Global C02Measurements

The rest of this contribution concentrates on recent and potential progress in the measurement of atmospheric C02 mixing ratios and related species. The challenge for such measurement programs is to monitor, with high precision, the temporal changes and/or spatial gradients of CO, and related species. Conventional methodologies for monitoring atmospheric C02, developed over the past 40 years, show a number of shortcomings when examined in the light of the requirements for improved estimates of regional fluxes from baseline atmospheric-composition measurements.

Measurements of carbon dioxide mixing ratios are made at over 100 globally distributed "baseline" sites (i.e., fixed or mobile sites for which measurements reflect CO, behavior over large spatial scales). The requirement for large-scale representation has heavily influenced global sampling strategies insofar as the great majority of sampling sites is located to access marine boundary-layer air. In fact, for the smaller sampling networks, zonal representation was a common assumption. Furthermore, data are still generally selected to reinforce the marine boundary layer bias, though this is changing. Most results are now reported to one or more data banks, including the Carbon Dioxide Information Analysis Center (CDIAC) World Data Centre—A, for Atmospheric Trace Gases, established in 1982 by the Oak Ridge National Laboratory, Tennessee, and the World Meteorological Organisation (WMO) World Data Centre for Greenhouse Gases (WD-CGG) in the Japan Meteorological Agency, established in 1990. In late 1995, a Co-operative Atmospheric Data Integration Project (CADIP-C02) was commenced in the NOAA Climate Monitoring and Diagnostics Laboratory (CMDL), U.S.A., using data from much the same sources, with the aim of providing an integrated "globally-consistent" data set, GL0BALVIEW-C02, for modeling studies. At the heart of GLOBALVIEW is a data extension and integration technique (Masarie and Tans, 1995) that addresses difficulties such as those related to missing data or introduction of new stations. However, interlaboratory calibration remains a problem.

Around 17 different laboratories from 12 nations are involved in the measurement and reporting of C02 data to these data banks. Historically, the WMO has taken responsibility for the in-tercalibration of measurements in different laboratories. Primary activities have involved the establishment of a Central Calibration Laboratory to maintain and provide access to "primary" CO,-in-air standards measured with high-precision manometric techniques, and initiation of blind "round-robin" intercalibrations involving the circulation of high-pressure cylinders containing CO,-in-air among participating laboratories. In addition, the WMO has provided a forum of "CO, Measurement Experts," now held once every two years to assess progress and plan future activities, with each meeting producing a WMO technical report.

Results from two recent WMO C02 round robins are summarized in Figure 2, adopted from WMO technical reports (Pearman, 1993; Peterson, 1997). As an example of the procedure, the most recent round-robin, (b), was proposed at the July 1995 8th WMO CO,-Experts Meeting in Boulder, Colorado, and was completed in time for an initial assessment at the 9th Meeting of Experts on the Measurement of Carbon Dioxide Concentration and Associated Tracers (endorsed by International Atomic Energy Agency), Aspendale, Australia, 1-4 September 1997. NOAA CMDL prepared three sets of three cylinders of air with nominal CO, mixing ratios of 345, 360, and 375 ppm. Each set was distributed to one of three groups of around eight laboratories (in North America and the Southern Hemisphere, Asia, and Europe). A target inter-laboratory precision of 0.05 ppm was identified by this community to achieve a "network precision" of 0.1 ppm. This precision is appropriate for the merging of data from different sites to estimate regional fluxes via synthesis inversion studies (WMO, 1987). This level of precision is comparable to that of an individual measurement in the better operational systems; the "target" precision of 0.05 ppm refers more to the requirement for precise average temporal values (e.g., annual or seasonal) and for precise large-scale values (e.g., GCM grid scale to hemispheric). Note that Fig. 2 results usually represent the average of multiple measurements on a cylinder.

The most important point to be drawn from Fig. 2 is that there are significant (>0.05 ppm) and variable calibration differences between laboratories, which are not currently accounted for in the CDIAC and WDCGG data bases, or in the GLOBALVIEW data assimilation. Another general observation is that there is a significant overall improvement going from the first to second round-robin (while the actual laboratories are not identified, the identification of country is sufficient to make this inference). However, in the second, more precise intercalibration a new concern about linearity emerges, with a majority of participants measuring lower than the low-mixing-ratio tank and higher than the high-mixing-ratio tank. The fact that the degree of "nonlinearity" varies widely suggests that this is an issue for many laboratories; it also argues for an independent verification of both the manometric technique and the scale propagation, e.g., by using gravimetric dilution techniques.

The unsatisfactory situation for C02 mixing-ratio intercalibration is also evident for 5' 3C of C02. The International Atomic Energy Agency (IAEA) conducts a co-operative research program on "Isotope-aided Studies of Atmospheric Carbon Dioxide and Other Greenhouse Gases" with an objective of providing whole-air standards for the measurement of greenhouse gas isotopes. Figure 3 shows preliminary results from the first circulation of "CLASSIC" (Circulation of Laboratory Air Standards for Stable Isotope inter-Comparisons) standards, where the initial round-robin has been restricted to four laboratories with the longest involvement in sampling the background atmosphere from a network of stations. Here the community has set a required target precision of 0.0 Woo for temporal or large-scale averages, which is even more demanding than the case with C02 mixing ratios since the typical precision on an individual measurement is around 0.03%o.

341 ppm 349 ppm 375 ppm

341 ppm 349 ppm 375 ppm

343 ppm S 358 ppm 1375 ppm

343 ppm S 358 ppm 1375 ppm

FIGURE 2 WMO round-robin intercalibrations of C02 measurement laboratories (identified by country only). Plotted are measured differences from mixing ratios assigned by NOAA CMDL. Data for (a) a circulation conducted between 1991-93, and (b) between 1995-97.

Preliminary results of this round robin are given in Fig. 3 (Allison et al, in press). Measured differences are reported with respect to initial measurements conducted at CSIRO in November 1996. CSIRO(2) refers to CSIRO measurements conducted after circulation in July 1998, confirming the stability of the tank standards. Measurements on pure C02 samples scatter by about ± 0.02%o, outside the required target. For the analyses of the whole-air standards the situation is much worse, with reported values scattered over a range of almost ± 0.1 %o, suggesting serious differences between pretreatments to extract C02 from air. Furthermore, there also appears to be a linearity problem with the CSIRO measurement compared to the other three laboratories.

The situation is even more serious than indicated by the round-robin comparisons. Since 1992, with the aim of confirming our ability to merge data from two different measuring laboratories, CMDL and CSIRO commenced an "operational intercalibration" (also referred to as the ICP, Inter-Comparison Program, also the "flask-air-sharing" comparison). Both CSIRO and CMDL

Lab. 1 OSIRO (2) Lab. 3

FIGURE 3 IAEA round robin intercalibrations of ö'-'C of CO, using both pure CO, (GS20) and whole air in high-pressure cylinders (in which 513C is related to CO, mixing-ratio difference from ambient values by about — 0.05%o ppm-1). USA and Japanese measurement laboratories are identified by number only. Plotted are measured differences from d<5'JC assigned by CSIRO prior to circulation. CSIRO(2) refers to analyses after circulation.

CNi o o

. • • • • • • •» • IF • •/• ... !

• • • •• • #%

• •

» •





A^ t*. *


A A .

1992 1993 1994 1995 1996 1997 1998 1999 FIGURE 4 (CSIRO-INSTAAR) measured differences on Cape Grim air from the same flask as a function of flask collection date, for CO, (red circles) and SUC (blue triangles).

networks collect pairs of flasks 3 or 4 times per month, from the Cape Grim station on the northwest tip of Tasmania. Approximately twice per month, one of a pair of CMDL flasks has been routinely routed through CSIRO's GASLAB for analysis prior to analysis in CMDL and Institute for Alpine and Arctic Research, University of Colorado (INSTAAR, for the isotopic measurements on CMDL flask samples). The process is facilitated by the unusually small sample requirements for precise analysis in GASLAB (Francey et al., 1996). Once per month, the results of the multi-comparisons (CO,, CH,„ CO, N20, H2, <5I3C, SlsO) in both routine flask sampling of Cape Grim air from each laboratory and from the ICP flasks are automatically processed and reported via ftp in both laboratories (Masarie et al., submitted).

No systematic influence of GASLAB measurements on CMDL flasks has been detected. Figure 4 shows the results of the ICP flask comparisons for C02 and for §"C. Compared to cylinder inter-comparisons, the precision on the ICP comparisons is low (individual measurements) but the frequency is high. The C02 results are startling. The Australian calibration scale was established to within ~0.01 ppm at ambient C02 mixing ratios by repeated analysis of 10 cylinders initially characterized by CMDL. Return of a subset of the cylinders after two years confirmed this agreement to within a few hundredths of a ppm, as have comparisons of other cylinders. Despite this agreement in calibration scales (see also Fig. 2b, Australia), there is a consistent mean difference (CSIRO-CMDL) in the ICP flasks of 0.17 ±0.17 ppm.

The reason for this offset in flasks compared to high-pressure cylinders is not yet fully understood. I Iowever, development of a low-flow (15 ml min-1), high-precision (~7 ppb), and highly

1992 1993 1994 1995 1996 1997 1998 1999 FIGURE 4 (CSIRO-INSTAAR) measured differences on Cape Grim air from the same flask as a function of flask collection date, for CO, (red circles) and SUC (blue triangles).

stable NDIR C02 analyzer at CSIRO (G. Da Costa and L. P. Steele, in preparation) has provided clues that high-pressure regulators are a likely contributor to such offsets.

The SUC comparison in Fig. 4 illustrates another advantage of the ICP. The (CSIRO-INSTAAR) difference begins at close to zero, or slightly negative, and early in 1994 jumps to a positive value. After 1994, the difference is consistent with highpressure cylinder intercomparisons included in Fig. 3. The discrepancy between the laboratories, if applied globally, translates into a partitioning error of around 1 Gt C year-1 between the two laboratories. The continuity of the ICP data has permitted detection of the onset of the problem with reasonable accuracy, and the identification of possible contributing factors that occurred around this time.

0 0

Post a comment