Currently GLOBALVIEW-CO is based solely on measurements made by the NOAA ESRL. ESRL analysis of air samples used gas chromatography followed by hot mercuric oxide reduction (RGA, Trace Analytical, Inc, Menlo Park, USA). The instrument response was defined bi-weekly using a suite of standards containing between 0 and 220 nmol mol-1. Beginning 2008, air samples were also analyzed using an instrument based on fluorescence of CO in the vacuum ultraviolet (VURF, Aero-Laser, Garmish Partenkirchen, Germany). The VURF provides a linear response over a wide range (0 to > 10,000 nmol mol-1 CO). A single standard with 125 nmol mol-1 was used as the reference. Because air samples are collected in pairs, agreement from two air samples collected simultaneously provides a strong indication of sample quality. Poor results are flagged based on pair agreement, and also because of analytical problems and deviation from background conditions. Details of the analysis and quality control are provided in Novelli et al., 1998, 2003.
WMO CO Mole Fraction Scale
The reference scale developed and used at NOAA ESRL consists of a series of primary CO-in-air mixtures prepared by a gravimetric method (Novelli et al., 1992, 2003). These primaries are the basis of a hierarchical scheme for the calibration of secondary and working standards. Gravimetric standards produced in 1989, 1992, 1997 and 2000 suggested long-term, relatively low, drift in ESRL secondary standards. All standards have been corrected for this drift (Novelli, 2003).
Prior to 2004, measurements were made using GC-HGO with working standards as the reference. Multipoint calibration curves, based upon secondary standards with CO mixing ratios between 50 and 200 nmol mol-1 by mole fraction, were determined biweekly to correct the non-linearity of the detector. In 2004, we moved CO calibrations to a VURF. A single working standard is used for calibration. These contain between 300 and 400 nmol mol-1 CO, based on comparison to multiple gravimetric and secondary standards (200 and 450 nmol mol-1). The uncertainty on these working standards is ~0.7%.