Carbon Monoxide (CO) WMO Scale

Scale update: WMO CO_X2025

Updated April 2026

The WMO CO in air scale has been updated to WMO CO X2025 and the range of the scale has been increased to cover the nominal mole fraction range 25-1000 ppb. This scale revision has been retroactively applied to all CO tank calibrations performed at NOAA since 1993. Here we provide a brief overview of the revision and suggested implementation strategies for external laboratories tied to the WMO CO scale. Further details are given in the supplemental co_scale_details.pdf file. This document is intended as internal documentation but are made available for those interested.

The new scale (WMO CO X2025) updates the X2014A scale to better account for growth of CO in the current primary standards (in use since 2011) and to improve assignments to older primary and secondary standards that defined the CO scale prior to 2011. The range of the official scale has been increased to cover the nominal range 25 - 1000 ppb to better serve the needs of the community engaged in monitoring efforts closer to high emission areas.

The WMO CO in air scale has been maintained by NOAA GML since 1997 under its role as the designated Central Calibration Laboratory for CO. Prior to 2011, GML maintained the scale by periodically making new sets of gravimetric standards that were used to value assign secondary standards. Limitations in this method developed because it was difficult to make gravimetric standards often enough to fully characterize the behavior of secondary standards which have finite lifetimes. In 2011, the CCL changed the method used to maintain the CO scale to align with other CCL trace gas scales and improve the stability of the CO scale over time. 14 gravimetric standards were made between July and December 2011 in 29 L aluminum cylinders. These standards (referred to as the 2011 primary standards) cover the nominal range 25 - 1000 ppb. They were designated as the primary standards for the CO scale with the intention of maintaining them for decades. To account for likely drift in the CO mole fraction in the primary standards, an internal tracer technique was developed to allow the CCL to measure the 2011 primary standards periodically and derive time dependent value assignments for individual standards. This method was first presented at GGMT-2015 (Crotwell et. al, 2015) and formed the basis for the X2014A scale which was released in December 2015. Additional updates on the CO scale have been given at GGMT-2017 (Crotwell et. al, 2017) and GGMT-2024 (Hall et. al, 2024).

The main driver for the X2025 revision is that extrapolation of the drift corrections applied to the 2011 primary standards for the X2014A scale, which were based on the information available in 2015, have now been determined to be incorrect. Measurements made subsequent to the X2014A release show that the drift corrections applied were generally too high at the low end of the scale leading to a slow time and mole fraction dependent bias developing in the scale implementation.

Figure 1 shows an example of the gravimetric and measurement history of one of the 2011 primary standards (serial number CA06531) along with its X2014A time dependent value assignment. The X2014A value assignment was developed using the gravimetric value and the first 5 internal tracer method measurement results. Subsequent measurements show that the X2014A value assignment is biased high for this 2011 primary standard.

In addition, there appears to be a small (~ 1-2 ppb) offset between the gravimetric value and the measurement results when they are extrapolated back to the production date. This was not detectable with the data available for X2014A but is now observed in the 2011 primary standards. One of the difficulties in the assessment of this offset is the data gap between the gravimetric production of the standards in 2011 and the first measurement using the internal tracer method in 2014. Using a separate set of gravimetric standards made in 2015 and measured regularly with the internal tracer method since then, we have concluded that the apparent offsets observed when the measurements are extrapolated back to the production date are likely due to an initial non-linear drift of CO. This work was presented at GGMT-2024 (Hall et. al, 2024) and is discussed further on the details page. X2025 uses this information to determine new time dependent value assignments on each 2011 primary standard. These new value assignments form the basis for the revised scale and are implemented through the entire measurement record by reprocessing the raw data for each calibration episode. Figure 2 shows the X2025 minus X2014A differences for tertiary standard calibrations since 2012 as a function of mole fraction, binned by analysis year. The bias in X2014A grew slowly since 2016 as the assignments deviated from the actual behavior of the standards. The bias is time and mole fraction dependent. In 2025, the X2014A scale was 5 ppb high at 50 ppb, showed no bias at 500 ppb, and was 2 ppb low at 700 ppb compared to X2025.

The re-assignment of the 2011 primary standards and the longer record of internal tracer measurements on older primary standards have implications for the CO scale prior to 2011. All historical primary standards that are still available have been measured vs the internal tracer technique and all historical secondary standards that are still available have been measured vs the 2011 primary standards. This new data has helped the CCL re-assess and improve the consistency of the CO scale implementation prior to 2011. Figure 3 shows the X2025 minus X2014A difference for tertiary standard calibrations since 1993. The changes prior to 2012 are due to increased understanding of the behavior of CO in the historical standards and through adherence to a more rigorous calibration hierarchy to improve the transparency of the scale.

All CCL CO calibrations of tertiary standards from 1993 – present have been reprocessed from the original raw data files using the new value assignments for primary and secondary standards on the CO X2025 scale. Raw data for measurements made before 1993 are not available and their results were not updated for X2014A nor have they been revised for X2025. X2025 results are available on the CCL website (https://gml.noaa.gov/ccl/). Historical tank calibration results on the X2014A scale have been archived by the CCL and are available upon request.

The CCL uses sets of target tanks analyzed at various frequency to evaluate the consistency of the scale dissemination over time. The long term target tanks are analyzed at the tertiary level once or twice per year and many of them have been in use since 2001 (see Table 1 and example Figure 4). The standard deviations of the residuals of a fit to the calibration histories for these tanks are an indication of the consistency of the scale transfer over time at the CCL. The average of the residuals for target tanks with mole fractions at or below 300 ppb is 0.5 ppb. Residuals are larger at higher mole fractions but are less than 1% up to 1000 ppb.

We recommend that the X2025 scale revision is implemented at laboratories by obtaining revised calibration data for all laboratory standards measured by the CCL (available at https://gml.noaa.gov/ccl/) and reprocessing. Drift in the standards, determined by re-calibrations at the CCL or by other lab specific procedures, should be corrected during the reprocessing. This is the most reliable way to account for the time and mole fraction dependent bias in the CO scale that have developed since the release of X2014A and to maintain traceability to the WMO CO scale.

We recognize that not all atmospheric measurement records can be reprocessed reliably and that some measurement programs may not need the highest measurement data quality to meet program goals. Due to the time dependence of the X2014A to X2025 revision, it is not possible for the CCL to provide a single scale conversion function that would be applicable to atmospheric data from all labs. Conversion functions would be related to when the lab had standards calibrated by the CCL rather than the date the atmospheric measurements were made. In addition, only around 30% of the CO standards produced by the CCL were ever returned for re-calibration. This indicates that many CO measurements are tied to the CCL through standards that may not be assessed for stability and that the scale implementation at the lab level may not be closely tracking the time dependent changes in the scale observed at the CCL.

In cases where the GGMT network compatibility goals (WMO, 2026) are required but data reprocessing is not possible, data providers will need to examine their tie to the CCL in detail to decide on appropriate atmospheric data conversion methods and related uncertainties. The CCL can help in these cases if the traceability path is documented.

When the GGMT extended network compatibility goal (WMO, 2026) is the relevant data quality objective we think it is appropriate to consider the two scales functionally equivalent as the differences are small relative to typical atmospheric signals in these records. The maximum bias expected from this assumption is around 5 ppb. This corresponds to an additional uncertainty term of 2.9 ppb in cases where knowledge of the traceability to the scale is limited.

We also note that the ability of the CCL to consistently disseminate the scale to external labs by value assigning CO in air mixtures in aluminum cylinders is limited by the stability of CO in the disseminated cylinders themselves. Aluminum cylinders show CO growth in most cases and the rate can be variable. Typical growth rate of CO in aluminum cylinders observed by the CCL is around 0.5 ppb per year, but can range up to ~1.5 ppb per year, based on cylinders calibrated by the CCL multiple times since 2011. To maintain a tight linkage to the WMO scale, labs should follow the GGMT recommendations and have their highest level CO standards recalibrated by the CCL every 3 years.

References:

Crotwell, A. M., P. Novelli, B. Hall, and E. Dlugokencky (2015), Tracking drift in WMO primary CO standards, 18th WMO/IAEA Meeting on Carbon Dioxide, Other Greenhouse Gases, and Related Tracers Measurement Techniques (GGMT-2015), 13-17 September 2015, Scrips Institution of Oceanography, La Jolla, CA, USA. Presentations available at https://wmo.int/events/ggmt-2015.

Crotwell, A. M., P. Novelli, B. Hall (2017), An update on the WMO CO2 calibration scale, 19th WMO/IAEA Meeting on Carbon Dioxide, Other Greenhouse Gases, and Related Tracers Measurement Techniques (GGMT-2017), 27-31 August 2017, Swiss Federal Laboratories for Materials Science and Technology (Empa), Dubendorf, Switzerland. Presentations available at https://community.wmo.int/events/19th-wmoiaea-meeting-carbon-dioxide-other-greenhouse-gases-and-related-measurement-techniques-ggmt.

Hall, B, A. Crotwell, G. Petron, S. DeVogel, T. Mefford, K. Thoning, and J. Mund (2024), Transition to laser spectroscopy for N2O and CO analysis at the CCL, and re-assessment of the X2014A CO scale, 22nd WMO/IAEA Meeting on Carbon Dioxide, Other Greenhouse Gases, and Related Tracers Measurement Techniques (GGMT-2024), 5-8 August 2024, Instituto Nacional de Pesquisas Espaciais (INPE) Sao Jose dos Campos, Brazil. Presentations available at https://community.wmo.int/events/22nd-wmoiaea-meeting-carbon-dioxide-other-greenhouse-gases-and-related-tracers-measurement.

Figure 1: Calibration history of 2011 primary standard CA06531 (one of 14) measured against standards produced using the internal tracer technique using various CCL analytical methods (VURF, OA-ICOS, QC-TILDAS). Solid line shows the X2014A time dependent value assignment for this standard which was derived from the gravimetric value and the average of the first 5 measurement episodes.

Figure 2: X2025 minus X2014A differences for tertiary standards calibrated by the CCL since 2012. Results are binned by analysis date in 2 year intervals.

Figure 3: X2025 minus X2014A differences for tertiary standards calibrated by the CCL since 1993 vs analysis date. Large differences prior to 2011 are most often associated with calibrations of high mole fraction CO cylinders that are well above the range of standards used at that time.

Figures 4a-c: Calibration histories of three of the long term target tanks used by the CCL to monitor the dissemination of the scale (ND15749 in 4a, ND17435 in 4b, and ND17431 in 4c). All three target tanks are N265 size aluminum cylinders and measurements are at the tertiary level in the calibration hierarchy. Error bars are the uncertainty of the calibration episode (68% CL).

Table 1: List of long term target tanks with the standard deviation of the residuals from a linear fit of their calibration histories.

Serial number	Fill code	Fill date	Nominal value (ppb)	SD residual	N
ND33962	A	2008-01-01	26	0.90	58
ND33423	A	2008-01-01	43	0.42	56
ND15749	A	2001-01-01	58	0.37	96
ND15807	A	2001-01-01	78	0.57	93
ND17445	A	2001-01-01	108	0.51	89
ND16443	A	2001-01-01	135	0.42	112
ND17435	A	2001-01-01	154	0.48	100
ND16439	B	2018-10-11	171	0.57	22
ND17431	A	2001-05-01	203	0.58	97
ND33961	A	2008-01-01	241	0.78	55
ND16416	A	2002-01-01	304	0.74	75
ND33960	C	2014-03-05	489	1.93	41
CB10910	B	2016-02-18	704	2.32	42
CA07339	B	2010-03-01	1003	6.89	45