A Preliminary Comparison of 13C Measurements in CO2
The 13C composition of atmospheric CO2 makes it possible to estimate biospheric fluxes of CO2, as plant photosynthesis discriminates against 13CO2, whereas isotopic fractionation during CO2 dissolution into the ocean is small. However, to infer the partitioning of anthropogenic CO2 between its oceanic and terrestrial sinks requires measurements of very high precision. Fluxes of CO2 that are readily measured via changes in the atmospheric CO2 mixing ratio have a smaller impact on 13C. For instance, adding 1 ppm (analytical precision is ~0.1 ppm) of purely "biogenic" CO2 to the atmosphere changes the 13C of that reservoir by only 0.05 (analytical precision is ~0.02). The isotopic effect of adding 1 ppm of "oceanic" CO2 is even smaller.
Although the internal precision of mass spectrometers is about 0.01, systematic experimental errors may bias measurements of the 13C of air samples [Francey et al., 1995]. Drift of the reference gases used to calibrate measurements can also introduce biases [Trolier, 1994]. Intercomparisons between different laboratories using independent calibration strategies and experimental protocols are crucial in order to assimilate their various data for use in models [Ciais et al., 1995]. One of the best ways to monitor the intercalibration of independent programs is to conduct parallel sampling of whole air on an ongoing basis and at a common site.
In this paper we report a preliminary intercomparison of 13C measurements of air samples independently collected at Mace Head Station (53.43ºN; -9.90ºW) in the North Atlantic by CMDL and by the Centre des Faibles Radioactivités at the Laboratoire de Modélisation du Climat et de L'Environnement (CFR-LMCE; France). The CMDL samples are measured for isotopic composition at the Institute of Arctic and Alpine Research (INSTAAR) at the University of Colorado. Mace Head is located in the vicinity of the North Atlantic oceanic sink of CO2 [Lefèvre, 1996], but it is also reached by continental air, especially in winter. The seasonal cycle of 13C at Mace Head has a fairly large amplitude (0.9), so 13C values can be compared over a rather wide range, from about -8.3 in winter to about -7.4 in spring and summer. The results of this intercomparison are summarized and discussed below.
CFR-LMCE began continuous atmospheric CO2 monitoring at Mace Head station in cooperation with the University College of Galway (Ireland) and International Science Consultants (United Kingdom) in July 1992. Meteorological data are also continuously recorded, enabling "background" sampling conditions to be distinguished from conditions that are influenced by local sources. Beginning in July 1993, flask samples (2-L glass flasks with Viton O-ring stopcocks, filled to 1 bar with air dried to a dewpoint of -55ºC) have been measured for the 13C composition of CO2. The usual sampling frequency is two pairs of flasks per month. During the period from May 31, 1993, to June 7, 1993, the sampling frequency was about one pair per day; 13C data for these samples are not shown here. A different standard was used for the isotopic analysis which increases the uncertainty.
CFR-LMCE has operated an isotope-ratio mass spectrometer (Finnigan MAT 252) since January 1993, collaborating with CSIRO (where a similar instrument is used) to characterize the effects of instrumental artifacts on 13C measurements. The recognized effects include [Francey et al., 1995] memory effects (when a sample has a very different 13C than the reference gas), size effects, which can affect linearity, and the "bleed correction," due to the fractionation that results from consumption of the reference gas. CFR recently made an accurate determination of the N2O interference. The CFR flask measure-ments are calibrated against one sample of carefully purified CO2, called SNAIL. SNAIL was calibrated against NBS-19, a carbonate provided by International Atomic Energy Agency (IAEA) [Hut, 1987] and two standard pure CO2 gases, GS19 and GS20, provided by the University of Groningen, Netherlands. The 13C of SNAIL standard was determined to -9.67 ± 0.04 (relative to VPDB-CO2).
NOAA began sampling at Mace Head in June 1991; the flasks (2.5-L glass flasks with Teflon o-ring stopcocks) are sampled in pairs and analyzed for the mixing ratios of CO2, CH4, and CO by CMDL, and for 13C and 18O of CO2 by INSTAAR. At INSTAAR, CO2 (with N2O) is extracted from about 750 bar·cm3 of whole air cryogenically, then analyzed for isotopic composition using a VG Sira Series II isotope-ratio mass spectrometer. The raw data (ratios of ion currents at masses 45 and 46 to mass 44) are corrected for the presence of N2O and for the contribution of species containing 17O to the ion currents. The experimental technique and data analysis are described by Trolier et al. . The INSTAAR flask measurements are calibrated against a suite of whole-air reference gases, which in turn are calibrated against VPDB-CO2 and V·SMOW. The estimated precision of individual 13C and 18O measurements are 0.03 and 0.05 respectively; the uncertainties in the absolute calibrations are ~0.02 and ~0.1 respectively.
Figure 1 shows the time series of 13C measurements from Mace Head from both groups. Overall, the agreement appears to be good with the data sets showing no large offset and comparable seasonal cycles. We have made three quantitative comparisons of the two data sets: (1) directly comparing flask samples obtained close in time by the two programs; (2) comparing CFR data to a smoothed curve representing the entire INSTAAR data set; and (3) comparing CFR data obtained during "background" atmospheric conditions to the same smoothed curve from the INSTAAR data set. Background conditions correspond to winds higher than 4 m s-1 in the wind sector within 200° and 300°.
Fig. 1. Time series of 13C of atmospheric CO2 measured
from samples of air taken at Mace Head, Ireland, by the CMDL Cooperative Flask
Sampling Network (open circles) and CFR-LMCE (filled circles). The CMDL samples
were analyzed by INSTAAR at the University of Colorado.
Because the two flask sampling programs are independent, their flasks are not necessarily filled close together in time. For example, there are only six instances between May 1993 and January 1995 for which flasks were obtained by both groups on the same day; in these cases, the flasks were sampled less than 1 hour apart. The average difference for these samples (CFR - INSTAAR) is -0.04 ± 0.09 (the error estimate is the standard deviation, 1, of the differences). In an attempt to compare the two records for flasks sampled on different days, a smooth curve [Thoning et al., 1989] was used to represent the INSTAAR record for days on which samples were not available. The smoothed curve is obtained by first fitting a curve consisting of the sum of a third-order polynomial trend and four-harmonics to the flask data; the residuals are then filtered in the time domain using a low-pass filter with a full width at half maximum of 100 days, and the smoothed residuals are added to the fitted curve. The smoothed curve was fitted to the INSTAAR data, and differences were calculated between the original flask data (for both INSTAAR and CFR) and the smoothed curve. These differences are shown in Figure 2. The mean difference between the CFR 13C values and the INSTAAR smoothed curve is -0.05 ± 0.08 for 26 samples. This analysis has been repeated using only INSTAAR data obtained under "background" conditions as defined by CFR; this eliminates ~50% of the INSTAAR measurements. This comparison of background samples appears in Figure 3. In this case, the average difference is -0.03 ± 0.07 for 18 samples. Each comparison suggests a slight offset between the calibration scales of the two groups, although in no case is the discrepancy greater than the error of the comparison.
Fig. 2. Differences in 13C between the INSTAAR and CFR flask samples, and a smoothed curve fit through the INSTAAR time series. The INSTAAR data are shown with open circles, and the CFR-LMCE data by filled circles.
Fig. 3. Same as Figure 2, except that only CFR/LMCE data collected under "background" atmospheric conditions are used in the analysis of the differences.
A preliminary comparison between the time series of 13C measurements at Mace Head, obtained by LMCE and INSTAAR-CMDL, covering approximately two annual cycles, is encouraging. An apparent offset of about -0.03 ± 0.07 exists between the groups, with the CFR data being lighter. The limited duration of parallel sampling, and the small number of simultaneous samples, prevents a quantitative evaluation of drifts between the groups or a comparison of the range of the measurements observed during an annual cycle. While the intercomparison will be strengthened by continuing parallel sampling, simultaneous sampling and independent checks would be useful additions. The latter point will be partially addressed by a round-robin intercalibration being organized by the IAEA [Allison et al., 1994] involving groups from Australia (CSIRO); the United States (INSTAAR, CMDL, and SIO); France (CFR-LMCE); and others.
Allison C.E., R.J. Francey, L.P. Steele, and F. de Silva, The IAEA interlaboratory calibration program: Air standards for the IAEA round-robin exercise, in Final Report on the IAEA Coordinated Research Program on Isotope Variations of Carbon Dioxide and Other Trace Gases in the Atmosphere, edited by K. Rozanski, IAEA, Vienna, Austria, 1994.
Ciais P., P.P. Tans, M. Trolier, J.W.C. White, and R.J. Francey, A large northern hemisphere terrestrial CO2 sink indicated by 13C/12C of atmospheric CO2, Science, in press, 1996.
Francey R.J., C.E. Allison, L.P. Steele, R.L. Langenfelds, E.D. Welch, J.W.C. White, M. Trolier, P.P. Tans, and K.A. Masarie, Intercomparison of stable isotope measurements of CO2, in Climate Monitoring and Diagnostics Laboratory, No. 23: Summary Report 1994, edited by J.T. Peterson and R.M. Rosson, NOAA Environmental Research Laboratories, Boulder, CO, pp. 106-110, 1995.
Hut, G., Consultant's group meeting on stable isotope reference samples for geochemical and hydrological investigations, IAEA report, Vienna, September 1985, 1987.
Lefèvre, N., A first step towards a reference P map for the North Atlantic ocean, IGBP-10, working paper 11, 1995.
Thoning, K.W., P.P. Tans, and W.D. Komhyr, Atmospheric carbon dioxide at Mauna Loa Observatory, 2: Analysis of the NOAA/GMCC data, 1974-1985, J. Geophys. Res., 94, 8549-8565, 1989.
Trolier, M., Calibrating the INSTAAR-NOAA/CMDL record of stable isotopic composition of atmospheric CO2, in Final Report on the IAEA Coordinated Research Program on Isotope Variations of Carbon Dioxide and Other Trace Gases in the Atmosphere, edited by K. Rozanski, IAEA, Vienna, Austria, 1994.
Trolier, M., J.W.C. White, P.P Tans, K.A. Masarie, and P.A. Gemery, Monitoring the isotopic composition of atmospheric CO2: Measurements from the NOAA global air sampling network, J. Geophys Res., in press, 1996.