Journal of the Meteorological Society of Japan, Vol. 91, No. 2, pp. 179̶192, 2013 DOI:10.2151/jmsj.2013-206
179
Development of a High Precision Continuous Measurement System for the Atmospheric O2/N2 Ratio and Its Application at Aobayama, Sendai, Japan Daisuke GOTO Center for Atmospheric and Oceanic Studies, Tohoku University, Sendai, Japan
Shinji MORIMOTO National Institute of Polar Research, Tokyo, Japan
Shigeyuki ISHIDOYA, Akinori OGI, Shuji AOKI, and Takakiyo NAKAZAWA Center for Atmospheric and Oceanic Studies, Tohoku University, Sendai, Japan (Manuscript received 2 July 2012, in final form 22 December 2012)
Abstract To contribute to a better understanding of the global carbon cycle, a high precision continuous measurement system for atmospheric O2/N2 ratio was developed using a fuel cell oxygen analyzer. To obtain highly precise values of the atmospheric O2/N2 ratio, pressure fluctuations of the sample and standard air were reduced to within ±0.005 Pa, with temperatures stabilized to 32.0 ± 0.1° C. The analytical precision of the system was estimated to be ±1.4 per meg for 24-minute measurement as the standard deviation (1σ) of replicate analyses of the same sample air. This analytical precision is sufficient for clearly detecting very small spatiotemporal variations of the atmospheric O2/N2 ratio. A new set of secondary and working standard gases with specified O2/N2 ratios were also prepared by drying natural air to dew points lower than −80° C using a specially designed H2O traps and then adjusting its amount of O2. The prepared five secondary standard gases were repeatedly calibrated against our primary standard, and their O2 /N2 ratios were confirmed to be stable with no appreciable trend for over 570 days at least. A non-dispersive infrared analyzer was also installed into the measurement system to allow simultaneous measurements of the atmospheric CO2 concentration. The analytical precision of the CO2 concentration was estimated to be ±0.03 ppm (1σ). Using the new system, we initiated a systematic observation of the atmospheric O2/N2 ratio at Aobayama, Sendai, Japan in February 2007. The observed measurements clearly showed seasonal and diurnal cycles, along with short-term variations on time scales of several hours to several days, caused by terrestrial biospheric and human activities. Keywords atmospheric O2; continuous measurement; O2/N2 exchange ratio
1.
Introduction
It is well known that O2 and CO2 fluxes are inversely correlated with each other during photosynthesis, respiration and fossil fuel combustion, while they have Corresponding author: Daisuke Goto, Center for Atmospheric and Oceanic Studies, Tohoku University, 6-3, Aramaki, Aoba, Sendai 980-8578, Japan. E-mail:
[email protected] ©2013, Meteorological Society of Japan
no correlation during air-sea exchange. Such a behavior of O2 and CO2 fluxes allows us to estimate land biotic and oceanic CO2 sinks separately by simultaneously observing the atmospheric O2 and CO2 concentrations for a long time (e.g. Manning and Keeling, 2006). A seasonal cycle of atmospheric O2 also has information about marine biological production, because a large amount of O2 is biologically produced and consumed in the surface ocean and the time scale for equilibration of O2 between the
180
Jounal of the Meteorological Society of Japan
atmosphere and the ocean is only a few weeks (Keeling and Shertz, 1992; Bender et al. 1996). To understand the global carbon cycle using atmospheric O2 variations, highly precise measurements are required (Keeling et al. 1993). For this purpose, various measurement systems have been developed employing, for example, an interferometer (Keeling, 1988), a mass spectrometer (Bender et al. 1994; Ishidoya et al. 2003) and a gas chromatograph (Tohjima et al. 2000). Using these systems, systematic observations of the atmospheric O2 and CO2 concentrations have been conducted since the early 1990s, in which air samples collected in flasks are analyzed in laboratories (Keeling and Sherzts, 1992; Bender et al. 2005; Langenfelds et al. 1999; Battle et al. 2006; Ishidoya et al. 2006, 2008a, 2008b; Manning and Keeling, 2006; Tohjima et al. 2005, 2008). These observations have revealed informative spatiotemporal variations in the atmospheric O2 concentration. More recently, several continuous measurement systems have been developed using a paramagnetic analyzer (Manning et al. 1999), a vacuum ultraviolet absorption analyzer (Stephens et al. 2003), a gas chromatograph (Yamagishi et al. 2008) and a fuel cell analyzer (Stephens et al. 2007). Continuous observations with these systems have been carried out at ground fixed stations (Lueker, 2004; Stephens et al. 2007; Yamagishi et al. 2008; van der Laan-Luijkx et al. 2010), and on board ship (Thompson et al. 2008; Yamagishi et al. 2012) and aircraft (Stephens et al. 2009), although some of them were made sporadically as part of campaign programs. These observations have allowed us to increase our understanding of the atmospheric O2 variations on short-time and smallspace scales. However, we still lack adequate levels of systematic continuous O2 measurements. In order to contribute to a better understanding of the global carbon cycle, we developed a new highprecision continuous measuring system using a differential fuel-cell oxygen analyzer and have been making simultaneous observations of atmospheric O2 and CO2 at a ground station, Aobayama, Japan since February 2007. In this paper, we present details of our newly developed measurement technique for the atmospheric O2 concentration. Preliminary results obtained at Aobayama are also presented and discussed. 2.
Continuous measurement system of the atmospheric O2/N2 ratio
2.1 System configuration In discussing variations of the atmospheric O2
Vol. 91, No. 2
concentration, a dilution effect of O2 caused by changes in the total number of molecules in the atmosphere is not negligible due to its abundance in the atmosphere. Therefore, the atmospheric O2 concentration is usually expressed as a change in the ratio of O2 to N2 relative to an arbitrary reference (Keeling and Sherzts, 1992): δ(O2/N2) =
(O2/N2)sample 6 − 1 × 10 (per meg), (O2/N2)reference
(1)
where (O2/N2)sample and (O2/N2)reference indicate the O2/N2 mole ratios of the sample and the reference air, respectively. The value of δ(O2/N2) defined by equation (1) is commonly reported in “per meg” unit, and 4.8 per meg is approximately equivalent to 1 μmol/mol (ppm) of the atmospheric O2 concentration. The continuous measurement system developed in this study is schematically illustrated in Fig. 1. The measurement system consists of a differential fuel-cell oxygen analyzer (Sable Systems Oxzilla II), a gas handling system and a data acquisition system. A nondispersive infrared CO2 analyzer (LI-COR LI-6252) is also installed in the system to simultaneously measure the CO2 concentration. The fuel-cell oxygen analyzer employed in this system is small and lightweight, so that it is suitable for on-site observations of atmospheric O2. The O2 analyzer has two symmetrical flow paths of air, and each path is equipped with a fuel cell as a detector. The sample or standard air is introduced into one cell and the reference air into the other cell. The fuel cell consists of a weak acid electrolyte, a gold cathode, a lead anode and a thin gaspermeable membrane, and the net reaction +
O2 + 2Pb + 4H → 2H2O + 2Pb
2+
(R1)
takes place in the fuel cell. This reaction produces an electric current that is linearly proportional to the partial pressure of O2 in the air at the surface of the membrane, and the analyzer continuously measures the output difference between the two fuel cells. Because the fuel cell is very sensitive to changes in the environmental temperature, both fuel cells are mounted inside a thermal insulation box. The analyzer is also thermally insulated and the inside temperature is maintained at 32.0 ± 0.1° C with a temperature regulator. In addition, to minimize possible different fractionations of O2 and N2 caused when the air passes through the analyzer, plastic tubes originally used for the analyzer tubing have all been removed and replaced with stainless-steel tubes with an outer diameter of 1/8 inch. To measure the CO2 concentration together with δ(O2 /N2 ), one outlet of the O2
April 2013
D. GOTO et al.
181
Fig. 1. Schematic diagram of the continuous measurement system developed by this study for atmospheric δ(O2/N2) and CO2.
analyzer is connected to a sample cell of the NDIR analyzer with a stainless-steel tube. The air with some amount of CO2 always flows through the reference cell of the NDIR at a rate of 5 mL/min. The gas handling system includes an air sampling system and a pressure regulation device. The air sampling system mainly comprises an air intake, a diaphragm pump, and three different types of air drier. We have adopted an aspirated intake for collecting the sample air, to prevent thermal fractionation of O2 and N2 due to solar radiation (Blaine et al. 2006); ambient air passes through the aspirated intake at a flow rate of 35̶40 L/min with the aid of an electric fan, and part of the flowing air is introduced into the O2 analyzer through a Dekabon tube with an inner diameter of 4 mm using the diaphragm pump. A buffer loop, which is a 4-m-long stainless-steel tube with an outer diameter of 1/4 inch, is connected to the outlet of the
diaphragm pump to minimize pressure fluctuations induced by the vibrating diaphragm. Two glass traps are immersed in an ethanol bath held below −80° C using a cryogenic cooler, and water vapor contained in the sample air is sufficiently removed by the traps. To reduce the number of times the traps need to be changed, a drier using a Peltier device and a Nafion tube are installed at the upstream of the pump. A stainless-steel tube with an outer diameter of 1/4 inch, of which one end is connected to the Dekabon tube, is cooled by the Peltier device at 0̶2° C, and the sample air passing through the stainless-steel tube is dried to dew points of 1̶2° C. Water removed from the sample air falls into a drain, and then is periodically discharged to the outside using a pump. The sample air is further dried to a dew point of approximately −40° C using the Nafion tube. In this drying system, ambient air dried with calcium sulfate (CaSO4) is forced to flow
182
Jounal of the Meteorological Society of Japan
along the outside of the tube in an opposite direction to the sample air at flow rates of 200̶300 mL/min. In order to achieve high-precision measurements of atmospheric δ(O2/N2), various fractionations of O2 and N2 in the sample air inside the tubing system, caused by fluctuations in pressure and temperature (Keeling et al. 1998), should be minimized. Therefore, the sample pressure in the traps is maintained at 0.1 MPa by controlling a flow regulation valve (PV1, HORIBA STEC, PV-1000) with the signal from a pressure gauge (P1, Honeywell, 40PC series). In addition, precise differential pressure gauges (P2 and P3, Setra systems Inc., model 239) are arranged at the respective outlets of the O2 analyzer, and the flow regulation valves (PV2 and PV3, HORIBA STEC, PV-1000) at the analyzer inlets are controlled so that the pressures of the air flowing through the analyzer are equalized as much as possible to that of the air inside a pressure reference volume at a gauge pressure of 20 kPa. This pressure regulation system, except for PV1, is thermally insulated and kept at 32.0±0.1° C. By employing these devices, the pressure fluctuations of the sample or standard air and the reference air in the fuel cells are reduced to within ±0.005 Pa, which greatly improves the measurement precision of atmospheric δ(O2/N2). High and low working standard air in 48 L highpressure aluminum cylinders are used to routinely calibrate the O2 analyzer. Their δ(O2 /N2 ) values are determined against our secondary standard air that will be described in section 2̶4. In accordance with Keeling et al. (1998, 2007), the standard and reference air cylinders are positioned horizontally in a thermally insulated rack to minimize the thermal and gravitational fractionations of O2 and N2 in the cylinders. A pressure regulator and a 2 μm air filter are connected to each cylinder, and the standard air from the cylinder is kept by the regulator at 0.1 MPa. All the standard air, the reference air, and the sample air are connected to a six-port valve (Valco Instruments Co. Inc., EC6W) through a manifold. The six-port valve has two positions so that the flow paths connected to the O2 analyzer can be alternated. While the O2 and NDIR analyzers measure the standard air for their calibration, the valve SV5 (Fig. 1) is opened so that the sample air always flows through the air sampling line at a rate of 80 mL/min. 2.2 Measurement procedures of O2 and CO2 Figure 2a shows an example of the output signals from the O2 analyzer obtained during actual atmospheric observations. The high working standard air (Hair) and the reference air (R-air) are first introduced into
Vol. 91, No. 2
Fig. 2. Examples of output signals of the O2 analyzer for (a) two-cycle measurements of atmospheric O2 and (b) for three minutes after alternating the air flow paths of the analyzer (expansion of the output surrounded by square in panel (a)). Vertical dashed lines represent the time when the flow paths are alternated.
the respective flow paths of the analyzer with a rate of 80 mL/min. To compensate for different output drifts of the respective fuel cells, the two flows are alternated three minutes later by switching the six-port valve and then measurement continues for another three minutes. The difference between the output signals before and after the switching is read as a measure of partial pressure difference of O2 between the H-air and the Rair. The output signals from the analyzer are collected every 2 seconds. However, because it takes approximately 120 seconds to stabilize their signals after each switching, as seen in Fig. 2b, the data taken for the first 120 seconds after switching are discarded, and an average of last 30 readings for the remaining 60 seconds is assigned to the analyzer output for the H-
April 2013
183
D. GOTO et al.
air. After measuring the H-air, the low standard air (Lair) is introduced into the O2 analyzer, and the same procedure as above is repeated to obtain the analyzer output for the L-air. Then, the sample air (S-air) is measured for 24 minutes, alternating the flow path eight times, and lastly the measurements of the H-air and the L-air are carried out again to confirm the output drift of the analyzer. As a result, it takes 48 minutes to perform one cycle analysis of the H-air, the L-air, the S-air, the H- and L-air, with 7 atmospheric δ(O2 /N2) measurements during each cycle. These procedures are repeated automatically by using a programmable logic controller. The δ(O2/N2) value of each sample air is calculated from its own analyzer output, in which the outputs obtained for the H-air and the L-air before and after sample measurements are averaged and then linearly interpolated with respect to δ(O2 /N2 ). The procedures will be described below in more detail. As mentioned above, the sample cell of the NDIR is connected to one outlet of the O2 analyzer with a stainless-steel tube. Therefore, the H-air, the R-air, the L-air and the R-air are introduced into the sample cell for three minutes in sequence, and subsequently the Sair and then the R-air flow through the cell, which is repeated four times. The H- and L-air also include the different amounts of CO2, so that the respective air acts as a CO2 standard gas. The output from the NDIR is also recorded once every two seconds and the average of the 30 readings for the last 60 seconds is used as the analyzer output for each air. Because this procedure allows us to take only four CO2 concentration data points during one cycle analysis of δ(O2 /N2), the measured values of the CO2 concentration are temporally interpolated linearly to estimate the corresponding value at the time when δ(O2 /N2) is measured. To convert the apparent atmospheric O2 mole fraction measured by the O2 analyzer to δ(O2/N2), it is necessary to correct for the dilution effect by CO2 contained in the sample air (Keeling et al. 1998). The apparent mole fraction differences (δXO2) of the working standard air used for this system are determined against our secondary standard air system (Section 2̶ 4). Therefore, δXO2 of the sample air can be obtained from its output signal of the O2 analyzer by linearly interpolating the signals from the two working standard air. The value, thus obtained, is converted to δ(O2 /N2) using the simultaneously measured CO2 concentration and the following equation (Manning et al. 1999):
Fig. 3. Occurrence distributions for the standard deviation values of (a) δ(O2/N2) and (b) CO2 concentration obtained by repeatedly analyzing the sample air from a cylinder using the continuous measurement system developed in this study.
δ(O2/N2) =
δXO2 + XO2([CO2]−355.01) , XO2(1−XO2)
(2)
where δXO2 is the apparent mole fraction of O2 in ppm , XO2 is the standard mole fraction of O2 in air (XO2 = 0.20945 (Machta and Hughes, 1970)), [CO2] is the measured CO2 concentration in ppm, and 355.01 (ppm) is the CO2 concentration of a primary standard air that defines zero for our δ(O2/N2) measurements. 2.3 Evaluation of analytical precision We evaluated the analytical precision of our measurement system by repeatedly determine the δ(O2/N2) and CO2 concentration values of the sample air from a high-pressure cylinder over a 24-hour period. As shown in Fig. 3a, the standard deviation of each set of seven values of δ(O2/N2) obtained every 24 minutes distributes mostly in a range of 1.2̶1.6 per meg, with an average of 1.4 ± 0.5 per meg which is equivalent to 0.3 ppm in the O2 mole fraction. We also found 1.9 per meg for the standard deviation of the δ(O2/N2) values obtained over a 24-hour period. From this result, we estimated our analytical precision of δ(O2/N2) to be 1.4 and 1.9 per meg for short-term and long-term measurements, respectively. Several laboratories have also developed continuous δ(O2 /N2) measurement systems employing different methods. Our analytical precision is close to 1.0 per meg obtained by Manning et al. (1999) using a paramagnetic analyzer and 1.4 per meg by Stephens et al. (2007) using the fuel cell analyzer, and is better than 2.5 per meg by Stephens et al. (2003) using a vacuum
184
Jounal of the Meteorological Society of Japan Table 1.
Vol. 91, No. 2
Secondary standard air prepared for continuous O2 measurements.
Cylinder No.
δ(O2/N2) (per meg)
CO2 (ppm)
Dew-point temperature (° C)
CQC00752 CQC00731 CQB15432 CQB07972 CQB09375
−104.7±6.4 −237.8±5.9 −380.3±2.7 −502.3±4.8 −633.1±4.0
402.96 403.82 406.52 412.85 430.16
−82.2 −81.7 −84.0 −80.6 −82.2
a,c
δXO2_ms (ppm)
−27.38±1.06 −49.61±0.97 −73.77±0.44 −95.31±0.79 −120.58±0.67
b,c
δXO2_fc (ppm)
−26.75±1.01 −50.34±1.33 −74.26±1.06 −96.00±0.44 −120.38±0.36
a
Mole fraction of O2 calculated from δ(O2/N2) and CO2 concentration Mole fraction of O2 obtained from the self-check analysis with the O2 analyzer using the δXO2_ms values c Uncertity of each value represents standard deviation (±1σ) b
ultraviolet absorption method and 6.0 per meg by Yamagishi et al. (2008) using a gas chromatographic method. The precision attained by this study is much better than those of the measurement systems developed for flask samples (e.g., 4.0 per meg by Bender et al. (1996) using a mass spectrometry method). The analytical precision of the CO2 concentration was also estimated in the same way as δ(O2 /N2 ). As seen in Fig. 3b, most of the standard deviation values over a 24-minute period distribute in a range of 0.03̶ 0.04 ppm, and their average is calculated to be 0.03 ± 0.01 ppm. On the other hand, the standard deviation over a 24-hour period was estimated to be 0.05 ppm. 2.4 Standard air for the O2 analyzer The standard air for our O2 measurements with the fuel-cell O2 analyzer is categorized into primary, secondary and working. The primary standard is a dried natural air, which was prepared at the beginning of our O2 research program (Ishidoya et al. 2003). The secondary standards were newly prepared from natural air, as summarized in Table 1. Their δ(O2/N2) values were determined against our primary standard using the mass spectrometer (Ishidoya et al. 2003), and the δXO2 values were calculated from their δ(O2 /N2 ) and CO2 concentrations using equation (2). In order to examine if the secondary standard air can be commonly used for the O2 analyzer and the mass spectrometer, a self-check analysis of these standard air with the O2 analyzer was carried out. In this analysis, the δXO2 values of the respective air were recalculated under an assumption that the relationship between the analyzer output and δXO2 is linear. The δXO2 values, thus obtained, are compared in Fig. 4 with the values calculated by the above-mentioned procedures using equation (2). A good linear relationship can be seen between the values from the two methods, the
Fig. 4. Plots of δXO2 obtained from self-check analysis of five secondary standard air with the O2 analyzer against δXO2 determined based on their mass spectrometer analyses.
differences being within 0.7 ppm. This agreement implies that the results from the O2 analyzer can be directly compared to those from the mass spectrometer. The working standard air that are routinely used for continuous O2 measurements were calibrated against the secondary standard air system using the O2 analyzer. The secondary and working standard air were prepared from ambient natural air. To collect ambient air, we used a booster pump with a maximum discharge pressure of 34 MPa. Figure 5 illustrates the air collection system developed in this study. The system consists mainly of water traps (T1 and T2), a
April 2013
D. GOTO et al.
Fig. 5.
185
Sampling system developed for preparing dried ambient air.
diaphragm pump and the booster pump. The T1 trap is comprised of a U-shaped Pyrex glass trap and two stainless-steel traps in which thin stainless-steel tubes are packed, and the T2 trap is composed of two Russian doll-type traps (hereafter referred to as RDT) (Brenninkmeijer and Röckmann, 1996). Both T1 and T2 are immersed in an ethanol bath held below −80° C. The air was taken from the aspirated intake mounted on the roof of a 40 m high laboratory building using the diaphragm pump at a flow rate of 12 L/min and introduced into the booster pump after passing through water traps to remove water vapor. To efficiently remove water vapor from a large amount of ambient air, we adopted two RDTs (T2 trap) in addition to the T1 trap to achieve high flow rates. RDT consists of three thimble-type membrane filters and a stainlesssteel cylinder with two stainless-tubes inserted inside. The thimble filters are set in the cylinder, as shown in Fig. 5, and the remaining water vapor in the air is removed by passing through the filters. It was confirmed that the dew point temperature of the ambient air after passing through the T2 trap is lower than −80° C at a flow rate of 12 L/min. The air dried by the T1 and T2 traps was compressed into 48̶L aluminum high-pressure cylinders to 15.0 MPa using the booster pump. Prior to use, all cylinders were evacuated using a turbo-molecular pump to a pressure −4 of 10 Pa and heated at 50° C for at least 12 hours. The δ(O2 /N2) values and CO2 concentrations of the collected air were determined using the mass spectrometer and the NDIR analyzer (HORIBA, VIA510R), respectively. Based on the measurement results, appropriate amounts of pure O2 were added to the air to adjust their δ(O2/N2) values. The dew point temperatures of all prepared standard air were confirmed to be lower than −80° C. If the removal of
water vapor is not complete, then the dilution effect by H2O on δ(O2/N2) has to be corrected. In our case, the effect was estimated to be approximately 0.4 per meg, which is negligibly small. We also confirmed that their δ(O2/N2) values determined using the O2 analyzer and the mass spectrometer are in excellent agreement with each other. In order to confirm the stability of δ(O2/N2) of our secondary standards, we repeated their calibration against our primary standard using the mass spectrometer. As seen in Fig. 6, their calibrated values of δ(O2/N2) show no significant trend for over 570 days; by assuming a linear trend, the average rates in change of δ(O2/N2) were found to be −4.5 ± 6.1, −0.4 ± 4.1, + 1.2 ± 1.9, + 4.4 ± 4.3 and −3.9 ± 3.3 per meg/yr for CQC00752, CQC00731, CQB15432, CQB07972, and CQB09375, respectively. The CO2 concentrations of the working standard air were determined using the NDIR analyzer (HORIBA VIR̶ 510R) against our gravimetrically prepared CO2 standard gas system (Tanaka et el. 1983). 3.
Preliminary observational results of atmospheric δ(O2/N2) in Sendai
Using the newly developed measurement system, we initiated systematic and continuous observations of atmospheric δ(O2/N2) and CO2 at Aobayama (38.25° N, 140.83° E) in the suburbs of Sendai, Japan in February 2007. The observation site is located on a hilltop west of the Sendai urban area and surrounded by deciduous shrubs. The aspirated air intake was set on the roof of our 40 m high laboratory building, and the sample air was drawn through the Dekabon tube using the diaphragm pump into an air-conditioned room in which our O2 measurement system was installed. Each working standard air in the 48 L aluminum cylinder
186
Jounal of the Meteorological Society of Japan
Fig. 6. δ(O2 /N2 ) values of five secondary standard air determined against primary standard using the mass spectrometer. Error bar of each symbol represents one standard deviation (±1σ).
was replaced with a new one when its pressure decreased below 3.0 MPa after use for approximately 7 months. By calibrating the δ(O2/N2) and CO2 concentration values of the working standard air against the secondary standards before and after their use, the drifts in the respective values were confirmed to be within 5 per meg and 0.06 ppm. Hourly mean values of the atmospheric δ(O2/N2) and CO2 concentration for the period February 2007̶ January 2010 are shown in Fig. 7, together with the results from the mass-spectrometric analysis of discrete flask sampling for the same period (Ishidoya et al. 2012). The δ(O2/N2) values obtained from the flask sampling are also compared in Fig. 8 with hourly means from the continuous measurements. Each hourly mean plotted in this figure is an average of continuously measured values for one hour including the time when the corresponding flask sampling was made. It is clearly seen from Fig. 8 that there is a good linear relationship between the δ(O2 /N2 ) values from the flask sampling and continuous measurements, with a correlation coefficient of 0.90. It is also found that the δ(O2/N2) values from the flask sampling measurements are marginally higher by 1.8 ± 5.5 per meg, on average, than those from the continuous measurements. However, since the difference is within our
Vol. 91, No. 2
Fig. 7. Hourly mean values (gray dots) of δ(O2/N2) and CO2 observed continuously at Aobayama, Sendai from February 2007 to January 2010. Solid and dashed lines represent best-fit curves and trends derived by applying the digital filtering technique to the observed daily mean values, respectively. The results obtained from flask sampling measurements at the site are also represented by open squares.
analytical precision, the values from both measurement methods are fundamentally the same. Best-fit curves and trends of δ(O2 /N2) and CO2 concentration were derived by applying a digital filtering technique (Nakazawa et al. 1997a) to the daily means of the continuously measured values. The daily mean values were calculated from the data obtained from 11:00 to 15:00 (LST). The CO2 concentration was relatively stable at low levels during this period, mainly due to strong atmospheric vertical mixing, as seen from its monthly averaged diurnal cycles to be discussed later (Fig. 9). In this filtering process, signals with periods of longer than 24 months were regarded as the trend, and an average seasonal cycle was approximated by fundamental and its first harmonics. As seen from the results shown in Fig. 7, δ(O2 /N2 ) decreases secularly with an average rate of −24.2 per meg/yr, accompanied by the seasonal cycle with a maximum in August and a minimum in late March. In contrast, the CO2 concentration increases secularly, showing an average rate of 3.3 ppm/yr, and varies seasonally in almost opposite phase with δ(O2/N2). The
April 2013
D. GOTO et al.
187
Fig. 8. Comparison of δ(O2 /N2 ) obtained from the flask sampling (δ(O2 /N2 )flask ) and continuous measurements (δ(O2/N2)1h) at Aobayama, Sendai from 6 March, 2007 to 5 January, 2010. Solid and dotted lines represent a linear regression line and one-to-one relationship between δ(O2/N2)flask and δ(O2/N2)1h, respectively.
seasonal peak-to-peak amplitudes of δ(O2/N2) and CO2 are approximately 117 per meg and 16 ppm, respectively. Keeling and Shertz (1992) found the respective seasonal amplitudes of δ(O2/N2) and CO2 to be 120 per meg and 13 ppm at La Jolla (32.9° N, 117.3° E ), California. Tohjima et al. (2003) also reported corresponding amplitudes of 146 per meg and 15.8 ppm at Cape Ochi-Ishi (43° N, 146° E), Japan. Their amplitudes observed at mid-northern latitudes are comparable to our results. The observed δ(O2 /N2) and CO2 also show clear short-term variations. Monthly averaged diurnal cycles of δ(O2 /N2 ) and CO2 are presented in Fig. 9. In the summer, both variables show clear diurnal cycles, and especially in August the peak-to-peak amplitudes of δ(O2/N2) and CO2 amount to 101 per meg and 21 ppm, respectively. Diurnal cycles can be seen for both variables even in the winter; δ(O2 /N2) and CO2 decreases and increases, respectively, around 9:00 and 19:00̶20:00 LST. Because these times correspond respectively to the start and end of office hours, such changes could be due to the increase in traffic around our laboratory building. In fact, an average −O2: CO2 exchange ratio, defined as the slope of a linear regression line between the measured values of
Fig. 9. Monthly averaged diurnal cycles of δ(O2 /N2 ) (left panel) and CO2 (right panel) concentration at Aobayama, Sendai for the period February 2007̶January 2010. Error bars represent the 95% confidence intervals for the respective mean values.
δ(O2/N2 ) and CO2 concentration, for these hours was calculated to be approximately 1.5 ppm/ppm, which is close to the value expected from the combustion of gasoline (Keeling, 1988). As an example of wintertime short-term variations, the values of δ(O2/N2) and CO2 concentration measured on 20̶26 February, 2009 are shown in Fig. 10a. During this period, the diurnal cycle is hardly observable for both variables, but large irregular variations are clearly seen twice, i.e. on 21̶22 and 25 February. As seen in Fig. 10c, δ(O2/N2) is negatively correlated with the CO2 concentration during both events. The −O2: CO2 exchange ratio was estimated to be 1.39 ± 0.04 ppm/ppm for 21̶22 February and 1.38 ± 0.03 ppm/ppm for 25 February. Keeling (1988) estimated the −O2: CO2 exchange ratios for fossil fuel combustion, by considering its typedifferent consumption amounts (natural gas, coal, oil, and natural gas flaring). Because CO2 emission data for
188
Jounal of the Meteorological Society of Japan
individual fossil fuel types in each country are available from the Carbon Dioxide Information Analysis Center (CDIAC), Oak Ridge National Laboratory (ORNL), U.S. Department of Energy (DOE) (http://cdiac.ornl.gov), we can estimate an average exchange ratio of −O2: CO2 for each country by using the same procedure as Keeling (1988) did. By doing so, we obtained a −O2: CO2 exchange ratio of 1.38 ppm/ppm for fossil fuel combustion in Japan. This exchange ratio is in good agreement with 1.39̶ 1.38 ppm/ppm observed at our site on 21̶22 and 25 February. We also examined other events in the winter and found the −O2: CO2 exchange ratio to be between 1.77 and 1.34 ppm/ppm, with an average of 1.41 ± 0.19 ppm/ppm. Therefore, the wintertime irregular variations observed in δ(O2 /N2 ) and CO2 would be attributable to the transport of air affected by the fossil fuel combustion in urban areas of Japan. Stephens et al. (2007) found a −O2: CO2 ratio of 1.46 ppm/ppm for some pollution events observed over woodlands (30̶396 m) in Wisconsin, U.S.A. in the winter, pointing out that it is consistent with an average exchange ratio expected from different types of fossil fuel consumed in the U.S.A. A slight deference between the −O2: CO2 exchange ratios for fossil fuel combustion in the U.S.A. and Japan is primarily due to the fact that the proportion of natural gas to fossil fuel combustion is higher in the U.S.A. than in Japan; natural gas yields a −O2: CO2 exchange ratio of 1.95 ppm/ppm which is higher than those for oil and coal (Keeling et al. 1988). By analyzing pollution events at Hateruma Island, Japan, Minejima et al. (2012) found that the observed −O2: CO2 exchange ratio is dependent on which country the air mass arriving at the island came from. These findings indicate that a simultaneous measurement of atmospheric O2 and CO2 would be a useful tool for differentiating various fossil fuel components consumed in different countries and locations. As an example of diurnally varying δ(O2 /N2 ) and CO2 concentration in the summer, their measured values on 11̶17 August, 2009 are shown in Fig. 10b. It is clearly seen that δ(O2 /N2) increases and the CO2 concentration decreases in the daytime and the situation is reversed in the nighttime. As seen from Fig. 10d, there is a strong negative correlation between the δ(O2 /N2) and CO2 variations, yielding a −O2: CO2 exchange ratio of 1.07 ± 0.02 ppm/ppm. Keeling (1988) estimated the −O2: CO2 exchange ratios to be 1.05 ± 0.05 ppm/ppm for photosynthesis and respiration by land plants. On the other hand, Severinghaus (1995) found a −O2: CO2 exchange ratio of 1.15
Vol. 91, No. 2
ppm/ppm for soil respiration. Considering these different results, Severinghaus (1995) proposed a −O2: CO2 exchange ratio value for land plant activities of 1.10 ± 0.05 ppm/ppm, which is currently adopted widely. In this connection, Stephens et al. (2007) also reported, from their observations of δ(O2 /N2 ) and CO2 concentration over woodlands in Wisconsin, a −O2: CO2 exchange ratio for land plant activities of 1.01̶1.10 ppm/ppm. Our results are close to their values, suggesting that the diurnal cycle of δ(O2 /N2) and CO2 observed at Aobayama in the summer is mainly produced by land plant activities around the site. To examine the exchange of O2 and CO2 associated with land plant activities in more detail, we calculated the −O2: CO2 exchange ratios using the δ(O2/N2) and CO2 concentration values observed in the summer of 2007̶2009. At first, we selected only those data taken under westerly winds during June̶August of each year. Because deciduous shrubs are situated mostly to the west of the site, it is expected that the δ(O2/N2) and CO2 concentration values measured under westerly winds will be much affected by land plant activities. Then, the −O2: CO2 exchange ratio was calculated from the slope of a linear regression line fitted to the measured values of δ(O2/N2) and CO2 concentration for the daytime (6:00̶15:00 LST) or nighttime (19:00̶ 4:00 LST) of each day, to differentiate the exchange ratio between photosynthesis and respiration. The results obtained are summarized in Table 2. It is obvious that there is no discernible difference between the −O2: CO2 exchange ratios derived for photosynthesis and respiration. The average value of the −O2: CO2 exchange ratio over the three summer periods was found to be 1.08 ± 0.09 ppm/ppm for the daytime and 1.08 ± 0.09 ppm/ppm for the nighttime, both ratios being identical to each other. Thus the present result is in excellent agreement with 1.10 ± 0.05 ppm/ppm estimated by Severinghaus (1995). We also calculated the −O2: CO2 exchange ratio under non-westerly winds for the same period. The result showed that the ratio values scatter between 1.0 and 1.5 ppm/ppm, suggesting that both land plant activities around the site and fossil fuel combustion in urban areas are responsible for temporal variations of the δ(O2/N2) and CO2 concentration, depending on the situation. 4.
Summary
We developed a high-precision continuous measurement system using a fuel-cell O2 analyzer to measure atmospheric δ(O2 /N2). To achieve highly precise measurements, pressure fluctuations of the
April 2013
189
D. GOTO et al.
Fig. 10. Temporal variations of δ(O2/N2) and CO2 concentration at Aobayama, Sendai on (a) 20̶26 February, 2009 and (b) 11̶17 August, 2009, and correlation plots of δ(O2/N2) and CO2 concentration on (c) 21̶22 and 25 February, 2009 and (d) 11̶17 August, 2009. Table 2. −O2: CO2 exchange ratios calculated from measured values of δ(O2/N2) and CO2 concentration at Aobayama, Sendai in the summer of 2007̶2009. −O2: CO2 exchange ratio (ppm/ppm) Daytime Nighttime
Year 2007 2008 2009 Average (2007̶2009)
1.07±0.09 1.09±0.10 1.07±0.09 1.08±0.09
sample and standard air in the fuel cells were minimized to within ±0.005 Pa using a flow regulation valve and a differential pressure gauge. Their temperatures were also stabilized to 32.0 ± 0.1° C not only by controlling the inside temperature of the analyzer but also by thermally insulating the analyzer and the pressure stabilization system. In addition, all the tubing of the analyzer was replaced
1.09±0.09 1.10±0.08 1.06±0.10 1.08±0.09
with a stainless-steel tube to minimize possible fractionation of O2 and N2. Furthermore, an aspirated air intake was adopted as an air inlet to prevent thermal fractionation of O2 and N2. As a result, a high analytical precision of ±1.4̶1.9 per meg was attained for continuous measurements of atmospheric δ(O2 /N2). The measurement system was also equipped with a NDIR analyzer to simultaneously measure the CO2
190
Jounal of the Meteorological Society of Japan
concentration. In order to calibrate the O2 analyzer, as well as to maintain the self-consistency of the measured values of atmospheric δ(O2/N2) over a long period, a standard air system consisting of primary, Secondary, and working was newly established. The secondary and working standard gases were prepared by drying ambient air to the dew-point temperatures of lower than −80° C with Russian doll-type traps, compressing into high-pressure aluminum cylinders with a booster pump and then adjusting their δ(O2 /N2) values by adding appropriate amounts of pure O2. The standard air thus prepared were confirmed to be stable in δ(O2/N2) over at least 570 days. The δ(O2/N2) values of the working standard air were determined using the secondary standards which were calibrated against the primary standard gas of dried natural air prepared at the beginning of our O2 research program. Using the newly developed measurement system, continuous observations of atmospheric δ(O2/N2) and CO2 concentration were initiated at Aobayama, Sendai in February, 2007. δ(O2 /N2 ) showed a clear seasonal cycle with a minimum value in late March to early April and a maximum value in late July to early August, superimposed on a secular decrease. The CO2 concentration increased secularly and varied seasonally in an opposite phase with δ(O2 /N2). Short-term variations on time scales of several hours to several days were also clearly observed. In the winter, it was often seen that δ(O2 /N2) sharply declined in a short time, accompanied by an increase in the CO2 concentration, and the low values lasted for several hours or days. The −O2: CO2 exchange ratio was found to be 1.39̶1.38 ppm/ppm for such wintertime short-term variations. Because these ratios are in good agreement with a mean value of the −O2: CO2 exchange ratio calculated for fossil fuel consumption in Japan, the observed decline in δ(O2 /N2) has been ascribed to the transport of urban air influenced by human activities. In the summer, a clear diurnal cycle was observable for both the atmospheric δ(O2/N2) and CO2 concentration, due mainly to plant activities near the site. The average −O2: CO2 exchange ratio over the summer periods of 2007̶2009 was found to be −1.08 ± 0.09 ppm/ppm for the daytime and −1.08 ± 0.09 ppm/ppm for the nighttime; these values are in excellent agreement with −1.10 ± 0.05 ppm/ppm reported by previous studies. For a better understanding not only of spatiotemporal variations of atmospheric O2 but also of the global carbon cycle, high precision continuous measurements of atmospheric δ(O2/N2) and CO2 concentration need to
Vol. 91, No. 2
be carried out systematically at many places, especially under baseline conditions. For that purpose, we also have started systematic and continuous observations with our system at Japanese Antarctic station, Syowa in January 2008 and at a small island in the northeastern part of the main island of Japan, Enoshima Island in October 2008. Their results will be soon published elsewhere. Acknowledgments We are grateful to H. Yashiro and T. Umezawa for their cooperation in developing the O2 measurement system and performing the continuous observations at Aobayama. This study was partly supported by the GRENE Arctic Climate Change Research Project, the JSPS Grants-in-Aid for Creative Scientific Research (2005/17GS0203), and the MEXT subsidized project “Formation of a virtual laboratory for diagnosing the earthʼs climate system”. References Battle, M., S. M. Fletcher, M. L. Bender, R. F. Keeling, A. C. Manning, N. Gruber, P. P. Tans, M. B. Hendrics, D. T. Ho, C. Simonds, R. Mika, and B. Paplawsky, 2006: Atmospheric potential oxygen: New observations ant their implications for some atmospheric and oceanic models. Global Biogeochem. Cycles, 20, GB1010, doi: 10.1029/2005GB002534. Bender, M. L., P. P. Tans, J. T. Ellis, J. Orchardo, and K. Habfast, 1994: High precision isotope ratio mass spectrometry method for measuring the O2 /N2 ratio of air. Geochim. Cosmochim. Acta, 58, 4751̶4758. Bender, M. L., D. T. Ho, M. B. Hendricks, R. Mika, M. O. Battle, P. P. Tans, T. J. Conway, B. Sturtevant, and N. Cassar 2005: Atmospheric O2/N2 changes, 1993̶2002: Implications for the partitioning of fossil fuel CO2 sequestration, Global Biogeochem. Cycles, 19, GB4017, doi:10.1029/2004GB002410. Blaine, T. W., R. F. Keeling, and W. J. Paplawsky, 2006: An improved inlet for precisely measuring the atmospheric Ar/N2 ratio, Atmos. Chem. Phys., 6, 1181̶1184. Brenninkmeijer, C. A. M., and T. Röckmann, 1996: Russian doll type cryogenic traps: Improved design and isotope separation effects. Anal. Chem., 68, 3050̶3053. Ishidoya, S., S. Aoki, and T. Nakazawa, 2003: High precision measurements of the atmospheric O2/N2 ratio on a mass spectrometer. J. Meteorol. Soc. Japan, 81, 127̶140. Ishidoya, S., S. Sugawara, G. Hashida, S. Morimoto, S. Aoki, T. Nakazawa, and T. Yamanouchi, 2006: Vertical profiles of the O2/N2 ratio in the stratosphere over Japan and Antarctica. Geophys. Res. Lett., 33, L13701, doi: 10.1029/2006GL025886. Ishidoya, S., S. Sugawara, S. Morimoto, S. Aoki, and T. Nakazawa, 2008a: Gravitational separation of major
April 2013
D. GOTO et al.
atmospheric components of nitrogen and oxygen in the stratosphere. Geophys. Res. Lett., 35, L03811, doi: 10.1029/2007GL030456. Ishidoya, S., S. Morimoto, S. Sugawara, T. Watai, T. Machida, S. Aoki, T. Nakazawa, and T. Yamanouchi, 2008b: 15 Gravitational separation suggested by O2/N2, d N of N2, 18 d O of O2, Ar/N2 observed in the lowermost part of the stratosphere at northern middle and high latitudes in the early spring of 2002. Geophys. Res. Lett., 35, L03812, doi:10.1029/2007GL031526. Ishidoya, S., S. Aoki, D. Goto, T. Nakazawa, S. Taguchi, and P. K. Patra, 2012: Time and space variations of the O2/N2 ratio in the troposphere over Japan and estimation of the global CO2 budget for the period 2000̶2010. Tellus, 64B, 18964. http://dx.doi.org/10.3402/tellusb. v64i0.18964. Keeling, R. F., 1988: Development of an interferometric oxygen analyzer for precise measurement of the atmospheric O2 mole fraction. Ph. D. thesis, Harvard University, 178 pp. Keeling, R. F., and S. R. Shertz, 1992: Seasonal and interannual variations in atmospheric oxygen and implications for the global carbon cycle. Nature, 358, 723̶727. Keeling, R. F., R. P. Najjar, M. L. Bender, and P. P. Tans, 1993: What atmospheric oxygen measurements can tell us about the global carbon cycle. Global Biogeochem. Cycles, 7, 37̶67. Keeling, R. F., B. B. Stephens, R. G. Najjar, S. C. Doney, D. Archer, and M. Heimann, 1998: Seasonal variations in the atmospheric O2/N2 ratio in relation to the kinetics of air-sea gas exchange. Global Biogeochem. Cycles, 12, 141̶163. Keeling R. F., A. C. Manning, W. J. Paplawsky, and A. C. Cox, 2007: On the long-term stability of reference gases for atmospheric O2/N2 and CO2 measurements. Tellus, 59B, 3̶14. Langenfelds, R. L., R. Francey, L. P. Steele, R. F. Keeling, M. Bender, M. Battle, and W. F. Budd, 1999: Measurements of O2/N2 ratio from the Cape Grim air archive and three independent flask sampling program. Baseline atmospheric program (Australia) 1996, 57̶70. Lueker, T. J, 2004: Coastal upwelling fluxes of O2, N2O, and CO2 assessed from continuous atmospheric observations at Trinidad, California. Biogeosciences, 1, 101̶111. Machta, L., and E. Hughes, 1970: Atmospheric oxygen in 1967 to 1970, Science, 168, 1582̶1584. Manning, A. C., R. F. Keeling, and J. P. Severinghaus, 1999: Precise atmospheric oxygen measurements with a paramagnetic oxygen analyzer. Global Biogeochem. Cycles, 13, 1107̶1115. Manning, A. C., and R. F. Keeling, 2006: Global oceanic and land biotic carbon sinks from the Scripps stmospheric oxygen flask sampling network. Tellus, 58B, 95̶116. Minejima, C., M. Kubo, Y. Tohjima, H. Yamagishi, Y. Koyama, S. Maksyutov, K. Kita, and H. Mukai, 2012:
191
Analysis of ΔO2 /ΔCO2 ratios for the pollution events observed at Hateruma Island. Japan. Atmos. Chem. Phys., 12, 2713̶2723. Severinghaus, J. P., 1995: Studies of the terrestrial molecular oxygen and carbon cycles in sand dune gases and in Biosphere 2. Ph. D. thesis, Columbia University. Stephens, B. B., R. F. Keeling, and W. J. Paplawsky, 2003: Shipboard measurements of atmospheric oxygen using a vacuum-ultraviolet absorption technique. Tellus, 55B, 857̶878. Stephens, B. B., P. S. Bakwin, P. P. Tans, R. M. TecLaw, and D. Baumann, 2007: Application of a differential FuelCell Analyzer for Measurung Atmospheric Oxygen Variations. J. Atmos. Onceanic Thech., 24, 82̶94. Stephens, B. B., S. R. Sherzts, A. S. Watt, J. D. Bent, R. F. Keeling, S. C. Wofsy, B. C. Daube, R. Jimenez, E. A. Kort, and S. E. Mikaloff-Fletcher, 2009: Airborne observation of atmospheric O2 and CO2 on regional to th global scale. Proceedings of the 8 International Carbon Dioxide Conference. Tanaka, M., T. Nakazawa, and S. Aoki, 1983: High-quality measurement of the concentration of atmospheric carbon dioxide. J. Meteorol. Soc. Japan, 61, 678̶685. Thompson, R. L., A. C. Manning, D. C. Lowe, and D. C. Wetherburn, 2007: A ship-based methodology for high precision atmospheric oxygen measurements and its application in the Southern Ocean region. Tellus, 59B, 643̶653. Thompson, R. L., M. Gloor, A. C. Manning, D. C. Lowe, C. Rödenbeck, and C. Le Quéré, 2008: Variability in atmospheric O2 and CO2 concentrations in the southern Pacific Ocean and their comparison with model estimates. J. Geophys. Res., 113, G02025, doi:10.1029/ 2007JG000554. Tohjima, Y., 2000: Method for measuring changes in the atmospheric O2 /N2 ratio by a gas chromatograph equipped with a thermal conductivity. J. Geophys. Res., 105, 14575̶14584. Tohjima, Y., H. Mukai, T. Machida, and Y. Nojiri, 2003: Gaschromatographic measurements of the atmospheric oxygen/nitrogen ratio at Hateruma Island and Cape Ochi-ishi, Japan. Geophys. Res. Lett., 30, 1653, doi: 10.1029/2003GL017282. Tohjima, Y., H. Mukai, T. Machida, Y. Nojiri, and M. Gloor, 2005: First measurements of the latitudinal atmospheric O2 and CO2 distributions across the western Pacific. Geophys. Res. Lett., 32, L17805. Tohjima, Y., H. Mukai, Y. Nojiri, H. Yamagishi, and T. Machida, 2008: Atmospheric O2/N2 ratio at two Japanese sites: esimation of global oceanic and land biotic carbon sinks and analysis of the variations in atmospheric potential oxygen (APO). Tellus, 60B, 213̶ 225. van der Laan-Luijkx, I. T, R. E. M. Neubert, S. van der Laan, and H. A. J. Meijer, 2010: Continuous measurements of the atmospheric oxygen and carbon dioxide on a North
192
Jounal of the Meteorological Society of Japan
Sea gas platform. Atmos. Meas. Tech., 3, 113̶125. Yamagishi, H., Y. Thojima, H. Mukai, and K. Sasaoka, 2008: Detection of regional scale sea-to-air oxygen emission related to spring bloom near Japan by using in-situ measurements of the atmospheric oxygen/nitrogen ratio. Atmos. Chem. Phys., 8, 3325̶3335.
Vol. 91, No. 2
Yamagihsi, H., Y. Tohjima, H. Mukai, Y. Nojiri, C. Miyazaki, and K. Katsumata, 2012: Observation of atmospheric oxygen/nitrogen ratio aboard a cargo ship using gas chromatography/thermal conductivity detector. J. Geophys. Res., 117, D04309, doi:10.1029/2011JD016939.
