The commercial development of laser-based instruments over the last decade that can measure real-time isotopic ratio variations of greenhouse gases, and notably CO2, has allowed their application across a wide range of scientific and technical disciplines. Precise measurements can be achieved, and appropriate calibration strategies [1] and standards need to be applied to achieve accurate results and consistency with traditional mass spectrometric measurement methods. For CO2, calibration strategies can be based on using CO2 in air standards with the same isotopic ratios but containing different amount fractions, or the same amount fraction and different isotope ratios. This has resulted in the availability of calibration standards containing different isotope ratios at different amount fractions, which may or may not contain nitrous oxide. An international comparison programme at the BIPM (BIPM.QM-K4) is in development to demonstrate the equivalence of such standards, which would allow them to be used interchangeably by operators.The BIPM’s comparison facility is based on a dual inlet isotope ratio mass spectrometer with a custom built (BIPM) Air Trapping system (BAT) to extract CO2 from air mixtures using cryogenic separation for determination of δ13Cand δ18O-CO2, with a correction for the N2O present in the sample. A procedure for regularly determining the relative ionization efficiency of N2O in relation to CO2 has been developed and is applied as a function of the amount fraction of N2O in the sample. Metrological traceability is achieved through a hierarchy of low-pressure CO2 standards with δ13C values nominally at -1 ‰, -35 ‰ and -43 ‰, calibrated on the VPDB scale via IAEA 603 carbonate standard material. Initial validation of the performance of the facility has been performed with the extraction of CO2 from gas mixtures within the range of 380 μmol mol−1 to 800 μmol mol−1 and δ13C and δ18O-CO2 values from 1 ‰ to -43 ‰ and -7 ‰ to -35 ‰, respectively. The method demonstrates excellent reproducibility, with standard deviations of 0.005% and 0.05%. for δ13C and δ18O-CO2, respectively. In addition, the robustness of the N2O correction has been demonstrated by comparing δ13C and δ18O-CO2 values from standards produced from the same CO2 source gas but at differing amount fractions. The performance and validation of the facility will be described. [1] Flores, E., Viallon, J., Moussay, P., Griffith, D. W. T. & Wielgosz, R. I. Calibration strategies for FT-IR and other isotope ratio infrared spectrometer instruments for accurate δ13C and δ18O measurements of CO2 in air. Anal. Chem. 89, 3648–3655 (2017).

The participants of the 12th International Atomic Energy Agency (IAEA) meeting on stable isotope reference materials reached a consensus, acknowledging the existence and use of two carbon isotope delta scales: the VPDB (Vienna Peedee belemnite) scale and the VPDB-LSVEC (LSVEC - lithium carbonate prepared by H. J. Svec). Conversion models between the two scales can be established and used but introduce uncertainty. A format for isotope delta scale definition was agreed upon and was used to define the two carbon isotope delta scales and the two main oxygen isotope delta scales, VSMOW-SLAP (Vienna Standard Mean Ocean Water-Standard Light Antarctic Precipitation) and VPDB. Confirmation or identification of a second-scale-defining point is still necessary for the nitrogen and sulfur isotope delta scales. Efforts are encouraged to improve consistency among laboratories in the isotopic analysis of "non-exchangeable hydrogen" in bulk organic materials and oxygen in carbonates using the phosphoric acid reaction. Additional topics discussed include (1) need for improvement in reference materials for accurate greenhouse gas isotopic analyses; (2) reference materials under production by the IAEA, the US Geological Survey (USGS), and the US National Institute of Standards and Technology (NIST); (3) methods for value and uncertainty assignment of reference materials; and (4) calculation of carbon-13 isotope delta and oxygen-18 isotope delta of CO2 measured by dual-inlet isotope ratio mass spectrometry.

Since 1999, Environment and Climate Change Canada (ECCC) has been coordinating a multi-laboratory comparison of measurements of long-lived greenhouse gases in whole air samples collected at the Global Atmosphere Watch (GAW) Alert Observatory located in the Canadian High Arctic (82∘28′ N, 62∘30′ W). In this paper, we evaluate the measurement agreement of atmospheric CO2, CH4, N2O, SF6, and stable isotopes of CO2 (δ13C, δ18O) between leading laboratories from seven independent international institutions. The measure of success is linked to target goals for network compatibility outlined by the World Meteorological Organization's (WMO) GAW greenhouse gas measurement community. Overall, based on ∼ 8000 discrete flask samples, we find that the co-located atmospheric CO2 and CH4 measurement records from Alert by CSIRO, MPI-BGC, SIO, UHEI-IUP, and ECCC versus NOAA (the designated reference laboratory) are generally consistent with the WMO compatibility goals of ± 0.1 ppm CO2 and ± 2 ppb CH4 over the 17-year period (1999–2016), although there are periods where differences exceed target levels and persist as systematic bias for months or years. Consistency with the WMO goals for N2O, SF6, and stable isotopes of CO2 (δ13C, δ18O) has not been demonstrated. Additional analysis of co-located comparison measurements between CSIRO and SIO versus NOAA or INSTAAR (for the isotopes of CO2) at other geographical sites suggests that the findings at Alert for CO2, CH4, N2O, and δ13C–CO2 could be extended across the CSIRO, SIO, and NOAA observing networks. The primary approach to estimate an overall measurement agreement level was carried out by pooling the differences of all individual laboratories versus the designated reference laboratory and determining the 95th percentile range of these data points. Using this approach over the entire data record, our best estimate of the measurement agreement range is −0.51 to +0.53 ppm for CO2, −0.09 ‰ to +0.07 ‰ for δ13C, −0.50 ‰ to +0.58 ‰ for δ18O, −4.86 to +6.16 ppb for CH4, −0.75 to +1.20 ppb for N2O, and −0.14 to +0.09 ppt for SF6. A secondary approach of using the average of 2 standard deviations of the means for all flask samples taken in each individual sampling episode provided similar results. These upper and lower limits represent our best estimate of the measurement agreement at the 95 % confidence level for these individual laboratories, providing more confidence for using these datasets in various scientific applications (e.g., long-term trend analysis).

Main text The pilot study CCQM-P204 was aimed at evaluating the level of compatibility of laboratories' measurement capabilities to value assign isotope ratios in samples of pure CO 2 gas, expressed as isotope delta values relative to the relevant international scale: δ 13 C VPDB and δ 18 O VPDB-CO2 . Pure CO 2 gas samples were prepared by the BIPM in batches of 10 samples of the same gas and circulated to participants for measurement. Each participant received four samples of CO 2 with a different nominal δ 13 C VPDB value: −1 ‰; −9 ‰; −35 ‰; and −42 ‰. The BIPM was also responsible for evaluating the homogeneity and stability of the samples. The co-coordinator IAEA received one sample per batch to confirm the batch-to-batch homogeneity. Within-batch and between-batch inhomogeneity was assessed and found to be negligible in comparison to the spread of results reported by participants. Participants used the analytical technique of their choice to measure the isotope delta values. They were requested to report the measurement results together with detailed information on their traceability, measurement methods and data treatment. Results of the comparison were to be compiled by the BIPM and evaluated jointly by the BIPM and the IAEA. The majority of participants reported results using DI-IRMS, and those that reported results based on laser spectroscopy techniques showed a very similar dispersion of results as for DI-IRMS, although generally with greater uncertainty. A total of nineteen participants reported their measurements, with two of them reporting results with different reference materials to provide more insight into the traceability of the measurements. The results were reported with traceability to three different VPDB scale realizations, notably VPDB, VPDB-LSVEC and VPDB2020, with 8, 7 and 6 results reported respectively for each of these. Participants agreed that results based on VPDB and VPDB2020 scale realizations should, in principle, lead to consistent results, whereas those based on VPDB-LSVEC should show a bias that increased as samples became more depleted in 13 C, with the bias approaching 0.2 ‰ for the most depleted sample. This bias was demonstrated by the participant reporting the most precise measurements based on the VPDB-LSVEC realizations, whereas for 2 participants using VPDB-LSVEC scale realizations other issues dominated the consistency of their results. The 3 laboratories using the NIST (8562,8563, 8564) reference materials (reported as on the VPDB-LSVEC scale), were highly consistent with each other, but the reported bias for the VPDB-LSVEC realization was not evident, with the historical method used for value assignment of the NIST RMs, and their relatively large uncertainty, being identified as possible causes for this. In general, for all results the dispersion was greater than expected based on the measurement uncertainties reported by participants. This dispersion increased as the samples became more depleted in 13 C, so that results that were traceable to realizations of the VPDB scale that could be considered equivalent (VPDB and VPDB2020) did not lead to ensembles that were fully consistent within their stated uncertainties. Either the reduced chi-squared or Birge Ratio provide easily calculated quantities to characterise lack of consistency in a data set, where consistent data would lead to values of unity for either of these, and discrepant data leading to increased values. This is most readily demonstrated by considering results based on DI-IRMS with traceability to the VPDB scale through either VPDB and VPDB 2020 realizations, where the standard deviation of 16 results was 0.043 ‰ and a Birge Ratio of 2.7 calculated for nominally −1‰ for δ 13 C, and the standard deviation was 0.12 ‰ and a Birge Ratio of 2.9 calculated at nominally −9 ‰ for δ 18 O. For the samples where the nominal δ 13 C value was −42 ‰, the standard deviation of 17 results was 0.085 ‰ and a Birge Ratio of 4.5 calculated for δ 13 C, and the standard deviation was 0.24 ‰ and a Birge Ratio of 3.4 calculated for δ 18 O at nominally −36 ‰. The observed magnitude of the standard deviation of results can also be compared to the standard uncertainty of the IAEA-603 materials certified values (0.01 ‰ for δ 13 C and 0.04 ‰ δ 18 O) and the smallest standard uncertainties reported by a participant (0.005 ‰ for δ 13 C, and 0.01 ‰ for δ 18 O). These results indicate an underestimation of uncertainty for reported results, especially for those with the smallest uncertainties. Components of uncertainty that should be reviewed before future comparisons include: the uncertainty contribution from reference materials; the uncertainty associated with the phosphoric acid reaction with carbonate reference materials; corrections and uncertainties related to cross-contamination effects in the IRMS; appropriate methods for combining uncertainties of multiple reference materials and accounting for their correlations. A retreatment of results, which normalizes results to the −1 ‰ and −42 ‰ δ 13 C samples, leads to improvement in the consistency of results as demonstrated for measurements on the nominally −35 ‰ δ 13 C, −30 ‰ δ18O samples for which the standard deviation is reduced to 0.034 ‰ and 0.057 ‰ for δ 13 C and δ 18 O respectively (from 0.072 ‰ and 0.198 ‰ without normalization). The results of the comparison indicate that once issues of non-ideal methods and use of LSVEC are removed, discrepancies in results arise from the challenges in accurately transferring delta values from carbonate reference materials to CO 2 gaseous samples, and that these issues can be reduced by having appropriate gaseous reference standards for calibration when measuring CO 2 gaseous samples. This is consistent with the identical treatment principle that is preferred in the isotope ratio measurement community. An analysis of results is presented in this report, with further consideration of the impact of the measurement method, the scale, and the reference materials. Uncertainties reported by participants are detailed and analysed, highlighting important differences in the uncertainty sources considered. Although CCQM-P204 was a comparison organised within the CCQM/GAWG and IRWG, no reference value was calculated, mainly because not all results appeared to be strictly on the same scale. Instead, a list of recommendations is provided to encourage more harmonised measurement practices and reach better consistency in future comparisons on similar materials. To reach the main text of this paper, click on Final Report . Note that this text is that which appears in Appendix B of the BIPM key comparison database https://www.bipm.org/kcdb/ . The final report has been peer-reviewed and approved for publication by the CCQM, according to the provisions of the CIPM Mutual Recognition Arrangement (CIPM MRA).

Abstract Stable hydrogen, carbon, nitrogen, oxygen and sulfur (HCNOS) isotope compositions expressed as isotope-delta values are typically reported relative to international standards such as Vienna Standard Mean Ocean Water (VSMOW), Vienna Peedee belemnite (VPDB) or Vienna Cañon Diablo Troilite (VCDT). These international standards are chosen by convention and the calibration methods used to realise them in practice undergo occasional changes. To ensure longevity and reusability of published data, a comprehensive description of (1) analytical procedure, (2) traceability, (3) data processing, and (4) uncertainty evaluation is required. Following earlier International Union of Pure and Applied Chemistry documents on terminology and notations, this paper proposes minimum requirements for publishing HCNOS stable-isotope delta results. Each of the requirements are presented with illustrative examples.

Rationale LSVEC, the second anchor Reference Material (RM) for the VPDB δ 13 C scale realisation, was introduced in 2006. In 2015, its δ 13 C value was found to be drifting and, in 2017, its use as an RM for δ 13 C was officially discontinued by IUPAC. New RMs of low uncertainty are needed. This paper describes the preparation and characterisation of IAEA‐610, IAEA‐611 and IAEA‐612 (calcium carbonate, of chemical origin) which shall serve as a set of RMs aimed at anchoring the VPDB scale at negative δ 13 C values. Methods The preparation and characterisation of IAEA‐610, IAEA‐611 and IAEA‐612 were performed by addressing the contemporary technical requirements for RM production and characterisation (ISO Guide 35:2017). The three RMs were produced in large quantities, and the first batch was sealed into ampoules (0.5 g) to ensure the integrity of the RM during storage; additional batches were sealed for long‐term storage. The most accurate method of CO 2 preparation and stable isotope measurements was used, namely carbonate‐H 3 PO 4 reaction under well‐controlled conditions combined with well‐tested stable isotope ratio mass spectrometry. Results The assigned values of δ 13 C and associated uncertainties are based on a large number of analyses (~10 mg aliquots) performed at IAEA and address all the known uncertainty components. For aliquots down to ~100 μg, the δ 13 C uncertainty is increased. The uncertainty components considered are as follows: (i) material homogeneity, (ii) value assignment against IAEA‐603, (iii) potential storage effects, (iv) effect of the 17 O correction, and (v) mass spectrometer linearity and cross‐contamination memory in the ion source. Conclusions The new RMs IAEA‐610, IAEA‐611 and IAEA‐612 have been characterised on the VPDB δ 13 C scale in a mutually consistent way. The use of three RMs will allow a consistent realisation of the VPDB δ 13 C scale with small uncertainty to be established, and to reach metrological compatibility of measurement results over several decades.

RationaleIn recent years, the primary reference material (RM) for the VPDB scale, NBS19, has become unavailable, and the RM used for low‐end scale‐anchoring, LSVEC, was found unsuitable due a drift in theδ13C value. Given these problems, new RMs aimed at realising the VPDBδ13C scale with low uncertainty were produced. Establishing the consistency of the new RMs with the “old” RMs prompted our revision of the underlying principles of RM value assignments, and the VPDBδ13C scale realisation and its long‐term sustainability.MethodsAnalysis of major developments of the VPDB scale, a review of the contemporary requirements for RMs, and comparison with well‐established measurement scales have been performed, with the aim of revising the VPDBδ13C scale, principles of RM value assignments, and calibrator hierarchy. Requirements for scale‐anchoring RMs with low uncertainty and measures to establish the scale sustainability have been formulated.ResultsThe revised scale realisation is based on multiple reference points, well‐defined calibration hierarchy and the use of well‐understood methods for value assignment. The realisation scheme includes the new primary RM IAEA‐603 and scale‐anchoring RMs IAEA‐610, IAEA‐611 and IAEA‐612, coveringδ13C from +2.46 to −36.7 ‰ VPDB, with uncertainties, including inhomogeneity and stability assessment, of less than 0.015 ‰. The values of these four RMs were assigned in a mutually consistent way; agreement between measurements made using this realisation with those made using the VPDB scale of 2006 has been demonstrated on NIST CO2RMs 8562–8564.ConclusionsMultipoint‐anchoring of the VPDBδ13C scale provides several distinct “points” on the scale as means for cross‐measurements to check the stability and viability of RMs and detect drift of values, if any. This ensures that theδ13C scale is suitable for the most demanding applications, and provides options for developing further RMs with high accuracy inside a robust scale realisation scheme.

Environmental contextChlorofluorocarbons (CFCs) are potent greenhouse and stratospheric ozone depleting trace gases. Their atmospheric concentrations are in decline, thanks to global production and consumption controls imposed by the Montreal Protocol. In recent years, the rates of decline of CFC atmospheric concentrations, especially for CFC-11 (CCl3F), are not as large as anticipated under the Protocol, resulting in renewed efforts to estimate CFC consumption and/or emissions to possibly identify new or poorly quantified sources. AbstractAustralian emissions of chlorofluorocarbons (CFCs) have been estimated from atmospheric CFC observations by both inverse modelling and interspecies correlation techniques, and from CFC production, import and consumption data compiled by industry and government. Australian and global CFC emissions show similar temporal behaviour, with emissions peaking in the late-1980s and then declining by ~10% per year through to the present. Australian CFC emissions since 1978 account for less than 1% of global emissions and therefore make a correspondingly small contribution to stratospheric ozone depletion. The current CFC emissions in Australia are likely from ‘banks’ of closed-cell foams, and refrigeration–air conditioning equipment now more than 20 years old. There is no evidence of renewed consumption or emissions of CFCs in Australia. The reduction in CFC emissions has made a significant contribution to reducing Australian greenhouse gas emissions.

We present the global research landscape which aims to deliver a measurement infrastructure to underpin atmospheric observations of key greenhouse gases governing changes in the Earth’s climate. These measurements present a significant challenge to the metrological community, analytical laboratories and major producers of reference materials. The review focuses on the progress made in the Gas Analysis Working Group of the Consultative Committee for Amount of Substance: Metrology in Chemistry and Biology (CCQM-GAWG) in establishing the primary realisation of the amount-of-substance fraction for carbon dioxide, methane and nitrous oxide in an air matrix. It also focuses on the importance of providing traceable measurements of isotopic composition of these components for commutability of reference materials and for isotope ratio measurements for greenhouse gas source attribution. The review examines the developments in the Global Atmosphere Watch (GAW) Programme of the World Meteorological Organization (WMO) for providing the framework for the development and implementation of integrated greenhouse gas observations, which is vital for understanding the global carbon cycle and the role greenhouse gases play in climate change. The developments in analytical techniques are also discussed which have shaped the direction of the metrology required to meet the evolving and future needs of stakeholders.

Abstract. Ice core records of the major atmospheric greenhouse gases (CO2, CH4, N2O) and their isotopologues covering recent centuries provide evidence of biogeochemical variations during the Late-Holocene and Pre-Industrial Periods and over the transition to the Industrial Period. These records come from a number of ice core and firn air sites, and have been measured in several laboratories around the world and show common features, but also unresolved differences. Here we present revised records, including new measurements, performed at the CSIRO Ice Core Extraction LABoratory (ICELAB) on air samples from ice obtained at the high accumulation site of Law Dome (East Antarctica). We are motivated by the increasing use of the records by the scientific community and by recent data-handling developments at CSIRO-ICELAB. A number of cores and firn air samples have been collected at Law Dome to provide high resolution records overlapping recent, direct atmospheric observations. The records have been updated through a dynamic link to the calibration scales used in the Global Atmospheric Sampling LABoratory (GASLAB) at CSIRO, which are periodically revised with information from the latest calibration experiments. The gas-age scales have been revised based on new ice-age scales, and the information derived from a new version of the CSIRO firn diffusion model. Additionally, the records have been revised with new, rule-based selection criteria and updated corrections for biases associated with the extraction procedure, and the effects of gravity and diffusion in the firn. All measurements carried out in ICELAB-GASLAB over the last 25 years are now managed through a database (the ICElab dataBASE or ICEBASE) which provides consistent data management, automatic corrections and selection of measurements, and a web-based user interface for data extraction. We present the new records, discuss their strengths and limitations and summarise their main features. The records reveal changes in the carbon cycle and atmospheric chemistry over the last two millennia, including the major changes of the anthropogenic era and the smaller, mainly natural variations beforehand. They provide the historical data to calibrate and test the next inter-comparison of models used to predict future climate change (Coupled Model Inter-comparison Project - phase 6, CMIP6). The datasets described in this paper, including spline fits, are available at https://doi.org/10.25919/5bfe29ff807fb (Rubino et al., 2018).

S1 ICELAB databaseThe ICELAB-GASLAB is a Microsoft SQL Server that consists of numerous tables linked together by a Universal Analysis Number (UAN) that is assigned to each sample prior to analysis.The UAN is contained in every table and serves as the index connecting all information about the sample.The structure of the database, ie hierarchy of tables, reflects the chronological sequence of steps involved in collecting a sample (ice-core, firn-air), preparing an ice sample, extracting air from that sample and analysing the concentration and isotopic composition of the extracted air (only the ICE-relevant tables and procedures are described here).The web-interface to the ICELAB database, developed with Microsoft Visual Studio, is shown in Fig. S1.It indicates the general structure and relationships crucial to the database, as described below.Samples are divided into three categories (ICE, Tests and FIRN), and data pertaining to these categories are stored in separate tables.Whenever a unique air sample is prepared (ICE, Test or FIRN) a unique UAN is created and stored in the relevant table ."ICE" contains the ice sample ID (e.g.: "DE08-256", or "DSS-144"), ice collection parameters, depth, age, mass and volume (calculated using an average ice density), along with comments pertaining to the visual, glaciological characteristics (such as "crusts", "melt layers", "cracks", etc.).Also included are the operator's initials and the preparation date."TEST" contains information about the type of test ("Blanks" and "BFI") and a sample comment (e.g.: "BFI no shaking", that is without shaking the ice grater, or "BFI Ar-sparged", that is BFI grown under a flow of Argon), reference gas used (that is the reference tank or flask of known composition used for the test), and physical parameters such as mass and volume.This also contains the operator's initials and the preparation date."FIRN" contains information about the sample collection (depth, age, volume of gas pumped, sampling date and time, pressure of the sample after sampling and the type of firn air sampling device used).

Abstract. The isotopic composition of carbon (Δ14C and δ13C) in atmospheric CO2 and in oceanic and terrestrial carbon reservoirs is influenced by anthropogenic emissions and by natural carbon exchanges, which can respond to and drive changes in climate. Simulations of 14C and 13C in the ocean and terrestrial components of Earth System Models (ESMs) present opportunities for model evaluation and for investigation of carbon cycling, including anthropogenic CO2 emissions and uptake. The use of carbon isotopes in novel evaluation of the ESMs' component ocean and terrestrial biosphere models and in new analyses of historical changes may improve predictions of future changes in the carbon cycle and climate system. We compile existing data to produce records of Δ14C and δ13C in atmospheric CO2 for the historical period 1850–2015. The primary motivation for this compilation is to provide the atmospheric boundary condition for historical simulations in the Coupled Model Intercomparison Project 6 (CMIP6) for models simulating carbon isotopes in their ocean or terrestrial biosphere models. The data may also be useful for other carbon cycle modelling activities.

The isotopic composition of carbon (Δ14C and δ13C) in atmospheric CO2 and in oceanic and terrestrial carbon reservoirs is influenced by anthropogenic emissions and by natural carbon exchanges, which can respond to and drive changes in climate. Simulations of 14C and 13C in the ocean and terrestrial components of Earth system models (ESMs) present opportunities for model evaluation and for investigation of carbon cycling, including anthropogenic CO2 emissions and uptake. The use of carbon isotopes in novel evaluation of the ESMs' component ocean and terrestrial biosphere models and in new analyses of historical changes may improve predictions of future changes in the carbon cycle and climate system. We compile existing data to produce records of Δ14C and δ13C in atmospheric CO2 for the historical period 1850–2015. The primary motivation for this compilation is to provide the atmospheric boundary condition for historical simulations in the Coupled Model Intercomparison Project 6 (CMIP6) for models simulating carbon isotopes in the ocean or terrestrial biosphere. The data may also be useful for other carbon cycle modelling activities.

Land carbon uptake reduced atmospheric CO2 levels during the Little Ice Age. Numerical simulations of atmospheric carbonyl sulfide levels and ice-core carbon isotope data reveal that temperature change, not land-cover change, was responsible. Low atmospheric carbon dioxide (CO2) concentration1 during the Little Ice Age has been used to derive the global carbon cycle sensitivity to temperature2. Recent evidence3 confirms earlier indications4 that the low CO2 was caused by increased terrestrial carbon storage. It remains unknown whether the terrestrial biosphere responded to temperature variations, or there was vegetation re-growth on abandoned farmland5. Here we present a global numerical simulation of atmospheric carbonyl sulfide concentrations in the pre-industrial period. Carbonyl sulfide concentration is linked to changes in gross primary production6 and shows a positive anomaly7 during the Little Ice Age. We show that a decrease in gross primary production and a larger decrease in ecosystem respiration is the most likely explanation for the decrease in atmospheric CO2 and increase in atmospheric carbonyl sulfide concentrations. Therefore, temperature change, not vegetation re-growth, was the main cause of the increased terrestrial carbon storage. We address the inconsistency between ice-core CO2 records from different sites8 measuring CO2 and δ13CO2 in ice from Dronning Maud Land (Antarctica). Our interpretation allows us to derive the temperature sensitivity of pre-industrial CO2 fluxes for the terrestrial biosphere (γL = −10 to −90 Pg C K−1), implying a positive climate feedback and providing a benchmark to reduce model uncertainties9.

We analyse global and regional changes in CO2 fluxes using two simple models, an airborne fraction of anthropogenic emissions and a linear relationship with CO2 concentrations. We show that both models are able to fit the non-anthropogenic (hereafter natural) flux over the length of the atmospheric concentration record. Analysis of the linear model (including its uncertainties) suggests no significant decrease in the response of the natural carbon cycle. Recent data points rather to an increase. We apply the same linear diagnostic to fluxes from atmospheric inversions. Flux responses show clear regional and seasonal patterns driven by terrestrial uptake in the northern summer. Ocean fluxes show little or no linear response. Terrestrial models show clear responses, agreeing globally with the inversion responses, however the spatial structure is quite different, with dominant responses in the tropics rather than the northern extratropics.