A new reference scale for atmospheric observations of CO is described here. The new scale, WMO CO X2014, is based on primary reference gases prepared in 2000, 2006 and 2011 by a gravimetric method (Novelli et al., 1991; Hall et al., 2007). This scale, while still considered preliminary, supersedes previous versions (WMO CO X2000 and WMO CO X2004) from 2000 to present. Its development is based on measurements of the three sets of primary standards. Three analytical methods were used: 1) gas chromatography with hot mercuric oxide reduction (GC-HgO, from 2000-2005), 2) resonance fluorescence in the vacuum ultraviolet (VUV, 2004-2010), and 3) off-axis Integrated Cavity Output Spectroscopy (ICOS, 2010 to present). The calibration of samples measured on either GC or ICOS used response curves derived from multiple standards. The VUV instrument which had a linear response used a single point calibration with reference gases between 300 and 400 nmol mol-1. The reference gases used on GC-HgO were revised to the new scale and the response curves re-determined. The new scale was evaluated for its consistency using calibration results for eight surveillance standards measured between 2001 and 2104 (Table 1). WMO CO X2014 extends the range of calibration to 500 nmol mol-1.

Samples measured on the GC with mole fractions > 200 nmol mol-1 have not yet been revised, nor have all measurements on VUV with CO > 350 nmol mol-1. Otherwise a preliminary revision to mole fractions determined between 2000 to present are available at www.esrl.noaa.gov/gmd/ccl/refgas.html. The uncertainty of the scale, estimated from calibrations of the surveillance cylinders, is approximately 1.2 nmol mol-1 or 0.6% of the mole fraction (k=2), whichever is greater. Measurement results prior to 2000 will also be revised to WMO CO X2014.

Table 1: Preliminary CO mole fractions assigned 8 surveillance cylinders on the X2014 scale. Mean CO mole fractions (nmol mol-1) with n=4 and uncertainty (k=2) of multiple measurements.

Tank ID GC-HgO*
2006 VUV#
01-03 2012
1 58.0 (0.9) 57.8 (0.8) 57.8 (0.2) 58.6 (0.2) 58.1 (0.8)
2 77.0 (1.3) 77.5 (0.8) 77.3 (0.4) 77.8 (0.3) 77.3 (0.8)
3 107.9 (0.8) 107.2 (0.5) 107.4 (0.2) 107.4 (0.4) 107.4 (0.8)
4 135.2 (0.7) 133.9 (0.3) 134.2 (0.2) 134.4 (0.2) 134.3 (0.8)
5 155.0 (1.4) 153.3 (0.3) 153.6 (0.2) 153.7 (0.3) 153.9 (1.1)
6 175.6 (1.2) 175.3 (0.3) 175.5 (0.8) 174.9 (0.2) 175.3 (0.9)
7 202.4 (1.5) 202.5 (0.7) 202.3 (0.3) 202.3 (0.3) 202.4 (1.0)
8 NA 301.5 (1.2) 302.4 (0.3) 302.9 (0.6) 302.3 (1.4)^

Since 2012 calibration results from the CCL have been drifting as a result of newly installed reference gases. These have been drifting upwards, leading to tank assignments trending low. The reference gases used for measurements on a Los Gatos ICOS instrument are secondary standards to a set of primaries prepared in late 2011 (www.esrl.noaa/gmd/ccl/co_scale.html). Both the primary and secondary standards are slowly increasing. Their drift rates have not yet been quantified. However the impact of drift on calibration results can be estimated from the long-term measurement histories of a eight surveillance cylinders (Figure 1). These are contained in 44 liter Al cylinders with CO ranging from 57 to 300 ppb and have shown to be relatively stable with time. Between 2004 and 2011 the eight tanks showed a mean increase of 0.1 ppb yr-1. Assuming they continued to increase at similar rates after 2012 combined with their decrease since 2012 of ≈ 0.25 ppb yr-1 , their negative rate of change since 2012 is approximately 0.35 ± 0.01 ppb yr-1.

A static dilution method which bootstraps dilution of CO with that of CH4 is being developed to aid in determining drift rates. Ultimately, rates of change in the NOAA standards will be determined taking into account changes in the surveillance cylinders, the results from the static dilution experiments and comparison to newly prepared primary standards.

A correction for drift in calibration results since 2012 will included in the release of the final of version WMO X2014 CO scale. Contact Paul Novelli (paul.c.novelli@noaa.gov) or Andrew Crotwell (andrew.crotwell@noaa.gov) for more information.

Figure 1. The measurement history for a surveillance cylinder calibrated by three methods and different reference gases. Data are the mean of a calibration event and the error bars are 1 σ. The thin vertical lines show when reference gases were changed.

For more information or to be added to the mailing list for users of the WMO CO scale please contact Paul Novelli at paul.c.novelli@noaa.gov