GMDGeoscientific Model DevelopmentGMDGeosci. Model Dev.1991-9603Copernicus PublicationsGöttingen, Germany10.5194/gmd-10-2057-2017Historical greenhouse gas concentrations for climate modelling (CMIP6)MeinshausenMaltemalte.meinshausen@unimelb.edu.auhttps://orcid.org/0000-0003-4048-3521VogelElisabethNauelsAlexanderhttps://orcid.org/0000-0003-1378-3377LorbacherKatjaMeinshausenNicolaiEtheridgeDavid M.https://orcid.org/0000-0001-7970-2002FraserPaul J.MontzkaStephen A.https://orcid.org/0000-0002-9396-0400RaynerPeter J.https://orcid.org/0000-0001-7707-6298TrudingerCathy M.https://orcid.org/0000-0002-4844-2153KrummelPaul B.https://orcid.org/0000-0002-4884-3678BeyerleUrshttps://orcid.org/0000-0002-6464-0838CanadellJosep G.https://orcid.org/0000-0002-8788-3218DanielJohn S.EntingIan G.https://orcid.org/0000-0002-9667-3271LawRachel M.https://orcid.org/0000-0002-7346-0927LunderChris R.O'DohertySimonhttps://orcid.org/0000-0002-4051-6760PrinnRon G.ReimannStefanhttps://orcid.org/0000-0002-9885-7138RubinoMaurohttps://orcid.org/0000-0002-8721-4508VeldersGuus J. M.https://orcid.org/0000-0002-6597-7088VollmerMartin K.https://orcid.org/0000-0001-5569-9718WangRay H. J.https://orcid.org/0000-0002-1550-3239WeissRayhttps://orcid.org/0000-0001-9551-7739Australian-German Climate & Energy College, The University of Melbourne, Parkville, Victoria, AustraliaDepartment of Earth Sciences, The University of Melbourne, Parkville, Victoria, AustraliaPotsdam Institute for Climate Impact Research, Potsdam, GermanySeminar for Statistics, Swiss Federal Institute of Technology (ETH Zurich), Zurich, SwitzerlandCSIRO Climate Science Centre, Oceans and Atmosphere, Aspendale, Victoria, AustraliaNOAA, Earth System Research Laboratory, Global Monitoring Division, Boulder, Colorado, USAInstitute for Atmospheric and Climate Science, Swiss Federal Institute of Technology (ETH Zurich), Zurich, SwitzerlandGlobal Carbon Project, CSIRO Oceans and Atmosphere, Canberra, ACT, AustraliaNOAA, Earth System Research Laboratory, Chemical Sciences Division, Boulder, Colorado, USAThe University of Melbourne, Victoria, AustraliaNorwegian Institute for Air Research, Kjeller, NorwayUniversity of Bristol, Bristol, UKMIT, Cambridge, Massachusetts, USAEmpa, Swiss Federal Laboratories for Materials Science and Technology, Laboratory for Air Pollution and Environmental Technology, Dübendorf, SwitzerlandDipartimento di matematica e fisica, Seconda Università degli studi di Napoli, Caserta, ItalyNational Institute for Public Health and the Environment (RIVM), Bilthoven, the NetherlandsSchool of Earth and Atmospheric Sciences, Georgia Institute of Technology, Atlanta, Georgia, USAScripps Institution of Oceanography, La Jolla, California, USAretiredMalte Meinshausen (malte.meinshausen@unimelb.edu.au)31May2017105205721162July20165August201623February201717March2017This work is licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by/3.0/This article is available from https://gmd.copernicus.org/articles/10/2057/2017/gmd-10-2057-2017.htmlThe full text article is available as a PDF file from https://gmd.copernicus.org/articles/10/2057/2017/gmd-10-2057-2017.pdf
Atmospheric greenhouse gas (GHG) concentrations are at
unprecedented, record-high levels compared to the last 800 000 years. Those
elevated GHG concentrations warm the planet and – partially offset by net
cooling effects by aerosols – are largely responsible for the observed
warming over the past 150 years. An accurate representation of GHG
concentrations is hence important to understand and model recent climate
change. So far, community efforts to create composite datasets of GHG
concentrations with seasonal and latitudinal information have focused on
marine boundary layer conditions and recent trends since the 1980s. Here, we
provide consolidated datasets of historical atmospheric concentrations (mole
fractions) of 43 GHGs to be used in the Climate Model Intercomparison Project
– Phase 6 (CMIP6) experiments. The presented datasets are based on AGAGE and
NOAA networks, firn and ice core data, and archived air data, and a large set
of published studies. In contrast to previous intercomparisons, the new
datasets are latitudinally resolved and include seasonality. We focus on the
period 1850–2014 for historical CMIP6 runs, but data are also provided for
the last 2000 years. We provide consolidated datasets in various
spatiotemporal resolutions for carbon dioxide (CO2), methane
(CH4) and nitrous oxide (N2O), as well as 40 other GHGs,
namely 17 ozone-depleting substances, 11 hydrofluorocarbons (HFCs),
9 perfluorocarbons (PFCs), sulfur hexafluoride (SF6), nitrogen
trifluoride (NF3) and sulfuryl fluoride (SO2F2). In
addition, we provide three equivalence species that aggregate concentrations
of GHGs other than CO2, CH4 and N2O, weighted by
their radiative forcing efficiencies. For the year 1850, which is used for
pre-industrial control runs, we estimate annual global-mean surface
concentrations of CO2 at 284.3 ppm, CH4 at
808.2 ppb and N2O at 273.0 ppb. The data are
available at https://esgf-node.llnl.gov/search/input4mips/ and
www.climatecollege.unimelb.edu.au/cmip6. While the minimum CMIP6
recommendation is to use the global- and annual-mean time series, modelling
groups can also choose our monthly and latitudinally resolved concentrations,
which imply a stronger radiative forcing in the Northern Hemisphere winter
(due to the latitudinal gradient and seasonality).
Introduction
Emissions from the burning of fossil fuels, deforestation,
agricultural activities and the production of synthetic greenhouse gases
(GHGs) are the primary reasons for the observed increases in GHG
concentrations, defined as mole fractions in dry air. The elevated GHG
concentrations induce a radiative forcing that in turn would cause more than
the observed recent global warming if it were not for the cooling effect by
aerosols (Fig. TS.10 in IPCC WG1 AR5; IPCC, 2013). An accurate quantification of anthropogenic and
natural climate drivers is crucial for general circulation and Earth system
models (ESMs). Simulations by these models for the historical time periods,
e.g. since 1850, can only be meaningfully compared to observations (e.g.
surface temperature, ocean heat uptake) to the degree that input forcings are
an accurate representation of the past. The difficulty with many
anthropogenic climate drivers is that their global-mean magnitude, their
latitudinal gradient and seasonal cycle are uncertain further back in time,
even for the main GHGs carbon dioxide (CO2), methane (CH4)
and nitrous oxide (N2O). Systematic observational efforts started in
1957–1958, measuring CO2 at the South Pole and Mauna Loa
observatories (Keeling et al., 2001). Measurements of archived air, firn air
and ice cores from both polar regions provide records for the
pre-observational time. To date, reconstructions of millennial global-mean
time series based on ice and firn data have been performed, e.g. for
CO2 over recent millennia (Ahn et al., 2012; MacFarling Meure et
al., 2006; Rubino et al., 2013). For the more recent past, several studies
investigated firn and ice data to constrain halocarbons (Buizert et
al., 2012; Martinerie et al., 2009; Mühle et al., 2010; Sturrock et
al., 2002; Trudinger et al., 2016), some of them with hemispheric resolution.
In terms of latitudinally resolved monthly data, there have only been a few
synthesis products, namely for CO2, CH4 and N2O over
the instrumental record over the past 20–40 years (NOAA, 2013; NOAA ESRL
GMD, 2014a, b, c). For this recent past, the World Data Centre for Greenhouse
Gases (WDCGG) (ds.data.jma.go.jp/gmd/wdcgg/) also provides a synthesis with
global and hemispheric means for CO2, CH4 and N2O
(Tsutsumi et al., 2009). In light of the observational gaps further back in
time, some studies, such as Keeling et al. (2011), used linear regressions
between fossil fuel use and latitudinal CO2 concentration trends to
separate natural from anthropogenically induced effects, which allows us to
infer latitudinal gradients back in time.
In previous climate model inter-comparison projects (Meehl et al., 2005),
global-mean concentrations have been prescribed (Meinshausen et al., 2011),
with some models constraining internally generated fields of GHG
concentrations to match those global-mean values. Here, we update those
global-mean and annual-mean GHG concentration time series for the historical
period over years 0–2014, with “historical” simulations in the CMIP6 model
intercomparison (Eyring et al., 2016) focussed on the most recent period,
1850–2014. In addition, we provide hemispheric and latitudinal
monthly-resolved fields for 43 GHGs in total. In the past, the large
latitudinal and seasonal gradient of GHG radiative forcing has not been
consistently applied to model radiative forcing and climate change. The new
datasets provide a more consistent starting point for climate model
experiments. The monthly and latitudinal resolution of this new GHG dataset
is designed to have a similar resolution to the monthly solar forcing
(Matthes et al., 2016) and monthly and latitudinally resolved ozone and
aerosol abundances. Many GHGs also have significant longitudinal (land–ocean)
and diurnal variations but we do not attempt to resolve them. Neither do we
provide vertical gradients of the GHG concentrations, and we only discuss
possible vertical extension methods (Sect. 4.1) in case models do not have
their own methods to derive vertical gradients.
In this study, we compile one possible reconstruction of latitudinally and
monthly resolved fields, as well as global annual means of surface GHG
concentrations for 43 gases from year 0 to 2014, as input for the forthcoming
model inter-comparison experiments that are part of the Phase-6 Coupled Model
Intercomparison Project (CMIP6) (Eyring et al., 2016). Specifically, we
provide the pre-industrial control runs at 1850 forcing levels (“picontrol”),
the experiment with abruptly quadrupled CO2 concentrations
(“abrupt4x”), the standard experiment of a 1 % annual CO2
concentration increase (“1pct2co2”), and the historical runs that are driven
with best-guess estimates of historical forcings since 1850. Species that are
radiatively less important than CO2, CH4 and N2O
(“importance” here being measured as radiative forcing exerted in year 2014
compared to 1750) are provided individually as well as aggregated as HFC-134a
and CFC-12 equivalent concentrations. The description of the datasets geared
towards CMIP6 modelling groups is provided in Sect. 4, including a
description of available data formats and CMIP6 minimum recommendations.
The design principle for this long-term dataset is to provide a plausible
reconstruction of past GHG concentrations to be used in climate models. Using
various gap-filling procedures, reconstruction and extensions, this dataset
aims to reflect observational evidence of both recent flask and in situ
observations from the worldwide network of NOAA ESRL and AGAGE stations, as
well as Antarctic and Greenland ice core and firn data over the last 2000
years, where available. Furthermore, many detailed literature studies (Arnold
et al., 2013, 2014; Aydin et al., 2010; Butler et al., 1999; Ivy et
al., 2012; Martinerie et al., 2009; Montzka et al., 2015; Mühle et
al., 2010; Oram et al., 2012; Sturrock et al., 2002; Trudinger et al., 2004,
2016; Velders and Daniel, 2014; Vollmer et al., 2016; Worton et al., 2006)
for radiatively less important species are compared with our data product in
the fact-sheet figures for the specific gases (Table 12 and Figs. S1–S40 in
Supplement), or synthesized where direct observational records from the above
networks were not available.
The predominant climate effect of GHG increases is captured by the global- and
annual-mean concentrations throughout the atmosphere. The surface global- and
annual-mean concentrations provided here, in combination with the models'
approximations for the vertical concentration profile, are the minimum
standard for CMIP6 models. Assimilating a latitudinally and seasonally
resolved data product serves two purposes. Firstly, to derive the global and
annual means from sparse observations rests on knowledge or assumptions about
spatial and seasonal distributions. Secondly, to open the opportunity for
some modelling groups to go beyond the prescription of global- and annual-mean
concentrations.
Undoubtedly, some of the assumptions stretch into unknown territory, such as
the seasonality of the CO2 concentrations in pre-observational times
or the time variability of latitudinal gradients, let alone the
higher-frequency fluctuations of global-mean concentrations during the time
when only ice core data are available. Errors in the historical forcing do
propagate and can hinder the comparison between observations and models. This
study therefore had to find a workable compromise between providing a
complete dataset that covers the whole time and space domain and being as
close as possible to sometimes sparse observations. Hence, the remaining
uncertainties in concentration gradients should be kept in mind, although
they might not be of primary concern in regard to the inter-comparison aspect
of the multi-model ensemble runs. Thus, while our CMIP6 community dataset
will improve on the global- and annual-mean time-series prescribed for the
last set of CMIP5 experiments on a number of key aspects, many research
questions remain open.
The underlying reasons for meridional gradients of annual-mean concentrations
are manifold (Keeling et al., 1989a, b; Tans et al., 1989). For one, the
sources of anthropogenic GHGs from fossil fuel burning and cement production
or industrial activities are not evenly distributed with latitude, but
concentrated in the mid-northern land masses. In the case of CO2,
emissions from deforestation are not uniformly distributed with latitude
either. The pattern of land-use-related emissions is even less stationary,
with CO2 uptakes and sources predominantly focussed in the
mid-northern latitudes up until earlier in the 20th century, shifting more
towards lower latitudes in recent decades (Hurtt et al., 2011). This study
uses an approach based on simple regressions that implicitly rest on the
assumption of a fixed pattern approximation (such as Keeling et al., 2011).
One complication to retrieving the latitudinal pre-industrial CO2
concentration profile is that CO2 fertilization and temperature
effects on the carbon cycle, over both ocean and land, change both the
magnitude and spatial patterns of natural CO2 fluxes. Lastly, both
the diurnal and seasonal cycle of photosynthesis and its covariance with
vertical atmospheric mixing can have a pronounced effect on measured surface
concentrations (the so-called “rectifier” effect), increasing annual mean
northern hemispheric CO2 surface concentrations by up to
2.5 ppm (Denning et al., 1999).
To dissect and analyse the different causes for temporal and spatial
heterogeneity in surface concentrations, a rich body of literature has
analysed observed latitudinal and seasonal gradients with various inversion
techniques. Recent research provides a clearer picture in regard to the
causes of the change in seasonality of CO2 concentrations (Forkel et
al., 2016), a topic researched already in 1989 (Kohlmaier et al., 1989) based
on the CO2 fertilization effect on northern hemispheric terrestrial
biota. Generally, the research into meridional and seasonal variations
employs various atmospheric inversion techniques (Enting and Mansbridge,
1991, 1989; Enting et al., 1995; Enting, 1998; Rayner et al., 1999) to match
observed concentrations with source and sink pattern estimates (Baker et
al., 2006; Enting et al., 1995; Gurney et al., 2002, 2003, 2004; Keeling et
al., 1989a, b; Peylin et al., 2013; Rayner et al., 1999; Tans et al., 1989,
1990a). Similarly to CO2, the spatial variation in CH4
concentrations is used for model inversions to infer sources and sinks (Fung
et al., 1991; Kirschke et al., 2013).
There is a substantial lack of observational evidence of both seasonality and
latitudinal CO2 gradients in pre-industrial times. Given that
atmospheric CO2 is not well preserved in the Greenland ice (Anklin et
al., 1995; Barnola et al., 1995), the pre-observational north–south gradient
cannot be inferred or derived from the Greenland and Antarctic ice core
records. Alternatively, understanding biospheric sink and source dynamics
could provide vital evidence to infer pre-industrial surface concentration
patterns. In this study, we do not employ any such inversion models or
results, and only note that our pre-industrial meridional and seasonal
variations should be regarded as highly uncertain. However, some plausibility
of the CO2 gradients is gained by comparison with some model studies
(Sect. 5). High-latitude records of CH4 are available from both
hemispheres (MacFarling Meure et al., 2006; Mitchell et al., 2013; Rhodes et
al., 2013), allowing us to estimate pre-industrial large-scale CH4
concentration gradients.
Methods
To achieve the goals of this study, several analytical steps were taken to
assimilate the observational data. Global-mean and annual-mean concentrations
are of primary interest, but the discussion also covers latitudinal and
seasonal variations. The assimilation procedure for sparse observational data
requires this spatio-temporal heterogeneity to be accounted for to derive
global and annual means.
We consider a total of 43 GHGs: CO2, CH4, N2O, a
group of 17 ozone-depleting substances (ODSs) made up of 5 CFCs (CFC-12, CFC-11,
CFC-113, CFC-114, CFC-115), 3 HCFCs (HCFC-22, HCFC-141b, HCFC-142b),
3 halons (Halon-1211, Halon-1301, Halon-2402), methyl chloroform
(CH3CCl3), carbon tetrachloride (CCl4), methyl chloride
(CH3Cl), methylene chloride (CH2Cl2), chloroform
(CHCl3), and methyl bromide (CH3Br), and 23 other fluorinated
compounds made up of 11 HFCs (HFC-134a, HFC-23, HFC-32, HFC-125, HFC-143a,
HFC-152a, HFC-227ea, HFC-236fa, HFC-245fa, HFC-365mfc, HFC-43-10mee), 9 PFCs (CF4, C2F6, C3F8, C4F10,
C5F12, C6F14, C7F16, C8F18, and
c-C4F8), NF3, SF6, and SO2F2.
All concentrations given here are dry air mole fractions and we use “mole
fractions” and “concentrations” interchangeably and synonymously with
“molar mixing ratios”. For simplicity, we denote the dry air mole fractions
“µmolmol-1”, “nmolmol-1” and
“pmolmol-1” as parts per million (ppm), parts per billion (ppb)
and parts per trillion (ppt), respectively. Note that dry air mole fractions
are independent of temperature and pressure, while volume mixing ratios (e.g.
ppmv) for mixtures of non-ideal real gases are not, and at standard
temperature and pressure conditions can differ significantly from their
corresponding mole ratios.
Summary of assimilation approach
We perform three consecutive steps to synthesize the global mole fraction
fields over the full-time horizon from year 0 to year 2014. First, we
aggregate the available observational data over the recent instrumental
period. Second, we estimate three components of the global surface
concentration fields from these data, namely global-mean mole fractions,
latitudinal gradients and seasonality. Third, we extend those components back
in time with – inter alia – ice core or firn data. The full historical GHG
concentration field can then be generated by the time-varying components.
Under this basic assimilation model, the concentration C^l,t at any point in time t and in a latitudinal band l can be
written as follows:
C^l,t=Cglobal‾t+S^l,my+L^ly,
where Cglobal‾t is the global-mean dry
air mole fraction at time t, S^l,m is the seasonality in each
latitude l and month m, and L^ly is the
latitudinal annual-mean deviation in year y at latitude l. With this
assimilation model, and the optimal low rank approximations of seasonality
and latitudinal gradients, a regularization of the data is performed by a
principal components analysis, which creates a degree of robustness against
data gaps or outliers. Other methods, like a harmonic representation of
station data, have, in principle, a similar smoothing and regularization
effect (Masarie and Tans, 1995), although quantitative differences exist
(Sect. 5.4).
A detailed data-flow diagram of how the historical GHG mole fractions are
derived in this study is provided in Fig. 1. The subsequent section will
describe the method step-by-step as indicated by the green circles in Fig. 1
and also tabulated for the three main GHGs in Table 1.
Data-flow diagram of how
historical GHG concentrations are derived in this study. See text.
Step 1: aggregating raw station data
Atmospheric measurements are taken in remote environments or locations that
are closer to pollution sources, in continental or marine areas, at different
times of the day or night, at different altitudes, and in different seasons of
the year, often using different calibration scales. This poses challenges for
any synthesis of observational data.
Derivation and construction of CMIP6 concentration fields for
CO2, CH4 and N2O, as shown in Fig. 1 and described in
Sect. 2.
GasTimeMain data sourceGlobal andSeasonalitySeasonalityLatitudinalperiodannual-meanS^l,mchange ΔSl,mgradient L^Cglobal‾CO21984 toNOAA ESRL Carbon CycleCalculated based onMean overLeading EOF ofTwo leading EOFs and2013/14Cooperative Global Airobservational data1984–2013residuals fromtheir scores derivedSampling Network,source (Sect. 2.1.3).period.observation.from residuals to1968–2014.1984–2013 observationsVersion: 2015-08-03,(2014: scores optimisedmonthly station averagesto match observations)(Dlugokencky, 2015b;NOAA ESRL GMD, 2014a, b, c).BeforeSee text.Optimized to matchKept constantRegressed againstThe score for1984Updated Law Domesmoothed medianas above.product of CO2EOF1 is regressed(Etheridge et al., 1998;approximation (Sect. 2.1.6)concentration andagainst global annualMacFarling Meure et al., 2006;of Law Dome recordsurface airfossil fuel andRubino et al., 2013) and(0–1966) and Mauna Loatemperature changeindustry emissionsannual-mean MLO station datarecord (1959–1984)since pre-industrial(Boden et al., 2013).(Keeling et al., 1976).with interpolationtimes.Score for EOF2 linearlybetween 1955 and 1958.returned to zero in 1850.See Fig. 9c.CH41985 toAGAGE monthly station means,Calculated based onMean overAssumed zero.Two leading EOFs and2013/14incl. pollution events (“.mop”)observational data1985–2013their scores derived(Cunnold et al., 2002), andsource (Sect. 2.1.3).period.from residuals fromNOAA ESRL monthly stationApplied asobservations (2014:data (Dlugokencky, 2015a).relativeoptimized to matchseasonality.observational data).BeforeUpdated Law DomeOptimized to matchThe score for EOF1 is1985(Etheridge et al., 1998;smoothed Law Domeregressed against globalMacFarling Meure et al., 2006)record and NEEMannual fossil fuel andand NEEM (Rhodes et al., 2013).firn data.industry emissions(Gütschow et al., 2016).Score for EOF2 keptconstant before in situinstrumental period.N2O1990 toAGAGE monthly station means,Calculated based onMean overAssumed zero.Two leading EOF2013/14incl. pollution eventsobservational data1990–2013and their scores(Prinn et al., 1990) andsource (Sect. 2.1.3).period.derived from residualscombined nitrous oxide dataApplied asfrom observations(monthly station averages)relative(2014: optimized tofrom the NOAA/ESRL Globalseasonality.match observational data).Monitoring Division.BeforeUpdated Law DomeOptimized to match smoothedScore for EOF1 and1990(MacFarling Meure et al., 2006)Law Dome record until 1968.2 kept constantuntil 1968.Interpolation until 1986 withbefore in situoptimization to sparseinstrumental period.observational data until 1990.
Raw data used for CO2 surface concentration field
derivation.
DatasetReference/URLStations/locationUsed forDescription/filteringNOAA ESRL GMDConway et al. (1988, 1994),81 stations of the surfaceObservational periodThis study usedSurface FlaskKomhyr et al. (1985, 1983),flask network*:estimation of globalmonthly average datadataTans et al. (1989, 1990a, b),ABP, ALT, AMS, AOC,mean, latitudinalthat uses all sampleThoning et al. (1995, 1989,ASC, ASK, AVI, AZR,gradient, seasonalitypoints which have1987),BAL, BHD, BKT, BME,and seasonality changean “accepted” flag,Zhao and Tans (2006)BMW, BRW, BSC, CBA,over 1984–2013.i.e. initial two dotsCGO, CHR,CIB, CMO,Optimization of global(“..*”) in the threeCPT, CRZ, DRP, DSI,mean and latitudinaldigit flag.EIC, GMI, GOZ, HBA,gradient in 2014HPB, HSU, HUN, ICE,and before 1984.IZO, KCO, KEY, KUM,KZD, KZM, LEF, LLB,LLN, LMP, MBC, MEX,MHD, MID, MKN, MLO,NAT, NMB, NWR, OPW,OXK, PAL, PAO, POC,PSA, PTA, RPB, SCS,SDZ, SEY, SGI, SGP,SHM, SMO, SPO, STC,STM, SUM, SYO, TAP,THD, TIK, USH, UTA,UUM, WIS, WLG, WPC, ZEPLaw DomeUpdated data fromLaw Dome ice coreUsed as inputEtheridge et al. (1998, 1996),for piecewiseRubino et al. (2013),third-degree polynomialMacFarling Meure et al. (2006)smoothing overremainder ofyears 0 to 1966.
* See station descriptions here:
http://www.esrl.noaa.gov/gmd/dv/site/site_table.html.
The observational station data over the recent decades used in this study are
predominantly sourced from the networks operated by NOAA (Earth System
Research Laboratories: ESRL) and AGAGE. In general, we use monthly station
data provided by the respective networks as a starting point. In the case of
the AGAGE network, monthly averages are provided with and without pollution
events (http://agage.eas.gatech.edu/data_archive/agage/ and
http://cdiac.ornl.gov/ftp/ale_gage_Agage/AGAGE/). We chose the monthly
averages that include pollution events (file-endings “.mop”, with the
exception of CH2Cl2, in which case data issues warranted the use of
monthly station averages without pollution events). The approach that we do
not restrict our source data to background conditions is consistent with our
approach elsewhere – and the NOAA network monthly station averages – which
do not screen out pollution events (although the dominant number of NOAA
flask measurements will likely be biased towards background conditions rather
than pollution events owing to their location and sampling protocols at most
sites focussed on collecting background air). In total, CO2 data from
81 stations from the NOAA flask network and 3 stations from the NOAA in situ
data stations are used (Table 2). For CH4, 87 sampling stations from
the NOAA flask network and 5 stations from the AGAGE in situ network are
compiled (Table 3). For N2O, data from flask and in situ measurements
at 13 stations of the NOAA HATS global network are combined with data from 5
stations from the AGAGE network (Table 4). For other gases, the AGAGE and
NOAA coverage and time frames vary, with individual station's codes provided
in the “f” panels of the individual gases' fact sheets (Figs. S1–S40). We provide
references to the used NOAA and AGAGE data in Table 12.
Raw data used for CH4 surface concentration field
derivation.
DatasetReference/URLStations/locationUsed forDescription/filteringNOAA ESRLDlugokencky et al. (2009,87 stations of the surfaceObservational periodThis study used monthlyGMD Surface1994a, c, 1998, 2001,flask networka:estimation of globalstation averages thatFlask data2005, 2015a),ABP, ALT, AMS, AMT,mean, latitudinalinclude all sample pointsLang (1990a, b, 1992),AOC, ASC, ASK, AVI,gradient, seasonalitywhich have an “accepted”Steele et al. (1987, 1991,AZR, BAL, BHD, BKT,and seasonality changeflag, i.e. initial two1992)BME, BMW, BRW, BSC,over 1984–2013.dots (“..*”) inCBA, CGO, CHR, CIB,Optimization of globalthe three-digit flag.CMO, CPT, CRZ, DRP,mean and latitudinalDSI, EIC, GMI, GOZ,gradient in 2014HBA, HPB, HSU, HUN,and before 1984.ICE, ITN, IZO, KCO,KEY, KPA, KUM, KZD,KZM, LEF, LLB, LLN,LMP, MBC, MCM, MEX,MHD, MID, MKN, MLO,NAT, NMB, NWR, NZL,OPW, OXK, PAL, PAO,POC, PSA, PTA, RPB,SCS, SDZ, SEY, SGI,SGP, SHM, SIO, SMO,SPO, STM, SUM, SYO,TAP, THD, TIK, USH,UTA, UUM, WIS, WKT,WLG, WPC, ZEPAGAGE GC-MDPrinn et al. (2000b)AGAGE GC-MD networkb:The monthly stationCGO, MHD, RPB,averages that includeSMO, THDpollution events(“.mop” fileendings in the case ofAGAGE) were used.Law DomeUpdated data fromLaw Dome ice core atLong-term high-latitudeEtheridge et al. (1998),-66.73∘ south.Southern HemisphereMacFarling Meure et al. (2006)reference point withpiecewise third-degreepolynomial smoothingsmoothing over years154 to 1974.EPICABarbante et al. (2006),Dronning Maud LandUsed as input forDronningCapron et al. (2010)ice corepiecewise third-degreeMaud Landpolynomial smoothingice coreover remainder ofof years 0 to 153.NEEMDahl-Jensen et al. (2013),NEEM ice coreUsed for optimizationGreenlandRhodes et al. (2013)Greenland dataof global mean andlatitudinal gradientscore of EOF1 overtimescale fromyear 0 to 1984, withlinear interpolationof the score in betweenavailable 5-yearlyNEEM data points(Sect. 2.1.4).
a NOAA station descriptions here:
http://www.esrl.noaa.gov/gmd/dv/site/site_table.html.
b AGAGE station descriptions here:
https://agage.mit.edu/global-network.
Raw data used for N2O surface concentration field
derivation.
DatasetReference/URLStations/locationUsed forDescription/filteringNOAA ESRLCombined N2O data from13 stations ofObservational period estimationThis study uses station averages, whichGMD Surfacethe NOAA/ESRL Globalthe NOAA HATS globala:of global mean, latitudinalinclude all sample points which haveFlask dataMonitoring Division,alt, brw, cgo, kum,gradient, seasonality andan “accepted” flag, i.e. initial twoftp://ftp.cmdl.noaa.gov/hats/n2o/,mhd, mlo, nwr, psa,seasonality change over 1990–2013.dots (“..*”) in the three-digit flag.file date: “Wednesday, Aug 19 2015,smo, spo, sum, tdf, thdOptimization of global mean and(“.mop” file endings in2:40:55 PM”latitudinal gradient in 2014.case of AGAGE).AGAGE GC-MDPrinn et al. (1990, 2000b)AGAGE GC-MD networkb:CGO, MHD, RPB, SMO, THDLaw DomeUpdated data fromLaw Dome ice core atLong-term high-latitude Southern HemisphereMacFarling Meure et al. (2006)-66.73∘ southreference point with piecewise third-degreepolynomial smoothing over years 155 to 1974.GapSparse data availability in the period 1968–1986 suggestedagainst optimizations of global-means with annualdata points, which is why an interpolation between 1968(starting from smoothed Law Dome record) and 1986 (endingwith optimized global mean to fit observational data) was assumed.
a NOAA station descriptions here:
http://www.esrl.noaa.gov/gmd/dv/site/site_table.html.
b AGAGE station descriptions here:
https://agage.mit.edu/global-network.
Calibration scales, i.e. the standardized gas mixtures that allow us to
calibrate the instrumentation used for in situ or flask measurements, differ between the NOAA and AGAGE networks. Gas measurements on different
measurement scales, even when using the same scales by different
laboratories, are subject to uncertainties (Hall et al., 2014). For
halocarbons, the difference in calibration scales has been estimated as
small, but not negligible, i.e. within 2.5 %, often within 1 %
(Rhoderick et al., 2015).
While we use the station data that have already been converted to the latest
scales of the respective networks, some older comparison data products use
previous scales (like the one published in the latest ozone assessment
report; WMO, 2014). Thus, where necessary, we convert those older data to the
newer scales. For 7 gases, we use scale conversion factors to convert to the
SIO14 scale, specifically 1.0826 for HFC-125 (from University of Bristol
scale: UB98), 1.1226 for HFC-227ea (from Empa-2005), 1.1970 for HFC-236fa
(from Empa-2009-p) and 1.1909 for HFC-245fa (from Empa-2005), 1.1079 for
HFC-365-mfc (from Empa-2003), 1.0485 for HFC-43-10-mee (from SIO-10-p) and
0.9903 for CH2Cl2 (from UB98), with all conversion factors taken
from the Appendix in WMO (2012).
Apart from those scale conversions to the latest NOAA and SIO scales
mentioned above, we only make sure that the three main gases are each on a
unified scale. In the case of CO2, we source all our CO2
station data from the NOAA network, which means no scale conversion is
necessary. In the case of CH4, we account for different calibration
scales by converting AGAGE CH4 data (Tohuko University scale) to the
NOAA scale (NOAA04) (multiplication by 1.0003). In the case of N2O,
both the AGAGE (SIO1998) and NOAA network calibration scales (NOAA-2006) are
compatible without the need for a conversion factor (WMO, 2012). The Law Dome
data used here (Etheridge et al., 1998, 1996; MacFarling Meure et al., 2006;
Rubino et al., 2013) have been updated for minor dating changes and placed on
current NOAA scales (http://www.esrl.noaa.gov/gmd/ccl/index.html).
Apart from those three main gases, we do not apply further scale conversions.
Thus, given that our results are based on a mixture of the AGAGE and NOAA
networks, they are de facto a weighted average between the respective two
standard scales (SIO and NOAA) for each gas. The effective weight in this
“weighted mean” depends on the station numbers and each network's station
distribution, given that our assimilation method implicitly gives less weight
to stations that are geographically close, i.e. in the same
latitude–longitude box. This mixture of scales is different from previous
studies that either applied empirical scale conversions (so that global-mean
or station averages are identical) or used both scales in parallel to
estimate a measurement uncertainty (WMO, 2014), for example when estimating
emissions with inverse techniques. Mathematically, our approach is similar to
an approach where a station-by-station scale conversion would be applied
towards an intermediate scale between NOAA and AGAGE. However, for some
applications, this approach is clearly a limitation as it hides the
uncertainty and would for example warrant a new data assimilation if one
network updates its scales (Sect. 6). The reason this “weighted mean”
approach is chosen in the context of this study is that we intend to
reconstruct a single concentration history making use of the station data
from both major measurement networks without giving preference to one or the
other measurement scale. Given that different scales between the two major
networks result in differences that are generally less than 2 % (and are
often for radiatively less important substances), this “middle of the road”
approach seems justified given the other uncertainties in climate model
forcings (vertical distributions, radiative forcing routines, other radiative
forcings such as aerosols). Any conversion to a single scale would ease
comparisons, but would not be able to address the inherent measurement
uncertainty, and might even face a stronger bias (if the two scales SIO and
NOAA are equally plausible representations of the “truth”) (Sect. 6).
However, in regard to the time of the day, month or year, we do not apply
interpolation or adjustment techniques other than a simple monthly binning of
all available data (see Sect. 2.1.2). The spatial and temporal coverages of
the raw data used in this study are depicted in Figs. 2, 3 and 4 for
CO2, CH4 and N2O data, respectively.
Availability of instrumental carbon dioxide data from 1968 to 2015
from the NOAA ESRL network, shown as data samples per month, per latitudinal
band (a–l) and per longitudinal bin within each latitudinal band.
Availability of instrumental CH4 data from 1983 to 2015 from
the AGAGE and NOAA ESRL networks, shown as data samples per month, per
latitudinal band (a–l) and per longitudinal bin within each
latitudinal band.
Availability of instrumental N2O data from 1983 to 2015 from
the AGAGE and NOAA ESRL networks, shown as data samples per month, per
latitudinal band (a–l) and per longitudinal bin within each
latitudinal band.
Step 2–4: binning and spatial interpolation
We employ a simple monthly mean binning of all available data, separately
averaged for each station. Stations with more than one measurement program,
e.g. with flask and in situ programs, are treated as distinct stations. Thus,
the monthly average of an in situ data series with 1000 measurement points
gets the same weight as the monthly average from a flask measurement program
with few observations. In each latitudinal–longitudinal box, all available monthly
mean station data are averaged, with the mean being assigned to the grid box
centre before employing a 2-D spatial interpolation to extend available data
points to longitudinal and latitudinal grid points that do not have observed
data for any particular month. Our method provides equal weight to each
station within a longitude–latitude box, no matter whether the station
reports a few flask measurement samples or sub-hourly in situ instrument
readings in each month. The chosen assimilation grid has 72 boxes with 12
equal-latitude bands of 15∘ and 6 longitudinal bands of 30∘.
Following the temporal monthly binning and subsequent spatial linear
interpolation, we average all data across the longitudes to obtain 12
latitudinally resolved monthly time series of surface concentrations.
Step 5: global-mean mole fractions
The annual global mean concentration Cglobal‾y is derived as the area-weighted arithmetic mean of the binned
latitudinal data (small grey “5” in Fig. 1). In addition to the annual
global mean, a time series of monthly values is derived as a smooth spline
interpolation between the annual data points, with the constraint of being
mean-preserving, i.e. that the average of the 12-monthly values is again the
global annual average value initially derived. Thus, the trend in the mole
fraction data is reflected in the global-mean time series from month to
month.
Step 6: latitudinal gradient
The annual-mean latitudinal gradients are derived as first and second
empirical orthogonal function (EOFs) from the annual-average residuals per
latitude after subtracting the global annual mean (step 6 in Fig. 1). Let
G be the n×m matrix of n years of observations and
m latitudinal boxes, then G can be decomposed into its EOFs and
scores by calculating the singular value decomposition of G=UDVT, where U and V are orthogonal
matrices in Rn and Rm, respectively, and D is the
n×m matrix with non-zero elements only on the diagonal. EOFi is
the ith column of V, and the score Si(y) of EOFi in
year y is given as the (y, i) entry of the UD matrix. In
other words, the EOFs are the eigenvectors of the Gram matrix 1/m×(G′G), and the scores are the projections of the
observations G onto the EOFs.
Those EOF scores are regressed with suitable predictors or extended as
constants. Thus, the term L^y is the optimal low rank
approximation of the latitudinal deviations from the global mean in year y.
It is composed of the leading EOFs of latitudinal annual-mean variation
multiplied with the observed or regressed scores S of that year y.
L^y=∑i=1imaxEOFiSi(y),
with imax being 1 or 2 if only the leading or the two leading EOFs are taken
into account, respectively.
Step 7–10: seasonality
The seasonality fulfils the condition that the sum of seasonal variations at
each latitude is zero over the year, i.e.
∑m=112S^l,m=0.
This seasonality S^l,m(t) at time t is calculated for most gases as
the relative seasonality
dS^l,mdCglobal, i.e. the
monthly deviation in mole fraction divided by the global-mean mole fraction,
multiplied by the global-mean mole fraction at time t (steps 7 and 10 in
Fig. 1).
An exception is the case of CO2 (steps 8 and 9 in Fig. 1). In this
case, the seasonality pattern over the observational period is held fixed as
absolute mole fractions, i.e. not relative to the global mean. However, the
residuals between this fixed seasonality and the seasonality, which is
derived from the observations by subtracting the latitudinal averages, are
used for a singular value decomposition. Let Rl,m(t) be the residuals at
latitude l and month m at time t; the optimal lower rank representation
of this seasonal change is then given by the first EOF of the gram matrix
1/n×R′R, with n being the number of observational
data points. The derived score, i.e. the projection of the residuals onto the
first EOF, is regressed against a time series P, a composite of global-mean
CO2 concentration and historical observed global-mean surface air
temperatures. This simplified choice is made because previous studies
identified warmer temperatures and elevated CO2 mole fractions as
dominant reasons for increased seasonality (Forkel et al., 2016; Graven et
al., 2013; Welp et al., 2016), although anthropogenically induced cropland
productivity increases are also suggested to play some role (Gray et
al., 2014). Specifically, P is assumed to be a composite of the product and
the sum of normed global-mean surface air temperature and normed CO2
mole fraction deviations from pre-industrial levels. The temperature and mole
fraction deviations are normalized such that the 2000–2010 deviation from
the 1850–1880 base period is set to 1. Thus, the regressor P can be
described as follows:
P(t)=ΔT(t)⋅ΔC(t)2+ΔT(t)+ΔC(t)2,
with ΔT being the temperature deviation from the 1850–1880 period,
specifically
ΔTt=Tt-∑t=18501880Tt∑t=20002010Tt-∑i=18501880Ti.
And ΔC being the normed mole fraction deviation. Note that this
regressor P is one of multiple options that were tested and could be
regarded as a plausible regressor for seasonality changes. Specifically, we
tested global-mean CO2 concentrations, global-mean annual average
surface air temperatures and lagged averages of surface air temperatures as
regressors (see Fig. 5). The R-squared values of the regressions over the
1984–2014 period are relatively similar across all regressors, around 0.8.
The marked difference is that the regression with only CO2
concentrations would result in a stronger reduction of seasonality around
1940–1960 and before 1900. By 1850, the reduction of summertime CO2
concentrations in the zonal band around 52.5∘ N would be around
8.6 ppm compared to 2014 (multiply the differences of the seasonality
scaling difference between 1850 and 2014, about 21, with the 0.41 ppm
maximum of the EOF pattern, shown in Fig. 9a.2). In contrast, the other
regression options would limit the maximal seasonality change to about
5.7 ppm, closer to the maximal seasonality change detected within the
period 1984–2014, of 4.5 ppm (cf. Fig. 5e). Given the uncertainty in
regard to pre-1960 seasonality, we opted for the more conservative
extrapolation method that implies a less significant change outside the
observational period and chose the regressor with the least variability,
namely our composite regressor combining temperature and CO2
concentrations.
Comparison of various scaling options for the change of seasonality
of CO2 concentrations over time. The first EOF of the residual fields
of observations minus the mean 1984–2014 CO2 seasonality (Fig. 9a.2)
is scaled with an EOF score. Before 1984, this EOF score is regressed against
a composite of global-mean CO2 concentrations and global-mean surface
air temperatures (see text and panel b). Alternative regressors
include global-mean CO2 concentrations (a), lagged averages
of monthly global-mean surface air temperatures (c) and raw
global-mean annual average surface air temperatures (HadCRUT4v) (Morice et
al., 2012) (d). The regressed EOF score back in time is shown
in (e). A comparison to the first CO2 measurements of higher
northern latitudes at so-called Station P (STP) and Point Barrow in Alaska
(PTB), where the seasonality change is most pronounced, is provided
in (f) and (g), respectively (see text for discussion).
Despite the differences in the regressors, it should be noted that early
CO2 observations are too sparse to come to a definite conclusion in
regard to which regressor is best suited – given that the induced differences
around the 1960s and 1970s are fairly small compared to the noise in the
observations (see Fig. 5f and g). Furthermore, seasonality changes in the
case of CO2 depend on a number of factors, inter alia complex
interaction of CO2 fertilization of temperate, seasonal gross primary
productivity, the influence of temperature, precipitation on biomass growth
and respiration, and directly human-induced changes in land use areas
and their productivity. Therefore, this extension of the observed seasonality
changes beyond the observational period based on a regression with
temperatures and CO2 concentrations is just that: a plausible
extrapolation that needs to be refined by further research to replace this
study's ad hoc assumption.
The measured seasonality of CH4 and N2O over the
observational time period is found to be closely approximated by our default
assumption of a seasonality that is proportional to global-mean mole fractions. For several other substances, however, seasonality has been
assumed to be zero – either because the diagnosed seasonality was very small
or due to a lack of observational data.
Step 11–13: extension of latitudinal gradients and global means with ice core and firn data
Historical GHG records from ice and firn provide high-latitude estimates of
atmospheric GHG mole fractions before the instrumental record from air
sampling stations. We rely mainly on the Law Dome data (Etheridge et
al., 1998, 1996; MacFarling Meure et al., 2006; Rubino et al., 2013), updated
for minor dating changes and placed on current NOAA scales, and, for northern
hemispheric CH4, Greenland NEEM ice core data (Rhodes et al., 2013).
Although we did not directly use their data, we acknowledge multiple other
efforts, including but not limited to Mitchell et al. (2013), Bauska et
al. (2015), Schilt et al. (2010b), Fluckiger et al. (2002) and Sowers et
al. (2003) (Fig. 6). Law Dome atmospheric composition records have the
advantage of a very narrow air age spread that provides measurements with
high temporal resolution and mean air ages up to the 1970s, where they
overlap with the beginning of atmospheric observations for many gases.
Having obtained estimates of the latitudinal gradients over the observational
period and having derived approximations back in time by regressing
latitudinal gradients EOF scores with emissions (step 11 in Fig. 1, Table 4),
we can estimate global-mean mole fractions based on the Law Dome data for
both CO2 and N2O (step 12 in Fig. 1). In the case of
CH4, the advantage is that there are northern hemispheric data
available from NEEM (Greenland) (Rhodes et al., 2013) over the past
2000 years. This NEEM record hence allows an optimization of both the EOF
scores and global means at past time points to match both the Law Dome and
NEEM records (step 13 in Fig. 1). Some data gaps in the NEEM record are
filled by linearly interpolating the optimized EOF scores of the latitudinal
gradient. With an interpolated EOF score, the global-mean mole fraction can
then be directly inferred from the Law Dome record. All optimizations are
performed by minimizing area-weighted squared residuals.
The Law Dome ice core data are smoothed with a piecewise local third-degree
polynomial median regression, using ad hoc expert judgement assumptions of
errors and smoothing window widths specific to each gas in order to
approximately reflect their long-term median evolution. In the case of
CO2, a random error of 2 ppm was assumed, a percentage age
error (reaching a maximum of 60 years at age 2000 years before present) with
a bagging of 250 ensembles, a kernel width of 120 years, minimal number of
data points of 7 and maximum of 25 (Fig. 8a). Likewise, CH4 Law Dome
data ice core data are smoothed with a third-degree polynomial median
regression with a maximum kernel width of 100 years, 4 minimal data points (a
constraint that overwrites the maximum kernel width, if necessary) and 10
maximal data points. As for CO2, 250 ensembles were averaged, after
adding noise of 3 ppb, and an age uncertainty of 50 years per
2000 years. For N2O, a kernel width of 300 years was chosen with a
minimum number of 7 and maximum number of 15 data points to be included in
the piecewise third-degree polynomial regression. As for CO2 and
CH4, 250 ensembles were used for bagging after injecting a random
noise of 3 ppb and an age-dependent x axis uncertainty of 90 years
per 2000 years. The higher age uncertainty for N2O in comparison to
CO2 and CH4 was chosen to account for the larger age gaps in
the N2O Law Dome data that required a stronger horizontal smoothing
for the median regression to converge. For CO2, the slightly higher
age uncertainty in comparison to CH4 was chosen so that the smoothed
record displays a comparable time evolution to the WAIS CO2 record
(Fig. 6).
Atmospheric CO2, CH4 and N2O concentrations
over different timescales, from 800 000 years ago until today (a),
over the last 2000 years (panel b) and over 1850–2014 (c, d, e).
The shown data is for CO2: Mauna Loa data by Keeling et al. (1976);
the Law Dome ice record (Etheridge et al., 1998; MacFarling Meure et
al., 2006; Rubino et al., 2013), updated for minor dating changes and placed
on current NOAA scales; NOAA ESRL station data (NOAA, 2013; NOAA ESRL GMD,
2014a, b, c); the EPICA composite data (Ahn and Brook, 2014; Bereiter et
al., 2015; Bereiter et al., 2012; Lüthi et al., 2008; MacFarling Meure et
al., 2006; Marcott et al., 2014; Monnin et al., 2004; Petit et al., 1999;
Rubino et al., 2013; Schneider et al., 2013; Siegenthaler et al., 2005); and
the WAIS data (Bauska et al., 2015). For CH4, the shown data is the
Law Dome data (Etheridge et al., 1998; MacFarling Meure et al., 2006), the
instrumental data from the NOAA and AGAGE networks (see Table 3), NEEM ice
core measurements (Rhodes et al., 2013), the EPICA Dronning Maud Land ice
core record (Barbante et al., 2006; Capron et al., 2010; Schilt et
al., 2010b), and the long record by Loulergue et al. (2008) as well as the
GISP2D, WDC05A and WDC06A records by Mitchell et al. (2013). In case of
N2O, the shown data is the Law Dome record (MacFarling Meure et
al., 2006), the Talos Dome record (Schilt et al., 2010b), the GISPII record
(Sowers et al., 2003) and the EPICA Dome C record (Fluckiger et al., 2002;
Schilt et al., 2010a; Spahni et al., 2005; Stauffer et al., 2002) in addition
to the H15 ice core record from Antarctica (Machida et al., 1995), the South
Pole firn record (Battle et al., 1996), the Law Dome firn record “Park”
(Park et al., 2012) and a modelling synthesis by Ishijima (2007). For data
sources behind “this study's” composite product, see Tables 2, 3
and 4.
The Greenland NEEM ice core CH4 data (Rhodes et al., 2013) exhibits
some outliers in the recent period (Fig. 6d) due to the incursion of modern air
into still-open pores of shallow ice. Spikes in deeper ice are likely due to
impurities. Hence we use the 5-year smoothed data provided by Rhodes et
al. (2013) as a proxy for Greenland atmospheric background mole fractions
(open red circles in Fig. 6b and d). We used the NEEM CH4 firn
measurements from Buizert et al. (2012) (2008 campaign), with effective ages
from Ghosh et al. (2015) based on the iterative dating method of Trudinger et
al. (2002b), corrected for the effect of gravity (as applied in other firn
data) and put onto the NOAA 2006 primary calibration scale.
Options for reducing the number of GHGs to be taken into account in
climate models to approximate full radiative forcing of all GHGs. The GHGs
are ranked by their radiative forcing, with CO2 having the highest
radiative effect change between 1750 and 2014. The stated percentages in this table's rows are cumulative, i.e. the radiative forcing stated in a row is the radiative forcing of the GHG stated in that row plus the sum of all higher-ranked GHGs in the rows above, expressed as percentage of total anthropogenic GHG radiative forcing. In Option 1, a climate model explicitly resolves
actual GHG concentrations. With 8 and 15 species, 99.1 and 99.7 % of the
total radiative effect can be captured, respectively. In Option 2, only
CFC-12 is modelled next to CO2, CH4 and N2O; all
other gases are summarized in a CFC-11-equivalence concentration. In
Option 3, all ODSs are summarized in a CFC-12-equivalence concentration, and
all other fluorinated substances are summarized in HFC-134a-equivalence
concentrations. Note that below shares are approximations, as linear
radiative forcing efficiencies are assumed here for all gases, and also for
CO2, N2O and CH4.
RankOption 1 Option 2 Option 3 The GHG contribution Using subset Summarizing all Summarizing all ODSs to climate change of actual gases of lower into CFC-12 eq. and since 1750. concentrations, importance than all other fluorinated no equivalent gases CFC-12 into CFC-11eq. gases into HFC134a eq. Shares of change of total Shares of total warming effect: approximate radiative effect compared to effect warming effect since 1750: of all GHGs (absolute in 2014, not relative to 1850). approx. Radiative forcing contribution between 1750 and 2014 relative to that of all GHGs 1CO264.0 %CO272.9 %CO272.9 %CO272.9 %2+CH479.5 %+N2O86.1 %+N2O86.1 %+N2O86.1 %3+CFC1286.0 %+CH495.0 %+CH495.0 %+CH495.0 %4+N2O92.2 %+CFC1297.2 %+CFC1297.2 %+CFC12 eq.99.5 %5+CFC1194.5 %+CFC1198.0 %+CFC11 eq.100.0 %+HFC134a eq.100 %6+HCFC2296.4 %+HCFC2298.6 %7+CFC11397.2 %+CFC11398.9 %8+CCl497.8 %+CCl499.1 %9+HFC134a98.3 %+HFC134a99.3 %10+CFC11498.5 %+CF499.4 %11+HFC2398.7 %+CH3Cl99.5 %12+SF698.8 %+CFC11499.5 %13+CF499.0 %+HFC2399.6 %14+HCFC142b99.2 %+SF699.7 %15+HCFC141b99.3 %+HCFC142b99.7 %…+28 additional GHGs100 %28 additional GHGs100 %Step 14: extension of latitudinal gradients and global means with literature
data
For several gases, including ODSs, halons and PFCs, the available AGAGE and
NOAA station data is spatially sparse. Before the start of systematic
instrumental measurements, we use literature studies which make use of
various data sources, such as air sample archives or firn records (step 14 in
Fig. 1). Specifically, if a global mean is provided, we use that global mean
in conjunction with our derived and regressed latitudinal gradients. In the
case of hemispheric data points, we adapt the latitudinal gradient to match
the literature studies, as in the case of C4F10, C5F12,
C6F14, C7F16 or C8F18, where we based both
the global mean and latitudinal gradients on the data of Ivy et al. (2012).
Other key studies used were Velders and Daniel (2014), the data underlying
the WMO Ozone Assessment Report (2014), Arnold et al. (2013, 2014), Trudinger
et al. (2004), Mühle et al. (2010, 2009), Montzka et al. (2011), updated
time series by Montzka et al. (1999) (updated at
ftp://ftp.cmdl.noaa.gov/hats/Total_Cl_Br/), the recent study by Vollmer
et al. (2016) in regard to Halons and by Trudinger et al. (2016) in regard to
PFCs, and others (Arnold et al., 2013, 2014; Butler et al., 1999; Ivy et
al., 2012; Montzka et al., 2015; Mühle et al., 2010; Oram et al., 2012;
Trudinger et al., 2016; Velders and Daniel, 2014; Vollmer et al., 2016;
Worton et al., 2007), as indicated in the gas-specific fact-sheet figures
(Figs. S1–S40 with references provided in Table 12). In the case of
N2O and CH2Cl2, we assumed a constant latitudinal gradient
back in time before ongoing measurement records are available (Figs. 12
and S7, respectively).
Step 15: extrapolation
For some limited data segments, an extrapolation has been used: either a
piecewise smoothing spline to converge concentrations back to zero or
pre-industrial background concentrations, e.g. before the WMO (2014) or
Velders and Daniel (2014) data started in 1978 or 1951, respectively. The
three radiatively most important fluorinated species, CFC-12, CFC-11 and
HCFC-22 (Table 5), follow the global mean concentrations provided by Velders
and Daniel (2014), in conjunction with separately derived latitudinal
gradients and seasonality. Furthermore, a linear extrapolation was applied
when there were not sufficient 2014 data available.
Step 16–19: creating the composite surface concentration field
Following Eq. (1), the surface mole fraction fields over the full-time span
are now synthesized from the lower rank representations of seasonality,
latitudinal gradient and the smooth monthly representation of global-mean mole fractions. As per the original station data aggregation, the latitudinal
resolution is 15∘ and the time resolution is monthly. In order to
assist with application in climate models with finer grids, we also produced
a finer grid interpolation to 0.5∘ latitudinal resolution using a
mean-preserving smoothing. This finer grid interpolation should not be
mistaken as a mole fraction field containing actual information at
0.5∘ level. The purpose is simply to offer a smooth interpolation
that avoids errors that will arise from, for example, a linear interpolation between
the provided 15∘ latitude points, as the mean across those (linearly)
interpolated values would not match the original field. The mean-preserving
smoothing code is available from the authors on request. Finally, the
15∘ fields are aggregated into global, Northern and Southern
Hemisphere monthly and annual means.
Step 20: aggregating equivalent mole fractions
It is computationally inefficient to model the radiative effect of 43
individual GHGs in today's ESMs or general circulation models. Climate models
use different pathways to approximate the radiative effects of the full set
of GHGs. As one strategy, only the radiatively major GHGs are explicitly
modelled, such as CO2, CH4, N2O, CFC-12 and CFC-11,
which together cause 94.5 % of the GHG warming effect (measured in
radiative forcing) in 2014 relative to 1750 and 98 % of the total
radiative effect compared to the full set of 43 GHGs (Table 5).
Alternatively, radiatively minor GHGs can be approximated by equivalent GHG
concentrations of a marker gas. In this way, the radiative effect of the
group of gases is expressed by a single gas mole fraction. One definitional
issue is whether the radiative forcing since 1750, i.e. only the changes
since pre-industrial levels, are expressed by the marker gas (here called
“marginal equivalence” Ceq,i). In this case, the marker gas'
concentrations Ceq,i are sought that would exert the same
aggregate radiative forcing since 1750 as the group of summarized gases.
Thus, let Cj(t) be the concentration (mole fraction in dry air) of a GHG
and C0,j the pre-industrial level, i.e. in year 1750, which is routinely
used as base year for radiative forcing (IPCC, 2013). A marker equivalence
mole fraction by gas Ceq,1 for group Cj with j=1,…n is then given by the following:
Ceq,it=Ri-1Ri(C0,i)+∑j=1nRjCjt-RjC0,j,
with Rj(C) being the radiative forcing function relating concentrations
C(t) at time t to radiative forcing for gas j, in the linear case
RjC=C⋅Ej, with Ej being the radiative
efficiency. Ri-1(F) is the inverse of this radiative forcing
function, so that the concentration C that corresponds to a forcing F is
given by C=Ri-1(F).
In contrast, equivalent concentrations can express the radiative effects of
the summarized GHGs including their natural background levels (here called
“full equivalence” C′eq,i).
C′eq,it=Ri-1∑j=1nRjCjt
While the former definition, “marginal equivalence”, is often used to express
the total GHG forcing in CO2 equivalence concentrations, the latter
“full equivalence” is the more appropriate quantity to drive climate models,
given that natural background concentrations of not-explicitly-considered
gases should nevertheless exert a radiative effect even in a pre-industrial
control, even though that radiative effect does not count under a radiative
forcing definition that looks at changes from 1750.
In the linear case, in which case radiative forcing is proportional to the
gas' concentrations, Eq. (7) can be written as follows:
C′eq,it=∑j=1nrjeff⋅Cjtrieff,
with rieff being the radiative efficiency of gas i (Wm-2 per ppb).
Thus, climate models have the option to reduce the complexity of 43 GHGs and
the associated computational burden by reducing the number of GHGs that are
taken into account. With the top 5 GHGs, CO2, CH4, N2O,
CFC-11 and CFC-12, climate models would capture 98 % of the total
radiative effect in year 2014 and 94.5 % of the radiative forcing since
1750, i.e. the change of the radiative effect between 1750 and 2014 (see
Table 5). As an alternative, there is the option to
use equivalent concentrations. For two such equivalence options, this study
provides input datasets. Modelling groups should indicate the combination
of files they employed:
Option 1: climate models implement a subset of 43 GHGs.
Option 2: climate models implement the four most important
GHGs with their actual mole fractions explicitly, namely CO2,
CH4, N2O and CFC-12, and summarize the effect of all other 39
gases in an equivalence concentration of CFC-11. For this purpose, we provide
CFC-11 eq. concentrations (“full equivalence”).
Option 3: like option 2, but with a different split up of gases
other than CO2, CH4 and N2O. Climate models implement
the three most important GHGs with their actual mole fractions explicitly,
namely CO2, CH4 and N2O, and summarize the radiative
effect of the ODSs in a CFC-12 eq. concentration and the
radiative effect of all other fluorinated gases as a HFC-134a eq.
concentration. For this purpose, we provide CFC-12 eq. and HFC-134a eq.
concentrations (“full equivalence”).
Data analysis for comparison with CMIP5 ESMs
We compare our results to various other datasets (see Sect. 5), inter alia to
CO2 fields from CMIP5 ESMs (Sect. 5.3). Here,
we briefly describe the analytical steps that we performed for retrieving the
ESM data. We analyse 10 CMIP5 ESMs that have an interactive carbon cycle
model and provided the mole fraction of carbon dioxide in the air as function
of different pressure surfaces for the esmhistorical experiment. We
diagnosed those esmhistorical experiments in terms of the simulated
CO2 mole fraction at surface pressure
(1 bar= 100 000 Pa) for 10 CMIP5 ESMs, for which data
were available: (1) BNU-ESM (BNU, China), (2) CanESM2 (CCCMA, Canada),
(3) CESM1-BGC (NSF-DOE-NCAR, USA), (4) FIO-ESM (FIO, China), (5) GFDL-ESM2G
(NOAA GFDL, USA), (6) GFDL-ESM2M (NOAA GFDL), (7) MIROC-ESM (MIROC, Japan),
(8) MPI-ESM-LR (MPI, Germany), (9) MRI-ESM1 (MRI, Japan) and (10) NorESM1-ME
(NCC, Norway). For the models CanESM2, MIROC-ESM and MPI-ESM-LR more than one
realization is available. We calculated an ensemble mean based on all
available ensemble members. The climatological seasonal cycle (Figs. S43
and S44) is calculated relative to the linear trend of the corresponding
30-year periods.
Comparison of 1950–1990 CO2 concentrations with early
Scripps station data (Keeling et al., 2001) for each 15∘ latitudinal
band. Also, the Law Dome ice record data is shown (k) with our
third-degree polynomial smoothing. This study's monthly CO2 zonal
means were derived from station data from 1984 onwards. Before that, this
study used Mauna Loa MLO annual average and smoothed Law Dome data (see
Table 1 and Sect. 2). The shown comparison with monthly Scripps station data
before 1984 is a qualitative validation of the applied methodology to regress
latitudinal gradient and seasonality changes to times before 1984. See text.
Historical GHG concentrations from 1750 to 2014 as global-mean
(right panels), northern hemispheric (middle panels) and southern hemispheric
averages (right panels). The top row comprises all GHGs, the middle row
comprises HFCs, PFCs, SF6, NF3 and SO2F2. The lower
row comprises all ozone-depleting substances, expressed as equivalent
CFC-12 eq. concentrations. In the narrow boxes, the last data year from
15 January 2014 to 15 December 2015 is shown, indicating the intra-annual
trend (top row), increasing gradient (middle row) or relatively flat
concentration levels (lower row).
Results
Here, we describe the historical concentrations of the main GHGs and provide
a fact sheet for all 43 individual gases.
Carbon dioxide
The 800 000-year EPICA composite ice core record (Ahn and Brook, 2014;
Bereiter et al., 2015, 2012; Lüthi et al., 2008; MacFarling Meure et
al., 2006; Marcott et al., 2014; Monnin et al., 2004; Petit et al., 1999;
Schneider et al., 2013; Siegenthaler et al., 2005. Available at
ftp://ftp.ncdc.noaa.gov/pub/data/paleo/icecore/antarctica/antarctica2015co2.xls)
indicates that CO2 concentrations have fluctuated between 170 and
270 ppm (Fig. 6a) in conjunction with glacial and inter-glacial
temperature variations. From the year 0 to 1000, our piecewise fit of the third-degree polynomial of Law Dome ice core data allows a derivation of global
mean concentrations of around 278.6 ppm (minimum–maximum range of
277.0–280.2 ppm).
Our smoothed Law Dome results do not reflect the higher-frequency variations
suggested by the individual data points (Etheridge et al., 1996; MacFarling
Meure et al., 2006; Rubino et al., 2013) and are comparable to the frequency
spectrum that would result from a smoothed median estimate of WAIS data by
Bauska et al. (2015) and Ahn et al. (2012). The WAIS record is generally
3–6 ppm higher than the Law Dome record and is also higher than
South Pole and EPICA DML ice cores (Ahn et al., 2012) and the Dronning Maud
Land ice (Rubino et al., 2016). The cause for this difference is not yet
known (Fig. 6b). The differences between the WAIS and the Law Dome record
persist in 1850–1890 with subsequent data points being more aligned with
each other (Fig. 6c). CMIP6 modelling groups might want to test an
alternative dataset that captures those higher-frequency characteristics of
the Law Dome record (data can be generated by the authors on request). In
that higher-frequency dataset, the minimum of global mean CO2
concentrations is close to 270 ppm around the year 1610. The smoother
version provided for CMIP6 has its minimum in year 1666 at 276.27 ppm
(Fig. 6b). The reason for the 1610 dip in the Law Dome record and why this
does not show in the WAIS record is not yet fully understood. The current
understanding of how the age kernel (to estimate the distribution of age of
air at the time of bubble trapping) is different for the two sites cannot yet
explain this difference in concentrations around 1610.
In regard to the latitudinal gradient, we explored various options. If we
regress the scores of the first EOF of the latitudinal gradient (Fig. 9d)
against global fossil CO2 emissions, the pre-industrial latitudinal
minimum of surface CO2 concentrations would be estimated in the
mid-northern latitudes (approximately 1.8 ppm below the global mean),
where the maximum was observed in recent decades (e.g. 4.8 ppm above
the global mean in 2010). Previously a similar regression approach between
concentrations and CO2 emissions was used by Keeling et al. (2011) to
separate the anthropogenic from the natural component in the concentration
difference between Mauna Loa and the South Pole. This approach is not perfect
due to the covariance of regional fossil fuel emissions with natural sinks
over the same period, different patterns of anthropogenic land-use emissions
and a latitudinal gradient component that merely results from seasonal
CO2 exchange (e.g. Denning et al., 1995). Nevertheless, this
regression approach can provide a first indication of the influence of
anthropogenic emissions on the latitudinal gradient. Furthermore, this
approach would result in an approximately 0.4 ppm higher
pre-industrial Antarctic CO2 concentration compared to the global
mean, coinciding with the assumption taken by Rubino et al. (2013). The
Scripps station data not used in this calibration (i.e. all Scripps stations
except for Mauna Loa) turn out to be matched rather closely by our approach
over the 1950–1990 period (see Fig. 7).
However, given the evidence by CMIP5 ESMs of a slight tropical local
maximum (Fig. 9b) and large uncertainties regarding
pre-industrial sinks and source distributions and hence the latitudinal
gradients of CO2, we assumed a zero pre-industrial latitudinal
gradient. Thus we performed a zero-intercept regression of the scores of the
latitudinal gradient EOF1 with global fossil CO2 emissions and
converged the score of the second EOF towards zero, resulting in a flat
latitudinal gradient in pre-industrial times.
The second EOF of the latitudinal gradient of CO2 does not exhibit
the same linearity over time as the first EOF, and the reasons are currently
unknown. Potential candidates for this pronounced spike (Fig. 9c) of
mid-northern latitude concentrations in the case of CO2 are a shift
in station sampling locations with more “polluted” land stations coming
online after 1995, the “rectifier” effect due to an enhanced seasonal cycle
(Denning et al., 1995) and the rise of Chinese emissions (the onset around
year 2003 of the recent surge in Chinese CO2 emissions is
approximately coinciding with the respective EOF score becoming strongly
positive; Fig. 9d; Francey et al., 2013). One suggested explanation for this
2010 change in north–south gradients are changes in interhemispheric
transport (Francey and Frederiksen, 2016). Recently, i.e. after 2010, this
spike in mid-latitude northern concentrations seemed to somewhat subside
again according to our analysis (see scores for EOF1 and EOF2 in Fig. 9d).
Future research could address the underlying reasons of this change in
latitudinal patterns, and a physical explanation will allow a more
appropriate backward extension in time.
The diagnosed average seasonality of atmospheric CO2 concentrations
over the observational period reflects the standard carbon cycle pattern of
strong CO2 uptake in spring and release in autumn due to
photosynthesis and heterotrophic respiration in the Northern Hemisphere's
ecosystems. Our EOF analysis of the residuals shows (Fig. 9a.2 and 9a.3) that
the seasonality has increased over recent decades in line with previous
studies, which explore the link to increased ecosystem productivity (Forkel
et al., 2016; Graven et al., 2013; Welp et al., 2016) and increased cropland
productivity (Gray et al., 2014). Specifically, our analysis shows a slight
shift of the seasonality to earlier months in the year, i.e. the negative and
positive deviations of the EOF pattern are shifted by a month compared to the
average seasonality (cf. Fig. 9a.1 and 9a.2). The strongest change in
CO2 seasonality is derived for the latitudinal bins centred at
37.5∘ N and up to 67.5∘ N bins with a maximum strengthening
of negative deviations in the latitudinal band centred at 52.5∘ N in
July by around 4 ppm over 1984–2013 (4 ppm results from multiplying
the EOF pattern value in July in the 52.5∘ bin with the EOF score
difference of around 10, see Fig. 9a.2 and a.3). The maximum strengthening of
the seasonal cycle happens in July in the 52.5∘ latitudinal band;
however, the maximum seasonal cycle deviation is still observed slightly
later in August and also extends slightly more towards the northern latitudes
(Fig. 9a.1).
In 1850, the start of the historical CMIP6 simulations, the estimated
global-mean CO2 concentration is 284.32 ppm, rising to
295.67 ppm in 1900, 312.82 ppm in 1950, 369.12 ppm in
year 2000 up to 397.55 ppm in 2014 (Table 6). Here and elsewhere
(e.g. Table 6) we provide more significant figures than is customary – not
to claim a 5-digit precision of the data, but to avoid unnecessary (even if
small) step changes in concentrations between the pre-industrial run and the
historical and other runs. Our methodology does not include a formal
uncertainty analysis. As a minimum uncertainty for the 1850 pre-industrial
values, we refer to the 1.2 ppm variability stated by Etheridge et
al. (1996), also used in Rubino et al. (2013) and Trudinger et al. (2002a),
as minimum uncertainty for that period.
Overview of historical CO2 concentrations.
(a.1) The average seasonality of CO2 over the observational
period, (a.2) the change of seasonality over time.
(a.3) The observationally derived and extended EOF score of the
seasonality change. The first EOF1's score is almost linearly increasing over
the time of instrumental data from 1984 to 2014. (b) The latitudinal
variation of mole fractions (dashed lines), shown for example years from 1500
to 2014, including (for comparison) the average of three CMIP5 ESMs (solid
lines). (c) The first and second EOF of latitudinal variation. The
second EOF exhibits a strong signal around mid-northern
latitudes (d), the EOF scores derived from the observational data
(dots) and regression (dashed line) as well as the ultimately used EOF score
(solid line). The second EOF's score indicates that the mid-latitude northern
spike was only a recent phenomenon and the score is here assumed to linearly
converge to zero. The first EOF's score is more linearly increasing, and
regressed against global fossil emissions. (e) The resulting
latitudinal-monthly concentration field, here shown between 1950 and 2014.
(f) Global and hemispheric means of the derived concentration field
over the same time period 1950–2014 in comparison to monthly station data
(grey dots), latitudinal average station data (coloured circles) and various
literature studies (see legend). (g) Same as (f), except
for time period 1750–2014. (h) Same as (f) but for time
period 2005–2010.
Annual growth rate of CO2 concentrations for global-mean,
northern hemispheric average and southern hemispheric average concentrations.
Before 1960, the smooth growth rate results from interpolated global mean
values. After 1960, the growth rates are diagnosed from the surface station
data, as shown in Fig. 9f. Noticeable are fluctuations of the annual growth
rate around 1973, 1981 and 1992.
Global-mean surface CO2 concentration growth slightly flattens off in
the 1930s, and a stronger flattening occurs during World War II until the
1950s (Bastos et al., 2016). The increase from 1970 onwards has a slightly
positive curvature (accelerating trend) with small deviations around 1973,
1981 and the temporary flattening of CO2 concentrations after the
1991 Pinatubo eruption (Jones and Cox, 2001; Peylin et al., 2005) (Figs. 9
and 10).
Historical: global- and annual-mean surface concentrations for the
historical CMIP6 experiments. The year-to-year and monthly resolved global,
hemispheric and latitudinally resolved concentrations for 43 GHGs and three
aggregate equivalent concentrations are provided in the accompanying datasets
over the time horizon year 0 (1 BC) to year 2014 AD. The complexity
reduction options for capturing all GHGs with fewer species than 43 are
indicated in the table as Option 1, Option 2 and Option 3, with “x”
denoting relevant columns under each option (Sect. 2.1.10).
YearsCO2CH4N2OCFC-12 eq.HFC-134a eq.CFC-11 eq.CFC-12OtherOption 1xxxxxOption 2xxxxxOption 3xxxxxUnitsppmppbppbpptpptpptpptAll or a subset of other 391750277.15731.41273.8716.5119.1532.110.00individual gases, available online1850284.32808.25273.0216.5119.1532.110.001851284.45808.41273.0916.5119.1532.110.001852284.60809.16273.1716.5119.1532.110.001853284.73810.40273.2616.5119.1532.110.001854284.85811.73273.3616.5119.1532.110.001855284.94813.33273.4716.5119.1532.110.001856285.05814.80273.5816.5119.1532.110.001857285.20816.45273.6816.5119.1532.110.001858285.37818.36273.7616.5119.1532.110.001859285.54820.40273.9016.5119.1532.110.001860285.74822.31274.0616.5119.1532.110.001861285.93824.40274.2416.5119.1532.110.001862286.10827.03274.4216.5119.1532.110.001863286.27830.17274.5716.5119.1532.110.001864286.44833.60274.7216.5119.1532.110.001865286.61836.89274.8816.5119.1532.110.001866286.78840.36275.0516.5119.1532.110.001867286.95844.00275.2116.5119.1532.110.001868287.10847.25275.3916.5119.1532.110.001869287.22850.13275.5616.5119.1532.110.001870287.35852.44275.7216.5119.1532.110.001871287.49853.99275.9016.5119.1532.110.001872287.66855.23276.0816.5119.1532.110.001873287.86856.17276.2516.5119.1532.110.001874288.06857.82276.4216.5119.1532.110.001875288.29859.47276.5916.5119.1532.110.001876288.52860.86276.7416.5119.1532.110.001877288.75862.38276.8616.5119.1532.110.001878288.99864.14277.0016.5119.1532.110.001879289.22866.28277.1316.5119.1532.110.001880289.47868.70277.2716.5119.1532.110.001881289.74870.98277.3716.5119.1532.110.001882290.02873.25277.4916.5119.1532.110.001883290.26875.60277.5916.5119.1532.110.001884290.51878.15277.7016.5119.1532.110.001885290.80881.03277.8016.5119.1532.110.001886291.10883.84277.8916.5119.1532.110.001887291.41886.93278.0016.5119.1532.110.001888291.76889.93278.0816.5119.1532.110.001889292.11893.16278.1916.5119.1532.110.001890292.46896.38278.2716.5119.1632.110.001891292.82899.67278.3516.5119.1632.110.001892293.17903.53278.4416.5119.1632.110.001893293.48907.27278.5516.5119.1632.110.001894293.79910.48278.6916.5119.1632.110.001895294.08913.23278.8316.5119.1632.110.001896294.36914.77278.9416.5119.1632.110.001897294.65916.27279.0516.5119.1632.110.001898294.95919.02279.1616.5119.1632.110.001899295.30922.28279.3116.5119.1632.110.00
Over the 800 000 years before year 0, atmospheric CH4 concentrations
varied between 348.7 and 728.4 ppb according to the EPICA ice core
composite (Barbante et al., 2006; Capron et al., 2010; Loulergue et
al., 2008) (Figs. 6c and 11). The Law Dome record (Etheridge et al., 1998;
MacFarling Meure et al., 2006) indicates an onset of increasing
concentrations around the year 1720 (Figs. 6d and 11). From year 1850 with
slightly higher than 800 ppb concentrations, a slight rise is observed until
the 1950s, when CH4 concentrations markedly increase first in the
latter half of the 1950s, then again from 1965 onwards. The Greenland firn
and ice core data (Rhodes et al., 2013) are more difficult to interpret
because part of the record is affected by high-frequency ice core CH4
signals, possibly of non-atmospheric origin. CH4 spikes are
accompanied by elevated concentrations of black carbon, ammonium and nitrate,
suggesting that biological in situ production may be responsible –
particularly in the later years of the record since 1940. We use the 5-yearly average measurement values, that have outliers removed (Rhodes et al., 2013), and which approximate the lower bounds of the raw data points until 1942. Using these values, we can then infer global gradients back in time and derive an estimate of global-mean concentrations. These global-mean
concentrations are estimated to be around 30 ppb higher than the Law
Dome record by 1850, with the difference growing to 45 ppb by 1940s,
increasing further from there (Fig. 6d). This approximately matches the
findings by Mitchell et al. (2013) of interpolar differences between about 35
and 45 ppb between 800 BC and 1700 AD.
Overview of historical CH4 concentrations.
(a.1) The relative seasonality of CH4 over the observational
period. (b) The latitudinal variation of concentrations (dashed
lines), shown for example years. (c) The first and second EOF of
latitudinal variation. (d) The EOF scores derived from the
observational data (dots) and regression against global emissions (dashed
line) as well as the ultimately used EOF score (solid line). (e) The
resulting latitudinal-monthly concentration field, here shown between 1950
and 2014. (f) Global and hemispheric means of the derived
concentration field over the same time period 1950–2014 in comparison to
monthly station data (grey dots), latitudinal average station data (coloured
circles), and various literature studies (see legend). (g) Same
as (f), except for time period 1750–2014. (h) Same
as (f) but for time period 2005–2010.
Our analysis of CH4 concentrations in the recent decades is based on
a large number of stations (Table 3 and Fig. 11f). While the annual increase
of global CH4 concentrations slowed over the 1980s and slowed
markedly after 1992 towards stabilized concentrations between 1999 and 2005,
CH4 increased again after 2006 at about 5.4 ppbyr-1
(Fig. 11f; Nisbet et al., 2016, 2014).
We retrieve a recent seasonal cycle of CH4 that is similar in the
latitudinal–temporal seasonality
pattern as that of CO2 (Fig. 11a). Each hemisphere exhibits its
lowest CH4 concentrations just after the summer solstice, up to
1.6% or 28 ppb lower than the global mean in the case of the
high-latitude northern summer (Fig. 11a). Quantifying the underlying reasons
is beyond the scope of this study, although the seasonally varying
atmospheric sink by OH oxidization is likely the main contributor to that
seasonal pattern – in combination with seasonally varying natural and
anthropogenic sources.
The latitudinal annual-mean gradient of CH4 concentrations is
separated into its first two EOFs, with the first EOF being a continuous
north–south gradient of about 90 ppb in the recent observational
period (combination of EOF and its score, see Fig. 11c and d). The second EOF
is a distinct mid-northern latitude local maximum with a high-latitude low,
showing a slight but marked rise in 2008 within the 1985–2014 observational
data window. Quantifying the reasons for this hump are again beyond the scope
of this study, with the possibility of a shift in locations of sampling
stations or coal-seam gas-fracking-related fugitive emissions being possible
contributors. We optimize the first EOF, which is the general north–south gradient, to match the firn and ice-core Greenland and Antarctic Law Dome data. The second EOF of the latitudinal gradient is kept constant at its 1985 value.
As a result of the constant extrapolation of the second EOF, and the
optimization of the first EOF's score (Fig. 11d), we yield a total
annual-mean meridional gradient for recent decades that features around
80 ppb higher surface CH4 concentrations in mid-to-high northern
latitudes compared to the global mean and around 60 ppb lower CH4
concentrations at the high southern latitudes (Fig. 11b). In pre-industrial
times, our approach of regressing the score of EOF1 with global emissions
(Gütschow et al., 2016) suggests that this gradient is smaller, with only
approximately 20–30 ppb higher northern and 20 ppb lower southern latitude
surface concentrations (Fig. 11b). These mean interpolar differences and
their variations have earlier been quantified by Etheridge et al. (1998) and
Mitchell et al. (2013), yielding similar results (between 30 and
60 ppb) compared to our 40–50 ppb estimate.
Nitrous oxide
N2O concentrations from ice cores dating back 800 000 years
(Fluckiger et al., 2002; Schilt et al., 2010b) varied approximately between
200 and 300 ppb, with most recent glacial concentration minima of
180 ppb around 23 000 years ago (Sowers et al., 2003) (Fig. 6a). The
ice core record over the last 2000 years indicates a marked difference
between the Law Dome and GISPII record (Sowers et al., 2003), with the latter
being up to 10 ppb lower. Here, as with CH4, we use again a
median quantile piecewise polynomial regression on the Law Dome record,
assuming constant N2O concentrations between year 0 and the first Law
Dome data point in year 154. In contrast to CH4, there is not a
monotonic increase of concentrations, but rather an initial slight decrease
until year 630 down to a minimum concentration of 265 ppb in our
smoothed time series, with a subsequent slow increase until the 9th century
AD, then a slight decrease until 1650 in the smoothed global-mean mole
fraction. A temporary local maximum indicated by individual Law Dome data in
the 15th century is not resolved by our smoothing, and a similar spike in the
17th century is only just reflected (Fig. 6b). Several data points indicate a small
decrease after a 1750 maximum with a minimum in 1850 of around
273.02 ppb. This maximum around 1750 and subsequent minimum around
1800–1850 is also apparent in the H15 ice core record by Machida et
al. (1995) (we scale-corrected the Machida data downwards by 1 ppb as
in Battle et al., 1996) (Fig. 6b). After 1850, N2O concentrations
increased markedly, reaching 1900, 1950, 2000 and 2014 values of 279.5,
289.7, 315.8 and 327.0 ppb, respectively (Table 6). Comparing the
different firn and ice records, the 1920–1940 period seems particularly
uncertain, with some high measurements close to and beyond 290 ppb
from both Law Dome and H15, while some of the Law Dome data is still at
levels around 285 ppb or even 280 ppb in the case of H15
(Fig. 6e). The South Pole firn data (Battle et al., 1996) suggest lower
N2O concentrations in the 1920s and around 1960 – compared to both
the smoothed Law Dome data (thin dashed line in Fig. 6e) and consequently our
even higher global-mean estimate. Although the Ishijima estimate (Ishijima et
al., 2007) (their Fig. 6a) around 1952 is almost identical to our global
mean, their modelling study suggests slightly lower values around 1960 before
being closely matched again from 1970 onwards. The Law Dome firn record (Park
et al., 2012) suggests slightly higher N2O concentrations for the
high southern latitudes compared to our global mean (Fig. 6e).
The variability of our derived N2O global-mean concentrations, in
particular the steps in the 1920s and 1940s, reflect the smoothing algorithm
choices to noisy data (Sect. 2.1.6), but should not be over-interpreted. Our
algorithm does not, for example, include information on the lifetime of
N2O that would guard against inferring too-rapid declines of
N2O mole fractions and mole fraction growth rates. The fit of the
smoothing algorithm was chosen to balance the resolution of smaller-scale
features with the uncertainty present in the input data sources for the
full-time horizon from year 0 to year 2014. Given overall uncertainties
(Fig. 6e), a smoother representation between 1900 and 1980 seems equally
justified.
Compared to CH4 and CO2, the seasonality and latitudinal
gradient of N2O are relatively small. The N2O seasonality is
only 0.1 % of global mole fractions and is almost symmetric and
seasonally time-synchronized between the Northern and Southern Hemispheres,
with minima in the Southern Hemisphere late autumn and northern Hemisphere
summer–autumn (Fig. 12a). The seasonality is currently of the same size as
the underlying trend, leading to global mean N2O mole fractions
increasing in the latter months of any year, with a subsequent flattening in
the first half of any calendar year (e.g. Fig. 12h). Given a
counter-intuitive slight decrease of the north–south gradient with flat or
slightly increasing global N2O emissions (Gütschow et al., 2016)
in recent years (Fig. 12d), we assumed constant scores for the latitudinal
gradient EOFs for times before 1996 (Fig. 12d). Due to measurement
fluctuations in the first years when systematic measurements started in 1978
that are larger compared to the recent period, we chose to interpolate
N2O global-mean mole fractions over 1966–1987. For the period
between 1978 and 1987, this interpolation is closely aligned with a smooth
representation of the atmospheric measurements (Fig. 12f, cf. ALE/GAGE/AGAGE
data as shown at
http://agage.eas.gatech.edu/data_archive/data_figures/gcmd_month/n2o_monS5.pdf).
Overview over historical N2O concentrations. As Fig. 11, but
for N2O.
Ozone-depleting substances (ODSs) and other chlorinated substances
ODSs, i.e. the substances destroying ozone and
being controlled under the Montreal Protocol, also have a large warming
effect (Velders et al., 2007, 2009). In particular, CFC-12 and CFC-11 are
important GHGs, as well as the replacement substance HCFC-22, which, unlike
CFCs, continues to increase in the atmosphere, albeit at a declining rate.
The radiative forcing of CFC-12 alone since 1750 is equivalent to that of
N2O, which is usually considered the third most important GHG after
CO2 and CH4 (Table 5). The impact of ODSs on climate is
somewhat complicated by their destruction of stratospheric ozone, which
induces dynamical effects on circulation patterns, and has a net cooling
effect on the global climate. The latest estimates suggest that this cooling
might offset roughly two-thirds of the warming of the entire class of ODSs
(Shindell et al., 2013). Note that here we also consider methylene chloride
and methyl chloride, although these chlorinated substances are not controlled
by the Montreal Protocol and are hence often not termed ODSs (WMO, 2014).
The most abundant ODSs in the atmosphere (in 2014) were
CFC-12 (520.6 ppt), CFC-11 (233.1 ppt) and HCFC-22
(229.5 ppt), with their mole fractions being about 6 orders of
magnitude lower than current measurements for CO2 (Table 7). In
addition, methyl chloride CH3Cl has a high mole fraction
(539.54 ppt), although it is not considered an ODS here as it is not
controlled by the Montreal Protocol. Out of the 17 considered chlorinated and
ODSs, only 6 have currently increasing concentrations.
Those are the three HCFCs, of which the increase in HCFC-22 alone has offset
the reducing radiative forcing of all other ODSs over the past decade
(Fig. 8m). The other three substances that are still increasing are
Halon-1301, methylene chloride (CH2Cl2) and chloroform
(CHCl3). Chloroform had been decreasing in the 1990s and stabilized
in the 2000s, but again recently showed an increase (Fig. S11).
Four of the considered chlorinated and ODSs are assumed
to have natural emissions and hence above-zero pre-industrial concentrations.
We estimate those pre-industrial natural background concentrations using a simple
budget equation under the assumption of a constant lifetime (IPCC, 2013) of
1 year for CH3Cl and 0.8 years for CH3Br – minimizing the
error term when taking into account anthropogenic emission and atmospheric
concentration estimates over 1950 to 1990 by Velders and Daniel (2014).
Specifically, methyl chloride (CH3Cl) is assumed to have
pre-industrial global-mean concentrations of 457 ppt, and methyl
bromide (CH3Br) with that of 5.3 ppt. Chloroform (CHCl3) is
assumed to have a pre-industrial concentration of about 6 ppt,
approximately in line with findings by Worton et al. (2006) and the
estimation by Aucott et al. (1999) that in 1990 CHCl3 was at about
8 ppt, with 80 % of emissions assumed to be of natural origin.
Lastly, in the absence of other information (a good understanding of the
natural versus anthropogenic source fraction or historical industrial production
records) the available firn measurements (e.g. Trudinger et al., 2004)
supplying information about methylene chloride (CH2Cl2) mole
fractions in the early 20th century are used to suggest a 6.9 ppt
pre-industrial mean concentration with a strong latitudinal gradient that
results in northern (southern) hemisphere average concentrations of 12.8
(1.0) ppt. The transition of concentrations of some species between
the observational station data and pre-industrial levels are also uncertain.
For CH2Cl2, our derivation is in line with the smooth trajectory of
Trudinger et al. (2004), indicating an almost monotonic transition between
1997 values and pre-industrial concentrations (Fig. S7f). Our assimilation
approach (which is based on the Walker et al., 2000 data) causes our carbon
tetrachloride (CCl4) reconstruction to have a near-zero
pre-industrial concentration of 0.025 ppt (0.025 % of its peak
value of 100 ppt). We note that Walker et al. (2000) suggest zero
pre-industrial concentrations before 1910, although the lowest empirical
evidence from firn records suggest < 5 ppt (Butler et al., 1999)
or 3–4 ppt as measured by S. Montzka for 1863 firn air and reported
in Liang et al. (2016).
The seasonal cycle of ODSs and other synthetic GHGs can
be influenced by seasonally varying stratospheric–tropospheric air
exchanges, interhemispheric transport, tropopause heights, emissions and, for
those substances with OH-related sinks, the seasonally varying OH
concentrations. For 11 out of the 17 considered ODSs we
find some indication of seasonal cycles based on the analysed station data,
namely for CCl4, CFC-11, CFC-12, CFC-113, CH2Cl2,
CH3Br, CH3CCl3, CH3Cl, CHCl3, Halon-1211
and HCFC-22. Our analysis indicates that HCFC-141b also shows some signs of a
seasonal cycle, although we here assumed a zero seasonal cycle due to data
sparsity (see Fig. S16a). We find the strongest seasonal cycles in the case of
the short-lived species CH3Cl, CHCl3, CH3Br and
CH2Cl2 with absolute maximal seasonal deviations of -11, -12 and
±9, -32 % compared to the annual mean, respectively. For the
radiatively important and longer-lived species CFC-12, CFC-11 and HCFC-22,
the seasonal cycle is much smaller, with ±0.2, ±0.4, ±0.8 %,
respectively.
Similar to the seasonality, the latitudinal gradient is found to be
especially pronounced for the short-lived substances. Specifically,
CH2Cl2 with a lifetime of 0.4 years, CH3Br with a lifetime
of 0.8 years, CH3CCl3 with a lifetime of 5 years, CH3Cl
with a lifetime of approximately 1 year and CHCl3 with a lifetime of
0.4 years show substantial latitudinal gradients due to spatially
heterogeneous sinks and sources (lifetimes following Table 8.A.1 in IPCC WG1
AR5, 2013). While chemicals with predominantly anthropogenic sources normally
exhibit the highest mole fractions at mid-northern to high northern latitudes, the
observations for several substances with substantial natural sources exhibit
highest mole fractions in the tropics or lower northern latitudes in the
recent observational period (e.g. CH3Cl in Fig. S10b and c).
Other fluorinated GHGs
The 23 other gases in this study are the hydrofluorocarbons (HFCs), which
have recently been added to the substances controlled under the Montreal
Protocol (Kigali amendment in October 2016), and those substances whose
production and consumption is not controlled under the Montreal Protocol,
namely perfluorocarbons (PFCs) as well as sulfur hexafluoride (SF6),
nitrogen trifluoride (NF3) and sulfuryl fluoride (SO2F2).
Except for the latter, the emissions of all these species are controlled
under the Kyoto Protocol and covered by most “nationally determined
contributions” (NDCs) under the Paris Agreement. However, currently the
aggregated greenhouse effect of this group of synthetic GHGs is still almost
a factor of 10 smaller compared to the ODSs (cf. Fig. 8g and m). In contrast
to the ODSs, nearly all of these other fluorinated gas concentrations are
still rising; the exception is HFC-152a, which has stopped growing since 2012
and may now be in decline, (Fig. S33f). Thus, a primary concern with these
gases is the potential for substantial climate forcing in the future if
uncontrolled growth continues.
The most abundant of these gases is the refrigerant HFC-134a, with 2014
concentrations estimated to be 80.5 ppt, followed by HFC-23
(26.9 ppt), HFC-125 (15.4 ppt) and HFC-143a
(15.2 ppt). At the other end of the concentration spectrum, we
include results from Ivy et al. (2012) for some PFCs that exhibit low
concentrations of 0.13 ppt (C5F12 and C7F16) or
0.09 ppt (C8F18) (Table 7). The only fluorinated gas
considered to have substantial natural sources, and hence a pre-industrial
background concentration, is CF4 with an assumed pre-industrial
concentration of 34.05 ppt (see Fig. S26), in line with findings by
Trudinger et al. (2016) and Mühle et al. (2010).
For a number of substances, especially the PFCs with lower abundances, there
were not sufficient data available to estimate the seasonality of atmospheric
concentrations. We consider seasonality only for 3 of the 23 species.
HFC-134a has a somewhat atypical pattern of lowest mole fractions in the
spring Northern Hemisphere (-2.6 % compared to annual mean) as other
gases normally show a summer or autumn low point of concentrations. This
spring minimum results from a seasonality of sources of this refrigerant
(Fig. S31a), although seasonality in loss also likely plays a role (Xiang et
al., 2014). Secondly, the short-lived HFC-152a (lifetime 1.5 years) shows
seasonal variations of up to ±13 % while the very long-lived
SF6 (lifetime of 3200 years) exhibits a much smaller seasonality of
up to ±0.5 %.
For most of the considered substances, the latitudinal gradient is rather
small. Exceptions are the shorter-lived species like HFC-32, whose
concentration has risen relatively quickly since 2000 due to rapidly increasing
northern hemispheric sources (Fig. S28b), HFC-152a and some other shorter-lived HFCs. For the three heavier PFCs with very low abundances of well below
1 ppt in 2014, namely C6F14, C7F16 and
C8F18, we incorporated hemispheric data from Ivy et al. (2012).
Before about 1990, those three gases are suggested to have reversed
latitudinal gradients with higher southern hemispheric concentrations. Due to
the very low mole fractions near the limit of measurement, future studies may
need to confirm whether those reverse gradients existed (and if so, why).
Given the negligible radiative forcing from these gases to date, this
uncertainty does not affect the overall results.
Global- and annual-mean GHG surface concentrations for year 2011 and
2014, including a comparison to 2011 NOAA, AGAGE and UCI estimates – as
provided in IPCC AR5 WG1. Unit is parts per trillion (ppt), unless otherwise stated.
20142011201120112011Rank of abundanceSpeciesCMIP6 (this study) UCISIO b/AGAGENOAA1CO2 (ppm)397.55390.94390.48±0.28390.44±0.162CH4 (ppb)1831.471813.071798.1±0.61803.1±4.81803.2±1.23N2O (ppb)326.99324.16324.0±0.1324.3±0.14CH3Cl539.54534.175CFC-12520.58528.53525.3±0.8529.5±0.2527.4±0.46CFC-11233.08238.25237.9±0.8236.9±0.1238.5±0.27HCFC-22229.54214.56209.0±1.2213.4±0.8213.2±1.28CCl483.0786.0687.8±0.685.0±0.186.5±0.39CF481.0979.0479.0±0.110HFC-134a80.5262.8563.4±0.962.4±0.363.0±0.611CFC-11372.7174.6474.9±0.674.29±0.0674.40±0.0412CH2Cl236.3529.4913HFC-2326.8924.1324.0±0.314HCFC-141b23.8121.5620.8±0.521.38±0.0921.4±0.215HCFC-142b22.0821.3521.0±0.521.35±0.0621.0±0.116CFC-11416.3116.3617HFC-12515.3610.469.58±0.0418HFC-143a15.2511.9212.04±0.0719CHCl39.908.9520CFC-1158.438.3921HFC-328.345.1722SF68.227.317.26±0.027.31±0.0223HFC-152a7.737.896.4±0.124CH3Br6.697.1125C2F64.404.174.16±0.0226Halon-12113.754.0527CH3CCl33.686.316.8±0.66.3±0.16.35±0.0728Halon-13013.303.2329HFC-245fa2.051.5630SO2F22.041.7431c-C4F81.341.2332NF31.240.8333HFC-227ea1.010.7434HFC-365mfc0.770.5635C3F80.600.5636Halon-24020.430.4537C6F140.280.2738HFC-43-10mee0.250.2239C4F100.180.1740HFC-236fa0.130.1041C5F120.130.1242C7F160.130.1243C8F180.090.09The CMIP6 recommendation and data format
We present the community CMIP6 datasets of historical GHG mole fractions. In
conjunction with other data, these GHG surface mole fraction datasets are to
be used in the historical concentration-driven runs for the Climate Model
Intercomparison Project Phase 6 (CMIP6) (Eyring et al., 2016). Depending on
the specific CMIP6 experiment, different protocols and recommendations can
apply. Modellers should hence also check the experiment specific descriptions
(see the special issue available at
https://gmd.copernicus.org/articles/special_issue590.html), including
protocols regarding the other important forcing input datasets like aerosols,
their emissions and optical properties, and land-use patterns, but also
short-lived GHGs like tropospheric and stratospheric ozone for models without
interactive ozone chemistry.
The historical GHG concentrations of this study are specifically designed to
be useful for the historical run, as well as the idealized runs of abrupt4x, 1pctCO2
and picontrol. Also, the PMIP4-related last-millennium experiment
will be based on the GHG concentrations of this study (Jungclaus et
al., 2017; Kageyama et al., 2016).
Regarding the historical runs of the DECK simulations, the CMIP6
recommendation as decided by the CMIP Panel is as follows: “In the
CO2-concentration-driven historical simulations, time-varying global
annual mean mole fractions for CO2 and other long-lived GHGs are
prescribed. If a modelling center decides to represent additional spatial and
seasonal variations in prescribed GHG forcings, this needs to be adequately
documented” (Eyring et al., 2016).
This study provides the data for both the global annual mean mole fractions
as well as the mole fraction histories that take latitudinal and seasonal
variations into account (see data description further below). CMIP6 modelling
groups should indicate which time and space resolution of the data version
they applied. All data are freely available via the PCMDI servers
(https://esgf-node.llnl.gov/search/input4mips/) as netcdf files. The
data is also available via ftp servers, and in multiple data formats (netcdf,
csv, xls and MATLAB mat) as described at climatecollege.unimelb.edu.au/cmip6.
picontrol: global- and annual-mean surface concentrations for the
picontrol CMIP6 experiment. The hemispheric and latitudinally resolved
concentrations for 43 GHGs and three aggregate equivalent concentrations are
provided in the accompanying historical run dataset for the year 1850. The
complexity reduction options for capturing all GHGs with fewer species than
43 are indicated in the table as Option 1, Option 2 and Option 3, with “x”
denoting relevant columns under each option.
YearsCO2CH4N2OCFC-12 eq.HFC-134a eq.CFC-11 eq.CFC-12OtherOption 1xxxxxOption 2xxxxxOption 3xxxxxUnitsppmppbppbpptpptpptppt1850284.317808.25273.0216.5119.1532.110.00All or a subset of other39 individual gases,available in Supplement
In terms of the spatio-temporal resolution, four files for each of the 43
GHGs and the three equivalence species CFC-12 eq., HFC-134a eq. and
CFC-11 eq. (Sect. 2.1.10) are provided as follows:
latitudinal 15∘ bins with monthly resolution (filename-code:
“_15degreelatXmonth”), with monthly means for each latitudinal band
provided at the centre of the box, i.e. -82.5, -67.5, … 67.5,
82.6;
interpolated latitudinal half-degree bins with monthly resolution
(filename-code: “_0p5degreelatXmonth”), with means for each latitudinal
band provided at the centre of the box, i.e. -89.75, -89.25,
… 89.25, 89.75. The area-weighted mean over 15∘ latitudinal
bands is the same as the files under variant I above;
global and hemispheric means with monthly resolution (filename-code
“_GMNHSHmeanXmonth”);
global and hemispheric means with annual resolution (filename-code
“_GMNHSHmeanXyear”).
Given that climate effects will vary depending on whether global, annual-mean
or seasonally varying latitudinally resolved surface mole fractions are
prescribed, modelling groups are asked to document which dataset(s) they
choose.
The CMIP6 recommendation for the picontrol experiment are to use the
1850 GHG mole fractions with annual means as provided in Table 8 (CO2
annual-mean mole fractions of 284.32 ppm, CH4 mole fractions
of 808.25 ppb and N2O mole fractions of 273.02 ppb).
Other gases are covered, depending on the choice of the modelling group by
either following Option 1, Option 2, or Option 3 described in Table 5, or an
equivalently suited method that aggregates the radiative effect of the
remaining 40 GHGs or a large fraction thereof.
The abrupt4x experiment should keep all GHG mole fractions unchanged
from the picontrol run except for the CO2 mole fractions, which
should be increased instantaneously in year 1 (= 1850) of the experiment
to 4 times the 1850 value, namely to 1137.27 ppm (Table 10).
The 1pctCO2 experiment should also keep all GHG mole fractions
unchanged from the picontrol run except for CO2 mole fractions.
Starting in year 1 of the experiment, CO2 mole fractions should
increase by 1 % per annum, reaching slightly over double the CO2
mole fractions in year 70 (or 1920, if the start year is set to 1850) with
570.56 and 1264.76 ppm in year 150 (or year 2000) (Table 9).
As with the abrupt4x and 1pctCO2 scenarios, the historical
experiment should diverge from the picontrol run. GHGs should then follow the
historical observations as derived in this study, reaching, for example, CO2
mole fractions of 397.55 ppm in 2014, and CH4 and N2O
mole fractions of 1831.47 and 326.99 ppb, respectively. Modelling
groups should document which spatial and temporal resolution (see above) of
the provided data they use, as the climate effect will likely be different
with different resolutions.
The future concentration pathways, the so-called “SSP-RCP” scenarios,
considered under ScenarioMIP (O'Neill et al., 2016) are planned to provide
the same data formats and spatio-temporal resolutions. The methodological
approach to derive and adapt both seasonality and latitudinal gradients in
this study was designed such that a future extrapolation will be possible.
1pctCO2: global-mean annual-mean surface CO2 concentrations
for idealized CMIP6 experiments 1pctCO2. All other gases, as in picontrol run
(see Table 8). The value 284.317 ppm with three-digit precision in year
1850 is increased by 1 % per year.
The purpose of our reconstructions is to provide radiative forcing for
climate models. This radiative forcing depends on the vertical as well as
horizontal distribution of a gases' mole fraction. Our reconstructions
describe only surface concentrations and modellers need some method for
calculating the 3-D distribution. If the model is capable of
calculating tracer transport, includes any sinks and sources in the free
atmosphere, and has an appropriate treatment of the boundary layer, we
recommend using this study's surface reconstruction as a mole fraction lower
boundary condition for a mass balance inversion. If this is not possible, we
propose a simple equation to reflect the relaxation of horizontal gradients
with height and the upward propagation of mole fraction changes from the
surface.
CMIP5 ESMs vertical mole fraction averages at the provided pressure
levels – averaged over the 30-year period 1976–2005. The black line
indicates surface mole fractions at the 1000 hPa pressure level. The red
bold line indicates mole fractions at the 100 hPa level (cf. Fig. 14a
and b).
Idealized vertical gradients recommended for implementation of
surface concentration fields. For parametric formulas, see text. Note that
tropospheric columns of non-CO2 gases are – for simplicity –
assumed to be well-mixed. The assumed age of air at the 1 hPa level for
CO2 is 5 years.
Comparison between the recommended annual global mean surface
concentrations of CO2, CH4 and N2O for CMIP5 and
CMIP6 historical experiments.
Comparison of global-mean and hemispheric monthly-average
concentrations of CO2(a), CH4(b) and
N2O(c) between the CMIP6 surface mole fractions (this
study), the NOAA Marine Layer Boundary products, the World Data Centre of
Greenhouse gases (WDCGG) products and the NASA AQUA satellite data of
tropospheric CO2 concentrations. For comparison, individual (monthly
average) NOAA and AGAGE station data across all latitudes is shown in the
background (grey dots).
Comparison of the CMIP6 historical CO2
emissions (a) with the NOAA Marine Boundary Layer (MBL) product from
1979 to 2014 (b). Differences indicate that a seasonal higher
CO2 concentration is implied by the CMIP6 data of up to 5 ppm
in mid-latitude northern bands, whereas some monthly tropical CO2
mole fractions tend to be slightly lower in the CMIP6 product compared to
NOAA MBL (c).
Comparison of the surface CH4 monthly mean concentrations
between CMIP6 (a) and the NOAA Marine Boundary Layer
product (b), and the difference between them (c). Since around 1992,
there are seasonal differences in the mid-northern latitudes with the CMIP6
data being up to 50 ppb higher than the NOAA MBL product. Similarly, higher
concentrations are apparent in the areas of tropical southern and lower
latitude southern areas, presumably due to differences of data over land
areas.
The comparison between latitudinal and monthly N2O
concentrations to the NOAA Marine Boundary Layer product (b). The
differences (c) show that the CMIP6 historical GHG concentrations
are slightly higher in the Southern Hemisphere (0.5 ppb) and slightly
lower in the tropics (0.5 ppb), as the stronger latitudinal gradient
from tropics to southern latitudes is not reproduced in CMIP6 data. Note:
data submitted by P. Tans, personal communication, 2016.
In the case of CO2, there are no sinks in the middle and upper
troposphere or stratosphere and only slight sources due to the oxidization of
CH4 and carbon monoxide (CO). Evidence from ESMs
(Fig. 13) indicates an almost well-mixed tropospheric column in the tropics
and little or partly reversed vertical gradient in the southern troposphere,
while the annual-mean gradient in the Northern Hemisphere is – depending on
the season – variable. The annual average vertical gradient in the Northern
Hemisphere is decreasing in all CMIP5 ESMs analysed here (Fig. 13).
abrupt4x: Global- and annual-mean surface concentrations for the
idealized abrupt4x CMIP6 experiment. The hemispheric and latitudinally
resolved concentrations for 43 GHGs and three aggregate equivalent
concentrations are provided in the accompanying historical-run dataset, with
the 1850 CO2 concentration of 284.317 being multiplied by 4. The
complexity reduction options for capturing all GHGs with fewer species than
43 are indicated in the table as Option 1, Option 2 and Option 3, with “x”
denoting relevant columns under each option.
YearsCO2CH4N2OCFC-12 eq.HFC-134a eq.CFC-11 eq.CFC-12OtherOption 1xxxxxOption 2xxxxxOption 3xxxxxUnitsppmppbppbpptpptpptppt0–1501137.268808.25273.0216.5119.1532.110.00All or a subset of other39 individual gases,available in Supplement
In order to enable the implementation of surface mole fractions in models
that do not have an inherent transport model to capture vertical gradients,
we offer here simplified parameterizations as default options. While an
assumption about a well-mixed atmospheric vertical column seems to be a justifiable
simplification, these simple vertical extensions could increase the realism,
vertical heating structure and overall climatic effect. Specifically,
modelling teams could use the following approximation to extend surface
concentration fields (at the 1000 hPa level) towards higher tropospheric and
stratospheric levels. First, a bell-shaped concentration distribution is
assumed at the 100 hPa level for the higher latitude tropopause and tropical
upper troposphere:
Cl,100hPa,t=C‾global,1000hPa,t…+(C‾global,1000hPa,t-5yrs-C‾global,1000hPa,t)⋅sin(l)22,
with C‾global,1000hPa,t
indicating global-average, annual-average concentrations at the surface
1000 hPa level at time t. Ideally, a smoothed mean-preserving monthly time
series of these annual-average global averages is used to prevent step
changes from calendar month 12 to 1. Equivalently, C‾global,1000hPa,t-5yrs indicates the
global-average, annual-average surface mole fraction 5 years earlier. The
sin(l)22 factor depends on the latitude l and results in the
bell-shaped concentration curve with concentrations at the tropical 100 hPa
level to be identical to the global average surface concentrations, while the
polar mole fractions are effectively of a medium age (2.5 years in the case
of linearly increasing concentration history). Having defined this 100 hPa
concentration level, the tropospheric mole fractions at latitude l and
pressure level p (with p> 100 hPa) can then be assumed to be a
simple linear interpolation between the surface mole fraction level at
latitude l and the 100 hPa level, so that
Cl,p,t=Cl,100hPa,t+Cl,1000hPa,t-Cl,100hPa,t⋅(p-100hPa)(1000hPa-100hPa).
Above 100 hPa, i.e. in the tropical upper troposphere and
stratosphere, the mole fraction is a simple linear interpolation between the
100 hPa level and the top-of-the atmosphere 1 hPa level that is assumed to
have a median age of air of 5 years, so that for p< 100 hPaCl,p,t=C‾global,1000hPa,t-5yrs…+Cl,100hPa,t-C‾global,1000hPa,t-5yrs…⋅(p-1hPa)(100hPa-1hPa),
with C‾global,1000hPa,t-5yrs being again the global-mean surface concentration (1000 hPa)
5 years ago and Cl,100hPa,t the
latitudinally dependent concentration at the 100 hPa level.
This equation captures the general form of the vertical CO2 mole
fraction gradient observed in CMIP5 ESMs – with the 100 hPa
being an approximate division line of the vertical CO2 gradient in
all CMIP5 models (see bold red line in Fig. 13). The annual-average vertical
gradient in the Northern Hemisphere will somewhat reduce the effect of
the strong surface latitudinal gradient. The idealized shaped of the above
parameterization for a hypothetical flat surface mole fraction of
100 ppm is shown in Fig. 14b. Assuming linearly increasing surface
mole fractions from a South Pole minimum towards a 3 ppm higher North Pole
maximum will – under this simplified parameterization - result in an almost
zero vertical tropospheric gradient in the Southern Hemisphere (Fig. 14a).
Exponents “s” to estimate vertical gradient of concentrations for
gases with stratospheric sinks in the stratospheric column – depending on
the latitude “lat”. See text. For HFC-134a and other species with
stratospheric lifetimes shorter than 30 years, the CH4 exponent
parameterization can be used as approximation. This exponent scale
parameterization is taken from the CESM, implemented by J. Kiehl.
Tropics and mid-latitudesMid- to high latitudes,ABS(LAT) < 45∘ABS(LAT) ≥ 45∘CH40.23530.2353+0.0225489× (abs(lat) - 45);N2O0.3478+0.00116× abs(lat)0.40+0.013333× (abs(lat) - 45)CFC-110.7273+0.00606× abs(lat)1.00+0.013333× (abs(lat) - 45);CFC-120.4000+0.00222× abs(lat)0.50+0.024444× (abs(lat) - 45)
For non-CO2 gases, we here suggest a scheme adapted from the CESM
current parameterization – in case that models do not have their own
vertical extrapolation methods. These parameterizations assumed a simplified
vertically well-mixed troposphere and define a tropopause height as follows:
ptropopause(l)=250hPa-150hPa⋅cos(l)2,
with ptropopause(l) being the tropopause height in hPa, depending
on the latitude l. Thus, below the tropopause, the zonal mean
concentrations are assumed to be well-mixed vertically, so that
Cl,p,t=Cl,1000hPa,tforp>ptropopause.
The stratospheric concentration can then be modelled for p<ptropopause as follows:
Cl,p,t=C‾global,1000hPa,t‾-1yrs⋅pptropopause(l)s,
with C‾global,1000hPa,t being
the global-mean and annual-mean surface mole fraction of the previous year,
p/ptropopause(l) being the ratio of the pressure at level p and
the tropopause pressure at that latitude, and s being a gas-dependent
scaling factor (Table 11).
As mentioned above, this simple vertical extrapolation option of the provided
surface data is only to be regarded as a simplified fall-back option in case there are no model-intrinsic parameterizations available or active
tracer transport part of the model. While this study provides the main step
from global-mean and annual-mean concentration histories towards zonally and
monthly resolved ones, future research will be needed to provide more robust
4-D fields of concentrations.
Discussion
We compare our results with a number of other data products. First, a
comparison with the previous CMIP5 recommendation for historical GHG
concentrations is provided (Sect. 5.1). Second, we analyse and compare our
CMIP6 recommendations to what the ESMs from the previous CMIP5
intercomparison produced in terms of CO2 concentration fields in the
emissions-driven runs (Sect. 5.3). Third, we compare our datasets to the
other global-mean, hemispheric and latitudinally resolved datasets, namely
the NOAA Marine Boundary Layer product and the WDCGG time series (Sect. 5.4).
Comparison to CMIP5 input datasets
For the CMIP5 inter-comparison, GHG concentrations were specified for
historical times until 2005, followed by the Representative Concentration Pathways (RCPs) and their extensions until
2300. The recommendations for GHG concentrations were global- and annual-mean
time series (Meinshausen et al., 2011), not including a seasonal cycle or
latitudinal gradient. Those historical time series were composite products of
existing ice core and instrumental data annual means (see references in
Meinshausen et al., 2011). Global, annual-mean CO2 concentrations
over 1975 to 2005 were very close (< 0.7 ppm different) to our
current recommendations for CMIP6. The CMIP5 time series did not show the
slight maximum in CO2 concentrations around 1973 (difference
1.2 ppm), and was generally lower between 1940 and 1956 at about the
time of the World War II, when CO2 concentrations briefly plateaued
(differences between 1.0 and 2.3 ppm) (Fig. 15). While the CMIP5
historical GHGs were an ad hoc extension to the RCP pathways, our CMIP6
recommendation advanced the integration of historical data by accounting for
latitudinal gradients (ice core data in CMIP5 has not been adjusted for the
latitudinal gradients) and by taking into account a large array of additional
data beyond a single network average for more recent times.
Recommended global-mean CH4 concentrations for CMIP5 were generally
lower than derived here, up to 50 ppb around 1910 and between 25 and
30 ppb more recently (2000–2005). The primary reason is that the
CMIP5 data did not take into account the strong latitudinal gradient of
CH4 concentrations. For N2O concentrations, the CMIP5
historical time series did not capture some higher-frequency variability,
which caused the CMIP6 recommendation for the picontrol 1850 global-mean
concentration being lower by around 2.5 ppb, and N2O
concentrations in the 1910s being higher by up to 2.3 ppb (Fig. 15).
Overall, CMIP5 and CMIP6 recommendations are relatively similar. The 1850
picontrol values at the time of CMIP5 were slightly higher for CO2
and N2O (0.14 % or 0.4 ppm and 0.87 % or
2.4 ppb, respectively), countered to some degree by slightly lower
values for CH4 (2.18 % or 17.3 ppb). This is equivalent
to a small net change in base year radiative forcing of
0.0065 Wm-2, when applying linear radiative efficiencies of IPCC
AR5 (Appendix 8.A in IPCC WG1 AR5).
Comparison to CO2 station data between 1958 and 1984
As our data synthesis used monthly station data only from 1984 onwards
(except for Mauna Loa annual averages back to 1958), a comparison to
available station data from before 1984 is useful to qualitatively validate
the extension method applied in this study. While latitudinal gradients (or
rather their first two EOFs) and seasonality changes are extended by
regression (Sects. 2.1.4–2.1.7), the CO2 fields' global-mean has
been optimized to match both the annual average Mauna Loa record and the Law
Dome ice record, specifically our smoothed version thereof (see Table 1 and
Fig. 7k). Thus, it is informative to compare our data product to available
station data from the period before 1984 both in terms of seasonality and the
absolute amplitude (which is derived from the global-mean and the regressed
latitudinal gradient) (see Fig. 7). We here use the Scripps CO2 data
series, available at
http://scrippsco2.ucsd.edu/data/atmospheric_co2/sampling_stations.
In general, the comparison suggests that this study's data product matches earlier station data
rather closely, thereby validating our chosen extension
approach to some degree. There are two noteworthy issues arising from this
comparison though. For high southern latitudes, both Law Dome as well as SPO
in situ and flask station data are available. It seems that our CMIP6 high-latitude data in the Southern Hemisphere could be ∼ 1 ppm too
low over the period 1959–1972 (Fig. 7l). Earlier, in 1958, and subsequently
from 1973 onwards, the match is rather close between SPO station data at
-90∘ and our latitudinal average for the -90 to -75∘
zonal mean. Given our data product matches the MLO record quite closely
(somewhat by design, given the optimization to match the annual-average MLO
record over that time), this points to a slightly exaggerated latitudinal
gradient between 1959 and 1972.
The second issue relates to a bump in the concentration series centred around
1974. In our data assimilation, this bump is a propagation of an anomaly in
the MLO record over that time and seems to a lesser degree to also show up in
other Northern Hemisphere records. However, the southern hemispheric SPO
record does not show (or only minimally shows) this slight upwards aberration from
1972 to 1974 and subsequent slowing and stagnating growth from 1974 to 1976
(while the lower precision Law Dome data would be consistent with that MLO
pattern, see Fig. 7k). To what extent this bump has been present in the
Southern Hemisphere is unknown, although earlier studies (Bacastow, 1976)
relate the increased atmospheric CO2 concentrations to decreased
oceanic uptake during the El Niño back then. Such a process explanation would
suggest that the atmospheric signal is also present throughout large parts of
the Southern Hemisphere, while a predominantly extra-tropical land-related
respiration increase during El Niño could imply that the signal is
predominantly present in the Northern Hemisphere. In summary, our
assimilation's hemispheric upwards anomaly around 1974 of around
∼ 2 ppm could largely be an artefact of our methodology, which
propagates the MLO anomaly globally under the assumption of exogenously
emission-regressed latitudinal gradients.
Comparison to CMIP5 ESM CO2 concentration fields
Several ESMs during CMIP5 used prescribed CO2 emissions instead of
CO2 concentrations and derived CO2 concentration fields
endogenously. For the year 1875, we see that models vary greatly, with some
showing reverse latitudinal gradients with higher concentrations in the south
(e.g. CanESM2), almost no gradient (CESM1-BCC), a local maximum in the
tropics with lower poleward concentrations (MIROC-ESM) and very heterogeneous
fields with high concentrations over the tropical rainforests (NorESM1-ME)
(see Fig. S41). Similarly, for 1990 (Fig. S42), the fields are dissimilar,
with some models exhibiting very strong north–south gradients (MPI-ESM-LR),
while others show only subtle gradients (CanESM2),
although all models indicate an increase of northern hemispheric
concentrations compared to the global mean between 1875 and 1990 (Fig. S45).
Though not as strong as NorESM1-ME, most models show a slight tropical
maximum in the latitudinal gradient (exceptions are CanESM2, MIROC-ESM) during both
1875 and 1990 (Figs. S46 and S47). The high-latitude southern
concentration deviations from the global-mean in the 1875 time slices have
different signs across the models, with some indicating clearly lower
concentrations (BNU-ESM, MPI-ESM-LR, NorESM1-ME) and others suggesting
slightly positive concentrations (CanESM2, MIROC-ESM in 1875). The average of
three CMIP5 ESMs with full CO2 data coverage at the surface 1000 hPa
level and global mean CO2 mole fraction values in line with
observational records (CanESM2, MPI-ESM-LR and NorESM1-ESM) shows a
latitudinal gradient for 1990 comparable to the observed one derived in this
study (Fig. 9b). Thus, given that the pre-industrial latitudinal gradient is
almost flat for the models with the highest skill to replicate current
observations, we assumed constant mole fractions with latitude for
pre-industrial times.
In general, all ESMs show climatological seasonal cycles of CO2
concentrations similar to the seasonality derived in this study (Fig. 9a).
The climatological 1861–1890 average concentrations across the models
clearly exhibit higher seasonality in the Northern Hemisphere, especially
above 40∘ N. While the seasonality in some models is weaker,
especially CESM1-BCC, others show variations of up to ±10 ppm
(MPI-ESM-LR). In addition, the latter model exhibits a larger Southern
Hemisphere seasonality than other models and what we observe. As expected
from our analysis of observational data, this seasonality strengthens up to
1990 across all models (cf. Figs. S43 and S44). The latitudinal spread of the
northern hemispheric minimum extends southwards towards the equator in August,
September and October as we observe (Fig. 9a), with the exception of the
BNU-ESM (Fig. S44), which indicates a northward propagation of the
minimum summer concentration values.
Overall, the basic features of the latitudinal gradient and seasonal cycle
are represented in the ESMs as seen in the observational data. However, the
variation across the models is substantial. This difference of several parts
per million (ppm) in the latitudinal gradient or seasonal cycles will lead
to follow-on differences in the climate response observed in those models.
As common input for the CMIP5 concentration-driven experiments, all models
were provided with the same historical global- and annual-mean CO2
concentrations. Some models had the capability to nudge internally generated
CO2 concentration fields to match the prescribed annual and global
mean CO2 concentrations. Nevertheless, the differences in those
internally generated fields can be substantial, as our analysis from CMIP5
shows, and different from the observations.
For future model inter-comparisons, it seems preferable that any
concentration-driven runs would use the same starting point. Of course, the
longer-term aspiration has to be that emission-driven ESMs reliably reproduce
observational concentration patterns. For CMIP6, modelling groups are
requested to document their choice of concentration input data, specifically
in relation to the chosen temporal and spatial resolutions.
Comparison of global means to NOAA marine boundary layer products and WDCGG
The primary observational data product with coverage across all latitudes is
the marine boundary layer (MBL) or GLOBALVIEW fields (NOAA, 2013; NOAA ESRL
GMD, 2014c), produced by the NOAA based on the Cooperative Global Air
Sampling Network (Conway et al., 1994; Dlugokencky et al., 1994b; Trolier et
al., 1996) for CO2, CH4 and N2O (available at
http://www.esrl.noaa.gov/gmd/ccgg/mbl/mbl.html, with N2O data
from P. Tans, personal communication, 2016). The aggregation method used to
produce this dataset is to first fit parametric functions to the weekly data
of each station, thereby providing a gap-filling method. In a next step, the
procedure fits smooth weekly latitudinal distributions to the various station
data points (Tans et al., 1989). These latitudinal distributions are then
combined into a 2-D field of latitude versus time, comparable to this study's
data product. The time period of these NOAA MBL data products is 1979–2014
for CO2, 1983–2014 for CH4 and 2001–2014 for N2O.
The four main methodological differences between the NOAA MBL data product
and ours are as follows: (1) the NOAA data product has a higher resolution in
time (weekly instead of monthly) and latitudes; (2) the NOAA MBL data product
includes only a subset of the NOAA network data (sites within the marine
boundary layer), while this study mixes both NOAA and AGAGE network data in
the case of CH4 and N2O; (3) this study characterizes the
global fields by lower rank representations (EOFs) of annual mean latitudinal
gradients and seasonality, while the NOAA product derives latitudinal
gradients (and seasonality thereby only implicitly) directly from the
observations at each time step; and (4) this study is extended by ice core and firn data,
regressions, and extrapolation or interpolation to span the full-time
period between year 0 and 2014. Thus, this study seamlessly merges in situ
observational, air archive, ice and firn data to generate a comprehensive
data product.
For several applications, the NOAA data product has clear advantages.
However, with the task of producing a continuous data product beyond the
instrumental observations, this study had to choose a method that was readily
extendable. Hence, this study chooses the characterization of global fields
into global means, latitudinal gradients and seasonality. This implies a high
degree of regularizations by relying on EOFs and corresponding scores. By
regression, these EOF scores for latitudinal gradients or seasonality changes
can be easily extended to cover the full-time period of interest. Hence, our
method allows an estimate of global-means even if there is only a single data
point (such as a Law Dome ice core record for a specific year), under the
assumption that latitudinal gradients and seasonality are captured by the
derived EOFs and regressed EOF scores.
Global-average time series of monthly GHG mole fractions are also provided by
the World Data Center for Greenhouse Gases (WDCGG) (Tsutsumi, 2009). The
WDCGG product uses similar smoothing techniques to the NOAA product, but
include, like this study, a broader set of measurement stations, both in
terms of regional coverage (including continental stations) and different
networks that use different calibration scales, sampling, gas handling, etc.
We compare the results of this study and NOAA MBL and WDCGG products.
Overall, our monthly hemispheric averages of CO2 closely match the
NOAA MBL product. The NOAA MBL product (which is not the same as NOAA network
monthly averages) suggests a slightly faster increase of northern hemispheric
concentrations in the latter months of each calendar year (cf. thick and thin
orange lines in Fig. 16a). Specifically, this difference results from the
mid-latitude northern hemispheric bands from about 1995 onwards (with
monthly-average differences of up to 4 ppm) where our study is higher
than the NOAA MBL product. This could be because this study does not screen
out land stations closer to the pollution sources, as the NOAA MBL product
does.
Likewise, the WDCGG includes a broader set of stations and matches very
closely with our global-mean time series, with our study being very close to
WDCGG or in between NOAA MBL and WDCGG (Fig. 16a). Given that the difference
between the NOAA study and our study has a strong seasonality, the nature of
those pollution sources and how they become mixed in the atmosphere, if these
effects contribute to the differences, could be a combination of
fossil-fuel-related and (more seasonally varying) biospheric sources
(Fig. 17c). The southern hemispheric means of our study and NOAA MBL are very
closely matched (cf. thick and thin blue lines in Fig. 16a). Consequently,
the global-mean concentrations from NOAA MBL and our study are closely
matched, although again our data suggests NH autumn concentrations rising
slightly faster than the NOAA MBL product, reflecting the northern
hemispheric difference (cf. thick and thin black lines in Fig. 16a).
For CH4, the differences between this study and the NOAA MBL data are
more systematic and stronger (∼ 10 ppb), with generally higher
surface CH4 concentrations implied by this study (Fig. 16b). Again,
this study's global mean matches the WDCGG closely or sits in between the
NOAA MBL and the WDCGG data products. There are some differences in the
seasonality compared to the NOAA MBL product though. The seasonal variation
is similarly shaped between our study and the NOAA MBL for the Southern
Hemisphere, although there seems to be a slight phase shift of about a month
with the NOAA MBL product in the Southern Hemisphere, assuming a slightly
earlier increase and decrease and slightly higher amplitude (Fig. 16b). This
phase shift of the Southern Hemisphere, together with sometimes lower peak
northern hemispheric concentrations in the NOAA MBL product, suggests
global-mean NOAA MBL CH4 concentrations that show a double peak within
any year, while our data assimilation and the WDCGG product suggests a
smoother single-peak oscillation of global-mean CH4 concentrations
(Fig. 16b). This peak results from the mid-northern latitudes, where in the
summer months, our study suggests up to 40 or 50 ppb higher concentrations
(Fig. 18c).
For N2O, the WDCGG global mean and our data match very closely, with
our implicit smoothing due to our lower rank representation of seasonal
cycles and latitudinal means resulting in a smoother global mean compared to
WDCGG (Fig. 16c). Similarly, the draft data product of the NOAA MBL indicates
almost identical mole fractions to our concentration fields over the
available time period from 2001 to 2014, with maximal differences being
0.8 ppb (Fig. 19).
In summary, our dataset closely matches the global means of WDCGG in many
years, but provides a complete 2-D field of mole fractions. In comparison to
the NOAA MBL products, there is one more systematic difference. Our CMIP6 GHG
concentration fields are meant to represent the mean monthly state of the
latitudinally averaged surface atmosphere, including land and polluted areas,
i.e. not confined to areas with background concentrations (Sect. 6). This is
a key difference to the NOAA Marine Boundary Layer product, which is a
consistent background concentration product, resulting in slightly lower
global-mean concentration estimates.
Comparison to mid-troposphere CO2 concentrations by NASA Aqua satellite
Since its launch in 2002, the Aqua satellite and its infrared sounder
provides an additional independent data product to estimate tropospheric
CO2 mole fractions. Rather than at ground level, this sensor provides
an estimate of tropospheric concentrations with a maximum sensitivity around
7 km height, i.e. in the mid-troposphere. In the tropics and the parts of the
Southern Hemisphere that are covered by the Aqua satellite product, the
agreement between our data and the AIRS level 3 data (available at
ftp://acdisc.gsfc.nasa.gov/ftp/data/s4pa/Aqua_AIRS_Level3/AIRX3C2M.005/)
is encouraging, although the overall gradient is lower in line with 3-D
atmospheric transport model results (Olsen and Randerson, 2004). In the
Northern Hemisphere, the difference in the phase and amplitude of the
seasonal cycle is most apparent, with satellite data showing a later onset of
the autumn concentration increase by about 4 months, while the drawdown of
concentrations seems closer in phase between mid-troposphere and surface
concentrations (Fig. 16a). Overall the amplitude is less than half of the
surface hemispheric mean amplitude, leading to seasonally higher winter and
lower summer concentrations of our surface data product in the Northern
Hemisphere by up to 10 ppm (Fig. 17e).
This systematic difference between ground-level and mid-atmosphere
concentrations, supported by 3-D transport modelling studies (Olsen and
Randerson, 2004), has ramifications for the implementation of vertical
concentration profiles in climate models. Without taking into account the
dampened seasonal cycle and latitudinal gradient in the mid- and higher
troposphere, the models could overestimate the variations in the radiative
effects, if our latitudinally and monthly resolved surface concentration
fields are prescribed. On the other hand, if global-annual mean values are
prescribed, the radiative forcing effect variations over latitudes and within
a year will obviously be underestimated.
Comparison to other literature studies
Our GHG derivations over the recent instrumental periods are based on the
AGAGE and NOAA station-by-station data and we extended our 2-D concentration
field results back in time by using, for example, global-mean estimates of
previous studies (Sect. 2). The AGAGE and NOAA networks themselves publish global-mean
results, and WMO as well as other literature studies produce composite
long-term global-mean and/or hemispheric concentration estimates. Thus, while
often not entirely independent, as the studies use the same original data
sources or we rely on some studies' previous derivations, we here provide a
comparison to a selection of the literature. Specifically, in addition to the
comparisons with NOAA marine boundary layer, WDCGG and NASA Aqua satellite
data, we discuss some instances where our results show substantial
differences compared to earlier studies that have derived hemispheric or
global means from instrumental data (Montzka et al., 2015; Rigby et
al., 2014), from firn data (Butler et al., 1999; Trudinger et al., 2016) or
are themselves composites of multiple data sources (Martinerie et al., 2009;
Velders and Daniel, 2014; WMO, 2014). The comparisons are shown in the
panels (f), (g), and (h) of the fact sheets for each gas (Figs. 9, 11, 12,
and S1–S40) with the comparison data described in Table 12. High-latitude
Northern Hemisphere data for atmospheric mole fractions is reported in the
supplement of Buizert et al. (2012), provided by Vas Petrenko and Patricia
Martinerie (Table 12). For CO2, the Petrenko dataset has, as expected
for the high northern latitudes, a very strong seasonal cycle, consistent
with our less pronounced northern-hemispheric-average cycle, as the data
represents higher northern latitudes (Fig. 9f, g, and h). The long-term
concentration trend over time in the Petrenko CO2 record seems
similar to the global CMIP5 dataset which in turn was based on previous Law
Dome data, indicating a slight local maximum in 1890 and lower 1940s plateau
(cf. Figs. 9g and 15).
For CH4, the Petrenko record shows a comparable, yet again stronger,
seasonality. The annual means are very comparable to our derivation (compare
the high-latitude red circles, indicating annual-mean station averages of our
analysis and Petrenko data as shown in Fig. 11f), although there are some
steps in annual means in the Petrenko dataset around 1956 and 1975, which
are not present in our dataset (Fig. 11f). For earlier times, i.e. between
1860 and the 1920s, the Petrenko annual mean is closer to our global mean,
not the high-latitude estimates, as our study assumes a large latitudinal
gradient based on the NEEM and Law Dome data differences (Sect. 2)
(Fig. 11g).
For CCl4, the Martinerie data show a lower increase from 1955 to the
late 1960s and strong increase around 1970. The firn data by Butler et
al. (1999) suggest an earlier start of atmospheric concentration increases
around 1890, and then slightly lower levels over 1960–1990 compared to the
WMO (2014) and Velders and Daniel (2014) time series which we use as an
optimization target for our 2-D fields. The difference between the Butler and
Velders datasets can probably be explained by the wider firn air age
distribution in the study by Butler. The findings by Sturrock et al. (2002)
suggest an onset of detectable atmospheric concentrations around 1920
(Fig. 5f therein). The NOAA global mean that is available from 1992 onwards
(Montzka et al., 1999 updated at
http://www.esrl.noaa.gov/gmd/hats/combined/CCl4.html) and indicates
initially slightly higher global mean estimates than our derivation, which is
for the instrumental period based on 6 AGAGE and 13 NOAA HATS stations
(Fig. S1f, g, h).
Description of data labels shown in fact sheets, namely Figs. 9, 11,
12, and S1 to S40.
LabelGasesDescription/sourceNOAA_SURFACE_FLASKCO2Atmospheric carbon dioxide dry air mole fractions fromthe NOAA ESRL Carbon Cycle Cooperative GlobalAir Sampling Network, 1968–2014,version: 2015-08-03Surface flask, available atdata ftp://aftp.cmdl.noaa.gov/data/trace_gases/co2/flask/surface/(Dlugokencky, 2015b)NOAA_SURFACE_INSITUCO2Atmospheric Carbon Dioxide Dry Air Mole Fractions from quasi-continuousmeasurements at Barrow, Alaska; Mauna Loa, Hawaii; American Samoa;and South Pole, 1973–2013; National Oceanic and AtmosphericAdministration (NOAA); Earth System Research Laboratory (ESRL),Global Monitoring Division (GMD),Carbon Cycle Greenhouse Gases (CCGG);version: 2014-11-10, available atftp://aftp.cmdl.noaa.gov/data/trace_gases/co2/in-situ/surface/(NOAA ESRL GMD, 2014a, b, c, d)NOAA_SURFACE_FLASKCH4Atmospheric methane dry air mole fractions fromthe NOAA ESRL GMD Carbon Cycle Cooperative Global Air,Sampling Network, 1983–2014,File versions: 2015-08-03, available atftp://aftp.cmdl.noaa.gov/data/trace_gases/ch4/flask/(Dlugokencky, 2015a)HATS_GLOBAL_COMBINEDN2O, CCl4, CFC-11,Combined data from the NOAA/ESRL Global Monitoring Division andCFC-113, CFC-12, SF6two or more measurement programs.Available at ftp://ftp.cmdl.noaa.gov/hats/n2o/combined/HATS_global_N2O.txt,ftp://ftp.cmdl.noaa.gov/hats/cfcs/cfc113/combined/HATS_global_F113.txt,ftp://ftp.cmdl.noaa.gov/hats/cfcs/cfc11/combined/HATS_global_F11.txt,ftp://ftp.cmdl.noaa.gov/hats/cfcs/cfc12/combined/HATS_global_F12.txt,ftp://ftp.cmdl.noaa.gov/hats/sf6/combined/HATS_global_SF6.txt,ftp://ftp.cmdl.noaa.gov/hats/solvents/CCl4/combined/HATS_global_CCl4.txtMONTZKA_NOAA_GMDCCl4, CFC-11, CFC-113,Flask data provided from the Global Monitoring Division of the NationalCH3CCl3, CH3Br, CH3Cl,Oceanic and Atmospheric Administration's Earth System ResearchCH2Cl2, HCFC-22,Laboratory (NOAA/ESRL/GMD) as a result of analysis on gas chromatographyHCFC-141b, HCFC-142b,with mass spectrometry instrumentation. Principal investigators S. MontzkaHFC-134a, HFC-152a,and James W. Elkins. Version 13 November 2015. Data available atHFC-32, HFC-125,ftp://ftp.cmdl.noaa.gov/hats/cfcs/cfc113/flasks/GCMS/CFC113_GCMS_flask.txt,HFC-143a, HFC-365mfc,ftp://ftp.cmdl.noaa.gov/hats/solvents/CH3CCl3/flasks/GCMS/CH3CCL3_GCMS_flask.txt,HFC-227ea, Halon-1211,ftp://ftp.cmdl.noaa.gov/hats/methylhalides/ch3br/flasks/CH3BR_GCMS_flask.txt,Halon-1301, Halon-2402ftp://ftp.cmdl.noaa.gov/hats/methylhalides/ch3cl/flasks/CH3Cl_GCMS_flask.txt,ftp://ftp.cmdl.noaa.gov/hats/solvents/CH2Cl2/flasks/ch2cl2_GCMS_flask.txt,ftp://ftp.cmdl.noaa.gov/hats/hcfcs/hcfc22/flasks/HCFC22_GCMS_flask.txt,ftp://ftp.cmdl.noaa.gov/hats/hcfcs/hcfc141b/HCFC141B_GCMS_flask.txt,ftp://ftp.cmdl.noaa.gov/hats/hcfcs/hcfc142b/flasks/HCFC142B_GCMS_flask.txt,ftp://ftp.cmdl.noaa.gov/hats/hfcs/hfc134a_GCMS_flask.txt,ftp://ftp.cmdl.noaa.gov/hats/hfcs/hf152a_GCMS_flask.txt,ftp://ftp.cmdl.noaa.gov/hats/hfcs/HFC-32_M2_MS_flask.txt,ftp://ftp.cmdl.noaa.gov/hats/hfcs/HFC-125_M2_MS_flask.txt,ftp://ftp.cmdl.noaa.gov/hats/hfcs/HFC-143a_M2_MS_flask.txt,ftp://ftp.cmdl.noaa.gov/hats/hfcs/HFC-365mfc_GCMS_flask.txt,ftp://ftp.cmdl.noaa.gov/hats/hfcs/HFC-227ea_GCMS_flask.txt,ftp://ftp.cmdl.noaa.gov/hats/halons/flasks/HAL1211_GCMS_flask.txt,ftp://ftp.cmdl.noaa.gov/hats/halons/flasks/H-1301_M2_MS_flask.txt,ftp://ftp.cmdl.noaa.gov/hats/halons/flasks/HAL2402_GCMS_flask.txtAGAGE_GC-MD_MONTHLYCFC-11, CFC-12, CH3CCl3,Chemical species measured by AGAGE GC-ECD/FID/MRD system.CCl4, N2O, CFC-113,Version 20 June 2015. Data available atCH4, CHCl3http://agage.eas.gatech.edu/data_archive/agage/gc-md/monthly/(Cunnold et al., 2002, 1997; Fraser et al., 1996;O'Doherty et al., 2001; Prinn et al., 1990, 2001, 2005;Reimann et al., 2005; Simmonds et al., 1998)AGAGE_GC-MS_MONTHLYHFC-134a, HCFC-22,Chemical compounds measured by AGAGE GC-MSHCFC-141b, HCFC-142b,(ADS) system. Version 20 June 2015.CH3Cl, CH3Br,Data available atHalon-1211, Halon-1301,http://agage.eas.gatech.edu/data archive/agage/gc-ms/monthly/HFC-152a, CH2Cl2,(Cox et al., 2003; Miller et al., 1998;CHClCCl3, CCl2CCl2O'Doherty et al., 2004; Simmonds et al., 2004)
Continued.
LabelGasesDescription/sourceAGAGE_GC-MS-MEDUSA_MONTHLYCFC-11, CFC-12, CFC-113,Chemical compounds measured by Medusa GCMS system.CFC-114, CFC-115, HCFC-22,Version 20 June 2015. Data available atHCFC-141b, HCFC-142b, HFC-125,http://agage.eas.gatech.edu/data archive/agage/gc-ms-medusa/monthly/HFC-134a, HFC-152a, HFC-365mfc,(Prinn et al., 2000a)HFC-23, HFC-4310mee, Halon-1211,Halon-1301, Halon-2402, CH3Cl,CH2Cl2, CHCl3, CH3Br,CH3CCl3, CCl4, SF6,SO2F2, NF3, PFC-14, PFC-116,PFC-218, HFC-32, HFC-143a,HFC-227ea HFC-236fa HFC-245faMONTZKA – NOAA ODSHCFC-22, CFC-113, CFC-11,Data from July 2015 update of NOAA compilation of monthly globalupdate 7/2015HCFC-141b, CCl4, CFC-12,mean concentrations, made available on web as “2015 updateHCFC-142b, CH3CCl3, H-1211,total Cl Br & F July update.xls” by S. Montzka atH-1301, H2402, CH3Br,ftp://ftp.cmdl.noaa.gov/hats/Total_Cl_Br/.HFC-134a, HFC-152a, HFC-143a,The substances HCFC-22, CFC-113, CFC-11, HCFC-141b, CCl4, CFC-12,HFC-125, HFC-32,HCFC-142b, CH3CCl3, Halon-1211, Halon-1301, are Halon-2402,HFC-365mfc, HFC-227eaare updated from data displayed in Fig. 1 in Montzka et al. (1999),with CH3Br data published in Montzka et al. (2003)and with HFC data published in Montzka et al. (2015).MARTINERIE-2010SF6, CFC-11, CFC-12,Monthly high-latitude Northern Hemisphere data by Patricia Martinerie,CFC-113, CCl4,made available as Supplement by Buizert et al. (2012)CH3CCl3, HFC-134ain files SCENARIO_NEEM08_XX.txtPETRENKO-2010CO2, CH4Monthly high-latitude Northern Hemisphere data by Vas Patrenko, madeavailable as Supplement by Buizert et al. (2012) in filesSCENARIO_NEEM08_CO2.txt and SCENARIO_NEEM08_CH4.txtWDCGG (2015)CO2, CH4, N2OData synthesis as available from the World Data Centreof Greenhouse Gas Emissions (Tsutsumi, 2009),available at http://ds.data.jma.go.jp/gmd/wdcgg/.Version: co2_monthly_20151109.csv, ch4_monthly_20151109.csvand n2o_monthly_20151109.csvNOAA MBLCO2, CH4NOAA Greenhouse Gas Marine Boundary Layer reference,derived from atmospheric carbon dioxide, methane andnitrous oxide concentrations, from the NOAA ESRLCarbon Cycle Cooperative Global Air, Sampling Network,available at http://www.esrl.noaa.gov/gmd/ccgg/mbl/for CO2 and CH4. Zonal means for SH and NH,as well as global means. File creation dates: 11 February 2016CMIP5 HIST.ManyThe global-mean annual average concentrations that were usedas default recommendation for concentration-driven runsin the CMIP5 experiment (Meinshausen et al., 2011).CMIP5 CTRL.ManyThe global-mean annual average concentrations in 1850that were recommended as picontrol concentrations inthe CMIP5 experiment (Meinshausen et al., 2011).FIRN –CFC-12, HFC-134a,“Southern Hemisphere atmospheric trace-gas histories usedMontzka-(2009)HCFC-22, CH3CCl3in the analysis of firn air” data compiled by Montzka in 2009(available at ftp://ftp.cmdl.noaa.gov/hats/firnair/ in file“SH Atmosphere Trace Gas Histories.xls”), based on several earlierstudies (Butler et al., 1999; Elkins et al., 1993; Montzka et al., 1993,1996, 2000; Prinn et al., 2005),and e.g. reported in Aydin et al. (2010) for CFC-12and underlying (Montzka et al., 2010).WMO (2014)CFC-11, CFC-12, CFC-113,Data from Table 5A2 in the 2014 Ozone Assessment (WMO, 2014),CFC-114, CFC-115, CCl4,starting with 5-year intervals from 1955 to 1980 then annually.CH3CCl3, HCFC-22,We interpolated the data to annual values using a localHCFC-141b, HCFC-142b, Halon-1211,polynomial regression between 1955 and 1980.Halon-1202, Halon-1301,Halon-2402, CH3Br, CH3ClWMO2014/AGAGEHFC-125, HFC-134a, HFC-152a,The network average global-mean mole fractions from“late”/“early”HFC-143a, HFC-32, HFC-245fa,the AGAGE network as shown in the WMO Ozone AssessmentHFC-365mfc, HFC-227ea, HFC-236fa,Report (WMO, 2014)CF4, HFC23, C2F6,C3F8, SF6, SO2F2, NF3
Continued.
LabelGasesDescription/sourceWMO2014/NOAAHFC-134a, HFC-152a, SF6NOAA global-mean annual average time series asshown in WMO Ozone Assessment Report (WMO, 2014).WMO2014/PFCC4F10, C5F12, C6F14,PFC data compiled and shown inC7F16, C8F18WMO Ozone Assessment Report (WMO, 2014).AGAGE – globalHFC-23, HFC-125, HFC-134a,Monthly global means of baseline datamonthly averageHFC-152a, HFC-227ea, HFC-236fa,derived from AGAGE measurements based onHFC-245fa, HFC-365mfc, HCFC-22,AGAGE GC-MS/Medusa measurementsHCFC-141b, HCFC-142b, H-1211,(from 2004 to current) from file global_mean_ms.txtH-1301, CH3Br, CH3Cl,available at http://agage.eas.gatech.edu/data_archive/global_mean/.CH2Cl2, CCl2CCl2, CHClCCl2,SF6, SO2F2, PFC-14,PFC-116, PFC-218, CFC-113,CFC-114, CFC-115, HFC-4310meeBinned annualAllThese are the monthly averages for each 15∘ zonal meanobservationsderived from the analysed station data points(with three-digit station names provided in the top leftcorner of panel f of each fact sheet). An “n/a” indicationbehind the latitude indicator means that not enough rawstation data points were available to create zonal meansfor that latitude. The estimate of the latitudinal gradient isthen based on the remainder of available latitudinal bands.Other labels, namely:VariousSee respective literature studies (Arnold et al., 2013, 2014;Montzka et al. (2015)Butler et al., 1999; Ivy et al., 2012;Velders et al. (2014)Montzka et al., 2015; Mühle et al., 2010;Mühle et al. (2010)Newland et al., 2013; Oram et al., 2012;Trudinger et al. (2016)Trudinger et al., 2016; Velders and Daniel, 2014;Ivy et al. (2012)Vollmer et al., 2016; Walker et al., 2000;Worton (2007)Worton et al., 2007). Note that the CCl4 dataButler et al. (1999)by Walker et al. (2000) is used as 1910–1950Arnold et al. (2013, 2014)amendment to the Velders and Daniel (2014) time series.Vollmer et al. (2016)Oram et al. (2012)Walker et al. (2000)Newland et al. (2013)
For CFC-11 (Fig. S2g), the NOAA Montzka-ODS reconstruction of the global-mean
is slightly higher (1 ppt) than ours, which is almost identical to
WMO (2014) and data by Velders and Daniel (2014). Those differences
presumably result from differences in station coverage, different calibration
scales, and air sampling and analysis techniques between the NOAA and AGAGE
networks. The seasonalities show comparable amplitudes, as they do for CFC-12
(Fig. S3h). With CFC-115, our study follows the historical shape of the WMO
(2014) record, with Velders and Daniel (2014) being slightly lower
(∼ 0.5 ppt) (Fig. S6f).
For CH2Cl2, the in situ instrumental record we use only reaches back
to 1994, although the Cape Grim air archive record goes back to 1978. From
1994 to 2003, the northern latitude measurements imply a mole fraction
reduction from 40 to 30 ppt, whereas the southern hemispheric
measurements are almost flat during that time (also shown in Trudinger et
al., 2004) (Fig. S7f). We note that there are substantial uncertainties in
the pre-1995 concentrations, as for example Koppmann et al. (1993) reported
18 and 36 ppt average concentrations for the southern hemispheric and
northern hemispheric measurements from a 1989 Atlantic transect ship
measurement campaign (not shown in the figure). This could imply a global
average value of approximately 27 ppt in 1989, instead of the
20 ppt assumed in this study – although different calibration scales
might contribute to this difference. Recent seasonality and increases of
CH2Cl2 are closely matching other time series, such as the AGAGE and
NOAA results from GCMS measurements (Fig. S7f). However, there is a slight
offset in the absolute level, possibly caused by our study not sorting out
data points from so-called pollution events in the case of AGAGE data for
CH2Cl2, whereas NOAA results are from flasks collected only in
baseline-air conditions (Spivakovsky et al., 2000).
For CH3Br, our CMIP6 recommendations match the NOAA very closely
(Montzka et al., 2003 updated on
ftp://ftp.cmdl.noaa.gov/hats/methylhalides/ch3br/flasks) and AGAGE
global means (2014) after 1995. Before then, the Butler et al. (1999)
global-mean firn reconstruction coincides closely with our southern
hemispheric mean. The 2004 firn reconstruction by Trudinger et al.(2004) is
close to the southern hemispheric mean, but shows somewhat more variation
than the smooth exponential increase assumed by this study, WMO (2014), and
Velders and Daniel (2014).
For CH3CCl3, the overall agreement between the different (although
not independent) studies considered here is excellent, for example the high
northern latitude data from Martinerie (Buizert et al., 2012; Martinerie et
al., 2009) in the South Pole firn data reconstruction (Montzka et al., 2010),
approximately in line also with the findings by Sturrock et al. (2002).
The atmospheric concentrations of CH3Cl show a strong seasonal cycle,
as is to be expected from the short lifetime due to the OH-related sink. As
in the case of methyl bromide (CH3Br), the pre-instrumental period
before 1995 implies a number of uncertainties in our CH3Cl time
series. Here, we follow again the WMO (2014) and (not independent) Velders
and Daniel (2014) reconstructions that are based on Butler et al. (1999) firn
reconstructions. However, we note that the more recent Trudinger et
al. (2004) CH3Cl reconstruction indicates both a significantly lower
concentration for southern latitudes in the 1970s and a smoother increase
compared to the more sudden rise of concentrations around 1940 as implied in
this study (Fig. S10g).
As briefly discussed in Sect. 3.4, the CHCl3 history in this study
relies on the Worton et al. (2006) reconstruction, whose shape is similar to
Trudinger et al. (2004), although the latter indicates lower global mean
concentrations and not the diminishing latitudinal gradient suggested by
Worton et al. (2006). As with other gases (e.g. CH2Cl2), the implied
pre-industrial value of around 6 ppt should be investigated in the
future (Fig. S11).
For Halon-1211, the recent study by Vollmer et al. (2016) and the earlier
study by Sturrock et al. (2002) (not shown) suggest slightly higher initial
concentrations (around 1975–1988) compared to the initially lower and then
larger exponential increase we assumed by following Velders and Daniel
(2014). We follow the global-mean derivation in the CSIRO inversion from
Vollmer et al. (2016) in the case of Halon-1211. After 1990 the southern
hemispheric reconstruction by the Bristol and CSIRO inversions (Vollmer et
al., 2016) are slightly lower and hence the latitudinal gradient slightly
larger than what we derived from the AGAGE and NOAA station data, but the
differences are small (Fig. S12f). The Cape Grim measurements analysed on the
UEA volumetric scale (Newland et al., 2013) are also in good agreement with
the small offset to our global mean, consistent with the derived latitudinal
gradient (Fig. S12f). Similarly to Halon-1211, the very early concentration
increases of the Halon-1301 between 1970 and 1978 are higher in the Vollmer
et al. (2016) study than in Velders and Daniel (2014), and again the more
recent years from 2007 onwards (Fig. S13h) are higher in Vollmer et
al. (2016). In those latter years, our aggregation of AGAGE and NOAA station
data, however, suggests slightly lower concentrations, although the absolute
difference (0.05 ppt) is within the measurement uncertainty and the
overall agreement is very good. The Newland et al. (2013) study of southern
hemispheric concentrations at Cape Grim would suggest slightly lower
concentrations, although part of the slight offset could be related to
differences in scales. However, our Halon-1301 record suffers from a
potentially inadequate scaling of the latitudinal gradient. A low gradient
around 2000–2002 (Fig. S13d and f) results from our scaling with global
emissions that are assumed to drop in that period (Velders and Daniel, 2014)
although subsequent station data suggest again a slightly stronger gradient.
Furthermore, a second issue with our Halon-1301 record is a slight drop in
the monthly data in year 2014 (Fig. S13f), which is likely an artefact of our
assimilation procedure, to be corrected by assimilations that consider
observational data beyond 2014.
Halon-2402 is likely the most obvious example where a shifting spatial coverage density of measurements can lead to small jumps in latitudinal
gradients or global means (Fig. S14f and h). The overall mole fractions are
very small and the early agreement between the WMO (2014) time series and the
Vollmer et al. (2016) findings is very good. In 2009, when data coverage
increased, the latitudinal gradient is suggested to suddenly decrease, which
is likely an artefact of the assimilation procedure that is only able to cope
with time-varying data coverage to a certain degree (Sect. 2). However,
overall, the implied shifts of 0.02 ppt are negligible in the larger
picture, and certainly negligible for radiative forcing, as the shift in
southern hemispheric radiative forcing is equivalent to only about
0.000003 Wm-2 (Fig. S14h). Halon-2402 is also an illustration of
how big differences in some measurement scales can potentially be. The Cape
Grim data analysed by Newland with a volumetric UEA scale indicates
10–15 % lower concentrations (Fig. S14f) (Newland et al., 2013).
For HCFC-142b our derived global-mean is in the middle of the AGAGE and NOAA
network averages, despite our study including those data points that are
subject to “pollution” events in the case of HCFC-142b, with large positive
outliers (Fig. S17f), similar to in the case of HFC-134a (Fig. S31f).
Pollution events might, however, be contributing to the difference between
our HFC-152a global-means and the two independently derived network global
means for AGAGE and NOAA, which largely exclude pollution events by using
statistical methods or conditional sampling (O'Doherty et al., 2001) (see
Fig. S33f). Two more issues can be observed with HCFC-142b data. Firstly, our
end of 2014 concentrations are somewhat uncertain and in this case possibly
incorrectly decreasing, which results from the smooth annual mean
representation and our assimilation procedure. The differences are again very
small and negligible in radiative forcing terms, but a smooth connection will
have to be designed for the adjacent datasets representing SSP-RCP scenarios.
Secondly, since 2010, our estimates for the HCFCs, namely HCFC-22
(Fig. S15f), HCFC-141b (Fig. S16f) and HCFC-142b (Fig. S17f), indicate
smaller increases than implied by the post-2010 non-observational scenario
data represented by Velders and Daniel (2014). As in the early study by
Sturrock et al. (2002), our study represents the slow onset of HCFC-142b
concentrations in between 1960 and 1990 as shown in WMO (2014) and Velders
and Daniel (2014).
For the three main PFCs, i.e. CF4 (Fig. S26), C2F6
(Fig. S18) and C3F8 (Fig. S19), we find a similar and good agreement
of the main studies. The outliers are the previously recommended CMIP5
concentrations (Meinshausen et al., 2011) for these gases, which were at the
time not yet based on either the Trudinger et al. (2016) or Mühle et
al. (2010) studies. As mentioned above, the concentrations of the lesser
important PFCs, C4F10 (Fig. S20), C5F12 (Fig. S21),
C6F14 (Fig. S22), C7F16 (Fig. S23) and C8F18
(Fig. S24) are based on the Ivy et al. (2012) reconstructions, with reversing
latitudinal gradients in the case of C6F14, C7F16, and
C8F18, which are unexplained so far and require further
confirmation. Our historical c-C4F8 concentrations are based on the
study by Oram et al. (2012) with assumed conversions of the Cape Grim
measurements to northern hemispheric and global averages.
For HFC-43-10mee, we based our trajectory on the Northern and Southern
Hemisphere estimates of Arnold et al. (2014) with relatively small
latitudinal gradient and hemispheric means being informed by the recently
available observations since 2010 from the AGAGE Medusa instruments
(Fig. S29f). Note that for HFC-365mfc data (Fig. S37), the difference between
the station data and those published in Montzka et al. (2015) reflects a
difference that is now much smaller after a calculation-related correction
was applied to the NOAA calibration scale after the publication of Montzka et
al. (2015). All studies are now in relatively close alignment with the shown
AGAGE network average, the Vollmer et al. (2011) study and our derivation
(which is slightly lower, < 0.1 ppt). In addition, the air archive
and AGAGE network analysis by Vollmer et al. (2011) investigated the HFCs
HFC-236fa, HFC-227ea and HFC-245fa. Those results are closely aligned
with the ones constructed here based on the WMO AGAGE network average
estimates (Figs. S35, S34, S36).
Like our study, there are also studies that assimilate a wide range of gases
with latitudinal and seasonal variation. For example, the AGAGE network
assimilation with a 12-box model and optimization approach to reconcile
emissions and concentrations (Rigby et al., 2011, 2013) produces four
semi-hemispheric concentration time series with three vertical levels (Rigby
et al., 2014). Those studies based on AGAGE data are more comprehensive than
this one, as both emissions and concentrations as well as lifetimes are
optimized and reconciled. In our case, we only assimilate AGAGE and NOAA
observations to derive atmospheric mole fractions in 15∘ latitudinal
bands (Sect. 2).
Limitations
Even though the presented dataset of historical surface GHG concentrations is
– to our knowledge – more comprehensive than other composite datasets
before it, there are several key limitations.
Specific use of dataset
First, the dataset was assimilated from several sources to provide a common
starting point for global climate models as part of the CMIP6 experiments.
Thus, for example, the data was not designed as a starting point for
inversion studies, which estimate emissions, or studies of biogeochemical
processes. Those studies tend to require pure observations, or at least
products with appropriate uncertainty information (including
auto-correlations) attached to it, rather than partly interpolated composite
products. As mentioned earlier, our assimilation does not incorporate early
atmospheric CO2 measurements from the South Pole, which might result
in a systematic bias for that latitude for some years of
∼ 1 ppm (Fig. 7l). This warning in terms of our data use is
especially important for the fine-grid interpolation we present. The
0.5∘ mean-preserving smooth interpolation should not be
misinterpreted to portray measurement information at such a fine scale.
No vertical and longitudinal resolution
The purpose of forcing climate models correctly would best be accomplished by
vertically resolved latitudinal and longitudinal fields, which (in the case
of CO2) even include a diurnal cycle. Our latitudinally and monthly
resolved dataset offers climate models an option to capture some key
variability compared to the global- and annual-mean CMIP5 concentration
recommendation (Meinshausen et al., 2011). However, a correct implementation
of this additional monthly and latitudinal variability is also dependent on
an appropriate propagation of the surface signal throughout the troposphere
and stratosphere. For example, some studies (Olsen and Randerson, 2004) find
that column CO2 is found to only exhibit roughly half of the
latitudinal gradient and seasonal variation compared to the surface
concentrations. In the CESM1 model (Hurrell et al., 2013) with prescribed
surface GHG concentrations, the vertical propagation of the CO2
concentration is assumed to be constant. In the case of the other GHGs
(CH4, N2O and CFCs) a constant concentration in the
troposphere and a decrease of the concentration in the stratosphere is
assumed in CESM1. In particular, the scale heights in the stratosphere of
these trace gases depend on latitude, which produces a more realistic
stratospheric distribution. We recommend vertical extensions to our surface
concentration reconstructions only in the case that the model has no
intrinsic transport model or extension parameterization. Furthermore, we do
not include the longitudinal variation. Again, specifically for CO2,
this longitudinal variation might be systematic given the land–ocean
contrast. For example, the MPI-ESM-LR model indicates systematically higher
surface CO2 concentration over land, which in turn would have a
radiative effect (Figs. S41 and S42).
Limited filtering of station measurements
Our assimilation procedure is a rather simple one and does not attempt to
offset potential biases due to day- and night-time sampling biases for
CO2 in the case of some flask measurements, or whether including
pollution events would bias the latitudinal averages towards higher-than-current-average values. In a world with continuing point sources, screening
out pollution effects might cause proposed averages to lag slightly behind
the true average concentration. The question is whether the correlation
between sampling locations and source locations will inherently bias the
average concentrations towards higher-than-true-average values in our
assimilation for species, where we include pollution events. For most
substances, we do not find any systematic difference between the network
averages from AGAGE or NOAA, although there are some species (e.g. HFC-152a,
see Fig. S33) for which our higher concentration reconstructions could in
part be explained by this different method.
The opposite might also be the case, i.e. that despite including some
pollution events, there could still be an inherent underestimation of true
zonal means. That is because the NOAA and AGAGE sampling stations, which we
are sourcing our raw data from, tend to be biased towards
remote, clean-air, or well-mixed conditions and this will have implications for
our latitudinal gradient and seasonal cycle. Where there are continental
sites, they are often at altitude, and when flasks are sampled, they are
generally for mid-afternoon when mixing is largest. Hence the fitted
latitudinal gradient for CO2 at least might be closer to the NOAA
marine boundary layer product than to a true zonal mean. Also, the seasonal
cycle will be more representative of marine conditions than continental ones
(where a diurnal rectifier could potentially dampen or offset seasonally low
concentrations in summer in the case of CO2). This bias towards
remote measurements tends to increase the further back in time we go.
Calibration scales
Another limitation of our study is related to the different calibration
scales of atmospheric gas measurements. In our data assimilation method with
no scale conversion between the SIO and NOAA scales of the AGAGE and NOAA
networks (Sect. 2), a time-varying difference between the scales or
time-varying coverage from one network to another can lead to spurious trends
in the derived concentrations. We argue that our “middle of the road” data
assimilation method across the two networks is, however, one justifiable
(albeit not the only viable) assimilation method. The reasons for our chosen
approach are as follows: (a) uncertainties in absolute mole fractions
estimates are small compared to other uncertainties that would affect the
radiative forcing in climate models; (b) alternative “pure” scale data
assimilation could only deal with the trend uncertainty, not with the
uncertainty arising for absolute mole fraction values (assuming that both the
SIO and NOAA scales are equally sound); (c) we intend to be
“network”-neutral; and (d) a single “in-between” concentration estimate
is likely the most appropriate for the primary application purpose
(historical simulations of climate models) of the provided data. However,
future researchers are encouraged to work directly with the principal
investigators of the two networks to devise data assimilation methods that
would be better suited for alternative applications, such as uncertainty
estimates of inverse emissions etc. A clear limitation of our data product is
hence our implicit “in between” scale, with time-varying influences from
measurements under one or the other network. Thus, differences to “pure”
SIO or NOAA scale will partly arise from this “scale” issue.
No uncertainty estimates
Another important limitation of our study is that we do not provide
uncertainty estimates. This is primarily related to the fact that the purpose
of this study was to provide a consolidated dataset for CMIP6 climate model
experiments. Those model experiments can only be performed a limited number
of times given today's computational resources. The experimental protocol
hence does not foresee an ability to vary GHG mole fractions within its
uncertainties, given that many aspects of climate models are affected by more
substantial uncertainties, such as aerosols. The original AGAGE and NOAA
(sometimes monthly averaged) sampling data points shown in the fact sheets
(see panels f, g and h) can, however, provide an indication of uncertainties
and the spread in observations.
Uncertain scaling of seasonality changes and latitudinal gradients back in time
Our choice of predictor for the CO2 seasonality change (namely the
product of CO2 concentration and global-mean temperature deviation
since pre-industrial times) is subjective, and using only CO2 concentration
or temperature would have yielded a larger seasonality difference between
current and pre-industrial times. Further research will be necessary to
obtain an optimal proxy for presumed pre-observational CO2
seasonality changes. Similarly, our common explanatory variable for
regressions of latitudinal gradients, i.e. global emissions (Boden et
al., 2013), is an approximation. Ideally, the time-changing latitudinal
distribution of emissions would be considered in those backward extensions of
the latitudinal gradient over time. More generally, further research into
observational and modelling-derived constraints regarding pre-1950
latitudinal gradients of CO2 could allow future studies to go beyond
our simplified assumption of a zero pre-industrial gradient in light of
the uncertainty.
Broad, but not comprehensive data coverage
For the recent instrumental period, our study is predominantly based on the
NOAA and the international AGAGE network data. Consistent quality control and
consistent scales are advantages of that approach. Ideally, however, our study
should have started out from a yet more inclusive representation, e.g.
including the multiple additional station datasets gathered and archived by
the WDCGG that are part of neither the
AGAGE nor NOAA networks. The WDCGG station raw data is available at
http://ds.data.jma.go.jp/gmd/wdcgg/cgi-bin/wdcgg/catalogue.cgi. While
the methodology of our study could be maintained or built upon, we hence
recommend for any future updates that those additional datasets are
considered – with the appropriate quality control and scale conversion
efforts.
Known issues
There is one known issue in the historical data series before the year 2002
for CF4, C2F6 and C3F8. We use the Trudinger et
al. (2016) datasets and our algorithm categorized them as mid-year values,
but the data were estimates for start-of-year values. Thus, while Trudinger
et al. (2016) is well aligned with the Mühle et al. (2010) over that time
period (given that the same in situ and archive data was used), our
historical time series suggest half a year's growth rate, i.e. up to maximum
0.63, 0.065 and 0.015 ppt, too-low mole fractions for CF4,
C2F6, C3F8, respectively for the pre-2002 time frame. In
terms of radiative forcing, this difference amounts to approximately 0.00022,
0.000016 and 0.0000043 Wm-2 in the years with the maximal growth
rates (1980, 1999 and 2002, respectively). Given that some CMIP6 models had
started using the historical data by the time of discovering this error
(which will have no significant effect on CMIP6 outputs), we opted for not
revising this study's CMIP6 datasets.
Conclusion
Ice core measurements over the past 800 000 years
reveal how atmospheric GHG concentrations of CO2, CH4 and
N2O varied. These variations indicate various feedback mechanisms
connected to the glacial and inter-glacial cycles driven by Milankovich
cycles. With the arrival of homo sapiens, the atmospheric composition
changed, initially through activities such as deforestation and agriculture,
and then through fossil-fuel driven industrial activities from the start of
the industrial revolution. Unprecedented over the 800 000 years of the ice
core record, CO2, CH4 and N2O concentrations suddenly
rose to record levels, with global-mean CO2 reaching a historical
mark of 400 ppm in 2015 (Fig. 6). Recently, synthetic GHGs arising
from refrigerants, solvents, foam-blowing agents and even gas-cushioned shoe
soles added to the warming effect, the radiative forcing. As the IPCC AR5
found, the most likely warming contribution from these GHGs is now higher
than the observed warming (Fig. TS.10 in IPCC AR5; IPCC, 2013). That means
that without the human activities that happen to cool the planet, namely the
aerosols we emit, observed warming would have been even greater than what has
already been experienced.
In this study, we compile a set of GHG histories over the last 2000 years –
based on numerous efforts by the scientific community to retrieve firn
samples and ice cores in the most remote places on Earth, unlock their
secrets by analysing the enclosed air and by investing in a large network of
in situ and flask measurement stations across the planet. Our understanding
of past climate change is vital to developing scenarios of the future and
designing humanity's response strategies in terms of mitigation and
adaptation. The ongoing efforts to retrieve and monitor the composition of
the planet's atmosphere efforts are sometimes threatened (Lewis, 2016).
Without those efforts, the future ahead of us would remain shrouded in even
greater uncertainty.
In this dataset, we attempted to provide a solid base for the next generation
of climate and ESMs to further our understanding of past and future climate
changes. Providing seasonal and latitudinal differences of the radiative
forcing that drives the climate change across the globe, we can hope for an
even more appropriate comparison between models and past land–ocean,
regional land and oceanic temperature observations. Ignoring these seasonal
and latitudinal differences can lead to different calculated climate impacts
of GHG emissions. Thus, accurately including this variability is a necessary
condition to accurately compare model calculations and observations and to
understand the reasons for the differences. Those agreements and
disagreements between what models and past observations tell us will then
allow us to calibrate our understanding of the Earth system, its
non-linearities and its many feedback cycles, the human influences and
natural variabilities – called “detection and attribution”.
We have been engaging in a unique experiment with our climate. In order to
stay below the warming limits, that were set forth in the Paris Agreement in
2015 (i.e. well below 2 and 1.5 ∘C relative to pre-industrial
levels), the next generation of climate models and the examination of their
response to climate drivers will be vital as an information basis for
decision makers. This study into the main past driver of human-induced
climate change will hence contribute to our collective examination of the
tremendous challenge in which we find ourselves.
The MATLAB and R code that was used to assimilate the raw
data is available from the authors on request.
A supplementary data table is available with global- and
annual-mean mole fractions. The complete dataset with latitudinally and
monthly resolved data in netcdf format is available via
https://esgf-node.llnl.gov/search/input4mips/. Additional data formats,
i.e. CSV, XLS, MATLAB .mat files of the same data, are also available via
www.climatecollege.unimelb.edu.au/cmip6. The respective raw data used
in this study is available from the original referenced data providers on
request or can be found at the web locations indicated in Table 12.
The Supplement related to this article is available online at doi:10.5194/gmd-10-2057-2017-supplement.
Malte Meinshausen designed the study. Elisabeth Vogel wrote
most of the data analysis and read-in routines together with Malte
Meinshausen. Katja Lorbacher analysed the CMIP5 ESMs and produced related
figures. Other figures and the fact sheets were produced by Malte
Meinshausen. Alexander Nauels provided a literature overview and data
intercomparison. All authors wrote, commented on and/or discussed the paper.
Nicolai Meinshausen designed the mean-preserving interpolation routines.
Multiple authors provided vital data.
The authors declare that they have no conflict of
interest.
Acknowledgements
The authors would first and foremost like to thank the large community of
scientists, research assistants and measurement technicians that were
involved in collecting the firn, ice core, and atmospheric in situ and flask
measurements across the world. Cape Grim is supported by CSIRO and the Bureau
of Meteorology, Australia. The primary networks AGAGE and the Cooperative Air
Sampling Network managed by NOAA deserve the utmost credit, including all its
individual researchers and the networks' policy to make the raw data
available to the broader scientific community. The Law Dome firn and ice core
program is supported by the Australian Antarctic Division. We thank in
particular the following researchers for invaluable efforts to collect,
screen and make available NOAA network data: Ed Dlugokencky, Pieter Tans,
David Nance, Bradley Hall, Geoff Dutton, James Elkins, Debra Mondeel,
Carolina Siso, Ben Miller. We would also like to thank the Editor
O. Morgenstern for helpful suggestions and Zebedee Nicholls for his comments.
Ray Langenfelds and Paul Steele (CSIRO) are thanked for their long-term
support of the Cape Grim, Cape Grim Air Archive and AGAGE
activities.
Furthermore, we thank the ESM CMIP5 modelling groups who contributed to the
5th Climate Model Intercomparison project and whose data we analysed. Also,
the reviewer comments from one anonymous reviewer and Piers Forster were
most helpful in improving the quality of the paper.
MM thankfully acknowledges the support by the Australian Research Council
Future Fellowship grant FT130100809. This work was undertaken in close
collaboration with partners in the European Union's Horizon 2020 research and
innovation programme CRESCENDO (grant no. 641816), of which the University of
Melbourne is an unfunded partner. CSIRO's contribution was supported by the
Australian Government, in part through the Australian Climate Change Science
Program.
Edited by: O. Morgenstern
Reviewed by: P. Forster and one anonymous referee
ReferencesAhn, J. and Brook, E. J.: Siple Dome ice reveals two modes of millennial
CO2 change during the last ice age, Nature Communications 5, 3723,
10.1038/ncomms4723, 2014.Ahn, J., Brook, E. J., Mitchell, L., Rosen, J., McConnell, J. R., Taylor, K.,
Etheridge, D., and Rubino, M.: Atmospheric CO2 over the last
1000 years: A high-resolution record from the West Antarctic Ice Sheet (WAIS)
Divide ice core, Global Biogeochem. Cy., 26, GB2027,
10.1029/2011GB004247, 2012.Anklin, M., Barnola, J., Schwander, J., Stauffer, B., and Raynaud, D.:
Processes affecting the CO2 concentrations measured in Greenland ice,
Tellus B, 47, 461–470, 1995.
Arnold, T., Harth, C. M., Mühle, J., Manning, A. J., Salameh, P. K., Kim,
J., Ivy, D. J., Steele, L. P., Petrenko, V. V., and Severinghaus, J. P.:
Nitrogen trifluoride global emissions estimated from updated atmospheric
measurements, P. Natl. Acad. Sci. USA, 110, 2029–2034, 2013.
Arnold, T., Ivy, D. J., Harth, C. M., Vollmer, M. K., Mühle, J., Salameh,
P. K., Paul Steele, L., Krummel, P. B., Wang, R. H., and Young, D.:
HFC-43-10mee atmospheric abundances and global emission estimates, Geophys.
Res. Lett., 41, 2228–2235, 2014.Aucott, M., McCulloch, A., Graedel, T., Kleiman, G., Midgley, P., and Li, Y.
F.: Anthropogenic emissions of trichloromethane (chloroform, CHCl3)
and chlorodifluoromethane (HCFC-22): Reactive chlorine emissions inventory,
J. Geophys. Res., 104, 8405–8415, 1999.Aydin, M., Montzka, S. A., Battle, M. O., Williams, M. B., De Bruyn, W. J.,
Butler, J. H., Verhulst, K. R., Tatum, C., Gun, B. K., Plotkin, D. A., Hall,
B. D., and Saltzman, E. S.: Post-coring entrapment of modern air in some
shallow ice cores collected near the firn-ice transition: evidence from
CFC-12 measurements in Antarctic firn air and ice cores, Atmos. Chem. Phys.,
10, 5135–5144, 10.5194/acp-10-5135-2010, 2010.
Bacastow, R.: Modulation of atmospheric carbon dioxide by the Southern
Oscillation, Nature, 261, 116–118, 1976.Baker, D., Law, R., Gurney, K., Rayner, P., Peylin, P., Denning, A.,
Bousquet, P., Bruhwiler, L., Chen, Y. H., and Ciais, P.: TransCom 3 inversion
intercomparison: Impact of transport model errors on the interannual
variability of regional CO2 fluxes, 1988–2003, Global Biogeochem.
Cy., 20, GB1002, 10.1029/2004GB002439, 2006.
Barbante, C., Barnola, J. M., Becagli, S., Beer, J., Bigler, M., Boutron, C.,
Blunier, T., Castellano, E., Cattani, O., Chappellaz, J., Dahl-Jensen, D.,
Debret, M., Delmonte, B., Dick, D., Falourd, S., Faria, S., Federer, U.,
Fischer, H., Freitag, J., Frenzel, A., Fritzsche, D., Fundel, F., Gabrielli,
P., Gaspari, V., Gersonde, R., Graf, W., Grigoriev, D., Hamann, I., Hansson,
M., Hoffmann, G., Hutterli, M. A., Huybrechts, P., Isaksson, E., Johnsen, S.,
Jouzel, J., Kaczmarska, M., Karlin, T., Kaufmann, P., Kipfstuhl, S., Kohno,
M., Lambert, F., Lambrecht, A., Lambrecht, A., Landais, A., Lawer, G.,
Leuenberger, M., Littot, G., Loulergue, L., Luthi, D., Maggi, V., Marino, F.,
Masson-Delmotte, V., Meyer, H., Miller, H., Mulvaney, R., Narcisi, B.,
Oerlemans, J., Oerter, H., Parrenin, F., Petit, J. R., Raisbeck, G., Raynaud,
D., Rothlisberger, R., Ruth, U., Rybak, O., Severi, M., Schmitt, J.,
Schwander, J., Siegenthaler, U., Siggaard-Andersen, M. L., Spahni, R.,
Steffensen, J. P., Stenni, B., Stocker, T. F., Tison, J. L., Traversi, R.,
Udisti, R., Valero-Delgado, F., van den Broeke, M. R., van de Wal, R. S. W.,
Wagenbach, D., Wegner, A., Weiler, K., Wilhelms, F., Winther, J. G., Wolff,
E., and Members, E. C.: One-to-one coupling of glacial climate variability in
Greenland and Antarctica, Nature, 444, 195–198, 2006.Barnola, J., Anklin, M., Porcheron, J., Raynaud, D., Schwander, J., and
Stauffer, B.: CO2 evolution during the last millennium as recorded by
Antarctic and Greenland ice, Tellus B, 47, 264–272, 1995.Bastos, A., Ciais, P., Barichivich, J., Bopp, L., Brovkin, V., Gasser, T.,
Peng, S., Pongratz, J., Viovy, N., and Trudinger, C. M.: Re-evaluating the
1940s CO2 plateau, Biogeosciences, 13, 4877–4897,
10.5194/bg-13-4877-2016, 2016.Battle, M., Bender, M., Sowerst, T., Tansi, P., Butleri, J., Elkinsi, J.,
Ellis, J., Conwayt, T., Zhangt, N., and Langt, P.: Atmospheric gas
concentrations over the past century measured in air from firn at the South
Pole, Nature, 383, 231–235, 10.1038/383231a0, 1996.
Bauska, T. K., Joos, F., Mix, A. C., Roth, R., Ahn, J., and Brook, E. J.:
Links between atmospheric carbon dioxide, the land carbon reservoir and
climate over the past millennium, Nat. Geosci., 8, 383–387, 2015.Bereiter, B., Lüthi, D., Siegrist, M., Schüpbach, S., Stocker, T. F.,
and Fischer, H.: Mode change of millennial CO2 variability during the
last glacial cycle associated with a bipolar marine carbon seesaw, P. Natl.
Acad. Sci. USA, 109, 9755–9760, 2012.Bereiter, B., Eggleston, S., Schmitt, J., Nehrbass-Ahles, C., Stocker, T. F.,
Fischer, H., Kipfstuhl, S., and Chappellaz, J.: Revision of the EPICA Dome C
CO2 record from 800 to 600 kyr before present, Geophys. Res. Lett.,
42, 542–549, 2015.Boden, T. A., Marland, G., and Andres, R. J.: Global, Regional, and National
Fossil-Fuel CO2 Emissions, Carbon Dioxide Information Analysis
Center, Oak Ridge National Laboratory, U.S. Department of Energy, Oak Ridge,
TN, USA, 2013.Buizert, C., Martinerie, P., Petrenko, V. V., Severinghaus, J. P., Trudinger,
C. M., Witrant, E., Rosen, J. L., Orsi, A. J., Rubino, M., Etheridge, D. M.,
Steele, L. P., Hogan, C., Laube, J. C., Sturges, W. T., Levchenko, V. A.,
Smith, A. M., Levin, I., Conway, T. J., Dlugokencky, E. J., Lang, P. M.,
Kawamura, K., Jenk, T. M., White, J. W. C., Sowers, T., Schwander, J., and
Blunier, T.: Gas transport in firn: multiple-tracer characterisation and
model intercomparison for NEEM, Northern Greenland, Atmos. Chem. Phys., 12,
4259–4277, 10.5194/acp-12-4259-2012, 2012.
Butler, J. H., Battle, M., Bender, M. L., Montzka, S. A., Clarke, A. D.,
Saltzman, E. S., Sucher, C. M., Severinghaus, J. P., and Elkins, J. W.:
A record of atmospheric halocarbons during the twentieth century from polar
firn air, Nature, 399, 749–755, 1999.Capron, E., Landais, A., Lemieux-Dudon, B., Schilt, A., Masson-Delmotte, V.,
Buiron, D., Chappellaz, J., Dahl-Jensen, D., Johnsen, S., and Leuenberger,
M.: Synchronising EDML and NorthGRIP ice cores using δ18O of
atmospheric oxygen (δ18O atm) and CH4 measurements over
MIS5 (80–123 kyr), Quaternary Sci. Rev., 29, 222–234, 2010.
Conway, T. J., Tans, P., Waterman, L. S., Thoning, K. W., Masarie, K. A., and
Gammon, R. H.: Atmospheric carbon dioxide measurements in the remote global
troposphere, 1981–1984, Tellus B, 40, 81–115, 1988.
Conway, T. J., Tans, P. P., Waterman, L. S., Thoning, K. W., Kitzis, D. R.,
Masarie, K. A., and Zhang, N.: Evidence for interannual variability of the
carbon cycle from the National Oceanic and Atmospheric Administration/Climate
Monitoring and Diagnostics Laboratory Global Air Sampling Network,
J. Geophys. Res., 99, 22831–22855, 1994.
Cox, M., Sturrock, G., Fraser, P., Siems, S., Krummel, P., and O'doherty, S.:
Regional sources of methyl chloride, chloroform and dichloromethane
identified from AGAGE observations at Cape Grim, Tasmania, 1998–2000,
J. Atmos. Chem., 45, 79–99, 2003.Cunnold, D., Weiss, R., Prinn, R., Hartley, D., Simmonds, P., Fraser, P.,
Miller, B., Alyea, F., and Porter, L.: GAGE/AGAGE measurements indicating
reductions in global emissions of CCl3F and CCl2F2 in
1992–1994, J. Geophys. Res., 102, 1259–1269, 1997.Cunnold, D., Steele, L., Fraser, P., Simmonds, P., Prinn, R., Weiss, R.,
Porter, L., O'Doherty, S., Langenfelds, R., and Krummel, P.: In situ
measurements of atmospheric methane at GAGE/AGAGE sites during 1985–2000 and
resulting source inferences, J. Geophys. Res., 107, ACH 20-1–ACH 20-18,
10.1029/2001JD001226, 2002.
Dahl-Jensen, D., Albert, M., Aldahan, A., Azuma, N., Balslev-Clausen, D.,
Baumgartner, M., Berggren, A.-M., Bigler, M., Binder, T., and Blunier, T.:
Eemian interglacial reconstructed from a Greenland folded ice core, Nature,
493, 489–494, 2013.Denning, A. S., Fung, I. Y., and Randall, D.: Latitudinal gradient of
atmospheric CO2 due to seasonal exchange with land biota, Nature,
376, 240–243, 1995.
Denning, A. S., Takahashi, T., and Friedlingstein, P.: Can a strong
atmospheric CO2 rectifier effect be reconciled with a “reasonable” carbon
budget?, Tellus B, 51, 249–253, 1999.
Dlugokencky, E. J., Masaire, K. A., Lang, P. M., Tans, P. P., Steele, L. P.,
and Nisbet, E. G.: A dramatic decrease in the growth rate of atmospheric
methane in the northern hemisphere during 1992, Geophys. Res. Lett., 21,
45–48, 1994a.
Dlugokencky, E. J., Steele, L. P., Lang, P. M., and Masarie, K. A.: The
Growth-Rate and Distribution of Atmospheric Methane, J. Geophys. Res., 99,
17021–17043, 1994b.
Dlugokencky, E. J., Steele, L. P., Lang, P. M., and Masarie, K. A.: The
growth rate and distribution of atmospheric methane, J. Geophys. Res., 99,
17021–17043, 1994c.
Dlugokencky, E. J., Masarie, K. A., Lang, P. M., and Tans, P. P.: Continuing
decline in the growth rate of the atmospheric methane burden, Nature, 393,
447–450, 1998.
Dlugokencky, E. J., Walter, B. P., Masarie, K. A., Lang, P. M., and
Kasischke, E. S.: Measurements of an anomalous global methane increase during
1998, Geophys. Res. Lett., 28, 499–502, 2001.Dlugokencky, E. J., Myers, R. C., Lang, P. M., Masarie, K. A., Crotwell,
A. M., Thoning, K. W., Hall, B. D., Elkins, J. W., and Steele, L. P.:
Conversion of NOAA atmospheric dry air CH4 mole fractions to a
gravimetrically prepared standard scale, J. Geophys. Res., 110, D18306,
10.1029/2005JD006035, 2005.Dlugokencky, E. J., Bruhwiler, L., White, J. W. C., Emmons, L. K., Novelli,
P. C., Montzka, S. A., Masarie, K. A., Lang, P. M., Crotwell, A. M., Miller,
J. B., and Gatti, L. V.: Observational constraints on recent increases in the
atmospheric CH4 burden, Geophys. Res. Lett., 36, L18803,
10.1029/2009GL039780, 2009.
Dlugokencky, E. J., Lang, P. M., Crotwell, A. M., Masarie, K. A., and
Crotwell, M. J.: Atmospheric Methane Dry Air Mole Fractions from the NOAA
ESRL Carbon Cycle Cooperative Global Air Sampling Network, 1983–2014, NOAA
ESRL Global Monitoring Division, Boulder, CO, USA, 2015a.
Dlugokencky, E. J., Lang, P. M., Masarie, K. A., Crotwell, A. M., and
Crotwell, M. J.: Atmospheric Carbon Dioxide Dry Air Mole Fractions from the
NOAA ESRL Carbon Cycle Cooperative Global Air Sampling Network, 1968–2014,
NOAA ESRL Global Monitoring Division, Boulder, CO, USA, 2015b.
Elkins, J., Thompson, T., Swanson, T., Butler, J., Hall, B., Cummings, S.,
Fisher, D., and Raffo, A.: Decrease in the growth rates of atmospheric
chlorofluorocarbons 11 and 12, Nature, 364, 780–783, 1993.Enting, I. G.: Green's Function Methods of Tracer Inversion, in: Inverse
Methods in Global Biogeochemical Cycles, edited by: Kasibhatla, P., Heimann,
M., Rayner, P., Mahowald, N.,Prinn, R. G., and Hartley, D. E., American
Geophysical Union, doi:10.1029/GM114p0019, 2013.Enting, I. G. and Mansbridge, J.: Seasonal sources and sinks of atmospheric
CO2 direct inversion of filtered data, Tellus B, 41, 111–126, 1989.Enting, I. G. and Mansbridge, J.: Latitudinal distribution of sources and
sinks of CO2: Results of an inversion study, Tellus B, 43, 156–170,
1991.Enting, I. G., Trudinger, C., and Francey, R.: A synthesis inversion of the
concentration and δ13C of atmospheric CO2, Tellus B, 47,
35–52, 1995.Etheridge, D. M., Steele, L. P., Langenfelds, R. L., Francey, R. J., Barnola,
J. M., and Morgan, V. I.: Natural and anthropogenic changes in atmospheric
CO2 over the last 1000 years from air in Antarctic ice and firn,
J. Geophys. Res., 101, 4115–4128, 1996.
Etheridge, D. M., Steele, L. P., Francey, R. J., and Langenfelds, R. L.:
Atmospheric methane between 1000 AD and present: Evidence of anthropogenic
emissions and climatic variability, J. Geophys. Res., 103, 15979–15993,
1998.Etheridge, D. M., Steele, L. P., Langenfelds, R. L., Francey, R. J., Barnola,
J.-M., and Morgan, V. I.: Natural and anthropogenic changes in atmospheric
CO2 over the last 1000 years from air in Antarctic ice and firn,
J. Geophys. Res., 101, 4115–4128, 1996.Eyring, V., Bony, S., Meehl, G. A., Senior, C. A., Stevens, B., Stouffer,
R. J., and Taylor, K. E.: Overview of the Coupled Model Intercomparison
Project Phase 6 (CMIP6) experimental design and organization, Geosci. Model
Dev., 9, 1937–1958, 10.5194/gmd-9-1937-2016, 2016.Fluckiger, J., Monnin, E., Stauffer, B., Schwander, J., Stocker, T. F.,
Chappellaz, J., Raynaud, D., and Barnola, J. M.: High-resolution Holocene
N2O ice core record and its relationship with CH4 and
CO2, Global Biogeochem. Cy., 16, 10-1–10-8, 2002.Forkel, M., Carvalhais, N., Rödenbeck, C., Keeling, R., Heimann, M.,
Thonicke, K., Zaehle, S., and Reichstein, M.: Enhanced seasonal CO2
exchange caused by amplified plant productivity in northern ecosystems,
Science, 351, 696–699, 10.1126/science.aac4971, 2016.Francey, R. J. and Frederiksen, J. S.: The 2009–2010 step in atmospheric
CO2 interhemispheric difference, Biogeosciences, 13, 873–885,
10.5194/bg-13-873-2016, 2016.Francey, R. J., Trudinger, C. M., van der Schoot, M., Law, R. M., Krummel,
P. B., Langenfelds, R. L., Steele, P. L., Allison, C. E., Stavert, A. R.,
Andres, R. J., and Rodenbeck, C.: Atmospheric verification of anthropogenic
CO2 emission trends, Nature Climate Change, 3, 520–524, 2013.
Fraser, P., Cunnold, D., Alyea, F., Weiss, R., Prinn, R., Simmonds, P.,
Miller, B., and Langenfelds, R.: Lifetime and emission estimates of 1, 1,
2-trichlorotrifluorethane (CFC-113) from daily global background observations
June 1982–June 1994, J. Geophys. Res., 101, 12585–12599, 1996.
Fung, I., John, J., Lerner, J., Matthews, E., Prather, M., Steele, L., and
Fraser, P.: Three-dimensional model synthesis of the global methane cycle,
J. Geophys. Res., 96, 13033–13065, 1991.Ghosh, A., Patra, P. K., Ishijima, K., Umezawa, T., Ito, A., Etheridge,
D. M., Sugawara, S., Kawamura, K., Miller, J. B., Dlugokencky, E. J.,
Krummel, P. B., Fraser, P. J., Steele, L. P., Langenfelds, R. L., Trudinger,
C. M., White, J. W. C., Vaughn, B., Saeki, T., Aoki, S., and Nakazawa, T.:
Variations in global methane sources and sinks during 1910–2010, Atmos.
Chem. Phys., 15, 2595–2612, 10.5194/acp-15-2595-2015, 2015.Graven, H. D., Keeling, R. F., Piper, S. C., Patra, P. K., Stephens, B. B.,
Wofsy, S. C., Welp, L. R., Sweeney, C., Tans, P. P., Kelley, J. J., Daube,
B. C., Kort, E. A., Santoni, G. W., and Bent, J. D.: Enhanced Seasonal
Exchange of CO2 by Northern Ecosystems Since 1960, Science, 341,
1085–1089, 2013.Gray, J. M., Frolking, S., Kort, E. A., Ray, D. K., Kucharik, C. J.,
Ramankutty, N., and Friedl, M. A.: Direct human influence on atmospheric
CO2 seasonality from increased cropland productivity, Nature, 515,
398–401, 2014.Gurney, K. R., Law, R. M., Denning, A. S., Rayner, P. J., Baker, D.,
Bousquet, P., Bruhwiler, L., Chen, Y.-H., Ciais, P., and Fan, S.: Towards
robust regional estimates of CO2 sources and sinks using atmospheric
transport models, Nature, 415, 626–630, 2002.Gurney, K. R., Law, R. M., Denning, A. S., Rayner, P. J., Baker, D.,
Bousquet, P., Bruhwiler, L., CHEN, Y. H., Ciais, P., and Fan, S.: TransCom 3
CO2 inversion intercomparison: 1. Annual mean control results and
sensitivity to transport and prior flux information, Tellus B, 55, 555–579,
2003.Gurney, K. R., Law, R. M., Denning, A. S., Rayner, P. J., Pak, B. C., Baker,
D., Bousquet, P., Bruhwiler, L., Chen, Y. H., and Ciais, P.: Transcom 3
inversion intercomparison: Model mean results for the estimation of seasonal
carbon sources and sinks, Global Biogeochem. Cy., 18, GB1010,
10.1029/2003GB002111, 2004.Gütschow, J., Jeffery, M. L., Gieseke, R., Gebel, R., Stevens, D., Krapp,
M., and Rocha, M.: The PRIMAP-hist national historical emissions time series,
Earth Syst. Sci. Data, 8, 571–603, 10.5194/essd-8-571-2016, 2016.Hall, B. D., Engel, A., Mühle, J., Elkins, J. W., Artuso, F., Atlas, E.,
Aydin, M., Blake, D., Brunke, E.-G., Chiavarini, S., Fraser, P. J., Happell,
J., Krummel, P. B., Levin, I., Loewenstein, M., Maione, M., Montzka, S. A.,
O'Doherty, S., Reimann, S., Rhoderick, G., Saltzman, E. S., Scheel, H. E.,
Steele, L. P., Vollmer, M. K., Weiss, R. F., Worthy, D., and Yokouchi, Y.:
Results from the International Halocarbons in Air Comparison Experiment
(IHALACE), Atmos. Meas. Tech., 7, 469–490, 10.5194/amt-7-469-2014,
2014.
Hurrell, J. W., Holland, M. M., Gent, P. R., Ghan, S., Kay, J. E., Kushner,
P. J., Lamarque, J.-F., Large, W. G., Lawrence, D., Lindsay, K., Lipscomb,
W. H., Long, M. C., Mahowald, N., Marsh, D. R., Neale, R. B., Rasch, P.,
Vavrus, S., Vertenstein, M., Bader, D., Collins, W. D., Hack, J. J., Kiehl,
J., and Marshall, S.: The Community Earth System Model: A Framework for
Collaborative Research, B. Am. Meteorol. Soc., 94, 1339–1360, 2013.
Hurtt, G., Chini, L. P., Frolking, S., Betts, R., Feddema, J., Fischer, G.,
Fisk, J., Hibbard, K., Houghton, R., and Janetos, A.: Harmonization of
land-use scenarios for the period 1500–2100: 600 years of global gridded
annual land-use transitions, wood harvest, and resulting secondary lands,
Climatic Change, 109, 117–161, 2011.
IPCC: Climate Change 2013: The Physical Science Basis. Contribution of
Working Group I to the Fifth Assessment Report of the Intergovernmental Panel
on Climate Change, edited by: Stocker, T. F., Qin, D., Plattner, G.-K.,
Tignor, M., Allen, S. K., Boschung, J., Nauels, A., Xia, Y., Bex, V., and
Midgley, P. M., Cambridge University Press, Cambridge, United Kingdom and New
York, NY, USA, 1535 pp., 2013.Ishijima, K., Sugawara, S., Kawamura, K., Hashida, G., Morimoto, S.,
Murayama, S., Aoki, S., and Nakazawa, T.: Temporal variations of the
atmospheric nitrous oxide concentration and its δ15N and δ18O for the latter half of the 20th century reconstructed from firn air
analyses, J. Geophys. Res., 112, D03305, 10.1029/2006JD007208, 2007.Ivy, D. J., Rigby, M., Baasandorj, M., Burkholder, J. B., and Prinn, R. G.:
Global emission estimates and radiative impact of C4F10,
C5F12, C6F14, C7F16 and C8F18,
Atmos. Chem. Phys., 12, 7635–7645, 10.5194/acp-12-7635-2012, 2012.Jones, C. D. and Cox, P. M.: Modeling the volcanic signal in the atmospheric
CO2 record, Global Biogeochem. Cy., 15, 453–465, 2001.Jungclaus, J. H., Bard, E., Baroni, M., Braconnot, P., Cao, J., Chini, L. P.,
Egorova, T., Evans, M., González-Rouco, J. F., Goosse, H., Hurtt, G. C.,
Joos, F., Kaplan, J. O., Khodri, M., Klein Goldewijk, K., Krivova, N.,
LeGrande, A. N., Lorenz, S. J., Luterbacher, J., Man, W., Meinshausen, M.,
Moberg, A., Nehrbass-Ahles, C., Otto-Bliesner, B. I., Phipps, S. J.,
Pongratz, J., Rozanov, E., Schmidt, G. A., Schmidt, H., Schmutz, W., Schurer,
A., Shapiro, A. I., Sigl, M., Smerdon, J. E., Solanki, S. K., Timmreck, C.,
Toohey, M., Usoskin, I. G., Wagner, S., Wu, C.-Y., Yeo, K. L., Zanchettin,
D., Zhang, Q., and Zorita, E.: The PMIP4 contribution to CMIP6 – Part 3: the
Last Millennium, Scientific Objective and Experimental Design for the PMIP4
past1000 simulations, Geosci. Model Dev. Discuss.,
10.5194/gmd-2016-278, in review, 2016.Kageyama, M., Braconnot, P., Harrison, S. P., Haywood, A. M., Jungclaus, J.,
Otto-Bliesner, B. L., Peterschmitt, J.-Y., Abe-Ouchi, A., Albani, S.,
Bartlein, P. J., Brierley, C., Crucifix, M., Dolan, A., Fernandez-Donado, L.,
Fischer, H., Hopcroft, P. O., Ivanovic, R. F., Lambert, F., Lunt, D. J.,
Mahowald, N. M., Peltier, W. R., Phipps, S. J., Roche, D. M., Schmidt, G. A.,
Tarasov, L., Valdes, P. J., Zhang, Q., and Zhou, T.: PMIP4–CMIP6: the
contribution of the Paleoclimate Modelling Intercomparison Project to CMIP6,
Geosci. Model Dev. Discuss., 10.5194/gmd-2016-106, in review, 2016.
Keeling, C. D., Bacastow, R. B., Bainbridge, A. E., Ekdahl, C. A., Guenther,
P. R., Waterman, L. S., and Chin, J. F.: Atmospheric carbon dioxide
variations at Mauna Loa observatory, Hawaii, Tellus, 28, 538–551, 1976.Keeling, C. D., Bacastow, R. B., Carter, A. F., Piper, S. C., Whorf, T. P.,
Heimann, M., Mook, W. G., and Roeloffzen, H.: A three-dimensional model of
atmospheric CO2 transport based on observed winds: 1. Analysis of
observational data, in Aspects of Climate Variability in the Pacific and the
Western Americas, edited by: Peterson, D. H., American Geophysical Union,
Washington, D.C., 165–236, 10.1029/GM055p0165, 1989a.Keeling, C. D., Piper, S. C., and Heimann, M.: A three-dimensional model of
atmospheric CO2 transport based on observed winds: 4. Mean annual
gradients and interannual variations, in: Aspects of Climate Variability in
the Pacific and the Western Americas, edited by: Peterson, D. H., American
Geophysical Union, Washington, D.C., 10.1029/GM055p0305, 305–363,
1989b.Keeling, C. D., Piper, S. C., Bacastow, R. B., Wahlen, M., Whorf, T. P.,
Heimann, M., and Meijer, H. A.: Exchanges of atmospheric CO2 and
13CO2 with the terrestrial biosphere and oceans from 1978 to
2000. I. Global aspects, Scripps Institution of Oceanography, 28 pp., 2001.Keeling, C. D., Piper, S. C., Whorf, T. P., and Keeling, R. F.: Evolution of
natural and anthropogenic fluxes of atmospheric CO2 from 1957 to
2003, Tellus B, 63, 1–22, 2011.
Kirschke, S., Bousquet, P., Ciais, P., Saunois, M., Canadell, J. G.,
Dlugokencky, E. J., Bergamaschi, P., Bergmann, D., Blake, D. R., Bruhwiler,
L., Cameron-Smith, P., Castaldi, S., Chevallier, F., Feng, L., Fraser, A.,
Heimann, M., Hodson, E. L., Houweling, S., Josse, B., Fraser, P. J., Krummel,
P. B., Lamarque, J.-F., Langenfelds, R. L., Le Quere, C., Naik, V.,
O'Doherty, S., Palmer, P. I., Pison, I., Plummer, D., Poulter, B., Prinn,
R. G., Rigby, M., Ringeval, B., Santini, M., Schmidt, M., Shindell, D. T.,
Simpson, I. J., Spahni, R., Steele, L. P., Strode, S. A., Sudo, K., Szopa,
S., van der Werf, G. R., Voulgarakis, A., van Weele, M., Weiss, R. F.,
Williams, J. E., and Zeng, G.: Three decades of global methane sources and
sinks, Nat. Geosci., 6, 813–823, 2013.Kohlmaier, G. H., Siré, E. O., Janecek, A., Keeling, C. D., Piper, S. C.,
and Revelle, R.: Modelling the seasonal contribution of a CO2
fertilization effect of the terrestrial vegetation to the amplitude increase
in atmospheric CO2 at Mauna Loa Observatory, Tellus B, 41, 487–510,
1989.Komhyr, W. D., Waterman, L. S., and Taylor, W. R.: Semiautomatic
nondispersive infrared analyzer apparatus for CO2 air sample
analyses, J. Geophys. Res., 88, 1315–1322, 1983.Komhyr, W. D., Gammon, R. H., Harris, T. B., Waterman, L. S., Conway, T. J.,
Taylor, W. R., and Thoning, K. W.: Global atmospheric CO2
distribution and variations from 1968–1982 NOAA/GMCC CO2 flask
sample data, J. Geophys. Res., 90, 5567–5596, 1985.
Koppmann, R., Johnen, F. J., Plass-Dülmer, C., and Rudolph, J.:
Distribution of methylchloride, dichloromethane, trichloroethene and
tetrachloroethene over the North and South Atlantic, J. Geophys. Res., 98,
20517–20526, 1993.
Lang, P. M., Steele, L. P., and Martin, R. C.: Atmospheric methane data for
the period 1986–1988 from the NOAA/CMDL global cooperative flask sampling
network, NOAA Technical Memorandum ERL CMDL-2, 1990a.
Lang, P. M., Steele, L. P., Martin, R. C., and Masarie, K. A.: Atmospheric
methane data for the period 1983–1985 from the NOAA/GMCC global cooperative
flask sampling network, NOAA Technical Memorandum ERL CMDL-1, 1990b.
Lang, P. M., Steele, L. P., Waterman, L. S., Martin, R. C., Masarie, K. A.,
and Dlugokencky, E. J.: NOAA/CMDL Atmospheric methane data for the period
1983–1990 from shipboard flask sampling, NOAA Technical Memorandum ERL
CMDL-4, 1992.Lewis, D.: Australian climate job cuts leave hole in Southern Hemisphere
research, Nature, 10.1038/nature.2016.19944, 2016.
Liang, Q., Newman, P., and Reimann, S.: SPARC Report on the Mystery of Carbon
Tetrachloride, SPARC Report No. 7, WCRP-13/2016, 2016.Loulergue, L., Schilt, A., Spahni, R., Masson-Delmotte, V., Blunier, T.,
Lemieux, B., Barnola, J.-M., Raynaud, D., Stocker, T. F., and Chappellaz, J.:
Orbital and millennial-scale features of atmospheric CH4 over the
past 800,000 years, Nature, 453, 383–386, 2008.
Lüthi, D., Le Floch, M., Bereiter, B., Blunier, T., Barnola, J.-M.,
Siegenthaler, U., Raynaud, D., Jouzel, J., Fischer, H., and Kawamura, K.:
High-resolution carbon dioxide concentration record 650,000–800,000 years
before present, Nature, 453, 379–382, 2008.MacFarling Meure, C., Etheridge, D., Trudinger, C., Steele, P., Langenfelds,
R., Van Ommen, T., Smith, A., and Elkins, J.: Law Dome CO2,
CH4 and N2O ice core records extended to 2000 years BP,
Geophys. Res. Lett., 33, L14810, 10.1029/2006GL026152, 2006.
Machida, T., Nakazawa, T., Fujii, Y., Aoki, S., and Watanabe, O.: Increase in
the atmospheric nitrous oxide concentration during the last 250 years,
Geophys. Res. Lett., 22, 2921–2924, 1995.
Marcott, S. A., Bauska, T. K., Buizert, C., Steig, E. J., Rosen, J. L.,
Cuffey, K. M., Fudge, T., Severinghaus, J. P., Ahn, J., and Kalk, M. L.:
Centennial-scale changes in the global carbon cycle during the last
deglaciation, Nature, 514, 616–619, 2014.Martinerie, P., Nourtier-Mazauric, E., Barnola, J.-M., Sturges, W. T.,
Worton, D. R., Atlas, E., Gohar, L. K., Shine, K. P., and Brasseur, G. P.:
Long-lived halocarbon trends and budgets from atmospheric chemistry modelling
constrained with measurements in polar firn, Atmos. Chem. Phys., 9,
3911–3934, 10.5194/acp-9-3911-2009, 2009.
Masarie, K. A. and Tans, P. P.: Extension and integration of atmospheric
carbon dioxide data into a globally consistent measurement record,
J. Geophys. Res., 100, 11593–11610, 1995.Matthes, K., Funke, B., Anderson, M. E., Barnard, L., Beer, J., Charbonneau,
P., Clilverd, M. A., Dudok de Wit, T., Haberreiter, M., Hendry, A., Jackman,
C. H., Kretschmar, M., Kruschke, T., Kunze, M., Langematz, U., Marsh, D. R.,
Maycock, A., Misios, S., Rodger, C. J., Scaife, A. A., Seppälä, A.,
Shangguan, M., Sinnhuber, M., Tourpali, K., Usoskin, I., van de Kamp, M.,
Verronen, P. T., and Versick, S.: Solar Forcing for CMIP6 (v3.1), Geosci.
Model Dev. Discuss., 10.5194/gmd-2016-91, in review, 2016.Meehl, G. A., Covey, C., McAvaney, B., Latif, M., and Stouffer, R. J.:
Overview of coupled model intercomparison project, B. Am. Meteorol. Soc., 86,
89–93, 10.1175/BAMS-86-1-89, 2005.
Meinshausen, M., Smith, S., Calvin, K., Daniel, J., Kainuma, M., Lamarque,
J. F., Matsumoto, K., Montzka, S., Raper, S., Riahi, K., Thomson, A.,
Velders, G., and van Vuuren, D. P.: The RCP greenhouse gas concentrations and
their extensions from 1765 to 2300, Climatic Change, 109, 213–241, 2011.
Miller, B., Huang, J., Weiss, R., Prinn, R., and Fraser, P.: Atmospheric
trend and lifetime of chlorodifluoromethane (HCFC-22) and the global
tropospheric OH concentration, J. Geophys. Res., 103, 13237–13248, 1998.
Mitchell, L., Brook, E., Lee, J. E., Buizert, C., and Sowers, T.: Constraints
on the Late Holocene anthropogenic contribution to the atmospheric methane
budget, Science, 342, 964–966, 2013.Monnin, E., Steig, E. J., Siegenthaler, U., Kawamura, K., Schwander, J.,
Stauffer, B., Stocker, T. F., Morse, D. L., Barnola, J. M., Bellier, B.,
Raynaud, D., and Fischer, H.: Evidence for substantial accumulation rate
variability in Antarctica during the Holocene, through synchronization of
CO2 in the Taylor Dome, Dome C and DML ice cores, Earth Planet. Sc.
Lett., 224, 45–54, 2004.
Montzka, S. A., Myers, R. C., Butler, J. H., Elkins, J. W., and Cummings,
S. O.: Global tropospheric distribution and calibration scale of HCFC-22,
Geophys. Res. Lett., 20, 703–706, 1993.
Montzka, S. A., Myers, R. C., Butler, J. H., Elkins, J. W., Lock, L. T.,
Clarke, A. D., and Goldstein, A. H.: Observations of HFC-134a in the remote
troposphere, Geophys. Res. Lett., 23, 169–172, 1996.
Montzka, S. A., Butler, J. H., Elkins, J. W., Thompson, T. M., Clarke, A. D.,
and Lock, L. T.: Present and future trends in the atmospheric burden of
ozone-depleting halogens, Nature, 398, 690–694, 1999.
Montzka, S. A., Spivakovsky, C. M., Butler, J. H., Elkins, J. W., Lock,
L. T., and Mondeel, D. J.: New observational constraints for atmospheric
hydroxyl on global and hemispheric scales, Science, 288, 500–503, 2000.Montzka, S., Butler, J., Hall, B., Mondeel, D., and Elkins, J.: A decline in
tropospheric organic bromine, Geophys. Res. Lett., 30, 1826,
10.1029/2003GL017745, 2003.Montzka, S., Kuijpers, L., Battle, M., Aydin, M., Verhulst, K., Saltzman, E.,
and Fahey, D.: Recent increases in global HFC-23 emissions, Geophys. Res.
Lett., 37, L02808, 10.1029/2009GL041195, 2010.Montzka, S. A., Dlugokencky, E. J., and Butler, J. H.: Non-CO2
greenhouse gases and climate change, Nature, 476, 43–50, 2011.
Montzka, S., McFarland, M., Andersen, S., Miller, B., Fahey, D., Hall, B.,
Hu, L., Siso, C., and Elkins, J.: Recent trends in global emissions of
hydrochlorofluorocarbons and hydrofluorocarbons: Reflecting on the 2007
adjustments to the Montreal Protocol, J. Phys. Chem. A, 119, 4439–4449,
2015.Morice, C. P., Kennedy, J. J., Rayner, N. A., and Jones, P. D.: Quantifying
uncertainties in global and regional temperature change using an ensemble of
observational estimates: The HadCRUT4 data set, J. Geophys. Res., 117,
D08101, 10.1029/2011JD017187, 2012.Mühle, J., Huang, J., Weiss, R., Prinn, R., Miller, B., Salameh, P.,
Harth, C., Fraser, P., Porter, L., and Greally, B.: Sulfuryl fluoride in the
global atmosphere, J. Geophys. Res., 114, D05306, 10.1029/2008JD011162,
2009.Mühle, J., Ganesan, A. L., Miller, B. R., Salameh, P. K., Harth, C. M.,
Greally, B. R., Rigby, M., Porter, L. W., Steele, L. P., Trudinger, C. M.,
Krummel, P. B., O'Doherty, S., Fraser, P. J., Simmonds, P. G., Prinn, R. G.,
and Weiss, R. F.: Perfluorocarbons in the global atmosphere:
tetrafluoromethane, hexafluoroethane, and octafluoropropane, Atmos. Chem.
Phys., 10, 5145–5164, 10.5194/acp-10-5145-2010, 2010.Newland, M. J., Reeves, C. E., Oram, D. E., Laube, J. C., Sturges, W. T.,
Hogan, C., Begley, P., and Fraser, P. J.: Southern hemispheric halon trends
and global halon emissions, 1978–2011, Atmos. Chem. Phys., 13, 5551–5565,
10.5194/acp-13-5551-2013, 2013.
Nisbet, E. G., Dlugokencky, E. J., and Bousquet, P.: Methane on the
rise–again, Science, 343, 493–495, 2014.
Nisbet, E. G., Dlugokencky, E., Manning, M., Lowry, D., Fisher, R., France,
J., Michel, S., Miller, J., White, J., and Vaughn, B.: Rising atmospheric
methane: 2007–2014 growth and isotopic shift, Global Biogeochem. Cy., 30,
1356–1370, 2016.
NOAA: Cooperative Global Atmospheric Data Integration Project:
Multi-laboratory compilation of synchronized and gap-filled atmospheric
carbon dioxide records for the period 1979–2012
(obspack_co2_1_GLOBALVIEW-CO2_2013_v1.0.4_2013-12-23),
edited by: NOAA Global Monitoring Division: Boulder, CO, USA, 2013.
NOAA ESRL GMD: Atmospheric Carbon Dioxide Dry Air Mole Fractions from
quasi-continuous measurements at American Samoa, compiled by: Thoning, K. W.,
Kitzis, D. R., and Crotwell, A., National Oceanic and Atmospheric
Administration (NOAA), Earth System Research Laboratory (ESRL), Global
Monitoring Division (GMD): Boulder, Colorado, USA, 2014a.
NOAA ESRL GMD: Atmospheric Carbon Dioxide Dry Air Mole Fractions from
quasi-continuous measurements at Barrow, Alaska, compiled by: Thoning, K. W.,
Kitzis, D. R., and Crotwell, A., National Oceanic and Atmospheric
Administration (NOAA), Earth System Research Laboratory (ESRL), Global
Monitoring Division (GMD): Boulder, Colorado, USA, 2014b.
NOAA ESRL GMD: Atmospheric Carbon Dioxide Dry Air Mole Fractions from
quasi-continuous measurements at Mauna Loa, Hawaii, compiled by: Thoning,
K. W., Kitzis, D. R., and Crotwell, A., National Oceanic and Atmospheric
Administration (NOAA), Earth System Research Laboratory (ESRL), Global
Monitoring Division (GMD): Boulder, Colorado, USA, 2014c.
NOAA ESRL GMD: Atmospheric Carbon Dioxide Dry Air Mole Fractions from
quasi-continuous measurements at South Pole, compiled by: Thoning, K. W.,
Kitzis, D. R., and Crotwell, A., National Oceanic and Atmospheric
Administration (NOAA), Earth System Research Laboratory (ESRL), Global
Monitoring Division (GMD): Boulder, Colorado, USA, 2014d.O'Doherty, S., Simmonds, P., Cunnold, D., Wang, H., Sturrock, G., Fraser, P.,
Ryall, D., Derwent, R., Weiss, R., and Salameh, P.: In situ chloroform
measurements at Advanced Global Atmospheric Gases Experiment atmospheric
research stations from 1994 to 1998, J. Geophys. Res., 106, 20429–20444,
10.1029/2000JD900792, 2001.O'Doherty, S., Cunnold, D., Manning, A., Miller, B., Wang, R., Krummel, P.,
Fraser, P., Simmonds, P., McCulloch, A., and Weiss, R.: Rapid growth of
hydrofluorocarbon 134a and hydrochlorofluorocarbons 141b, 142b, and 22 from
Advanced Global Atmospheric Gases Experiment (AGAGE) observations at Cape
Grim, Tasmania, and Mace Head, Ireland, J. Geophys. Res., 109, D06310,
10.1029/2003JD004277, 2004.Olsen, S. C. and Randerson, J. T.: Differences between surface and column
atmospheric CO2 and implications for carbon cycle research,
J. Geophys. Res., 109, D02301, 10.1029/2003JD003968, 2004.O'Neill, B. C., Tebaldi, C., van Vuuren, D. P., Eyring, V., Friedlingstein,
P., Hurtt, G., Knutti, R., Kriegler, E., Lamarque, J.-F., Lowe, J., Meehl,
G. A., Moss, R., Riahi, K., and Sanderson, B. M.: The Scenario Model
Intercomparison Project (ScenarioMIP) for CMIP6, Geosci. Model Dev., 9,
3461–3482, 10.5194/gmd-9-3461-2016, 2016.Oram, D. E., Mani, F. S., Laube, J. C., Newland, M. J., Reeves, C. E.,
Sturges, W. T., Penkett, S. A., Brenninkmeijer, C. A. M., Röckmann, T.,
and Fraser, P. J.: Long-term tropospheric trend of octafluorocyclobutane
(c-C4F8 or PFC-318), Atmos. Chem. Phys., 12, 261–269,
10.5194/acp-12-261-2012, 2012.
Park, S., Croteau, P., Boering, K., Etheridge, D., Ferretti, D., Fraser, P.,
Kim, K.-R., Krummel, P., Langenfelds, R., and Van Ommen, T.: Trends and
seasonal cycles in the isotopic composition of nitrous oxide since 1940, Nat.
Geosci., 5, 261–265, 2012.
Petit, J.-R., Jouzel, J., Raynaud, D., Barkov, N. I., Barnola, J.-M., Basile,
I., Bender, M., Chappellaz, J., Davis, M., and Delaygue, G.: Climate and
atmospheric history of the past 420,000 years from the Vostok ice core,
Antarctica, Nature, 399, 429–436, 1999.Peylin, P., Bousquet, P., Le Quéré, C., Sitch, S., Friedlingstein,
P., McKinley, G., Gruber, N., Rayner, P., and Ciais, P.: Multiple constraints
on regional CO2 flux variations over land and oceans, Global
Biogeochem. Cy., 19, GB1011, 10.1029/2003GB002214, 2005.Peylin, P., Law, R. M., Gurney, K. R., Chevallier, F., Jacobson, A. R., Maki,
T., Niwa, Y., Patra, P. K., Peters, W., Rayner, P. J., Rödenbeck, C.,
van der Laan-Luijkx, I. T., and Zhang, X.: Global atmospheric carbon budget:
results from an ensemble of atmospheric CO2 inversions,
Biogeosciences, 10, 6699–6720, 10.5194/bg-10-6699-2013, 2013.
Prinn, R., Cunnold, D., Rasmussen, R., Simmonds, P., Alyea, F., Crawford, A.,
Fraser, P., and Rosen, R.: Atmospheric emissions and trends of nitrous oxide
deduced from 10 years of ALE–GAGE data, J. Geophys. Res., 95, 18369–18385,
1990.
Prinn, R., Weiss, R., Fraser, P., Simmonds, P., Cunnold, D., Alyea, F.,
O'Doherty, S., Salameh, P., Miller, B., and Huang, J.: A history of
chemically and radiatively important gases in air deduced from
ALE/GAGE/AGAGE, J. Geophys. Res., 105, 17751–17792, 2000a.
Prinn, R. G., Weiss, R. F., Fraser, P. J., Simmonds, P. G., Cunnold, D. M.,
Alyea, F. N., O'Doherty, S., Salameh, P., Miller, B. R., Huang, J., Wang,
R. H. J., Hartley, D. E., Harth, C., Steele, L. P., Sturrock, G., Midgley,
P. M., and McCulloch, A.: A history of chemically and radiatively important
gases in air deduced from ALE/GAGE/AGAGE, J. Geophys. Res., 105,
17751–17792, 2000b.
Prinn, R., Huang, J., Weiss, R., Cunnold, D., Fraser, P., Simmonds, P.,
McCulloch, A., Harth, C., Salameh, P., and O'Doherty, S.: Evidence for
substantial variations of atmospheric hydroxyl radicals in the past two
decades, Science, 292, 1882–1888, 2001.Prinn, R., Huang, J., Weiss, R., Cunnold, D., Fraser, P., Simmonds, P.,
McCulloch, A., Harth, C., Reimann, S., and Salameh, P.: Evidence for
variability of atmospheric hydroxyl radicals over the past quarter century,
Geophys. Res. Lett., 32, L07809, 10.1029/2004GL022228, 2005.Rayner, P., Enting, I., Francey, R., and Langenfelds, R.: Reconstructing the
recent carbon cycle from atmospheric CO2, δ13C and
O2/N2 observations, Tellus B, 51, 213–232, 1999.
Reimann, S., Manning, A. J., Simmonds, P. G., Cunnold, D. M., Wang, R. H.,
Li, J., McCulloch, A., Prinn, R. G., Huang, J., and Weiss, R. F.: Low
European methyl chloroform emissions inferred from long-term atmospheric
measurements, Nature, 433, 506–508, 2005.Rhoderick, G. C., Hall, B. D., Harth, C. M., Kim, J. S., Lee, J., Montzka,
S. A., Mühle, J., Reimann, S., Vollmer, M. K., and Weiss, R. F.:
Comparison of halocarbon measurements in an atmospheric dry whole air sample,
Elem. Sci. Anth., 3, 75, 10.12952/journal.elementa.000075, 2015.
Rhodes, R. H., Faïn, X., Stowasser, C., Blunier, T., Chappellaz, J.,
McConnell, J. R., Romanini, D., Mitchell, L. E., and Brook, E. J.: Continuous
methane measurements from a late Holocene Greenland ice core: Atmospheric and
in-situ signals, Earth Planet. Sc. Lett., 368, 9–19, 2013.Rigby, M., Ganesan, A. L., and Prinn, R. G.: Deriving emissions time series
from sparse atmospheric mole fractions, J. Geophys. Res., 116, D08306,
10.1029/2010JD015401, 2011.Rigby, M., Prinn, R. G., O'Doherty, S., Montzka, S. A., McCulloch, A., Harth,
C. M., Mühle, J., Salameh, P. K., Weiss, R. F., Young, D., Simmonds,
P. G., Hall, B. D., Dutton, G. S., Nance, D., Mondeel, D. J., Elkins, J. W.,
Krummel, P. B., Steele, L. P., and Fraser, P. J.: Re-evaluation of the
lifetimes of the major CFCs and CH3CCl3 using atmospheric trends,
Atmos. Chem. Phys., 13, 2691–2702, 10.5194/acp-13-2691-2013, 2013.
Rigby, M., Prinn, R. G., O'Doherty, S., Miller, B. R., Ivy, D., Mühle,
J., Harth, C. M., Salameh, P. K., Arnold, T., Weiss, R. F., Krummel, P. B.,
Steele, L. P., Fraser, P. J., Young, D., and Simmonds, P. G.: Recent and
future trends in synthetic greenhouse gas radiative forcing, Geophys. Res.
Lett., 41, 2623–2630, 2014.Rubino, M., Etheridge, D., Trudinger, C., Allison, C., Battle, M.,
Langenfelds, R., Steele, L., Curran, M., Bender, M., and White, J.: A revised
1000 year atmospheric δ13C-CO2 record from Law Dome and
South Pole, Antarctica, J. Geophys. Res.-Atmos., 118, 8482–8499, 2013.Rubino, M., Etheridge, D. M., Trudinger, C. M., Allison, C. E., Rayner,
P. J., Enting, I., Mulvaney, R., Steele, L. P., Langenfelds, R. L., Sturges,
W. T., Curran, M. A. J., and Smith, A. M.: Low atmospheric CO2 levels
during the Little Ice Age due to cooling-induced terrestrial uptake, Nat.
Geosci., 9, 691–694, 2016.
Schilt, A., Baumgartner, M., Blunier, T., Schwander, J., Spahni, R., Fischer,
H., and Stocker, T. F.: Glacial–interglacial and millennial-scale variations
in the atmospheric nitrous oxide concentration during the last 800,000 years,
Quaternary Sci. Rev., 29, 182–192, 2010a.
Schilt, A., Baumgartner, M., Schwander, J., Buiron, D., Capron, E.,
Chappellaz, J., Loulergue, L., Schüpbach, S., Spahni, R., Fischer, H.,
and Stocker, T. F.: Atmospheric nitrous oxide during the last 140,000 years,
Earth Planet. Sc. Lett., 300, 33–43, 2010b.Schneider, R., Schmitt, J., Köhler, P., Joos, F., and Fischer, H.: A
reconstruction of atmospheric carbon dioxide and its stable carbon isotopic
composition from the penultimate glacial maximum to the last glacial
inception, Clim. Past, 9, 2507–2523, 10.5194/cp-9-2507-2013, 2013.
Shindell, D., Faluvegi, G., Nazarenko, L., Bowman, K., Lamarque, J.-F.,
Voulgarakis, A., Schmidt, G. A., Pechony, O., and Ruedy, R.: Attribution of
historical ozone forcing to anthropogenic emissions, Nature Climate Change,
3, 567–570, 2013.
Siegenthaler, U., Stocker, T. F., Monnin, E., Lüthi, D., Schwander, J.,
Stauffer, B., Raynaud, D., Barnola, J.-M., Fischer, H., and Masson-Delmotte,
V.: Stable carbon cycle–climate relationship during the late Pleistocene,
Science, 310, 1313–1317, 2005.Simmonds, P. G., Cunnold, D. M., Weiss, R. F., Prinn, R. G., Fraser, P. J.,
McCulloch, A., Alyea, F. N., and O'Doherty, S.: Global trends and emission
estimates of CCl4 from in situ background observations from July 1978
to June 1996, J. Geophys. Res., 103, 16017–16027, 1998.
Simmonds, P. G., Derwent, R. G., Manning, A. J., Fraser, P. J., Krummel,
P. B., O'doherty, S., Prinn, R. G., Cunnold, D. M., Miller, B. R., Wang,
H. J., Ryall, D. B., Porter, L. W., Weiss, R. F., and Salameh, P. K.: AGAGE
observations of methyl bromide and methyl chloride at Mace Head, Ireland, and
Cape Grim, Tasmania, 1998–2001, J. Atmos. Chem., 47, 243–269, 2004.Sowers, T., Alley, R. B., and Jubenville, J.: Ice core records of atmospheric
N2O covering the last 106,000 years, Science, 301, 945–948, 2003.
Spahni, R., Chappellaz, J., Stocker, T. F., Loulergue, L., Hausammann, G.,
Kawamura, K., Flückiger, J., Schwander, J., Raynaud, D., and
Masson-Delmotte, V.: Atmospheric methane and nitrous oxide of the late
Pleistocene from Antarctic ice cores, Science, 310, 1317-1321, 2005.
Spivakovsky, C., Logan, J., Montzka, S., Balkanski, Y., Foreman-Fowler, M.,
Jones, D., Horowitz, L., Fusco, A., Brenninkmeijer, C., and Prather, M.:
Three-dimensional climatological distribution of tropospheric OH: Update and
evaluation, J. Geophys. Res., 105, 8931–8980, 2000.Stauffer, B., Fluckiger, J., Monnin, E., Schwander, J., Barnola, J.-M., and
Chappellaz, J.: Atmospheric CO2, CH4 and N2O records
over the past 60 000 years based on the comparison of different polar ice
cores, Ann. Glaciol., 35, 202–208, 2002.
Steele, L. P., Fraser, P. J., Rasmussen, R. A., Khalil, M. A. K., Conway,
T. J., Crawford, A. J., Gammon, R. H., Masarie, K. A., and Thoning, K. W.:
The global distribution of methane in the troposphere, J. Atmos. Chem., 5,
125–171, 1987.
Steele, L. P., Lang, P. M., and Sepanski, R. J.: Atmospheric methane
concentrations – the NOAA/CMDL global cooperative flask sampling network,
1983–1988, Technical Report, Oak Ridge National Lab., TN, USA, 1991.
Steele, L. P., Dlugokencky, E. J., Lang, P. M., Tans, P. P., Martin, R. C.,
and Masarie, K. A.: Slowing down of the global accumulation of atmospheric
methane during the 1980s, Nature, 358, 313–316, 1992.Sturrock, G. A., Etheridge, D. M., Trudinger, C. M., Fraser, P. J., and
Smith, A. M.: Atmospheric histories of halocarbons from analysis of Antarctic
firn air: Major Montreal Protocol species, J. Geophys. Res., 107, 4765,
10.1029/2002JD002548, 2002.
Tans, P. P., Conway, T. J., and Nakazawa, T.: Latitudinal distribution of the
sources and sinks of atmospheric carbon dioxide derived from surface
observations and an atmospheric transport model, J. Geophys. Res., 94,
5151–5172, 1989.Tans, P. P., Fung, I. Y., and Takahashi, T.: Observational contrains on the
global atmospheric CO2 budget, Science, 247, 1431–1438, 1990a.Tans, P. P., Thoning, K. W., Elliott, W. P., and Conway, T. J.: Error
estimates of background atmospheric CO2 patterns from weekly flask
samples, J. Geophys. Res., 95, 14063–14070, 1990b.
Thoning, K. W., Tans, P., Conway, T. J., and Waterman, L. S.: NOAA/GMCC
calibrations of CO2-in-air reference gases: 1979–1985, NOAA Tech. Memo. ERL
ARL-150, NOAA Environmental Research Laboratories, Boulder, CO, 63 pp., 1987.
Thoning, K. W., Tans, P. P., and Komhyr, W. D.: Atmospheric carbon dioxide at
Mauna Loa Observatory: 2. Analysis of the NOAA GMCC data, 1974–1985,
J. Geophys. Res., 94, 8549–8565, 1989.Thoning, K. W., Conway, T. J., Zhang, N., and Kitzis, D.: Analysis System for
Measurement of CO2 Mixing Ratios in Flask Air Samples, J. Atmos.
Ocean. Tech., 12, 1349–1356, 1995.Trolier, M., White, J. W. C., Tans, P. P., Masarie, K. A., and Gemery, P. A.:
Monitoring the isotopic composition of atmospheric CO2: Measurements
from the NOAA Global Air Sampling Network, J. Geophys. Res., 101,
25897–25916, 1996.Trudinger, C. M., Enting, I. G., Rayner, P. J., and Francey, R. J.: Kalman
filter analysis of ice core data, 2, Double deconvolution of CO2 and
δ13C measurements, J. Geophys. Res., 107, 4423,
10.1029/2001JD001112, 2002a.Trudinger, C. M., Etheridge, D. M., Rayner, P. J., Enting, I. G., Sturrock,
G. A., and Langenfelds, R. L.: Reconstructing atmospheric histories from
measurements of air composition in firn, J. Geophys. Res., 107, 4780,
10.1029/2002JD002545, 2002b.Trudinger, C. M., Etheridge, D. M., Sturrock, G. A., Fraser, P. J., Krummel,
P. B., and McCulloch, A.: Atmospheric histories of halocarbons from analysis
of Antarctic firn air: Methyl bromide, methyl chloride, chloroform, and
dichloromethane, J. Geophys. Res., 109, D22310, 10.1029/2004JD004932,
2004.Trudinger, C. M., Fraser, P. J., Etheridge, D. M., Sturges, W. T., Vollmer,
M. K., Rigby, M., Martinerie, P., Mühle, J., Worton, D. R., Krummel,
P. B., Steele, L. P., Miller, B. R., Laube, J., Mani, F. S., Rayner, P. J.,
Harth, C. M., Witrant, E., Blunier, T., Schwander, J., O'Doherty, S., and
Battle, M.: Atmospheric abundance and global emissions of perfluorocarbons
CF4, C2F6 and C3F8 since 1800 inferred from ice
core, firn, air archive and in situ measurements, Atmos. Chem. Phys., 16,
11733–11754, 10.5194/acp-16-11733-2016, 2016.Tsutsumi, Y., Mori, K., Hirahara, T., Ikegami, M., and Conway, T. J.:
Technical report of global analysis method for major greenhouse gases by the
World Data Center for greenhouse gases, WMO/TD, GAW Report No. 184, available
at:
https://www.wmo.int/pages/prog/arep/gaw/documents/TD_1473_GAW184_web.pdf,
last access: 20 May 2017, 2009.Velders, G. J. M. and Daniel, J. S.: Uncertainty analysis of projections of
ozone-depleting substances: mixing ratios, EESC, ODPs, and GWPs, Atmos. Chem.
Phys., 14, 2757–2776, 10.5194/acp-14-2757-2014, 2014.
Velders, G. J. M., Andersen, S. O., Daniel, J. S., Fahey, D. W., and
McFarland, M.: The importance of the Montreal Protocol in protecting climate,
P. Natl. Acad. Sci. USA, 104, 4814–4819, 2007.
Velders, G. J. M., Fahey, D. W., Daniel, J. S., McFarland, M., and Andersen,
S. O.: The large contribution of projected HFC emissions to future climate
forcing, P. Natl. Acad. Sci. USA, 106, 10949–10954, 2009.Vollmer, M. K., Miller, B. R., Rigby, M., Reimann, S., Mühle, J.,
Krummel, P. B., O'Doherty, S., Kim, J., Rhee, T. S., Weiss, R. F., Fraser,
P. J., Simmonds, P. G., Salameh, P. K., Harth, C. M., Wang, R. H. J., Steele,
L. P., Young, D., Lunder, C. R., Hermansen, O., Ivy, D., Arnold, T.,
Schmidbauer, N., Kim, K.-R., Greally, B. R., Hill, M., Leist, M., Wenger, A.,
and Prinn, R. G.: Atmospheric histories and global emissions of the
anthropogenic hydrofluorocarbons HFC-365mfc, HFC-245fa, HFC-227ea, and
HFC-236fa, J. Geophys. Res., 116, D08304, 10.1029/2010JD015309, 2011.Vollmer, M. K., Mühle, J., Trudinger, C. M., Rigby, M., Montzka, S. A.,
Harth, C. M., Miller, B. R., Henne, S., Krummel, P. B., Hall, B. D., Young,
D., Kim, J., Arduini, J., Wenger, A., Yao, B., Reimann, S., O'Doherty, S.,
Maione, M., Etheridge, D. M., Li, S., Verdonik, D. P., Park, S., Dutton, G.,
Steele, L. P., Lunder, C. R., Rhee, T. S., Hermansen, O., Schmidbauer, N.,
Wang, R. H. J., Hill, M., Salameh, P. K., Langenfelds, R. L., Zhou, L.,
Blunier, T., Schwander, J., Elkins, J. W., Butler, J. H., Simmonds, P. G.,
Weiss, R. F., Prinn, R. G., and Fraser, P. J.: Atmospheric histories and
global emissions of halons H-1211 (CBrClF2), H-1301 (CBrF3),
and H-2402 (CBrF2CBrF2), J. Geophys. Res.-Atmos., 121, 3663-3686,
10.1002/2015JD024488, 2016.
Walker, S. J., Weiss, R. F., and Salameh, P. K.: Reconstructed histories of
the annual mean atmospheric mole fractions for the halocarbons CFC-11 CFC-12,
CFC-113, and carbon tetrachloride, J. Geophys. Res., 105, 14285–14296, 2000.Welp, L. R., Patra, P. K., Rödenbeck, C., Nemani, R., Bi, J., Piper,
S. C., and Keeling, R. F.: Increasing summer net CO2 uptake in high
northern ecosystems inferred from atmospheric inversions and comparisons to
remote-sensing NDVI, Atmos. Chem. Phys., 16, 9047–9066,
10.5194/acp-16-9047-2016, 2016.
WMO: WMO WDCGG Data Summary 36, Japan Meterological Agency and WMO, 100 pp.,
2012.
WMO: Scientific Assessment of Ozone Depletion: 2014, World Meteorological
Organization, Geneva, Switzerland, 416 pp., 2014.Worton, D. R., Sturges, W. T., Schwander, J., Mulvaney, R., Barnola, J.-M.,
and Chappellaz, J.: 20th century trends and budget implications of chloroform
and related tri-and dihalomethanes inferred from firn air, Atmos. Chem.
Phys., 6, 2847–2863, 10.5194/acp-6-2847-2006, 2006.
Worton, D. R., Sturges, W. T., Gohar, L. K., Shine, K. P., Martinerie, P.,
Oram, D. E., Humphrey, S. P., Begley, P., Gunn, L., Barnola, J. M.,
Schwander, J., and Mulvaney, R.: Atmospheric trends and radiative forcings of
CF4 and C2F6 inferred from firn air, Environ. Sci. Technol., 41, 2184–2189,
2007.
Xiang, B., Patra, P. K., Montzka, S. A., Miller, S. M., Elkins, J. W., Moore,
F. L., Atlas, E. L., Miller, B. R., Weiss, R. F., Prinn, R. G., and Wofsy,
S. C.: Global emissions of refrigerants HCFC-22 and HFC-134a: Unforeseen
seasonal contributions, P. Natl. Acad. Sci. USA, 111, 17379–17384, 2014.Zhao, C. L. and Tans, P. P.: Estimating uncertainty of the WMO mole fraction
scale for carbon dioxide in air, J. Geophys. Res., 111, D08S09,
10.1029/2005JD006003, 2006.