This does not mean the seasonal cycle of δ 18 O is solely caused by the cross-tropopause exchange [cf. ref . 58]. Instead, in addition to natural biogeochemical cycle that results in maximal δ 18 O in ~April, elevated δ 18 O from the stratosphere is to modify the seasonal cycle to move the peak from April to June 57 . The presence of frequent deep intrusions over Tibet was shown recently 28 . However, to fully resolve the source of summertime O 3 , tracers like Δ 17 O that are seriously affected by stratospheric processes are essential. In this work, the size of Δ 17 O 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 CO 2 ) anomaly 45,59 from ~100 mbar to 1000 mbar. Attempts to utilize Δ 17 O for stratospheric and surface flux estimates are made below.
Given that CO 2 is chemically inert in the troposphere, assuming steady state, we have and reduced δ 13 C on that day (see Table 1). Taking daytime photosynthetic flux of 10 15 molecules cm −2 s −1 from a direct flux measurement for CO 2 in a subtropical forest 63 and following the same assumption as Hoag et al. 38 45,59). F str 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 F sur remains poorly understood. Hence similarly, if one can get an improved understanding for F str from, for example, extensive mid-tropospheric measurements 58 , F sur can be better determined. We expect the utilization of multiple tracers (such as N 2 O) obtained by the CARIBIC project 58 along with Δ 17 O in CO 2 and a global model 45  In short, stratosphere-troposphere exchange carries stratospheric air to the troposphere. The air from the stratosphere has oxygen isotope signature of CO 2 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 Δ 17 O in the near surface air CO 2 . The detection of Δ 17 O 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 CO 2 at the surface potentially provides an additional constraint to refine our view of carbon cycle involving CO 2 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 Δ 17 O at such a high sampling frequency.