2.2. CARBON DIOXIDE
2.2.1. IN SITU CARBON DIOXIDE MEASUREMENTS
The mixing ratio of atmospheric CO2 was measured with continuously
operating nondispersive infrared (NDIR) analyzers at the four CMDL observatories
during 1994 and 1995 as in previous years. Monthly and annual mean CO2
concentrations (in the WMO 1993 mole fraction scale (X93)) are given in Table
2.1. These values are provisional, pending final calibrations of station standards.
Preliminary selected monthly average CO
mixing ratios for the entire record through 1995 are plotted versus time for
the four observatories in Figure 2.1.
TABLE 2.1. Provisional 1994 and 1995 Monthly Mean CO2 Mixing
Ratios From Continuous Analyzer Data
(micromol/mol, abbreviated as ppm, relative to dry air WMO X85 mole fraction scale)
Fig. 2.1. Preliminary selected monthly mean CO2 mixing ratios
expressed in micromol/mol at the four CMDL observatories.
The CO2 in situ systems operated during 1994 and 1995 at 94.0% and 91.2% at BRW; 94.2% and 95.1% at MLO; 91.0% and 78.7% at SMO; and 90.9% and 92.9% at SPO. The maximum percentage expected is 95.8% based on missing data due to reference gas calibrations during the year. The majority of the loss of data at SMO in 1995 was due to a failure in the CO2 NDIR analyzer in November that lasted until late December.
A new data acquisition and control system was installed at MLO in December 1995. This system uses a HewlettPackard Unix workstation for controlling not only the CO2 NDIR measurements, but the CH4 and CO in situ gas chromatograph systems as well. Data are downloaded from MLO to Boulder daily over the Internet, as well as recorded on optical disks at MLO for backup. New data acquisition and control systems will be installed at the remaining observatories during 1996.
In addition to the new data system, the CO2 NDIR analyzer was fitted with smaller optical cells, 60 mm in length compared with the original 180 mm length cells. The glass H2O cryotrap was relocated between the inlet air pumps and the gas manifold with a smaller auxiliary cryotrap added in between the gas manifold and the NDIR analyzer. With this setup, the first cryotrap dried only the ambient air samples, and the second cryotrap dried the reference gases and the dried ambient air. The volume of gas that the reference gases need to flush away after each gas change was greatly reduced so that lower flow rates were possible. The flow rate was reduced to ~150 cc min-1 from 300 cc min-1.
2.2.2. FLASK SAMPLE CARBON DIOXIDE MEASUREMENTS
Measurements of the distribution and variations of atmospheric CO2 continued during 1994 and 1995 using samples collected throughout the CMDL global air sampling network. In January and February 1994, sampling began at Easter Island, Chile (29°09'S, 109°26'W; site code: EIC) and Ny-Alesund, Svalbard (78°54'N, 11°53'E; ZEP), respectively. The flask sampling at Easter Island is through the cooperation of the Chilean Meteorological Service. Ny-Alesund is a collaboration with the Stockholm University in situ CO2 measurement program and is intended to study the North Atlantic/subarctic marine sink for CO2. In September 1994, flask sampling began near Ushuaia, Argentina (54°52'S, 68°29'W; TDF) in support of the new Global Atmosphere Watch (GAW) observatory. In October 1994, sampling was initiated at Constanta, Romania (44°10'N, 28°41'E; BSC) on the western shore of the Black Sea, and the sampling program on Terceira Island, Azores (38°46'N, 27°23'W; AZR) was revived after a 2-year interruption. In September 1995 flask sampling began at Assekrem, Algeria (23°11'N, 5°25'E; ASK) in cooperation with the GAW observatory at Tamanrasset, Algeria. Finally, in November 1995 CCG began receiving samples collected in the Negev Desert, Israel (31°08'N, 34°53'E; WIS) in cooperation with the Weizmann Institute of Science. These new sites are shown with the rest of the air sampling network in Figure 2.2. Annual mean mixing ratios for 41 sites for 1993, 1994, and 1995 are given in Table 2.2. The 1995 values are based on preliminary editing and data selection.
Fig. 2.2. Network of continuing measurements by the Carbon Cycle Group.
TABLE 2.2. Provisional 1993-1995 Annual Mean CO2 Mixing Ratios From Network Sites
|ALT||Alert, N.W.T., Canada||357.7||359.8||361.1|
|AZR||Terceira Island, Azores||[ ]||[ ]||359.5|
|BME||Bermuda (east coast)||356.8||358.8||361.3|
|BMW||Bermuda (west coast)||357.3||359.8||361.0|
|BSC||Constanta, Romania||[ ]||364.6|
|CBA||Cold Bay, Alaska||357.8||359.1||361.4|
|CGO||Cape Grim, Tasmania||354.5||356.1||357.9|
|CHR||Christmas Island||357.4||[ ]||[ ]|
|CMO||Cape Meares, Oregon||358.5||361.7||[ ]|
|EIC||Easter Island, Chile||355.7||357.6|
|GMI||Guam, Mariana Islands||356.6||358.5||360.6|
|GOZ||Gozo Island, Malta||[ ]||359.6||362.2|
|HBA||Halley Bay, Antarctica||355.1||356.9||358.1|
|HUN||Hegyhatsal, Hungary||[ ]||362.2||366.6|
|IZO||Izana Observatory, Tenerife||357.5||358.6||361.4|
|KEY||Key Biscayne, Florida||358.4||359.3||362.1|
|KUM||Cape Kumukahi, Hawaii||357.1||359.1||360.9|
|MBC||Mould Bay, Canada||357.8||359.9||361.3|
|MHT||Mace Head, Ireland||356.7||358.6||360.7|
|MLO||Mauna Loa, Hawaii||356.9||358.5||360.6|
|NWR||Niwot Ridge, Colorado||357.4||359.5||361.2|
|PSA||Palmer Station, Antarctica||355.1||356.4||358.1|
|QPC||Qinghai Province, China||357.3||359.3||[ ]|
|RPB||Ragged Point, Barbados||356.7||358.0||360.2|
|SEY||Mahe Island, Seychelles||356.0||356.5||358.1|
|SHM||Shemya Island, Alaska||357.7||360.6||360.9|
|SPO||South Pole, Antarctica||354.8||356.2||357.7|
|STM||Ocean Station M||357.5||359.2||360.4|
|SYO||Syowa Station, Antarctica||354.5||356.1||358.1|
|TAP||Tae-ahn Peninsula, S. Korea||360.4||361.2||363.6|
|TDF||Tierra del Fuego, Argentina||[ ]||[ ]|
|UTA||Wendover, Utah||[ ]||361.2||361.2|
|UUM||Ulaan Uul, Mongolia||357.1||359.3||360.3|
The 1994 and 1995 annual means have been adjusted upward by 0.25 ppm to correct
for a systematic loss of CO2 in the flask analysis apparatus.
Air samples were collected in evacuated flasks at 5 degree latitude intervals
over the Pacific Ocean aboard the California Star (OPC) during 1993 through
1995. In 1995 sampling began on a second ship, the Brisbane Star (OPB).
Annual averages calculated from merged data from both ships (POC) are given
for 14 latitude intervals in Table 2.3. Flask samples were also collected from
ships in the South China Sea (SCS). Annual averages for seven 3 degree latitude
intervals are given in Table 2.4.
TABLE 2.3. Provisional 1993-1995 Annual Mean CO2 Mixing Ratios from Pacific Ocean Cruises
TABLE 2.4. Provisional 1993-1995 Annual Mean CO2 Mixing Ratios from South China Sea
The globally-averaged CO2 growth rate determined from the air sampling
network data is shown in Figure 2.3. The CO2 growth rate declined
from a high of ~2.6 mmol (abbreviated ppm) per year
in 1987 to a low of ~0.6 ppm yr-1 in 1992. The growth rate in 1994
was above the 1981-1995 average of ~1.4 ppm yr-1, and in 1995 the
growth rate was still above the decadal average.
Fig. 2.3. Global CO2 growth rate.
A two-dimensional model was developed to use the flask CO2 data, together with measurements of the 13C/12C ratio in CO2 from the same air samples, to partition CO2 sources and sinks into terrestrial and marine components as a function of latitude [Ciais et al., 1995a]. An application of this model to data through 1993 attributed a large fraction of the northern hemisphere sink to the terrestrial biosphere [Ciais et al., 1995b]. This is an important result because it is not known which processes account for this sink and also because carbon stored as biomass has the potential to be returned to the atmosphere on short time scales. A preliminary application of the model through 1995 shows that the total global CO2 sink in 1992 was a factor of 2 larger than in 1995. This analysis also showed significant interannual variations in both the marine and terrestrial components of this sink.
2.2.3. CARBON DIOXIDE REFERENCE GAS CALIBRATIONS
The calibration of CO2inair reference gas tanks continued in 1995; 407 tanks were calibrated using the NDIR analyzer. All CO2inair reference gas tanks used by CMDL are filled with clean dry ambient air from Niwot Ridge, Colorado, in aluminum tanks.
A manometric system was developed for performing primary calibrations of the absolute mole fraction of CO2 in a carrier (air) gas. The manometric apparatus itself is principally made of glass. It consists essentially of a 6-L glass flask connected by means of a manifold to a 10-mL glass small volume. A pressure gauge of quartz spiral Bourdon tube type as a primary manometer in the manometric system is used to measure the CO2 and the carrier gas pressures precisely. The apparatus is enclosed in an oven. The temperature of the oven is uniform and controlled to an accuracy of 0.01°C during the calibration process. The quartz spiral pressure gauge is regularly calibrated with a dead-weight pressure calibration tester.
To begin the manometric calibrating process the glass manometric chamber, including the 6-L large volume, the manifold, and 10-mL volume, are evacuated to a residual pressure of less than 1 millitorr. A sample of air from a cylinder to be analyzed is dried at 70°C to remove water vapor and then fills the evacuated glass chamber to the ambient pressure. After the sample air in the chamber has come to equilibrium, the temperature is measured by platinum resistance thermometers, and pressure is measured by the quartz spiral gauge. The sample air from the chamber is then slowly pumped out through two liquid nitrogen traps, freezing out CO2, N2O, and residual water vapor. Upon completion of the extraction, the CO2 frozen in the traps is dried with a dry icealcohol mixture and then transferred to the small 10 mL volume by placing liquid nitrogen around it. After the 10-mL small volume containing the collected pure CO2 (and N2O) is thawed, the temperature and pressure are continuously measured while equilibrium is reached. Because the volume ratio of the small and the large volumes is known accurately, the molar ratio of the CO2 in the original air sample can be calculated with the virial equation of state, taking real gas compressibility into account, and correcting for the N2O contribution.
From December 1995 to February 1996, the CO2 concentrations of three cylinders with CO2inair mixtures were determined by the manometric calibration system. The results of the tests are presented in Table 2.5. For comparison, the CO2mole fractions measured by a NDIR analyzer using reference gases calibrated by the Scripps Institution of Oceanography (SIO) are also shown in the table. The reproducibility of the manometric system indicated in Table 2.5 as the standard deviation is about 0.06 mol for a total of 20 measurements. The largest mean difference of measurements between the NDIR and the manometric system is 0.08 m for the three CO2inair mixture cylinders.
TABLE 2.5. Results of Tests Using Manometric Calibration System
|Dec. 6, 1995||71568||386.33|
|Dec. 6, 1995||71568||386.30|
|Dec. 8, 1995||71568||386.37||386.33||0.035||386.25||0.08|
|Feb. 1, 1996||56797||352.67|
|Feb. 2, 1996||56797||352.78|
|Feb. 5, 1996||56797||352.83|
|Feb. 13, 1996||56797||352.62|
|Feb. 16, 1996||56797||352.77|
|Feb. 17, 1996||56797||352.77|
|Feb. 19, 1996||56797||352.76|
|Feb. 20, 1996||56797||352.62|
|Feb. 20, 1996||56797||352.77|
|Feb. 21, 1996||56797||352.63|
|Feb. 21, 1996||56797||352.89||352.74||0.090||352.77||0.03|
|Feb. 23, 1996||114997||314.97|
|Feb. 24, 1996||114997||314.95|
|Feb. 26, 1996||114997||315.02|
|Feb. 28, 1996||114997||314.90|
|Feb. 28, 1996||114997||315.02|
|Feb. 29, 1996||114997||315.08||314.99||0.063||315.05||0.06|
2.2.4. MEASUREMENTS OF STABLE ISOTOPES OF CO2
Since 1990, the Stable Isotope Laboratory at INSTAAR has been measuring the stable isotopic composition of CO2 from flask samples from the CMDL global air sampling network. The natural ratio of 13C to 12C is about 1.1% everywhere, but biogeochemical processes (such as photosynthesis or atmosphereocean exchange) can sustain small but readily measurable differences in that ratio between different carbon reservoirs. For example, plants discriminate against 13C during photosynthetic uptake, therefore the 13C/12C ratio in plant carbon (and, by derivation, in soils and fossil fuels) is depleted relative to the atmosphere, typically by about 20 (per mil, or parts per thousand)which in turn leaves the atmosphere subtly enriched in 13C. Observing such a 13C signature allows exchanges of CO2 with the biosphere to be distinguished from oceanic fluxes because the latter do not carry a significant isotopic signature [e.g., Keeling et al., 1995]. The 18O composition of atmospheric CO2 ultimately derives from its equilibration with liquid water, providing a link between the global carbon and hydrologic cycles. For example, CO2 exposed to water within the leaves of plants, but diffusing out of the leaf before being incorporated, carries the isotopic signature of leaf water back to the atmosphere. Because CO2 "remembers" the 18O signature of the water reservoir it has most recently visited, this tracer may prove to be useful in quantifying the gross annual uptake of CO2 by photosynthesis and its release by respiration.
INSTAAR currently measures d13C (the normalized difference between the isotopic ratios of a sample and standard) and d18O in CO2 for almost all of the CMDL network flasks, having begun with a selection of only six sites and two ships in 1990. The growth of the effort, reflected by the number of sites and flasks measured each year, is shown in Figure 2.4. Each measurement is made by first cryogenically extracting CO2 from about 750 standard cm3 of dried air, then measuring the relative abundance of isotopic species of masses 44, 45, and 46 using a triplecollector isotoperatio mass spectrometer [Trolier et al., 1996]; precisions of 0.03 and 0.06 are obtained for d13C and d18O respectively. Small numerical corrections account for the presence of N2O trapped with CO2 and for the presence of isotopic species including 17O. The isotopic data are fully integrated into the Carbon Cycle Group's trace gas data base.
Fig. 2.4. Statistics of the isotope measuring effort. The extent of the NOAA CO2 monitoring program is shown for comparison. Solid bars represent land sites, hatched bars represent latitude bands from shipboard sampling. (a) Number of sites measured during each year. (b) Number of flasks analyzed each year. (c) Percentage of "good" flask pairs for d13C and d18O.
A sample of the isotope data is given in Figure 2.5, which shows time series of CO2 mixing ratio, d13C, and d18O from Barrow, Alaska, from 1990 through 1995. There is a striking anticorrelation between the seasonal cycles of mixing ratio and <d13C, reflecting the strong influence of the annual cycle of photosynthesis and respiration imposed on the atmosphere by the terrestrial biosphere in the northern hemisphere. Whereas the mixing ratio shows an increasing longterm trend due to the use of fossil fuels, the trend of d13C is to lighter values, reflecting the depletion in d13C of fossil fuel relative to the atmosphere. The seasonal cycle of d18O lags behind CO2 and d13C, and while its interannual variability shows no steady trend, it can change its level dramatically from year to year, most likely because of the large exchanges of CO2 between biosphere and atmosphere that are subject to efficient oxygen isotope exchange.
Fig. 2.5. Time series of CO2 (upper panel), d13C of CO2 (middle panel), and d18O (lower panel) from Point Barrow, Alaska.
The INSTAAR isotope data were recently described in detail [Trolier et al., 1996]. The INSTAAR d13C time series, though beginning only in 1990, were used in concert with the longer d13C time series from Cape Grim, Australia, obtained by CSIRO, to identify a global flattening of the longterm d13C trend during 19881992 [Francey et al., 1995a]. The decadal average trend observed during the 1980s, about 0.025 yr-1, was apparently offset during these years by anomalously high uptake of CO2 by the global biosphere. Similarly, the INSTAAR d13C data, supplemented by CSIRO data from the southern hemisphere, definitively identify a strong northern hemisphere biospheric sink equivalent to nearly half the annual anthropogenic source during 1992 and 1993 [Ciais et al., 1995b]. Interpretive work using the d18O data is underway. In addition to these scientific analyses, the measurements are actively intercompared with other atmospheric monitoring laboratories measuring CO2 isotopic composition [Francey et al., 1995b; Gaudry et al., 1996].
INSTAAR recently obtained a more precise isotoperatio mass spectrometer, a VG Optima, for analyzing the CMDL flasks. The instrument is currently being tested and it will come on-line for flask analysis during 1996. This instrument will be devoted entirely to analysis of atmospheric samples and will allow us to focus more attention on calibration. It is expected that the Optima will improve the analytical precision by about a factor of 3.
2.2.5. THE AIRKIT SAMPLER
Field testing of a new prototype air sampling apparatus began at SMO in September 1994 and Cape Kumukahi, Hawaii (KUM) in May 1995. The new Airkit (Air Kitzis sampler) differs from the currently used MAKS (Martin and Kitzis Sampler) in two important ways: (1) It has a thermoelectrically cooled condenser to remove water vapor from the air stream, and (2) It has a microprocessor to control the sampling process so that collecting the sample is more automated and less subject to operator error. The effect of drying the air sample is most dramatic for the measurement of 18O/16O in CO2 (Figures 2.6 and 2.7). In samples collected at humid, tropical locations without drying, the 18O/16O measurements are highly variable and consistently more depleted in 18O due to the exchange of oxygen atoms between CO2 and H2O molecules. It was established through systematic tests at INSTAAR [Gemery, 1993] that the exchange takes place during storage in the flasks when the relative humidity of the air sample is above 50%. Overlapped sampling with the Airkit and MAKS at SMO and KUM shows that this effect is eliminated with the Airkit and that the measurement of other species is not affected by the drying (Table 2.6). The pair agreement improves from 0.73 (1s) to 0.09.
Fig. 2.6. Oxygen-18 in flask samples from Cape Matatula, Samoa. The
majority of the "wet" samples were rejected due to poor pair agreement. The
few fortuitously retained pairs tend to be isotopically "light." Since samples
have been dried with the Airkit, however, a first glimpse of the true d18O
signature of CO2 at equatorial latitudes from the CMDL network has
Fig. 2.7. Oxygen-18 in flask samples from Cape Kumukahi, Hawaii. The
comparison between Airkit and MAKS is very similar to Samoa.
TABLE 2.6. Comparison of Airkit to MAKS Sampler
|Site||Species||No. Pairs||Average||Standard Deviation||Units|
2.2.6. CALIBRATION OF MEASUREMENTS OF STABLE ISOTOPES OF CO2
The INSTAAR stable isotope data are reported as isotopic composition relative to VPDBCO2 for both d13C and d18O. Calibration of this record has two distinct facets. The first relies on comparatively precise intercomparisons of samples of CO2 extracted from air. Carbon dioxide from individual flask samples is always compared with CO2 from a "working reference" cylinder. The isotopic composition of the working reference cylinder itself is currently tracked on a monthly basis by comparison with a suite of secondary reference cylinders. The second facet of calibration consists of the comparison of the isotopic scale established by the reference cylinders to accepted international standards ("absolute" calibration). These measurements require the comparison of CO2 derived from different materials using different preparation systems and to date have been subject to larger uncertainties. Both carbonate standards (NBS19, NBS20, and INSTAAR laboratory standards) and water standards (VSMOW, SLAP, and laboratory standards) have been used for absolute calibrations.
Comparisons among the reference cylinders, though few in the early years of the program, have expanded to provide a fairly strong constraint on the consistency of the working reference gas scales for d13C and d18O. Working reference gas cylinders are used for up to 3 years; transitions between working reference cylinders provide the unfortunate possibility of a step shift in calibration. Such shifts could be too subtle to detect by tooseldom and toonoisy comparisons with carbonates and waters but can be tracked by other cylinders (provided, of course, that the suite of cylinders is not drifting in parallel).
Figure 2.8 shows a summary of the calibration data available for the INSTAAR isotope data for <d13C and d18O. The central feature of these plots are the scales defined by the sequence of working reference gases shown by solid lines plotted for each cylinder during its lifetime. This scale determines the values that are assigned to individual flask samples. Dotted lines represent the cylinders' values outside their period of use as the working reference. The relative values of any two working reference cylinders are determined by extensive inter-comparison preceding the transition. Also shown are the values for the working cylinders determined from other reference gases (open symbols) and carbonate or water standards (closed symbols); error bars (1s) are shown where more than one determination is made in a month. Uncertainties clearly persist, particularly for the absolute assignment of the d18O scale. The assigned scales are to some extent based on subjective evaluations of the reliability of different calibration methods. However, drifts and step shifts within the working reference scales seem unlikely beyond (pessimistically) 0.02U for d13C and 0.04 for d18O.
Fig. 2.8. Calibration of the NOAA-CU data set. The upper panel shows
calibrations for d13C, the lower for d18O.
Reference gases are identified by cylinder name. Monthly means of calibrations
of reference gases are plotted using open symbols; where more than one determination
was made in a month, error bars are plotted at twice the standard deviation
of the monthly values. Mean values for working reference gases are plotted as
thick solid lines, for secondary reference gases as thin lines. Calibrations
of the working reference scale using primary isotope reference materials are
plotted as solid symbols (NBS-19, squares; NBS-20, diamonds; SMOW, circles).