2.4. CARBON MONOXIDE
2.4.1. IN SITU CARBON MONOXIDE MEASUREMENTS
In situ measurements of CO continued at BRW and MLO during 1994 and 1995. For the analysis, a Reduction Gas Analyzer (RGA) (Trace Analytical, Inc.) was used. This measures CO using the gas chromatography-mercuric oxide reduction technique (previously described in Peterson and Rosson, 1993). The instruments operating at both observatories are identical and provided CO concentrations for four to five air samples per hour. The CO content of air samples was quantified by comparison to standards that reflected the range of concentrations seen at each site: 80 to 220 ppb at BRW and 60 to 180 ppb at MLO. All standards were referenced to the CMDL CO reference scale [Novelli et al., 1991]. To account for a nonlinear detector response common to the RGAs, a 3-point linear calibration (three standards) was used. This approach fits a linear regression to the two standards closest in instrument response to that of the sample, the regression coefficient, then used to calculate the sample CO mixing ratio.
Preliminary CO hourly-average mixing ratios measured at BRW and MLO during 1994 and 1995 are presented in Figure 2.15. These data have not been filtered for instrument performance or selected for background conditions. Work is currently underway to develop an expert system, based upon chromatographic parameters, that will automatically identify and flag periods when the instrument was not operating satisfactorily. The unselected time series, show features of the local and regional atmosphere. The timing of the seasonal cycles agrees well with that previously reported at these sites [Seiler et al., 1976; Novelli et al., 1992]. Maximum CO mixing ratios occur in late winter/early spring and the minimum occurs in summer. Periods of low variability are interrupted by short-term increases or decreases. These events reflect both the impact of local sources and the transport of air parcels from other locations.
Fig. 2.15. Preliminary in situ hourly average CO mixing ratios during 1994-1995 at (a) BRW and (b) MLO.
The annual mean CO mixing ratio determined from the in situ measurements made at BRW during 1994 and 1995 were 141.9 and 138.6 nanomol/mol, abbreviated as ppb, respectively. The annual means at MLO were 88.9 and 88.4 ppb. Breaks in the time series of about 2 weeks extent, occurred at MLO in 1995 due primarily to problems related to data storage. In spite of the high frequency variation seen in the in situ record, the annual average CO mixing ratios agree well with those determined from weekly flask samples that are collected to represent background conditions (Table 2.8).
TABLE 2.8. Preliminary 1994 and 1995 Mean CO Mixing Ratios From Land Sites
|Annual Mean CO (ppb)|
|BRW||Pt. Barrow, Alaska||141.6||131.3|
|CBA||Cold Bay, Canada||139.5||126.5|
|CGO||Cape Grim, Tasmania||51.3||51.7|
|CHR||Christmas Island||73.9||[ ]|
|CMO||Cape Meares, Oregon||151.0||[ ]|
|EIC||Easter Island, Chile||55.6||57.7|
|GMI||Marianas Island, Guam||90.5||94.5|
|ICE||Vestmanaeyjar, Iceland||137.4||[ ]|
|ITN||Grifton, N. Carolina||182.0||171.6|
|KEY||Biscayne, Florida||103.1||[ ]|
|KUM||Cape Kumukahi, Hawaii||110.7||102.2|
|MBC||Mold Bay, Canada||140.0||129.3|
|MHT||Mace Head, Ireland||137.2||124.1|
|MLO||Mauna Loa, Hawaii||95.1||90.2|
|NWR||Niwot Ridge, Colorado||121.7||119.2|
|PSA||Palmer Station||[ ]||48.6|
|QPC||Qinghai Prov., China||131.2||127.8|
|RPB||Ragged Point, Barbados||93.9||89.7|
|TAP||Tae-ahn Peninsula, S. Korea||226.1||204.4|
|UUM||Ulaan Uul, Mongolia||161.6||141.4|
|ZEP||Ny-Alesund, Spitzbergen||[ ]||132.6|
Comparison of CO mixing ratios determined using the in situ measurements to those measured from weekly flask samples provide a means to assure the quality of the former. There is strong confidence in the flask measurements because CMDL has better control over the characteristics of the analytical system and the stability of the CO standards used for flask analysis. Figure 2.16 compares CO mixing ratios measured in weekly flask samples of air to the corresponding hourly mean mixing ratio determined in situ. The results from the two sampling approaches agree well (r values >0.97). There is no significant difference between the flask concentrations and those measured in situ at BRW. However, the slight positive Y intercept in the regression of the MLO data suggests a small positive offset. It is unlikely that this is due to the calibration gases, because all standards were referenced against the CMDL working standards. If the instrument zero has increased (as observed before with these instruments) and is not accounted for, the calculated in situ CO mixing ratios could be slightly underestimated.
Fig. 2.16. Comparison of CO hourly averages measured in situ to those measured using flask sampling at (a) BRW and (b) MLO.
2.4.2. FLASK MEASUREMENTS OF CARBON MONOXIDE
Carbon monoxide mixing ratios were measured in a subset of flasks collected as part of the cooperative air sampling network. It was previously reported [Novelli et al., 1992] that the stability of CO in a container is dependent upon the flask materials and geometry. Only glass flasks fitted with glass piston stopcocks were used to measure CO. Over the lifetime of the CO program, the number of sampling locations has gradually increased as new sites in the network are started and the type of flasks used at older sites are converted to glass flasks for CO measurements. Analysis of air from flasks for CO and H2 were made on a semiautomated RGA. The response characteristics of the instrument used for flask analysis were nonlinear for CO over the range of atmospheric values. Therefore, a multipoint calibration (six to eight standards) was used to quantify the sample CO content [Peterson and Rosson, 1993; Novelli et al., 1994]. The precision of the CO method, estimated as the difference of mixing ratios determined for each flask in a simultaneously collected pair of flasks, was typically better than 2 ppb. A data selection routine flagged flask pairs having a difference of greater than 3 ppb. As before, hydrogen was referenced to an arbitrary scale. A set of H standards was prepared using gravimetric methods in collaboration with the NOAH Group. The H working standards are now being evaluated against the gravimetric standards.
Table 2.8 provides the land-based sites at which CO was measured in 1994 and 1995, and, whenever possible, the 1994 and 1995 annual mean values for these sites are shown. Samples for CO were also collected on trans-Pacific and South China Sea cruises; the annual mean CO mixing ratios are presented in Tables 2.9 and 2.10. These mean values were calculated from a curve fit to the total time series [Thoning et al., 1989].
TABLE 2.9. Preliminary 1994 and 1995 Mean CO Mixing Ratios From Combined Pacific Ocean Cruises
|Annual Mean CO (ppb)|
TABLE 2.10. Preliminary 1994 and 1995 Mean CO Mixing Ratios From South China Sea Cruise
|Annual Mean CO (ppb)|
Over the past several years new sites located near areas of human activity have been added to the CMDL air sampling network and these are expected to represent the regionally-polluted atmosphere. Comparison of these sites to "background" sites located at similar latitude illustrates the impact of economic development on atmospheric composition and are important constraints on models of global trace gas budgets. The difference in CO levels at two sites in Europe: Mace Head, Ireland (MHT), and the middle of the Baltic Sea (BAL), show the effect of human activities on regional-scale surface CO levels. MHT is a coastal site (53°20'N, 9°54'W), and winds are typically off the north Atlantic. BAL, located about 2000 km to the northeast (55°30'N, 16°40'E), is polluted from combustion of fossil fuels in Europe. Carbon monoxide time series measured at BAL is much noisier and mixing ratios are consistently higher than at MHT. In winter, CO mixing ratios at BAL are often 100 ppb greater than those at MHT, while in the summer the difference is 25 to 75 ppb. At BAL carbon dioxide (CO), another combustion product, was also enhanced relative to mixing ratios observed at MHT. However, there are also times when the CO and CO differences between the two sites are quite small, suggesting that BAL experiences periods of relatively unpolluted air.
Similarly, comparison of CO mixing ratios measured as part of the shipboard sampling programs in the Pacific and in the South China Sea (Tables 2.9 and 2.10) show the effects of human activities on CO in the boundary layer. Whereas the Pacific cruises sample air representative of the background marine boundary layer, the SCS cruises encounter pollution from the highly developed coast of southeastern Asia. CO mixing ratios along coastal Asia are typically 50 to 100% greater than those found in the Pacific. At the lower latitude SCS sites, isentropic back-trajectories suggested that during periods in October 1994, air was transported to these sites from areas in the southern hemisphere where fires had been observed. The high levels of CO seen in these regions may then result from both fossil fuel combustion in industrialized areas plus emission of CO from biomass burning in less developed areas.
2.4.3. THE MAPS PROGRAM
As part of the CMDL collaboration with the Measurement of Air Pollution from Satellites (MAPS) program (National Aeronautics and Space Administration-Langley Research Center), nearly real-time data from BRW and MLO were provided to the MAPS team during April and October 1994. Because the MAPS instrument provides a maximum signal in the middle troposphere [Reichle et al., 1990], measurements from mountain sites above the boundary layer were used as a quick test of the radiances measured by the space-borne instrument and the associated retrieval calculations. During March to November 1994, a CO instrument was installed at Niwot Ridge, Colorado, and the CMDL aircraft program flew vertical profiles above the site during the MAPS missions. These data have proved very valuable in the validation of the MAPS measurements. The MAPS measurements have also been compared with other ground based and aircraft measurements supported by a program of reference gas standard intercomparisons (section 2.4.4). CMDL coordinated the correlative measurements team for the 1994 flights of MAPS. This team provided MAPS with CO data from more than 60 sites worldwide. These data were used to validate measurements made by MAPS and to provide a unique picture of CO in the lower troposphere during April and October 1994.
2.4.4. CARBON MONOXIDE STANDARDS
The primary CMDL CO standards were prepared gravimetrically during 1988-1989 and then propagated to a set of working standards [Novelli et al., 1991]. These working standards were re-evaluated using a new set of gravimetric standards in March 1992. Comparisons of values assigned to working standards using the original gravimetrics, those produced in 1992, and the working standards themselves, suggest that the accuracy of propagation and stability of the scale has been within about 1% [Novelli et al., 1994].
It is now well known that CO standards used by one laboratory can be significantly
different from those used in another [Weeks et al., 1989]. Therefore,
it has been difficult to combine CO measurements made by different laboratories.
Under the MAPS program, an inter-comparison of CO measurements made by 11 laboratories
in 8 countries was organized. The round-robin inter-comparison was organized
with four standards having approximate mixing ratios of 50, 100, 150, and 200
ppb in air (levels that represent the range of global CO mixing ratios in the
unpolluted atmosphere). The experiment began in July 1993 and was completed
in October 1995. The participating laboratories used either gas chromatography
with HgO reduction detection or gas filter correlation radiometry and standards
from several sources, including CMDL, National Institute of Standards Technology
(NIST), the Fraunhofer Institute (Germany), and the Chemical Instrument Testing
Institute (Japan). Differences between participants ranged to 20%. These could
not be explained solely by differences in calibration gases and indicate the
effect of different calibration procedures and instrument configuration on the