Since the first estimate of global CO2 emissions was
published in 1894, important progress has been made in the development of
estimation methods while the number of available datasets has grown. The
existence of parallel efforts should lead to improved accuracy and
understanding of emissions estimates, but there remains significant
deviation between estimates and relatively poor understanding of the reasons
for this. Here I describe the most important global emissions datasets
available today and – by way of global, large-emitter, and case examples – quantitatively compare their estimates, exploring the reasons for
differences. In many cases differences in emissions come down to differences
in system boundaries: which emissions sources are included and which are
omitted. With minimal work in harmonising these system boundaries across
datasets, the range of estimates of global emissions drops to 5 %, and
further work on harmonisation would likely result in an even lower range,
without changing the data. Some potential errors were found, and some
discrepancies remain unexplained, but it is shown to be inappropriate to
conclude that uncertainty in emissions is high simply because estimates
exhibit a wide range. While “true” emissions cannot be known, by comparing
different datasets methodically, differences that result from system
boundaries and allocation approaches can be highlighted and set aside to
enable identification of true differences, and potential errors. This must
be an important way forward in improving global datasets of CO2
emissions. Data used to generate Figs. 3–18 are available at
10.5281/zenodo.3687042 (Andrew, 2020).
Introduction
Since the first known estimate of global anthropogenic emissions of CO2
was made in the early 1890s, methods have substantially improved, detail has
increased, and additional emissions sources have been included. Meanwhile,
with international agreements to mitigate climate change, the production of
such estimates has grown beyond the realm of scientific enquiry to become a
critical input to policy.
Fossil CO2 emissions occur when fossil carbon compounds are broken down
via combustion or other oxidation processes. Most of these fossil compounds
are in the form of fossil fuels, such as coal, oil, and natural gas. In
addition are fossil carbonates, such as calcium carbonate and magnesium
carbonate, which are used as feedstocks in several important industrial
processes – including cement production – and whose decomposition also
leads to emissions of CO2.
Every year several global emissions datasets are updated and present
different estimates of CO2 at both national and global levels, but
there is little information available about why these estimates differ, and
sometimes the range between these estimates is merely assumed to represent
uncertainty, suggesting that no more can be known. In fact, there are core
reasons why the estimates from these different datasets differ, but these
are largely buried in the documentation and have not previously been
comprehensively analysed; these will be discussed in detail in this article.
The accumulated global emissions of carbon dioxide are drawing precariously
close to the best estimates of the total budget available before the world
crosses certain temperatures agreed to in international negotiations
(Nauels et al., 2019). A detailed understanding of these different annual
estimates – to unravel the “range is uncertainty” tangle – is therefore
important in efforts to correctly represent uncertainty and improve our
understanding of the requirements before global society.
There are identifiable reasons why estimates differ between datasets, but a
close examination of the datasets' documentation and quantitative comparison
of the data themselves are required to determine the significance of
differences in sources and methods used. In particular, not all datasets
attempt to be comprehensive either geographically or by including all
emissions sources.
In this article I will discuss datasets whose spatial resolution is at the
country level, but there exist several datasets that are further
disaggregated to grids of varying resolution. These include CDIAC (Andres
et al., 1997, 2016a), ODIAC (Oda and Maksyutov, 2011; Oda
et al., 2018), EDGAR (Janssens-Maenhout et al., 2012), FFDAS
(Asefi-Najafabady et al., 2014), CEDS
(Hoesly et al., 2018), and
PKU-FUEL (Chen et al., 2016)
CDIAC: Carbon Dioxide
Information Analysis Center; ODIAC: Open-source Data Inventory for Anthropogenic
CO2; EDGAR: Emissions Database for Global Atmospheric Research;
FFDAS: Fossil Fuel Data Assimilation System; CEDS: Community Emissions Data
System; PKU-FUEL: Peking University Fuel emissions.
. Many of these gridded
datasets use existing country-level datasets as primary input data and will
therefore have similar attributes to the datasets discussed in this article.
An overview of some gridded dataset was presented by
Andres et al. (2012), and recent
assessments of uncertainty in gridded emissions are presented by
Andres et al. (2016b) and Oda et al. (2019).
Because fossil fuel CO2 emissions are largely connected with energy,
which is a closely tracked commodity group with its critical role in
economic activity, there is a wealth of underlying data that can be used for
estimating emissions. However, differences in collection, treatment,
interpretation, inclusion, and various factors such as carbon contents and
fractions of oxidised carbon, can lead to significant differences in
estimates of emissions between datasets.
Several comparisons have been performed before, notably Marland
et al. (1999), who compared CDIAC and EDGAR,
Andres et al. (2012), who made a high-level
comparison of several datasets, and Ciais et
al. (2010), who compared datasets for the EU. The most complete comparison
of global datasets to date is that of Macknick (2011), who made
quantitative comparisons of CDIAC, EDGAR, the IEA, BP, and the EIA, including
discussion of the differences in underlying energy datasets. That study is
now almost 10 years old, and methodologies of each of those datasets have
changed in the intervening time, while new datasets are also available.
Given the introduction of a temperature-based goal in the Copenhagen Accord
(UNFCCC, 2009), and in particular the much more proximal 1.5 ∘C goal in the Paris Agreement (UN, 2015), tracking global emissions is
highly important, requiring reliable estimates of uncertainty.
This paper summarises early efforts to quantify global CO2 emissions, before
moving to current datasets, starting by discussing differences that can be
expected a priori, describing the important IPCC inventory guidelines, summarising
the publicly available emissions datasets, and then comparing these in some
detail and explaining quantitative differences where possible. Because many
of these products have appeared and will continue to appear in Earth System Science Data, readers can apply this review as a detailed guide to most
recent emissions compilations and products as published in this journal.
Early estimates
In the earliest years, the carbon cycle of interest was the long-term
balancing of geological, extraterrestrial, and natural-system inputs and
outputs, primarily to understand the mysteries of the ice ages. While
geological interest remains, interest in the human perturbation of the
global carbon cycle has grown as anthropogenic emissions – once considered
negligible – have grown to a very significant degree. Accurately gauging
the magnitude of this perturbation is of key importance in understanding how
current and future climatic changes relate to our historical and current
emissions. Over more than a century, the quality of estimates of global
CO2 emissions has steadily improved as greater understanding and
greater effort have been brought to bear on the problem.
In the early 1890s, Swedish geochemist Arvid Högbom was possibly the
first to consider the global geochemical carbon cycle, and he presented some of
his thoughts to the Swedish Chemical Society, later published in the Swedish
Chemistry Journal (Högbom, 1894). Högbom briefly
considered whether combustion of fossil fuels might perturb the carbon
cycle, estimating that emissions at that time were 0.5 Gt C (see
Supplement for details) and determining that this was
insufficient to have any effect on atmospheric concentrations because it
would merely compensate for CO2 consumed in the continuous formation of
carbonates.
Svante Arrhenius, inspired by Högbom's lecture, initially accepted his
main conclusion that the short-term carbon cycle was in balance, but
he considered what might happen if fossil emissions were to further increase
(Arrhenius, 1908). In so doing he presents estimates of emissions
from the global combustion of coal of 0.51 Gt C in 1890, 0.55 Gt C in 1899,
and 0.89 Gt C in 1904.
Guy Callendar (1938), investigating the influence of
fossil CO2 on temperature, stated without support that annual emissions
of CO2 at the time amounted to 4.3 Gt CO2 and that the cumulative
emissions over the previous 50 years were 150 Gt CO2.
Gilbert Plass (1956) stated, again without support, that
the combustion of fossil fuels at that time was adding 6 Gt CO2
annually. He also listed other human activities that release carbon,
unfortunately without any quantification: “the clearance of forests, the
drainage and cultivation of lands, and industrial processes such as lime
burning and fermentation” (Plass, 1956, p. 379).
In 1957, Roger Revelle and Hans Suess, interested in the fate of the carbon
dioxide added to the atmosphere by human activities, estimated emissions
from fossil fuel combustion per decade, from the 1860s through 1940s
(Revelle and Suess, 1957). The methods were not given, but
reference was elsewhere given to a recent United Nations report “World
requirements of energy, 1975–2000” presented at the International
Conference on Peaceful Uses of Atomic Energy, held in Geneva in 1955
(United Nations, 1956), and this is most likely the source of the energy
data.
Importantly, all the foregoing estimates appear to have implicitly assumed
in their calculations that fossil fuels were composed almost entirely of
carbon. It was not until the 1965 report by the President's Science Advisory
Committee Panel on Environmental Pollution that sources and carbon contents
by fuel type are provided (Revelle et al., 1965). The panel's report
also broke emissions down by main fuel category (coal, oil, gas). In
addition, specific sources of energy data are clearly stated.
Baxter and Walton (1970), looking to explain the decline
in the fraction of isotope 14C in the atmosphere, presented annual
fossil carbon emissions, including, for the first time, from cement
production. While methods and sources are reasonably clearly presented, some
errors in interpretation of sources led to highly inflated estimates of
emissions from lignite.
Broecker et al. (1971), looking at uptake of CO2 by the oceans,
present decadal global emissions by coal, oil, and gas, and they were the first
to make explicit their assumption of complete oxidation of all fossil fuels.
Keeling (1973) was dissatisfied with the lack of rigour in
previous emissions assessments and substantially increased the detail of
analysis. He identified that earlier studies had greatly overestimated
emissions from coal and lignite by using inflated carbon contents,
introduced adjustments for non-energy uses of fuels and losses, and
performed an uncertainty assessment. Keeling's methods shaped the methods
for estimating CO2 emissions for the following decades.
Rotty (1973) introduced estimates for flaring and venting
of natural gas and CO2, although because of data limitations venting of
natural gas could not be separated out, and the methane content was therefore
assumed to be oxidised immediately to CO2, an assumption that continued
in later datasets. While most previous estimates had relied on energy data
from the UN, Rotty pointed to an alternative source of energy data,
demonstrating that the two sources were in agreement when close attention
was paid to definitions. Later, Rotty (1983) presented the first
estimates of sub-global CO2 emissions using apparent consumption of
energy – production adjusted for international trade – rather than the
energy production data used in all previous studies.
Marland and Rotty (1984) re-examined the method of
Keeling (1973) and produced slightly revised emission factors,
still time-invariant, an assumption that appeared valid given the
observational data available at the time. Importantly, they were the first
to make use of new energy data from the United Nations that were already in
units of energy, having been converted from physical units using
country-specific factors, and this avoided the use of global-average,
time-invariant energy conversion factors that previous estimates had been
based on. The method developed by Marland and Rotty (1984) has
been used – with some modifications and improvements – by CDIAC right
through until its 2019 release.
Potential reasons for differences between datasets
Before turning to an exploration of the actual differences between emissions
datasets, I first discuss some of the reasons for which datasets should be
expected a priori to differ based on what is already known. Many of these hinge on what can be called “system boundaries”.
The term system boundary is found in the life-cycle assessment (LCA)
literature where it describes the scope of analysis: which activities in a
process or supply chain are included and which are omitted (Baumann
and Tillman, 2004). Here I use the term to describe the categories of
emissions that are included in each dataset and the way in which they are
distinguished when presented in more detail. There are many aspects to
these, which I will discuss in turn, limiting myself to CO2 emissions
data.
There are three main physical sources of anthropogenic carbon dioxide:
oxidation of fossil fuels, land-use change (e.g. deforestation), and
decomposition of (fossil) carbonates (Friedlingstein et al., 2019). All
emissions datasets include fossil fuels, while fewer include either land-use
change or carbonates.
Decomposition of carbonates occurs in production of cement, lime, and glass,
but also in steel manufacturing where carbonates are used as a flux agent to
facilitate removal of impurities, and in flue gas desulfurisation
(Córdoba, 2015). Datasets may exclude carbonate
emissions entirely or include emissions only from cement production (e.g.
CDIAC) or from all carbonate decomposition (e.g. EDGAR).
Several emissions datasets are relatively simple extensions of energy
datasets (e.g. IEA, EIA, BP), and their primary purpose is to show the
emissions associated with consumption of energy, rather than to provide a
comprehensive picture of all emissions of CO2. Most emissions of
CO2 are from fossil fuels, with emissions from land-use change and
carbonates currently amounting to about 13 % and 5 % of the global
total, respectively (Friedlingstein et al., 2019; Crippa et al., 2019).
While most datasets focus on combustion of fossil fuels, some extend the
definition to all oxidation of fossil fuels. While combustion is one form of
oxidation, other forms exist, such as in chemical processes where
hydrocarbons are used as a source of carbon or as a reducing agent. This
distinction generally hinges on whether the fossil fuel is primarily
required as an energy source (energy released by combustion) or as an agent
in a chemical process.
In particular coal consumption in the metal industry can be considered to be
either combustion or a reducing agent. Coke used in refining iron ore is
critical as a reducing agent but also serves as an energy source. However,
when coal and oil are used to make carbon anodes for aluminium smelting, the
oxidation of the anode that occurs during smelting is not considered
combustion.
The entities to which emissions are assigned vary between datasets. For
example, some parties have different geographic and economic extents under
the Kyoto Protocol and the UNFCCC, and therefore they submit more than one
inventory to the UNFCCC. These include Denmark, France, and the United
Kingdom (EEA, 2019). These make differences of less than 2 % for
individual countries. The European Union also submits two sets of
inventories: EUA (convention: strictly EU territory) and EUC (Kyoto
Protocol: also includes Iceland and overseas territories of member states).
Similarly, the United States reports include Puerto Rico and other
territories when submitted to the UNFCCC.
Moreover, emissions can be limited to either geographical areas or economic
activities. Inventories, as for example submitted to the UNFCCC, cover
geographical areas (akin to energy balances), while accounts cover economic
activities (akin to energy accounts) (UNSD, 2018). Accounts must be
adjusted for the activities of foreign nationals and companies within the
territory (e.g. emissions from tourists driving cars), and activities of
nationals and national companies in other territories. Accounts follow the
definitions of the System of National Accounts, used, among other things,
for calculating gross domestic product (GDP; European Commission, International Monetary Fund, Organisation for Economic
Co-operation and Development, United Nations, and World Bank, 2009).
With regard to these country definitions, the allocation of emissions from
combustion of international bunker fuels has been particularly problematic.
While energy data are collated as to which country sells bunker fuels, this
is very poorly related to which country has responsibility for the
combustion of those fuels. Various methods have been proposed to allocate
these emissions, such as to the country whose flag a ship operates under, or
that which the owner of the ship is a tax resident in, or those that operate
the ship, of even those who purchase the goods borne by the ship
(Heitmann and Khalilian, 2011). However, none of these are clearly
superior to the others, and they can result in very different distributions
of these emissions. This is in effect why the international aviation and
maritime industries have been largely excluded from negotiations and are
acting partly independently on a global basis (UNFCCC, 2020). See
Supplement for further details on the inclusion of bunker
fuels.
Further methods of allocating emissions have been devised, such as
reallocating through economic supply chains to the point of final
consumption, so-called consumption-based emissions, and variants (e.g.
Davis and Caldeira, 2010; Andrew et al., 2013). These alternatives have not
yet obtained international acceptance, although official reporting at
national level does occur in, e.g. the UK (DEFRA, 2019), Sweden
(Björk et al., 2018), and France (SOeS, 2012;
I4CE, 2018).
The ways in which national or global emissions are presented in more
detailed form can vary substantially between datasets. While the IPCC
Guidelines set a clear method for differentiating between “sectors of [the]
economy” (Penman et al., 2006, p. 4), these sectors are quite
different to those understood by economists. The energy sector, for example,
includes most combustion of energy, whether the activities are undertaken by
enterprises whose main activity is energy production or not. All household
combustion of gasoline in private transportation is included in the energy
sector, whereas under economic accounts such activities would be accounted
for in the household sector. Agricultural emissions, under the IPCC methodology,
do not include such activities as driving tractors or heating glasshouses.
So, while in economics a “sector” is a grouping of similar economic actors,
in the IPCC Guidelines a sector is a grouping of activities. Some other
datasets do assign emissions to economic activities. Further, breakdown by
type of fossil fuel can vary, with the use of solid, liquid, and gaseous
fuel categories as distinct from coal, oil, and natural gas (see Sect. 5.4).
The time period over which emissions are accounted for can vary. While all
modern datasets present annual emissions, some also report sub-annual
periods. More importantly, while most countries' data are reported for
calendar years (from 1 January to 31 December), some are reported for
financial years. In the IEA's data, which are probably representative of most
datasets because of non-independent original sources, non-calendar year data
are reported for Bangladesh, Egypt, India, and Nepal. For India, by far the most
significant of these, the IEA's data for 2016 represent the financial year 1 April 2016–31 March 2017 (the majority of this period falling in 2016),
which would be called the 2017 year in India (the financial year ending in
2017, FY17).
One final category of system boundaries is the inclusion of confidential
data. At detailed levels some countries may withhold and aggregate reporting
of emissions from certain activities for strategic reasons. While these are
generally included at aggregate level, emissions from military activities
are known sometimes to be withheld entirely. The IEA “has found that in
practice most countries consider information on military consumption as
confidential and therefore either combine it with other information or do
not include it at all” (IEA, 2018c, p. 55). While some confidential
emissions might be excluded from national accounts, energy production
statistics most likely cover all energy produced so that estimates of global
fossil CO2 emissions would not be expected to exclude military energy
consumption.
Different methods can be used to estimate emissions, based on different
original data sources. The most important distinction is between the
sectoral approach and the reference approach, applying only to emissions
from fossil fuels. While the sectoral approach is based on detailed
demand-side energy data (a bottom-up calculation, starting with as much
detail as possible, typically sales data), the reference approach is based
on much less detailed supply-side data (a top-down calculation, typically
using national production, international trade, and stock change data). At
the national level, estimates generated by the sectoral approach are
definitive, while those under the reference approach are used as a partially
independent cross check (Treanton et al., 2006b).
IPCC inventory guidelines
The Intergovernmental Panel on Climate Change (IPCC) provides comprehensive
guidance for compiling emissions inventories for all sources of emissions,
which must be used in reporting to the UNFCCC (2019b).
The guidelines are built on decades of efforts and expertise in compiling
emissions estimates and are designed to be flexible to suit countries'
specific needs. Work began on the guidelines in 1991 by Working Group 1 of
the IPCC under the IPCC/OECD/IEA Programme on National Greenhouse Gas
Inventories, with the first edition approved in 1994 and adopted the
following year. A revision to these was published in 1996
(Houghton et al., 1996), and a new edition was published
in 2006 (Eggleston et al., 2006), with later amendments (e.g. the wetlands supplement; Hiraishi et al., 2013). In
2019 a “refinement” to the guidelines was released, although with little
material consequence for fossil emissions beyond some changes to fugitive
emissions calculations and clarifications on biofuel emissions
(IPCC, 2019).
The methodology is divided into three “tiers”, where Tier 1 uses supplied
default emission factors applied to national activity data, Tier 2 uses
national emission factors, and Tier 3 uses national models and/or direct
measurements. The Tier 1 approach is often used by compilers of
international inventories because they can do so using existing
international datasets of activity, such as energy or agriculture databases.
The IPCC Guidelines divide emissions into sectors, the most important of
which for CO2 emissions are the energy sector and the industrial
processes and product use (IPPU) sector. The Energy sector includes
emissions from activities in the energy industry (e.g. electricity
production, flaring on oil platforms) and all uses of energy sources
primarily for energy purposes. The IPPU sector, in contrast, includes
emissions from both decomposition of carbonates (e.g. cement production) and
non-energy uses of fossil fuels (e.g. carbon anodes in aluminium smelting).
Smaller amounts of CO2 are also reported in the agriculture sector, as
emissions from use of urea and lime.
While the 1996 guidelines included a default fraction of carbon stored
(sequestered) from non-energy use (allowing for some to be oxidised at some
point), the 2006 guidelines removed these, effectively setting the fraction
stored to 1.0 for all products. This was because “in most instances,
emission inventory compilers had no `real' information as to whether this
correction was actually applicable” (IEA, 2018a, p. I.22).
Detailed descriptions of emissions data sources
Figure 1 shows the main global energy datasets,
primary global emissions datasets, and secondary (derived from primary)
emissions datasets. The most important data source type for emissions
estimates is energy data, which are ultimately derived from heterogeneous
national sources. The IEA and Eurostat have developed questionnaires that
are sent to at least 61 countries: all members of the Organisation for
Economic Co-operation and Development (OECD), European Union (EU), United
Nations Economic Commission for Europe (UNECE), “and a few others”. These
identically completed questionnaires are returned to the IEA, the UN, and
(for certain countries) Eurostat (IEA, 2019g).
Dependencies of selected global energy and CO2 emissions
datasets. Here a “primary” emissions dataset is one that calculated
emissions directly from energy data, rather than collating emissions
estimates from other sources. In addition to energy data sources, some
emissions datasets include emissions from carbonates, which rely on other
data sources. Some national data are first collated by regional
organisations. “UN stats” is the United Nations Statistics Office (not
UNFCCC).
In the following sections I will describe each of the main global emissions
datasets in turn. Table 1 summarises the datasets
and the versions analysed in detail.
Summary of datasets included in this analysis.
ShortDatasetnameVersionReferenceDOI/URLGlobal Carbon ProjectGCP2019Friedlingstein et al. (2019)10.18160/GCP-2019Energy Information Admin-istration International EnergyStatisticsEIA2020EIA (2020)https://www.eia.gov/international/data/worldInternational Energy Agency CO2 Emissions from FuelCombustionIEA2019IEA (2019b)https://webstore.iea.org/co2-emissions-from-fuel-combustion-2019-highlightsBP Statistical Review of WorldEnergyBP2019BP (2019)https://www.bp.com/en/global/corporate/energy-economics/statistical-review-of-world-energy.htmlCommunity Emissions DataSystemCEDSv_2019_12_23Hoesly et al. (2018)10.5281/zenodo.3606753Emissions Database for GlobalAtmospheric ResearchEDGARv5.0Crippa et al. (2019)https://edgar.jrc.ec.europa.eu/overview.php?v=50_GHGAnnex I Common ReportingFormatCRF20-Apr-19UNFCCC (2019a)https://unfccc.int/process-and-meetings/transparency-and-reporting/reporting-and-review-under-the-convention/greenhouse-gas-inventories-annex-i-parties/national-inventory-submissions-2019CDIAC Global, Regional,and National Fossil-Fuel CO2 EmissionsCDIAC2019Gilfillan et al. (2019)https://energy.appstate.edu/research/work-areas/cdiac-appstatePotsdam Real-time IntegratedModel for probabilistic Assessment of emissions PathsPRIMAP-hist2.1Gütschow et al. (2019b)10.5880/PIK.2019.018BP's Statistical Review of World Energy
BP produced its first limited-circulation Statistical Review of World Energy
in 1952 (BP, 2011). In recent years the BP review has been highly
anticipated primarily because it is the earliest data release to cover
global energy and fossil fuel CO2 emissions, published in June of each
year with data up to the previous year. The dataset is widely used,
something that is facilitated by its being freely available, and its
publication in full in Microsoft Excel format. Since 2007 the review has
been produced in collaboration with the Centre for Energy Economics Research
and Policy (CEERP) at Heriot-Watt University, Scotland (Heriot Watt
University, 2017).
Energy data are sourced directly from countries, although there is little
documentation of specific sources. In the most recent edition, data were
reported for 80 separate countries in addition to further regional
groupings, from 1965 to 2017 (BP, 2018).
It appears that emissions were first included in the second release of the
2009 edition. Prior to the 2016 edition, emissions of CO2 were
calculated simply using a single emission factor for oil, gas, and
coal each, taking no account of consumption for non-combustion purposes (e.g.
bitumen). From 2016 this has been revised to use the default emission
factors for each product type from the IPCC 2006 Guidelines, with biofuels
assumed to be carbon neutral. In addition, non-combusted energy is now
removed using shares from the IEA's World Energy Balances (BP, 2017). The main consequence of
this change is a decrease in emissions from oil in particular, because of
high non-fuel use of oil.
BP provides national total emissions, without a breakdown by fuel type or
sector. International bunker fuels are not separately reported but included
in national emissions. Emissions from venting and flaring are excluded,
while own use (e.g. on oil rigs) is included (CEERP, personal communication, February 2019).
In house, BP have energy data at a more disaggregated level than those
reported (coal, oil, and natural gas), and emissions are calculated from
these more disaggregated data. While oil is divided into a number of
petroleum products, coal is divided into hard coal and brown coal (coke is
assigned to hard coal), and gas is just natural gas (CEERP, personal communication,
February 2019). This means that both the non-fuel-use (NFU) shares and the
emission factors are applied at these disaggregated levels. A consequence of
this is that emissions from oil and natural gas should lie close to the IEA's
estimates, with differences deriving from differences in source energy data,
while coal might deviate because of a lower level of disaggregation. The NFU
share from the most recent IEA data year is used in cases where BP data
extend beyond the period of the IEA data (CEERP, personal communication, February 2019).
Because BP's dataset is the first to come out with the previous year's
energy data, growth rates derived from its data can be used to extend
emissions datasets (Myhre et al., 2009), and this is done by CDIAC,
the GCP, EDGAR, and PRIMAP-hist.
BP's global emissions estimate is calculated as the sum of all country
estimates.
BP provides no quantitative assessment of uncertainty associated with its
emissions dataset.
Climate Analysis Indicators Tool (CAIT)
The World Resources Institute (WRI) developed CAIT, collating data from
other emissions datasets. The 2015 edition included 185 countries
(WRI, 2015), with emissions by sector for 1990–2014, and country
total emissions for 1850–2014. A new edition released in December 2019
extended the time series to 2016 using the same methodology.
For CO2, the IEA's (International Energy Agency) sectoral approach
emissions estimates are used directly for the 135 countries covered by that
dataset, starting from 1971. CDIAC is used from 1850 to 1970 for all
countries – with estimates prior to 1850 deemed to have insufficient
geographic coverage – and from 1971 to 2011/2012 for countries not present
in the IEA's dataset. Because Lesotho's data in CDIAC begin only in 1990, CAIT
uses EIA (US Energy Information Administration) data for Lesotho for 1980 to
2012. EIA data are also used for 2012 for all countries for which 2012 was
not present in either IEA or CDIAC data, potentially introducing
discontinuities. EIA data are also used for emissions from flaring. UNFCCC
inventories are not used in the main dataset because of their limited
geographic coverage but are presented separately.
The WRI refers readers to the underlying data sources for information on
uncertainty and makes no assessment of uncertainty in their assembled
dataset.
Community Emissions Data System (CEDS)
The Community Emissions Data System
(Hoesly et al., 2018) is
intended to be an open-source emissions data production system, although the
full system requires access to the IEA's energy data. It produces annual
national, sectoral, and monthly gridded emissions of a number of greenhouse
gases and pollutants, including CO2. CEDS estimates are to be used in
the sixth round of Coupled Model Intercomparison Project Phase 6
(CMIP6) for climate models and were initially limited to anthropogenic aerosol and aerosol and
ozone precursor compounds (Smith et al., 2015) but have since
expanded. In addition to gridded estimates, CEDS also produces country-level
estimates, which are those discussed here.
Emissions of CO2 are primarily derived from the IEA's World Energy
Statistics in physical units, using emission factors from CDIAC and the EIA.
Cement emissions are taken directly from CDIAC. Estimates for countries for
which official (or near-official) estimates are available are scaled to
those official estimates during the periods they are available; this
scaling maintains the proportions of any greater sectoral disaggregation
available in the IEA energy data.
Emission factors from CDIAC are applied for coal and natural gas combustion,
from Boden et al. (1995) (and therefore also Marland and
Rotty, 1984; Marland and Boden, 1993). For China a lower coal oxidation
factor was used, based on specific research there (Liu et
al., 2015). For liquid fuels (heavy, medium, and light oils) and coal coke,
emission factors are taken from the EIA. Emission factors are modified by
fuel-specific fractions oxidised, following CDIAC's documented methodology.
CEDS uses an independent estimate of international marine bunker fuel
emissions, combining estimates from several sources that have used bottom-up
methods based on ship activity and fuel consumption rather than on reported
sales of bunker fuels. CEDS' estimates of these emissions in recent years is
more than 50 % higher than those reported by the IEA in some years (Fig. S28). There is however the potential for double counting of emissions here
based on the possibly incorrect assumption that underestimated emissions
from international marine bunkers means that those emissions are omitted in
other sources, rather than that they are misallocated. See the Supplement for more discussion of bunker fuels.
Emissions calculated using IEA energy data are “default emissions” for
1960/71–2014. These are then scaled to EDGAR, then to “national
inventories” where available, and then extrapolated historically using CDIAC –
with some minor corrections to CDIAC's data – and proxy activity data. For
China, the emissions dataset MEIC (Li
et al., 2017) is considered a national inventory and China's emissions are
scaled to MEIC for the years 2008, 2010, and 2012.
CEDS reports a breakdown of emissions by fuel type at the global level, but
not at the national level, due to rights restrictions tied to the energy data
from the IEA.
CEDS is one of two global CO2 datasets (the other being EDGAR) that
include global estimates of emissions from all carbonate decomposition, not
just cement production, and thereby covers all fossil carbon sources.
CEDS' global emissions estimate is calculated as the sum of all country
estimates. The period covered by the dataset is 1750–2014.
Hoesly et al. (2018) discuss
uncertainty at some length, but quantitative estimates of uncertainty have not
been completed yet.
CDIAC
The emissions dataset of the Carbon Dioxide Information Analysis Center
(CDIAC) at Oak Ridge National Laboratory has been widely used, and some
aspects of its construction methodology were incorporated into the Tier 1
approach in the first IPCC Guidelines
(Haukås et al., 1997). The IPCC's Fifth
Assessment Report used CDIAC's emissions estimates when reporting both
long-term and short-term emissions trends (Ciais et al., 2013).
The dataset has a long heritage (see Sect. 2), and
its long pedigree and long time series with a consistent methodology are
probably the core reasons the CDIAC dataset remains widely used.
The CDIAC dataset has been updated annually, with the most recent release in
2019 with data for 1751–2016. In 2016 it was announced that the US
Department of Defense would be withdrawing funding for CDIAC, throwing the
dataset's future into doubt, but it has since been taken up again by
Appalachian State University (ASU). The 2018 and 2019 releases were made
available on the ASU website (Boden et al., 2018; Gilfillan
et al., 2019), and plans are in place to continue regular updates in future
(Gregg Marland, personal communication, April 2019).
CDIAC's estimates are primarily derived from UN energy data, which in more
recent years were in most cases identical to IEA data, except for the UN's
addition of data for a number of small countries (see
Fig. 1). The emissions data include estimates in
five categories: solid, liquid, gas, cement, and flaring, with cement
emissions derived from USGS cement production data (USGS, 2020).
Separate methods are used to derive global and national estimates, and these
methods have evolved since the dataset's documenting articles. For global
estimates, CDIAC uses energy production data rather than consumption data
based on the assumption that consumption data are more uncertain because
they rely on more uncertain international trade data. Through the 2018
edition, global energy production was not adjusted for stock changes, but in
the 2019 edition an adjustment for global stock changes was introduced for
historical data back to 1992 in light of very high coal stock changes in
2016 (Gregg Marland, personal communication, August 2019). A fixed fraction of 6.7 %
of liquid fuels is assumed to be “stored” (not oxidised) following Table 8
of Marland and Rotty (1984).
For national emissions, apparent gross energy consumption is calculated from
production plus imports, minus exports, minus supply to international bunkers,
and adjusted for changes in stocks. While CDIAC used to use estimates of
oxidation rates for national estimates as well (Marland and Rotty,
1984), UN data at the national level subsequently improved and non-fuel uses
have been removed from national estimates according to reported data since
the 2009 edition (Dennis Gilfillan, personal communication, January 2020), but only for
liquid fuels. This means that emissions from, for example, natural gas used
as a feedstock in fertiliser production are included, but that those from
oxidation of, for example, petroleum coke used as a chemical reagent may not
be. Originally CDIAC converted national energy data in original units for
all three fuels directly to emissions using carbon contents from Table 13 of
Marland and Rotty energy contents for solid fuels as part of their energy
data, and these are now used.
For both global and national emissions estimates, carbon contents from Table 13 of Marland and Rotty (1984) are used for gaseous and liquid
fuels, while for solid fuels two separate carbon contents are used for hard
coal and others (Dennis Gilfillan, personal communication, January 2020).
For cement, production in tonnes from the USGS is multiplied by 0.136 g C g-1 of cement (Boden et al., 1995). A number of authors have raised
questions about the accuracy of CDIAC's cement emissions estimates,
particularly for China (e.g. Lei, 2012; Ke et al., 2013; Liu et al.,
2015). Andrew (2019) discussed the reasons for this method producing
inflated emissions estimates, and CDIAC is actively pursuing solutions
(Gregg Marland, personal communication, August 2019).
Other emissions from decomposition of carbonates are not included, but EDGAR
data indicate that process emissions in cement production amounted to 78 %
and 80 % of global carbonate emissions in 2014 and 2018, respectively. Put
another way, non-cement carbonate emissions make up about 1.3 % of global
fossil CO2 emissions.
According to Marland and Rotty (1984), emissions from flared
natural gas starting in 1971 are calculated from data provided by the US
Department of the Interior and Department of Energy, while earlier estimates
are taken directly from Rotty (1974). Rotty (1974) had access to data on flared gas in the USA from 1935, but for other
countries they used a regression approach based on quantity of oil produced.
CDIAC also assumes that vented natural gas is oxidised within the same
calendar year because of a lack of data separating flared from vented gas
(Andres et al., 2012).
Andres et al. (2012) reported uncertainty
on global emissions as ±10 % at the 95 % confidence interval or 2
standard deviations, while this was subsequently updated by
Andres et al. (2014) to ±8.4 % at 95 % or 2σ.
The Emissions Database for Global Atmospheric Research (EDGAR)
The Netherlands Environment Agency (PBL) published the first version of
EDGAR in 1995, limited to emissions from aviation, spatially distributed on
a 5∘×5∘ grid for the year 1990 (Olivier,
1995). Version 2.0 was published the following year on a 1∘×1∘ grid for 1990, with sectoral, grid, and per-country data
(Olivier et al., 1996, 1999).
EDGAR is developed and maintained by the Joint Research Centre of the
European Commission, with continued input by PBL, and is used by the IPCC
(e.g. Blanco et al., 2014). The methodology is fully documented by
Janssens-Maenhout et al. (2019), describing v4.3.2,
presenting emissions for 1970–2012.
The EDGAR CO2 emissions database is released in more than one format,
with the fully disaggregated dataset updated less frequently. A “Fast Track”
version is produced every year, partly using extrapolation based on activity
data, and is released at a much more aggregated level of detail. The most
recently published version at the time of writing was v5.0_FT2018, used by Crippa et al. (2019), for which the
publicly available version includes total fossil CO2 (fossil fuels and
carbonates) for five sectors, 208 countries, plus bunker fuels, for
1970–2018. The more complete v5.0 provides estimates for 35 sectors, 231 countries, plus bunker fuels, for 1970–2018; data are not provided by fuel
type.
Emissions from fossil fuels are derived using the IPCC Tier 1 approach
according to the 2006 IPCC Guidelines from the IEA's energy data (v4.3.2
used the IEA's 2014 edition, while v5.0_FT2018 used the 2017
edition), and for this reason they are identical to those in the IEA's emissions
database in most years, noting that – for reasons of timing and consistency
– EDGAR's release relies on previous years' energy data releases from the IEA,
and later years are extrapolated using growth rates from BP data (Monica Crippa, personal communication, January 2020). Carbonate emissions are largely based
on production data from the USGS in addition to extrapolation using activity
data such as crude steel production (Crippa et al.,
2019).
EDGAR is one of two global CO2 datasets (the other being CEDS) that
include global estimates of emissions from all carbonate decomposition, not
just cement production, and thereby covers all fossil carbon sources.
EDGAR's global emissions estimate is calculated as the sum of all country
estimates.
EDGAR estimates uncertainty on CO2 emissions by assigning 2σ
values that vary by country/region and time (Table 2b in Janssens-Maenhout et al., 2019), and this is in active development.
Global uncertainty on CO2 emissions, assuming all errors are
independent, is calculated to be ±9 % at 2σ.
US Energy Information Administration (EIA)
The EIA is a US federal statistical agency formed in the wake of the energy
crises of the 1970s to collect and disseminate energy information
(EIA, 2019c). As with BP and the IEA, its primary concern is with energy,
but it thereby has all the data required to produce estimates of CO2
emissions from energy consumption. The EIA collects international energy data
directly from a large number of sources (EIA, 2020).
The dataset “Total Carbon Dioxide Emissions from the Consumption of Energy”
is part of their International Energy Statistics (EIA, 2019a). The product
was in “beta” from May 2015 through January 2020, when the beta designation
was removed (EIA, 2020), but it is not clear what update frequency is
intended. At time of writing the data covered the period 1980–2016 for 228
countries and territories.
In the beta version, the EIA's emissions estimation methodology was stated
to be documented in a 2008 report (EIA, 2019b), although this document
specifically deals with emissions in the US and makes no mention of
international emissions (EIA, 2008). The documentation of the 2020
version of the international dataset at the time of writing is incomplete but
appears also to describe only estimates for the US, and there appear to be
some problems with the dataset (see Supplement). The EIA is
working towards a new process that will include more transparency and closer
links with the data used in its International Energy Outlook (Perry Lindstrom, personal communication,
March 2019).
The EIA's international energy dataset does not report non-fuel uses, so the
(beta) methodology for estimating emissions first adjusts for these. Carbon
in natural gas used for manufacturing nitrogenous fertiliser is assumed
emitted, otherwise all non-fuel use is assumed to be sequestered. Coke use
in metallurgy is taken to be combusted, rather than a non-fuel use.
In the beta release, national emissions included bunker fuels along with
flared natural gas and vented CO2, with emissions from flared gas
reported separately. In the new version released in 2020, emissions from
flared natural gas are no longer reported separately and are included in
the natural gas emissions category.
The EIA provides no quantitative assessment of uncertainty associated with
its emissions dataset.
Global Carbon Project (GCP)
The GCP is an international collaboration whose main purpose is to
understand the global carbon cycle. Since 2005 it has released a Global
Carbon Budget, later to become an annual publication whose release is
usually timed to coincide with the UNFCCC Conference of the Parties, and one
component of this publication is a fossil CO2 emissions dataset
(Friedlingstein et al., 2019).
The GCP's fossil CO2 dataset is based primarily on CDIAC, because of that
dataset's wide use in the carbon-cycle community for many years. In
addition, the GCP prioritises data from the Annex I countries' inventory reports
to the UNFCCC, overwrites cement emissions from Andrew (2019), and
uses energy growth rates from BP and the USGS to extend time series where
applicable.
Combining inventory data with CDIAC for Annex I countries (38 % of
CO2 global emissions in 2017) introduces problems of consistency of
system boundaries. That which CDIAC includes as emissions for each fuel
category is not the same as that which the IPCC Guidelines indicate for the
energy sector (see Sect. 5.4). The GCP uses official
inventory estimates for national total (all sectors) CO2 emissions and
maps these to CDIAC's reported system boundaries to reduce inconsistencies
and discontinuities (Friedlingstein et al., 2019). For example, the GCP's
natural gas estimates include not only combustion of natural gas but also
any use of natural gas that under CDIAC's methodology is assumed to be
oxidised in the short term, such as use as a feedstock in fertiliser
manufacture. For liquid fuels, this includes not only combustion of
petroleum products, but also incineration of plastics, similarly for coal. In
addition, the GCP includes an “other” category for CO2 emissions in the CRF
reports that are outside of CDIAC's system boundary, such as those in
quicklime production, urea application, and combustion of peat.
In some cases, the GCP overwrites other CDIAC data, for example in the case
of Norway. In 2019, the GCP used the previous year's edition of CDIAC for China
and Saudi Arabia because of apparent problems in the 2019 edition of CDIAC.
Further, in 2019 the GCP recalculated global emissions as the sum of countries'
emissions, rather than using CDIAC's global estimates (Friedlingstein et
al., 2019).
The GCP's data period is (in the 2019 edition) from 1750 to 2018, with a global
projection to 2019. The dataset is released annually and available both as
an Excel/CSV download and via a web-based interface (Global Carbon
Project, 2018). Documentation is updated annually through the “living
data” process at the journal Earth System Science Data.
The GCP assesses uncertainty on global emissions to be ±10 % at the 95 % /2σ level, after Andres et
al. (2012), with uncertainty for developed countries at ±10 % and
for developing countries at ±20 % (Friedlingstein et al., 2019).
The GCP reports uncertainties for all components at the 68 % /1σ level
(Friedlingstein et al., 2019).
International Energy Agency (IEA)
The IEA, established in 1974, is an intergovernmental organisation whose
membership draws from members of the Organisation for Economic Cooperation
and Development (OECD) (Scott, 1994). The IEA's data coverage
originally extended only to the OECD countries, and early publications were
continuations of publications made by the OECD (Scott, 1994), but in
1994 this coverage was expanded (IEA, 2019f) and today it collects energy
data for about 150 countries (Coënt, 2017). Data for remaining
countries are obtained from the UN Statistics Division (Francesco Mattion, personal communication,
February 2020).
The IEA currently has about 30 staff dedicated to energy statistics, works
directly and iteratively with national energy data providers, and partners
with regional energy data organisations (Roberta Quadrelli, personal communication,
December 2019).
There are five annual questionnaires that members of the OECD, EU, and UNECE
are obliged to return: one each for coal, oil, natural gas, electricity and
heat, and renewables. The completed questionnaires are submitted directly to
the IEA, UN, and (if European) Eurostat. For coal, which includes peat and oil
shale/sands, data are submitted in mass terms along with both net and gross
calorific values (NCV and GCV; energy contents); oil data are submitted in mass
terms along with energy in NCV; natural gas data are submitted in both
volume and energy terms in both NCV and GCV (IEA, 2019g).
Questionnaire data are supplemented with information directly from national
administrations, and, where necessary, industry (Coënt, 2017).
For all other countries, the “commodity balances … are based on
national energy data of heterogeneous nature, converted and adapted to fit
the IEA format and methodology” (IEA, 2018b, p. I.17).
The IEA has reported global CO2 emissions since the 1997 release of its
“CO2 Emissions from Fuel Combustion” report (Bamberger, 2004),
calculating emissions directly from energy data in terajoules using (since
the 2015 edition) IPCC 2006 Tier 1 methods and default CO2 emission
factors from Table 2.2 of the 2006 guidelines, with both bioenergy
combustion and non-energy use resulting in zero emissions. IEA energy and
emissions datasets exclude flaring and venting (IEA, 2019d, a).
In the 2006 IPCC Guidelines, emissions from “non-energy uses of fossil
fuels” were reallocated to IPPU from the energy sector (Eggleston,
2008, p. 15). However, some of these emissions are simultaneously non-energy
(i.e. chemical use of carbon) and energy use of fossil fuels, for example
the use of coke in iron production. Because of this, and to maintain
consistency with previous work, the IEA includes in its estimates of
emissions from fuel combustion those emissions which would be reported under
IPPU but which constitute use of energy. All such emissions occur in the
metal production industries.
The IEA uses the term flow to describe what happens to energy products, largely
categories of production, transformation, or use. Flows are presented at
different levels of detail in different datasets.
There are three variants of the IEA's emissions database, all released annually:
detailed estimates (IEA, 2019c, a) include emissions from 1960 (for OECD members, and 1971 for non-members),
148 countries/35 regions, 47 products, and 41 flows and is a paid service;
2006 guidelines (IEA, 2019a) include emissions from 1970 (for OECD members, and 1971 for non-members),
185 countries/regions, 5 products, and 13 flows and is a paid service;
CO2highlights (IEA, 2019b) include emissions from 1971, 143 countries/24 regions, and 6 flows
(total fuel combustion emissions by the sectoral approach, emissions by
coal, oil, gas, and emissions from aviation and marine bunkers). A limited
sectoral breakdown is available for the most recent data year. It is freely
available.
The IEA has three main definitions for total emissions (IEA, 2018c)
(see Fig. 2).
CO2fuel combustion. This includes both that which would be included in IPCC category
1A (Energy: Fuel combustion) and any fuel combustion in IPPU. This is the
IEA's headline definition.
CO2sectoral approach. This includes fuel combustion only in IPCC category 1A
(Energy: Fuel combustion). This excludes certain emissions in the metals
industry.
CO2reference approach. This includes fuel combustion calculated using energy supply
data. These therefore include some fugitive emissions (e.g. from
refineries), in addition to differing by “statistical differences” (which
represent the mismatch between supply-side and demand-side data). In general
the IEA expects their reference approach emissions to be an overestimate of fuel
combustion emissions (IEA, 2018c).
Schematic explanation of the IEA's emissions estimate totals. The IEA
reports emissions from fuel combustion, using a reference approach, based on
energy supply-side data (production, trade, stock changes), and a sectoral
approach, based on energy demand-side data (largely sales). The latter is
also subdivided between emissions that fall under the IPCC's energy sector
and those that fall under the industrial processes and other product use
(IPPU) sector (largely in the metal industry).
The process of data harmonisation and training actively driven by the IEA
with its close interaction with national agencies helps to develop skills
and processes, directly benefitting other organisations that collect
directly from these national agencies.
The IEA's global emissions estimate is calculated as the sum of all country
estimates.
The IEA provides no quantitative assessment of uncertainty associated with
its emissions dataset.
PRIMAP-hist
The Potsdam Real-time Integrated Model for probabilistic Assessment of
emissions Paths (PRIMAP) historical emissions dataset (PRIMAP-hist), led by the
Potsdam Institute for Climate Impact Research (PIK), is constructed based on
a prioritisation scheme from other emissions datasets
(Gütschow et al., 2016).
For fossil CO2 emissions, data sources include CDIAC, EDGAR, BP, and
the CRFs for Annex I Parties to the UNFCCC. In addition, some emissions for
non-Annex I parties were obtained from the UNFCCC's “Detailed data by
party” web interface, along with selected biennial update reports. Unique
to this dataset, data from non-Annex I parties' biennial update reports
(BURs) are included. BURs are PDF-format reports, and data presentation is
highly inconsistent. Emissions from international marine and aviation bunker
fuels are not included in PRIMAP-hist.
The dataset breaks down emissions by 17 IPCC sectors (in addition to
aggregates), which are called “categories”.
Version 2.0 was released in January 2019 and includes emissions for
1850–2016, with updated data sources in addition to the use of cement
emissions from Andrew (2019). The sector disaggregation is updated to
follow the 2006 IPCC Guidelines, and an additional time series is included
that prioritises third-party data over country-reported data
(Gütschow et al., 2019a). Version 2.1 is a minor update,
released in September 2019 (Gütschow et al., 2019b).
PRIMAP-hist includes two variants, which they call “scenarios”: “HISTCR” is
assembled by giving country-reported data (e.g. CRF, BUR) priority over data
from third parties (e.g. CDIAC, FAO), while “HISTTP” is the reverse.
UNFCCC: CRFs
Since 1996, Annex I parties to the UN Framework Convention on Climate Change
(UNFCCC) have been required to submit emissions inventories for at least the
period from 1990 to 2 years before the submission year, following a decision
made at the first Conference of the Parties (COP) (UNFCCC, 1995). At
the fifth COP in 1999, it was agreed that Annex I parties would report
quantitative inventories in both electronic form and in hard copy via a
detailed Common Reporting Format (CRF), due 15 April every year
(UNFCCC, 2000), and an Excel tool for this purpose was made available
by the UNFCCC Secretariat in January 2000 (Temertekov et al.,
2003). In 2019, 45 countries submitted, in addition to the EU's combined
submission, since the EU is also a party to the UNFCCC in its own right.
At the fifth COP in 1999 it was also agreed that Annex I Parties'
inventories would be reviewed annually by expert review teams, primarily a
desk-based review, with in-country reviews every 5 years (UNFCCC,
2000). Inventories are assessed according to five general criteria:
transparency, consistency, comparability, completeness, and accuracy
(UNFCCC, 2014). Activity data are compared with data from “relevant
external authoritative sources” (p. 112, UNFCCC, 2000), which
include the International Energy Agency, Food and Agriculture Organization of
the United Nations, World Bank, Montreal Protocol, and UN Statistical
Division (Olsson, 2008).
CRF data files are generated using software developed by the UNFCCC
Secretariat following the structure presented by the IPCC Guidelines (see
Sect. 4) but are in a relatively poor format for
comprehensive analysis, with the 2019 edition's dataset spread over 132 050
spreadsheets in 1390 separate Excel files (UNFCCC, 2019c). However,
Jeffery et al. (2018) have produced a flat-record format
dataset from the Excel files, updated by Gütschow et al. (2020), and data available through the UNFCCC's online “detailed data by
party” interface (UNFCCC, 2019d) have been re-packaged in machine-readable
format by Gieseke and Gütschow (2020), facilitating more
widespread analysis. Revised inventories are typically submitted several
times through the year as corrections are made. Revisions from year to year
of historical estimates do sometimes result in significant changes (see
Supplement).
Energy emissions are estimated using a bottom-up, demand-side, energy
consumption approach, called the sectoral approach. In addition, a reference
approach, using coarser, national-level, supply-side data, is used to produce
a cross check. The reference approach serves as a quality check using
somewhat independent data and a simplified methodology. Parties are required
to compare the two approaches in their inventory report (UNFCCC,
2014). In countries that take part in the EU Emissions Trading System (ETS),
considerable data are sourced from detailed company-level ETS reporting.
Non-Annex I parties (parties to the UNFCCC that are not listed in Annex I of
the convention treaty text) are requested to submit national communications
(NCs) and biennial update reports (BURs), for which the requirements are less
stringent than the CRF. In particular, most of these reports do not include
time series of emissions, but rather a single year.
Each Annex I party estimates uncertainties associated with each emissions
estimate, and aggregated uncertainties for totals, following the uncertainty
guidelines provided by the IPCC (Frey et al., 2006).
Henceforth in this work I will analyse Annex I emissions inventories and
refer to them as “CRF”.
Other datasetsHYDE
The IMAGE 2 “hundred year” (1890–1990) database of the global environment
included historical CO2 emissions for 13 regions (Klein
Goldewijk and Battjes, 1995, 1997), but the CO2 data are no longer
updated (Kees Klein Goldewijk, personal communication, February 2019). It calculated
emissions for 1860–1949 directly from energy production data (Etemad and
Luciani, 1991). HYDE's documentation makes no mention of adjusting for
energy trade, which Etemad and Luciani (1991) did not include. For the
period 1950–1990 emissions were taken directly from
Marland et al. (1994).
MATCH
The Modelling and Assessment of Contributions to Climate Change (MATCH)
expert group was established by the UNFCCC in 2001 to generate a historical
emissions time series in the wake of Brazil's proposal to include historical
emissions in negotiations (Höhne et al., 2011; MATCH, 2019). It was
updated by den Elzen et al. (2013) with data from EDGAR but is
no longer updated.
Comparison of some important general characteristics of nine
emissions datasets, with green indicating a characteristic that might be
considered a strength. The partial use of IPCC default emissions factors by
UNFCCC CRFs is considered a strength because when default factors are not
used more accurate country-specific factors are used.
The dataset included emissions from energy and industry (CO2, CH4, N2O), agriculture and waste (CH4, N2O), and land-use change and
forestry (CO2). Emissions were collated from other datasets with the
following order of prioritisation:
UNFCCC submissions (Annex I: 1990–2004, Non-Annex I: 1994 and earlier where
available);
IEA CO2 from fuel combustion 1970–2004 + cement emissions from CDIAC;
US EPA 1990–2005 for CH4 and N2O;
CDIAC 1751–2003;
EDGAR/HYDE 1890–1990, all sectors, 17 regions;
MNP/RIVM IMAGE 2.2: 1970–2100, all gases, all sectors, 17 regions.
Summary of selected datasets
Table 2 provides an overview of some general
characteristics of nine emissions datasets. Primary source indicates whether
or not the dataset relies on other emissions datasets at all or whether all
emissions are entirely derived from activity data; “Reports bunkers
separately” refers to whether emissions from international bunker fuels are
explicitly reported or included in totals; “By fuel type” indicates whether
the dataset publicly reports a breakdown by fuel type (e.g. solid, liquid,
gas).
Comparison of emissions datasets
In this section I turn to a quantitative comparison of the major datasets
reported above and exploration of the specific reasons for those
differences. Data used to generate Figs. 3–18 are available at
https://doi.org/10.5281/zenodo.3687042 (Andrew, 2020).
Global emissions
Total global CO2 emissions across the datasets range from 32.4 to
36.3 Gt in 2014, the most recent year for which all datasets are represented
(Fig. 3). But these datasets have varying system
boundaries beyond the energy sector, with inclusion of carbonate emissions
varying from none to all and inclusion of non-energy fossil fuel emissions
varying from some to all. In addition, three datasets include the small
amounts of CO2 emitted in the agriculture sector.
Annual global CO2 emissions from various sources: total
emissions. Sources: see Table 1.
By selecting subsets of data, it is possible to bring the system boundaries
closer together (Fig. 4). This reconciliation is
limited by the breakdowns available in each dataset: with CDIAC and the GCP,
cement emissions are removed; with EDGAR, CEDS, and PRIMAP-hist, IPCC sectors
1, 2B, 2C, 2D, and 2H are selected
Sector 2A includes only
emissions from carbonate decomposition, and sectors 2E, 2F, and 2G include
only non-CO2 emissions; 2H was only present in CEDS and PRIMAP-hist.
;
and BP, the IEA, and the EIA do not permit further adjustment. But with this
adjustment to more similar system boundaries, the datasets range in 2014
from 32.4 to 34.2 Gt, with the EIA being a clear outlier at 35.4 Gt. Thus, the
range of these emissions estimates declines from 3.9 to 3.0 Gt, or
considerably further to 1.7 Gt if the EIA is considered an outlier.
Annual global CO2 emissions from various sources using
similar system boundaries.
To explore these differences in detail, Fig. 5
compares global emissions in the year 2014 by fuel type across those
datasets that provide such a breakdown, in addition to more aggregated
emissions from BP and EDGAR. For BP, the method description allows for
emissions from natural gas to be calculated from BP's energy data, but the
data for solid and liquid fuels are insufficiently disaggregated to allow
replication of BP's emissions calculation method for those fuels. The year
2014 is chosen to maximise the number of datasets that can be compared.
Fugitive emissions are those from venting of CO2 and flaring of
methane.
For solid fuels, emissions from CDIAC, the GCP, and the IEA agree to within 0.2 Gt,
but the EIA's are substantially higher, requiring further investigation (Sect. 6.1.1). CEDS' emissions from solid fuels are very
low as a result of this dataset only including energy emissions in the
breakdown by fuel type: process emissions such as use of coal in the iron
and steel industry are included in others. The sum of solid- and liquid-fuel emissions in BP is similar to those of CDIAC, the GCP, and the IEA.
There is a large spread of about 1.1 Gt in emissions from liquid fuels, from
10.9 Gt in the IEA to 12.0 Gt in the EIA and CDIAC. CDIAC and the GCP are expected to
have higher emissions in liquid fuels than the IEA because they include
additional emissions from liquid fuels not used as energy sources, but the GCP
sources some emissions from Annex I parties' official reports: for Annex I
countries, emissions from liquid fuels are about 360 Mt CO2 higher in
the CRFs than in the IEA, and the GCP's reallocation of emissions in Annex 1
countries' IPPU sector to liquid fuels contributes about a further 210 Mt
(Fig. 7). CDIAC uses energy production data to
estimate global emissions, and these are higher than energy consumption data
because of statistical differences, particularly in liquid fuels, where the
sum of CDIAC's country-level estimates comes to 10.4 Gt CO2 in 2014,
much lower than its global estimate of 12.0 Gt.
Emissions from natural gas range from 6.3 Gt in BP to 7.0 Gt in the EIA. The
somewhat higher emissions in the EIA, CDIAC, and the GCP result from inclusion of
eventual emissions from use of fertilisers derived from natural gas, and
possibly other feedstock uses of natural gas that result in CO2
emissions. The EIA also includes flared (fugitive) natural gas emissions in its
natural gas category, contributing about 0.2 Gt.
CDIAC includes only cement emissions in carbonates, but according to
Andrew (2019), these are inflated. Since CEDS uses cement emissions
estimates directly from CDIAC, these are also inflated, and its others
category also includes oxidation of all fossil fuels used as feedstocks.
EDGAR's v5.0 release introduced an error in cement emissions,
unintentionally inflating them by more than 20 % for the years 2016–2018
(Jos Olivier, personal communication, December 2019). BP, the EIA, and the IEA do not include
emissions from carbonates.
Comparison of global CO2 emissions for the year 2014 from
seven datasets, broken down by fuel type. BP and EDGAR do not report
emissions by fuel type.
Underlying energy data
All CO2 emissions estimates are ultimately largely derived from energy
data, and this is therefore a key potential source of deviation. It is also
a large potential source of uncertainty, relying as it does on national
administrations to report correctly.
Figure 6 compares global energy consumption from
three energy datasets (IEA, BP, EIA), demonstrating general coherence but
with some deviations. These three datasets are used to estimate emissions by
BP, the EIA, CEDS, EDGAR, and the IEA (see Fig. 1). CDIAC
and the GCP rely on UN energy data, which are likely to be very similar to IEA
energy data but are not freely available.
To make the three datasets comparable, it was necessary to convert them all
to the same units, exajoules of net calorific value (NCV). The EIA presents all its
energy data in gross calorific value terms (also known as the higher heating
value), and to convert to NCV I use the three simple factors suggested by
the IPCC: coal 0.95, oil 0.95, and gas 0.90 (Eggleston et al.,
2006).
The figure shows a significant deviation of the EIA's coal energy data after
2010 from the other two sources here and something of a diverging oil trend
between the three after about 2005.
BP's oil consumption numbers lie slightly below those of the IEA and EIA
over recent years.
Global energy consumption estimates from three sources. The EIA
reports energy production in gross calorific value (GCV), and this has
been converted to net calorific value (NCV) using the standard factors of
0.9 for natural gas and 0.95 for oil and coal.
Further analysis shows that the reason for the significant divergence in
coal consumption between EIA and the other two sources here is largely in
China and appears to result from a difference in treatment of the waste
products from washing of coal, which are assumed to have zero energy content
by both the IEA and China's National Bureau of Statistics, but not by the
EIA (see Supplement for a full analysis). In 2012, the EIA's
reported coal consumption for China is 9.1 EJ (11 %) higher than the
IEA's, contributing directly to the EIA's estimate of global emissions from
solid fuels being 1.5 Gt CO2 higher in that year.
Annex 1 countries
The Annex I parties to the UNFCCC contribute about 40 % of global fossil
CO2 emissions, and they all submit official estimates of territorial
emissions (Sect. 5.10), allowing comparison of
official estimates with third-party estimates. Here I compare the emissions
datasets for 38 of these countries, allowing inclusion of eight datasets in
the comparison.
The official estimates (CRF) indicate total emissions of 14.0 Gt CO2 in
these 38 Annex I countries, excluding emissions from international bunker
fuels (Fig. 7). The CRF others category here
includes all carbonate emissions as well as non-energy uses of fossil fuels.
The GCP's total matches the CRF total exactly, by design. The totals from EDGAR
and the EIA are also very close, although the EIA's estimate excludes emissions from
carbonates. EDGAR has almost the same system boundary as the CRFs.
Emissions from solid fuels range between 4.1 Gt in CRF and 4.4 Gt in
the GCP. The GCP reallocates some of CRF's others emissions back to their original
fossil fuels. Both CDIAC and the IEA also include some non-energy emissions in
this category, for example from the use of coking coal in the iron and steel
industries.
The spread for liquid-fuel emissions is much higher, at 1.1 Gt CO2, with the
EIA's estimate 24 % higher than the IEA's. About 0.5 Gt of this may be
because the EIA includes bunker fuels in with liquid fuels, and this approximately
matches the difference between the EIA and GCP. The GCP takes CRF estimates and
reallocates them, which added about 0.2 Gt to liquid-fuel emissions in this
manner. The IEA's estimate is over 0.3 Gt lower than the official CRF estimates,
more than might be expected, and Sect. 6.3
demonstrates that this is largely a difference in estimates for the USA.
Emissions from gaseous fuels range from 3.7 Gt CO2 to almost 4.0 Gt.
Much of this is a result of the inclusion of emissions from feedstock uses
of natural gas, known to be the case in CDIAC, the GCP, and the EIA. Further, the EIA
includes fugitive emissions from the flaring of natural gas; the IEA does
not include fugitive emissions.
The datasets generally show consistent trends in recent years (Fig. S30).
Comparison of fossil CO2 emissions in most Annex 1 countries
for the year 2014 from eight datasets. Smaller emitters (Cyprus, Israel,
Liechtenstein, Malta) are excluded to increase the number of datasets in the
comparison.
USA
Figure 8 compares emissions estimates across eight
datasets for the year 2014. Total emissions vary here from 5.0 Gt in CEDS
and the IEA to 5.6 Gt in CRF and the GCP. It is surprising that CEDS' estimate is so
similar to the IEA's, given that CEDS includes all emissions sources while
the IEA does not, and also given that CEDS uses the IEA's energy data as a primary
source, but CEDS does not provide a fuel breakdown at the country level,
hindering further investigation. The CRF, GCP, CEDS, and EDGAR all include
all emissions sources but vary considerably in their total emissions
estimates. BP and the EIA both exclude emissions from decomposition of
carbonates and fugitive sources.
Emissions estimates from solid fuels vary by about 80 Mt, or about 5 %,
indicating relatively good agreement between the datasets. The GCP's estimate is
highest, a result of mapping some process emissions in IPPU back to the coal
used as a feedstock.
However, emissions estimates from liquid fuels vary by about 350 Mt, or more
than 15 %. According to the available documentation, the EIA's estimate
includes bunker fuels, which in the CRF are about 100 Mt. With or without
bunker fuels, there is a large gap between the estimates from the EIA, GCP, and
CRF and those from CDIAC and, particularly, the IEA. Liquid-fuel emissions in
the GCP are higher than those in the CRF because of reallocation of some IPPU
(others) emissions to primary fuels, but in the case of the US this makes
a difference of only 1 %. CDIAC's underlying energy data should be similar
to the IEA data, but CDIAC uses different emission factors. The divergence
in estimates of emissions from liquid fuels is significant and this warrants
further investigation.
Estimates of emissions from gaseous fuels vary by about 50 Mt, or 4 %, and
part of this is because of the reallocation of the use of natural gas as a
feedstock from IPPU back to the gas, known to be the case with both the GCP and
CDIAC and also thought to be the case with the EIA.
The datasets generally show consistent trends in recent years (Fig. S31).
Comparison of CO2 emissions estimates in the USA from eight
datasets, 2014.
Figure 9 compares the sectoral approach estimates
from the CRF and the IEA, showing a significant gap between the estimates from
liquid fuels. The gap varies over time, and reaches almost 300 Mt, or
15 %, in later years.
Comparison of emissions estimates for the USA from the IEA and UNFCCC.
Natural gas emissions are usually well estimated in all datasets because they
usually describe a single energy product, and because energy content
exhibits limited regional variation. Oil products, in contrast, are highly
variable, with significant variation in energy contents.
Figure 10 focusses on emissions from US consumption
of liquid fuels, comparing estimates from the EIA and IEA with those in the
CRF (by the US EPA, who submits the CRF). Clearly the approach used (sectoral
vs. reference) has little to say in the differences between data sources. The
gap between the two reference approach estimates is about 200 Mt CO2 in
later years, and between the sectoral approach estimates it is about 300 Mt,
as noted above.
Initial analysis (not shown here) indicates several potential reasons for
this gap.
The EPA's use of US-specific emissions factors rather than IPCC default factors
adds about 20 Mt CO2.
Inclusion of additional overseas territories in the CRF could add up to 20 Mt CO2, and this is probably largely limited to liquid fuels.
The EPA's imports of crude oil appear to be 2 %–3 % higher than those
reported by the IEA in energy units. This is probably due to classification
methods, such that imports of some commodities are included or excluded in
the two datasets as crude oil.
Different energy contents and densities for products are used: the EIA reports
most liquid-fuel data only in volumetric units.
Mapping from the EIA's product categories to categories used for emission
factors appears to be different. For example, the EIA oil product motor
gasoline blending components is mapped to other liquids by the EPA but to motor
gasoline by the IEA, and these two products have both different energy contents
and different emission factors.
Further resolution of this gap would be possible knowing the conversion
factors used by the IEA from the EIA's physical units to energy units, but these
were not available.
Comparison of estimates for CO2 emissions from US liquid
fuels, from the EIA, CRF (EPA), and the IEA.
European Union
For the combined 28 members of the European Union, total CO2 emissions
vary across the datasets from 3.2 Gt in the IEA to 3.7 Gt in the EIA, a range of
about 15 % (Fig. 11). However, the EIA is known to
include at least some bunker fuels, significantly increasing its estimate of
emissions from liquid fuels; BP also includes bunker fuels in its estimates.
Looking only at the datasets that include all sources of CO2 emissions,
but exclude bunker fuels (i.e. CEDS, EDGAR, GCP, CRF), these exhibit a much
smaller range from 3.4 to 3.5 Gt. While the total of the GCP is by design
equal to that of CRF, and therefore does not provide additional information,
the estimates by CEDS and EDGAR are independent and lie very close to the
CRF total.
When looking at total emissions from fossil fuel combustion, those in the IEA,
CDIAC, and CRF are very close, varying by only 41 Mt out of about 3.1 Gt. As
noted previously, the GCP's estimates by fossil fuels are larger because they
reallocate some IPPU emissions sources back to fuel categories. The “all
fuels” totals for CEDS and EDGAR are all emissions deriving from fossil
fuels, in both energy and IPPU, and these totals are very similar to those
in the GCP.
Emissions from both solid and gaseous fuels are very similar across
datasets. Solid-fuel combustion emissions estimates (excluding the GCP) vary by
only 26 Mt, about 2.5 %. Gas fuel combustion estimates vary by 40 Mt,
about 5 %, with CDIAC, the GCP, and the EIA all known to include on-farm emissions
from fertilisers made from natural gas, excluded by the IEA and CRF from fuel
emissions (the IEA excludes these entirely, while CRF reports these in the
agriculture sector, in the figure included with others).
The variation in emissions from liquid fuels is much larger. However, in the
case of the GCP this is a result of the reallocation of IPPU emissions to liquid
fuels, and in the case of the EIA there are bunker fuels included, increasing
both estimates. Where these factors do not occur, in the IEA, CDIAC, and CRF,
the range of emissions from liquid fuels is only 53 Mt, about 4 %.
There is substantially higher agreement across datasets for the EU than for
the USA. One possible explanation for this is that the data methods used by
EU members states lie very close to those used by the IEA, a result of
inter-country collaboration leading to learning and standardisation of
methods, coordinated via regular reporting to the European Environment
Agency (EEA). This includes the use of common units of measurement, such
that no third party need make assumptions about energy contents of different
fuel types. As a result, the underlying energy data used by each emissions
dataset are much more uniform than those used in generating US emissions
estimates.
The differences between datasets in 2014 are consistent across time, apart
from the EIA, which has diverged since 2006, as shown in Fig. S32.
Comparison of fossil CO2 emissions from eight sources for
the European Union (28 members).
China
China is not an Annex I party to the UNFCCC, and therefore it has different
reporting lines and requirements. Furthermore as an emerging economy China is
expected to have less institutional experience and capacity in national
statistics. It also has a very large population and a rapidly developing and
changing economic structure, making the collection of accurate statistics
across millions of enterprises a significant challenge
(Korsbakken et al., 2016). China does not report annual
emissions but has commenced biennial reporting to the UNFCCC, each report
for a single year. In lieu of a CRF, here I compare the global datasets with
the official second biennial update report (BUR) from China, which presents
estimates for the year 2014 (Anonymous, 2018).
At first glance there is substantial variation in estimates from different
datasets, with total CO2 emissions ranging from 9.2 to 10.8 Gt, a
variation of about 15 %. There are two main reasons for this large
variation. The first is that the EIA's emissions from solid fuels are
significantly larger than other datasets, as was discussed in Sect. 6.1.1, with further details given in the
Supplement. The second is that China has very large emissions
from sources other than combustion of fossil fuels, and system boundary
differences therefore explain most of the difference between the lower
estimates (BP and the IEA) and the higher estimates.
When looking at total emissions from fossil fuels, and excluding the EIA, the
estimates vary much less. The total emissions from CEDS and CDIAC are very
similar to those in the BUR, with EDGAR somewhat higher and the GCP a little
lower. The GCP's fossil fuel emissions here are by definition identical to
CDIAC, so their similarity does not tell us anything about independent
consistency across datasets, but it uses an independent lower estimate of
emissions from cement production. While CDIAC's fossil fuel emissions are
very similar to the fuel emissions reported in the BUR, the BUR's others
emissions here include about 410 Mt from non-energy use of fossil fuels.
Moreover, CDIAC excludes emissions from decomposition of carbonates outside
of cement production, which have been estimated to be about 450 Mt in 2014
(Cui et al., 2019). Both CDIAC and CEDS use what are thought to
be inflated estimates of China's cement emissions (see Sect. 5.4), but their emissions sources other than fossil
fuels are perhaps coincidentally very similar to those in the BUR.
There are only two estimates of emissions from coal, again excluding the
EIA: 7425 Mt from CDIAC and 7591 Mt from the IEA, a difference of 166 Mt, or
about 2 %. For liquid fuels, the estimates range from 1173 Mt in the IEA to
1500 Mt in the EIA, and it is unclear why the EIA estimate is so high. Reported
bunker fuels are surprisingly small in China, given the size of its
international trade. The BUR reports emissions of only 51 Mt CO2
(Anonymous, 2018), insufficient to explain the EIA's high liquid-fuel
emissions. For natural gas, estimates range from 291 Mt in BP to 355 Mt in
the EIA.
The absolute range of estimates has grown as emissions have grown, and there
are some considerable differences in trend in some recent years (Fig. S33).
Comparison of fossil CO2 emissions from seven sources for
China.
Other illustrative cases
Here I present several cases to compare datasets and explore the specific
reasons for differences between estimates. For this I use estimates broken
down by major fuel category. As before, the year 2014 is chosen to maximise
the number of datasets in the comparison.
BENELUX
The BENELUX group of countries consists of Belgium, the Netherlands, and Luxembourg
(Fig. 13). For emissions from solid fuels, the
datasets agree well, with a maximum variation of 8 % around the median.
The outlier is the GCP, which reassigns some of the others emissions from the
CRFs to solid fuels. For liquid fuels there are much larger variations,
notably the very high estimate presented by BP, a direct result of this
dataset's inclusion of bunker fuels (about 77 Mt) in liquid fuels. The EIA's
new (2020) dataset does not appear to include bunker fuels for the
Netherlands in 2014 but does for other years (see Fig. S26). Even so,
the EIA's estimate for liquid fuels is very high compared to liquid fuels in the
other datasets. Among the other datasets, the GCP is about 9 % higher than the
CRFs, again because of reassignment of some others to the source fossil
fuel. For emissions from gaseous fuels, one can see more of the consequences
of these different mappings, but in addition the mapping of emissions from
natural gas used in production of artificial fertilisers mapped to gas
emissions for the EIA, CDIAC, and GCP, but to others in the CRFs, and absent in
both BP and the IEA. The IEA has low others emissions because these include
only emissions from the use of fossil fuels as reagents in the iron and
steel industry, while CDIAC's low others emissions include only process
emissions from cement production. EDGAR and the CRFs include all non-energy
emissions in others, while the GCP includes only emissions from carbonates
(including cement production) and maps other process emissions back to the
fossil fuel sources. The GCP's total emissions are by design equal to those in
the CRF, while EDGAR's use of bottom-up data gives a very close match to the
CRF total. CDIAC's total is also very close in this example to the CRF, and
the IEA is slightly lower because of its slightly smaller system boundary,
particularly its exclusion of emissions from carbonates. EDGAR does not
report country-level emissions from bunker fuels.
Comparison of fossil CO2 emissions from seven sources for
the Netherlands, Belgium, and Luxembourg.
Estonia
The next example is for Estonia (Fig. 14). For
solid fuels there are significant differences, with both CEDS and the EIA having
very low estimates. Estonia uses oil shale, an energy-rich sedimentary rock,
for about 70 % of its primary energy supply, and according to the CRF its
combustion contributed about 90 % of energy emissions in 2017. IEA energy
data indicate that about two-thirds of solid oil shale is used directly to
produce electricity, with the remainder converted to liquid and gaseous
fuels (IEA, 2019e). The EIA does not include oil shale in its energy or
emissions data for Estonia, resulting in a very low estimate of total
emissions.
Comparison of fossil CO2 emissions from seven sources for
Estonia.
Iceland
Iceland has a large metal production industry, resulting in large industrial
process emissions from the decomposition of carbon anodes in aluminium
production as well as emissions in iron and steel production, and the CRF
indicates that these make up almost half of Iceland's total CO2
emissions in 2016 (Fig. 15). Non-energy emissions
from aluminium production are excluded from the system boundaries of the IEA,
CDIAC, BP, and the EIA, resulting in substantially lower total emissions from
these datasets (BP does not present data for Iceland, so is not shown on the
chart). The GCP reassigns these emissions in the aluminium industry to liquid
and solid on the assumption that the carbon used to make the anodes is
sourced 20 % from coal and 80 % from petroleum coke. The EIA, again,
includes bunker fuels in their estimate of emissions from liquid fuels, in
the case of Iceland having a substantial effect on total emissions. The
fugitive emissions indicated in the CRF are replicated in the GCP, but not
present in other datasets, and this is because these emissions are from
geothermal energy generation, which is not included in the other datasets.
For example, EDGAR's estimates of fugitive emissions appear to be based only
on fossil fuel production (Janssens-Maenhout et
al., 2019).
Comparison of fossil CO2 emissions from six sources for
Iceland.
Norway
Norway produces both oil and natural gas – about 90 % of which are
exported – and also has a significant metal industry. CDIAC's estimates for
emissions from liquid fuels are particularly high, a result of its method
being based on the reference approach, calculating apparent consumption from
production, net imports, and stock changes, rather than using reported
energy consumption. Norway's official estimates using the reference approach and
sectoral approach differ by as much as 45 % (Miljødirektoratet,
2019). Norway has repeatedly been asked by the expert review team (ERT) to
explain this difference (UNFCCC, 2019a) and has worked since 2011 to
reduce the inconsistency. As stated by the IPCC in its guidelines, “for
countries that produce and export large amounts of fuel, the uncertainty on
the residual supply may be significant and could affect the reference
approach” (p. 6.13, Treanton et al., 2006a). That the EIA's
estimate of liquid-fuel emissions is so close to those from CDIAC strongly
suggests that the EIA has also based their estimate on apparent consumption,
something that is not documented. The IEA's reported emissions from solid
fuels include use of coal as a feedstock in the iron and steel industry,
which the CRF reports under IPPU (here others). BP's lower emissions
estimate for natural gas results directly from a lower estimate of natural
gas consumption, the reason for which is unclear.
Comparison of fossil CO2 emissions from eight sources for
Norway.
Nigeria
For Nigeria, there are substantial differences between datasets. Both CEDS
and EDGAR have much lower emissions from fuel combustion than the other
datasets, and this is because they have used an earlier energy data source,
as shown by the addition of the 2017 edition of the IEA's emissions dataset for
this comparison. Liquid-fuel consumption data have since been revised
significantly upward.
The EIA's emissions from natural gas are much higher than in other datasets due
to the inclusion of flaring in natural gas, which for Nigeria makes a
substantial difference, and the EIA's natural gas emissions here are very
similar to the total of gas and fugitive emissions in CDIAC.
Comparison of fossil CO2 emissions from seven sources for
Nigeria.
Cumulative global emissions
It has been shown that for cumulative global emissions up to 6000 Gt CO2, global temperature rise can be expected to be approximately
linearly related to cumulative emissions, while cumulative emissions since
1850 are on the order of 2400 Gt CO2, including emissions from land-use
change (Friedlingstein et al., 2019; Matthews et al., 2009; Allen et al.,
2009; IPCC, 2014). This has been used to determine so-called “carbon
budgets”, i.e. the total cumulative CO2 emissions corresponding to
specific temperature targets, such as 2 ∘C
(Rogelj et al., 2016). For this it is clearly important to
have reliable estimates of cumulative global CO2 emissions.
Figure 18 shows the cumulative difference between
global fossil carbon emissions for PRIMAP, CEDS, and the GCP compared with CDIAC;
these are the four emissions datasets that have long time series. I compare
with CDIAC because that is the longest standing global emissions dataset
and is used by the IPCC to report cumulative emissions (Stocker et al.,
2013), without suggesting that it is more accurate than the other datasets
here. The differences between the GCP and CDIAC are relatively small, never more
than 10 Gt, and result partly from the GCP's calculation of global emissions as
the sum of countries' emissions while CDIAC uses an independent global
production method. CEDS deviates much more, reaching a peak of about 40 Gt
cumulative difference in recent years. This is partly explained by higher
estimates for emissions from bunker fuels, but also emissions in Europe are
higher in CEDS for 1960–1990, for reasons that are unclear. PRIMAP, whose
emissions prior to about 1970 are reported to be based largely on CDIAC,
nevertheless diverges strongly from CDIAC's cumulative emissions, reaching a
peak of almost 75 Gt cumulative difference. This amounts to about 5.1 %
more CO2 over this period.
Difference of cumulative global CO2 emissions from various
datasets from those of CDIAC, starting in 1850.
Discussion
In this article I have discussed the historical development of estimating
global CO2 emissions and summarised the methods used by the most
important current datasets. Given the complexity of the methodologies and
source datasets involved, with subtle but important differences, it is
unsurprising that documentation in some cases is difficult to follow. In
addition, some methods have changed since documentation was published,
including that of CDIAC. There is certainly room for improvement in
describing the methods and data sources used for construction of emissions
estimates.
All emissions datasets build from data on energy and other activities, such
as cement production, but it is emissions from fossil fuels and therefore
energy data that are most important. There are several major energy datasets
with global coverage: the IEA, the EIA, BP, and the UN. While some differences in
emissions datasets are due to the use of different energy datasets,
different processing methods can result in different emissions estimates
even when emissions datasets rely on the same underlying energy data.
Fundamentally, energy datasets are not independent, with all relying on the
same underlying energy data in physical units reported by national agencies,
with few exceptions. However, differences in interpreting these data,
converting to the energy units required for estimating CO2 emissions,
emissions factors used, and of course errors, all result in different
emissions estimates.
An additional core reason for differing emissions estimates is simply the
coverage of the dataset, whether geographic coverage or type of emitting
activity. Some datasets do not include emissions from the decomposition of
fossil carbonates, such as in the production of cement, while others include
international bunker fuels in national estimates. The inclusion of emissions
from non-energy uses of fossil fuels also varies across datasets.
Close comparison of emissions datasets in this article revealed that
estimates of global emissions vary considerably less when they are adjusted
to match a common system boundary: which emissions sources are included.
Excluding the apparently overestimated estimates from the EIA, and making the
best efforts to harmonise system boundaries, global emissions in 2014 varied
by 1.7 Gt CO2 (about 5 %), and in that comparison residual system
boundary differences remained. The simple range of total emissions across
datasets is therefore a poor measure of uncertainty. Many of the differences
between datasets can be explained from the methods used, the approaches used
to allocate to categories, and known system boundaries, and these can make
substantial differences between datasets for specific countries, as was
shown in the cases of the BENELUX countries, Iceland, Estonia, Norway, and
Nigeria.
That said, some discrepancies were found, such as the EIA's apparent
overestimate of China's consumption of coal energy, and the sizable
differences between estimates of US emissions could not be fully explained.
Emissions and energy datasets are complex, and errors are bound to occur,
requiring careful checks and comparison with other datasets.
Of the three large emitters investigated here (USA, EU, China), estimates of
the European Union's emissions varied the least, after known differences in
coverage were accounted for. It is expected that this results from the
considerable effort put into energy and emissions statistics in the EU,
combined with close collaboration both between countries and with the EEA
and IEA. The EU's Emissions Trading System also acts as a valuable data
source.
In contrast, the large discrepancies in estimates of emissions from liquid
fuels in the USA warrant further investigation beyond what was possible in
this article. The reporting of energy in physical units and use of gross
calorific values rather than net calorific values, among other things,
hamper a quantitative comparison between datasets.
For China there was relatively good agreement between datasets for emissions
from fossil fuels, apart from the EIA, mentioned earlier. While there is
good agreement, revisions have previously led to substantial changes in all
datasets. This problem is not limited to China, though, as the example of
Nigeria showed.
Little effort has been put into creating independent estimates of emissions
prior to 1950, with only two examples known, and no previous comparison of
these two was found in the literature. Two examples of potentially overestimated
emissions in the early 20th century were highlighted. Given the global
carbon imbalance in the middle of the 20th century, further
investigation of emissions estimates in this period is warranted.
Cumulative emissions are important for climate modelling and comparisons
with remaining carbon budgets, which tie emissions to temperature targets.
Brief analysis here of datasets with long time series shows considerable
divergence, at up to 5 % cumulative emissions over 1850–2014. Further
analysis of the causes of these long-term divergences is required given the
importance of cumulative emissions estimates.
Much of the difference between different estimates of CO2 emissions
results from differences in system boundaries. Given this, and an observed
lack of awareness of this fact in the carbon community, it behoves data
providers to find a way to be more explicit about what is and what is not
included in datasets. Simply saying “energy-related emissions”, for example,
without clarifying what is excluded is likely to leave many readers
uninformed.
Reconciling estimates of emissions from different sources is an important
goal, but these estimates are much less independent than many believe
because the energy data that underlie all estimates originally come from
single sources in each country. While it is appropriate to place a certain
amount of faith in these agencies, errors and omissions do happen, and
partially independent methods such as atmospheric inversions do have a place
in increasing our confidence in the level of global emissions. The use of
atmospheric inversion modelling, derived from satellite observations but
still requiring a priori estimates, is an active area of research.
Given the inconsistent system boundaries across emissions datasets, one
could conceive of a “carbon emissions dataset intercomparison project”, or
CEDIP, along the lines of the Coupled Model Intercomparison Project (CMIP)
and other related model comparison projects. A core part of these
intercomparison projects is the requirement that participants report model
outputs to a specified and very clear template such that the issue of system
boundary differences is removed. For example, an estimate for global total
CO2 emissions could not be reported for a dataset if it excluded
sources or countries. In effect, a CEDIP would extend the work done in this
article, allowing each data provider to submit according to their superior
understanding of their own datasets, permitting more robust comparison, and,
critically, allowing lessons to be gained such that estimates can be
improved.
The process undergone in this article is essentially one of verification.
While “true” emissions cannot be known, by comparing different datasets
methodically, differences that result from system boundaries and allocation
approaches can be highlighted and set aside to enable identification of true
differences, and potential errors. This must be an important way forward in
improving global datasets of CO2 emissions.
Data availability
Data used to generate Figs. 4–19 are available at 10.5281/zenodo.3687042 (Andrew,
2020).
The IEA World Energy Balances, World Energy Statistics, and Detailed
CO2 Emissions were used with the kind permission of the IEA, http://www.iea.org/statistics (last access: 7 November 2019), all rights reserved.
The supplement related to this article is available online at: https://doi.org/10.5194/essd-12-1437-2020-supplement.
Competing interests
The author declares that there is no conflict of interest.
Acknowledgements
I would like to express my gratitude to Jan Ivar
Korsbakken (CICERO) for his assistance in investigating the differences
between the IEA, NBS, and EIA data on Chinese coal; to Roberta Quadrelli,
Francesco Mattion, and Julia Guyon (IEA) for their assistance in
understanding IEA data; to Dennis Gilfillan and Gregg Marland (Appalachian
State University) for their assistance in understanding the methodology now
used to produce CDIAC data; to Monica Crippa and Efisio Solazzo (JRC) for
their assistance with understanding EDGAR methods; to Steve Smith
(PNNL-JGCRI) for assistance in understanding the CEDS database; to Johannes
Gütschow (PIK) for assistance in understanding the PRIMAP-hist dataset;
to the VERIFY WP5 team for useful discussions; and to Glen Peters (CICERO)
for numerous discussions over the years on the complex details of emissions
data.
Financial support
This research has been supported by the Horizon 2020 (VERIFY, grant no. 776810).
Review statement
This paper was edited by David Carlson and reviewed by two anonymous referees.
ReferencesAllen, M. R., Frame, D. J., Huntingford, C., Jones, C. D., Lowe, J. A.,
Meinshausen, M., and Meinshausen, N.: Warming caused by cumulative carbon
emissions towards the trillionth tonne, Nature, 458, 1163–1166, 10.1038/nature08019, 2009.
Andres, R. J., Marland, G., Fung, I., Matthews, E., and Brenkert, A. L.:
Geographic Patterns of Carbon Dioxide Emissions from Fossil-Fuel Burning,
Hydraulic Cement Production, and Gas Flaring on a One Degree by One Degree
Grid Cell Basis: 1950 to 1990, Carbon Dioxide Information Analysis Center
(CDIAC); Oak Ridge National Laboratory, NDP: 058,
1997.Andres, R. J., Boden, T. A., Bréon, F.-M., Ciais, P., Davis, S., Erickson, D., Gregg, J. S., Jacobson, A., Marland, G., Miller, J., Oda, T., Olivier, J. G. J., Raupach, M. R., Rayner, P., and Treanton, K.: A synthesis of carbon dioxide emissions from fossil-fuel combustion, Biogeosciences, 9, 1845–1871, 10.5194/bg-9-1845-2012, 2012.Andres, R. J., Boden, T. A., and Higdon, D.: A new evaluation of the
uncertainty associated with CDIAC estimates of fossil fuel carbon dioxide
emission, Tellus B, 66, 23616, 10.3402/tellusb.v66.23616, 2014.Andres, R. J., Boden, T., and Marland, G.: Annual Fossil-Fuel CO2
Emissions: Mass of Emissions Gridded by One Degree Latitude by One Degree
Longitude, Carbon Dioxide Information Analysis Center, Oak Ridge National
Laboratory, available at: https://cdiac.ess-dive.lbl.gov/epubs/ndp/ndp058/ndp058_v2016.html (last access: 6 December 2019), 2016a.Andres, R. J., Boden, T. A., and Higdon, D. M.: Gridded uncertainty in fossil fuel carbon dioxide emission maps, a CDIAC example, Atmos. Chem. Phys., 16, 14979–14995, 10.5194/acp-16-14979-2016, 2016b.Andrew, R.: A comparison of estimates of global carbon dioxide emissions
from fossil carbon sources, Zenodo, 10.5281/zenodo.3687042, 2020.Andrew, R. M.: Global CO2 emissions from cement production, 1928–2018, Earth Syst. Sci. Data, 11, 1675–1710, 10.5194/essd-11-1675-2019, 2019.Andrew, R. M., Davis, S. J., and Peters, G. P.: Climate Policy and
Dependence on Traded Carbon, Environ. Res. Lett., 8, 034011,
10.1088/1748-9326/8/3/034011, 2013.Anonymous: The People's Republic of China: Second Biennial Update Report on
Climate Change, available at: https://unfccc.int/BURs (last access: 10
February 2020), 2018.
Arrhenius, S.: Worlds in the Making: the Evolution of the Universe
(Translated by Dr. H. Borns), Harper & Brothers Publishers, New York and
London, 1908.Asefi-Najafabady, S., Rayner, P. J., Gurney, K. R., McRobert, A., Song, Y.,
Coltin, K., Huang, J., Elvidge, C., and Baugh, K.: A multiyear, global
gridded fossil fuel CO2 emission data product: Evaluation and analysis of
results, J. Geophys. Res.-Atmospheres, 119,
10213–10231, 10.1002/2013jd021296, 2014.Bamberger, C. S.: The History of the IEA: The First 20 Years, vol. 4,
Supplement to Volumes I, II & III, OECD/IEA, Paris, available at: https://www.iea.org/about/history (last access: 13 February 2020), 2004.
Baumann, H. and Tillman, A. M.: The Hitch Hiker's Guide to LCA: An
Orientation in Life Cycle Assessment Methodology and Application,
Professional Publishing House, 2004.Baxter, M. S. and Walton, A.: A Theoretical Approach to the Suess Effect,
P. Roy. Soc. Lond. A Mat., 318, 213–230, 10.1098/rspa.1970.0141, 1970.Björk, I., Steinbach, N., and Berglund, M.: Metodbeskrivning av
beräkning av konsumtionens miljöpåverkan – växthusgaser,
Enheten för Miljöekonomi och miljö, Statistiska centralbyrån
(SCB), available at: https://www.naturvardsverket.se/upload/sa-mar-miljon/statistik-a-till-o/vaxthusgaser/2018/metodbeskrivning-konsumtion.pdf
(last access: 14 December 2019), 2018.Blanco, G., Gerlagh, R., Suh, S., Barrett, J., Coninck, H. C. d., Morejon,
C. F. D., Mathur, R., Nakicenovic, N., Ahenkora, A. O., Pan, J., Pathak, H.,
Rice, J., Richels, R., Smith, S. J., Stern, D. I., Toth, F. L., and Zhou,
P.: Drivers, Trends and Mitigation, in: Climate Change 2014: Mitigation of
Climate Change, Contribution of Working Group III to the Fifth Assessment
Report of the Intergovernmental Panel on Climate Change, edited by:
Edenhofer, O., Pichs-Madruga, R., Sokona, Y., Farahani, E., Kadner, S.,
Seyboth, K., Adler, A., Baum, I., Brunner, S., Eickemeier, P., Kriemann, B.,
Savolainen, J., Schlömer, S., Stechow, C. v., Zwickel, T., and Minx, J.
C., Cambridge University Press, Cambridge, UK, New York, NY,
USA, available at: http://www.ipcc.ch/report/ar5/wg3/ (last access: 25
April 2015), 2014.Boden, T., Marland, G., and Andres, R. J.: Estimates of Global, Regional,
and Naitonal Annual CO2 Emissions from Fossil-Fuel Burning, Hydraulic
Cement Production, and Gas Flaring: 1950–1992, Carbon Dioxide Information
Analysis Center (CDIAC), Oak Ridge National Laboratory, Oak Ridge,
Tennessee, available at: http://cdiac.esd.ornl.gov/epubs/ndp/ndp030/ndp0301 (last access: 28 June
2017), 1995.Boden, T., Marland, G., and Andres, R. J.: Global Carbon Dioxide emissions
from fossil fuel consumption and cement production, available at: https://energy.appstate.edu/CDIAC (last access: 2 August 2019), 2018.BP: 60 Years BP Statistical Review of World Energy: 1951–2011, available at:
https://www.bp.com/en/global/corporate/energy-economics/statistical-review-of-world-energy/downloads.html
(last access: 8 February 2019), 2011.BP: Methodology for calculating CO2 emissions from energy use, available at:
https://www.bp.com/en/global/corporate/energy-economics/statistical-review-of-world-energy/co2-emissions.html
(last access: 8 February 2019), 2017.BP: BP Statistical Review of World Energy June 2018, available at: https://www.bp.com/en/global/corporate/energy-economics/statistical-review-of-world-energy.html,
last access: 14 June 2018.
Broecker, W. S., Li, Y.-H., and Peng, T.-H.: Carbon Dioxide – Man's Unseen
Artifact, in: Impingement of Man on the Oceans, John Wiley & Sons, Inc.,
1971.
Callendar, G. S.: The artificial production of carbon dioxide and its
influence on temperature, Q. J. Roy. Meteor.
Soc., 64, 223–240, 1938.Chen, H., Huang, Y., Shen, H., Chen, Y., Ru, M., Chen, Y., Lin, N., Su, S.,
Zhuo, S., Zhong, Q., Wang, X., Liu, J., Li, B., and Tao, S.: Modeling
temporal variations in global residential energy consumption and pollutant
emissions, Appl. Energ., 184, 820–829, 10.1016/j.apenergy.2015.10.185, 2016.Ciais, P., Paris, J. D., Marland, G., Peylin, P., Piao, S. L., Levin, I.,
Pregger, T., Scholz, Y., Friedrich, R., Rivier, L., Houwelling, S., Schulze,
E. D., and members of the CARBOEUROPE SYNTHESIS TEAM: The European carbon balance. Part 1: fossil
fuel emissions, Glob. Change Biol., 16, 1395–1408, 10.1111/j.1365-2486.2009.02098.x, 2010.Ciais, P., Sabine, C., Bala, G., Bopp, L., Brovkin, V., Canadell, J.,
Chhabra, A., DeFries, R., Galloway, J., Heimann, M., Jones, C.,
Quéré, C. L., Myneni, R. B., Piao, S., and Thornton, P.: Carbon and
Other Biogeochemical Cycles, in: 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, UK,
New York, NY, USA, available at: http://www.ipcc.ch/report/ar5/wg1/ (last access: 28 April 2015), 2013.Coënt, L.: The IEA energy data collection and CO2 estimates: an
overview, International Energy Agency, available at: https://www.transparency-partnership.net/sites/default/files/u2620/the_iea_ energy_data_collection_and_co2_estimates_an_ overview__ iea_ _ coent.pdf (last access:
2 August 2019), 2017.Córdoba, P.: Status of Flue Gas Desulphurisation (FGD) systems from
coal-fired power plants: Overview of the physic-chemical control processes
of wet limestone FGDs, Fuel, 144, 274–286, 10.1016/j.fuel.2014.12.065, 2015.Crippa, M., Oreggioni, G., Guizzardi, D., Muntean, M., Schaaf, E., Lo Vullo,
E., Solazzo, E., Monforti-Ferrario, F., Olivier, J. G. J., and Vignati, E.:
Fossil CO2 and GHG emissions of all world countries: 2019 report,
Publications Office of the European Union, Luxembourg, EUR 29849 EN,
JRC117610, ISBN 978-92-76-11100-9, available at: https://edgar.jrc.ec.europa.eu/overview.php?v=booklet2019
(last access: 14 April 2020), 2019.Cui, D., Deng, Z., and Liu, Z.: China's non-fossil fuel CO2 emissions from
industrial processes, Appl. Energ., 254, 113537, 10.1016/j.apenergy.2019.113537, 2019.Davis, S. J. and Caldeira, K.: Consumption-based accounting of CO2
emissions, P. Natl. Acad. Sci. USA, 107, 5687–5692, 10.1073/pnas.0906974107, 2010.DEFRA: UK's carbon footprint: Annual carbon dioxide emissions relating to UK
consumption, Department for Environment, Food & Rural Affairs 2019, available at:
https://www.gov.uk/government/statistics/uks-carbon-footprint, last access: 30 July 2019.den Elzen, M. J., Olivier, J. J., Höhne, N., and Janssens-Maenhout, G.:
Countries' contributions to climate change: effect of accounting for all
greenhouse gases, recent trends, basic needs and technological progress,
Clim. Change, 121, 397–412, 10.1007/s10584-013-0865-6, 2013.EEA: Annual European Union greenhouse gas inventory 1990–2017 and inventory
report 2019, available at: https://unfccc.int/process-and-meetings/transparency-and-reporting/reporting-and-review-under-the-convention/greenhouse-gas-inventories-annex-i-parties/national-inventory-submissions-2019
(last access: 30 January 2020), 2019.Eggleston, S., Buendia, L., Miwa, K., Ngara, T., and Tanabe, K.: Guidelines for National Greenhouse Gas Inventories, Prepared by the National Greenhouse Gas Inventories Programme, edited by: IPCC, Institute for Global Environmental Strategies, Hayama, Kanagawa, Japan, available at: http://www.ipcc-nggip.iges.or.jp/public/2006gl/index.html (last access: 21 May 2017), 2006.Eggleston, S.: The IPCC 2006 Guidelines and their evolution from the Revised
1996 Guidelines, Sessional period of the subsidiary bodies, SB28, Bonn,
Germany, 2–13 June 2008, available at: https://unfccc.int/files/meetings/ad_hoc_working_ groups/kp/application/pdf/sb28_ipcc_simon.pdf (last access: 4 September 2019), 2008.EIA: Documentation for Emissions of Greenhouse Gases in the United States
2006, Energy Information Administration, Washington, DC, available at: https://www.eia.gov/environment/emissions/archive/ghg/gg07rpt/pdf/0573(2006).pdf
(last access: 8 February 2019), 2008.EIA: International Energy Statistics, Energy Information Administration, available at:
https://www.eia.gov/beta/international/data/browser/, last
access: 31 August 2019a.EIA: International Energy Statistics: Table Notes, Energy Information
Administration , available at: https://www.eia.gov/beta/international/data/browser/views/partials/table_notes.html, last access: 2 August 2019b.EIA: About EIA: Legislative Timeline, Energy Information Administration, available at:
https://www.eia.gov/about/legislative_timeline.php, last access: 2 August 2019c.EIA: International Energy Statistics, Energy Information Administration,
available at: https://www.eia.gov/international/data/world, last
access: 4 February 2020.
Etemad, B. and Luciani, J.: World Energy Production 1800–1985, Librairie
Droz, 1991.European Commission, International Monetary Fund, Organisation for Economic
Co-operation and Development, United Nations, and World Bank: System of
National Accounts 2008, available at: http://unstats.un.org/unsd/nationalaccount/sna2008.asp (last access: 7 May
2013), 2009.Frey, C., Penman, J., Hanle, L., Monni, S., and Ogle, S.: Uncertainties, in:
2006 IPCC Guidelines for National Greenhouse Gas Inventories, edited by:
Eggleston, S., Buendia, L., Miwa, K., Ngara, T., and Tanabe, K., IGES,
Japan, available at: http://www.ipcc-nggip.iges.or.jp/public/2006gl/index.html (last access: 21 May 2017), 2006.Friedlingstein, P., Jones, M. W., O'Sullivan, M., Andrew, R. M., Hauck, J., Peters, G. P., Peters, W., Pongratz, J., Sitch, S., Le Quéré, C., Bakker, D. C. E., Canadell, J. G., Ciais, P., Jackson, R. B., Anthoni, P., Barbero, L., Bastos, A., Bastrikov, V., Becker, M., Bopp, L., Buitenhuis, E., Chandra, N., Chevallier, F., Chini, L. P., Currie, K. I., Feely, R. A., Gehlen, M., Gilfillan, D., Gkritzalis, T., Goll, D. S., Gruber, N., Gutekunst, S., Harris, I., Haverd, V., Houghton, R. A., Hurtt, G., Ilyina, T., Jain, A. K., Joetzjer, E., Kaplan, J. O., Kato, E., Klein Goldewijk, K., Korsbakken, J. I., Landschützer, P., Lauvset, S. K., Lefèvre, N., Lenton, A., Lienert, S., Lombardozzi, D., Marland, G., McGuire, P. C., Melton, J. R., Metzl, N., Munro, D. R., Nabel, J. E. M. S., Nakaoka, S.-I., Neill, C., Omar, A. M., Ono, T., Peregon, A., Pierrot, D., Poulter, B., Rehder, G., Resplandy, L., Robertson, E., Rödenbeck, C., Séférian, R., Schwinger, J., Smith, N., Tans, P. P., Tian, H., Tilbrook, B., Tubiello, F. N., van der Werf, G. R., Wiltshire, A. J., and Zaehle, S.: Global Carbon Budget 2019, Earth Syst. Sci. Data, 11, 1783–1838, 10.5194/essd-11-1783-2019, 2019.Gieseke, R. and Gütschow, J.: UNFCCC Emissions data from the Detailed
Data By Party interface, available at: 10.5281/zenodo.3763019, last access: 14 May 2020.Gilfillan, D., Marland, G., Boden, T., and Andres, R.: Global, Regional, and
National Fossil-Fuel CO2 Emissions: 1751–2016, Appalachian Energy Center,
Appalachian State University, available at: https://energy.appstate.edu/research/work-areas/cdiac-appstate, last
access: 6 December 2019.Global Carbon Project: Global Carbon Atlas, available at: http://www.globalcarbonatlas.org/en/CO2-emissions (last access: 2 August
2019), 2018.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.Gütschow, J., Jeffery, L., and Gieseke, R.: The PRIMAP-hist national
historical emissions time series (1850–2016), V.2.0, GFZ Data Services, 10.5880/PIK.2019.001, 2019a.Gütschow, J., Jeffery, L., Gieseke, R., and Günther, A.: The
PRIMAP-hist national historical emissions time series (1850–2017), V.2.1,
GFZ Data Services, 10.5880/PIK.2019.018, 2019b.Gütschow, J., Jeffery, M. L., and Günther, A.: PRIMAP-crf: UNFCCC
CRF data in IPCC categories (PRIMAP-crf-2019-v2) (Version 2019v2), Zenodo,
10.5281/zenodo.3775575, 2020.Haukås, H. T., Jaques, A., Neitzert, F., Rosland, A., Rypdal, K., and
Weidemann, F.: Industrial Processes, in: Revised 1996 IPCC
Guidelines for National Greenhouse Gas Inventories: Reference Manual, chap. 2, edited
by: Houghton, J. T., Meira Filho, L. G., Lim, B., Tréanton, K., Mamaty,
I., Bonduki, Y., Griggs, D. J., and Callander, B. A., IPCC/OECD/IEA, Paris, available at: http://www.ipcc-nggip.iges.or.jp/public/gl/invs1.html
(last access: 27 October 2014),
1997.Heitmann, N. and Khalilian, S.: Accounting for carbon dioxide emissions
from international shipping: Burden sharing under different UNFCCC
allocation options and regime scenarios, Mar. Policy, 35, 682–691,
10.1016/j.marpol.2011.02.009, 2011.Heriot Watt University: Heriot-Watt sets global data landscape for BP
Statistical Review, available at: https://www.hw.ac.uk/about/news/2017/heriot-watt-sets-global-data-landscape-for.htm
(last access: 12 March 2019), 2017.Hiraishi, T., Krug, T., Tanabe, K., Srivastava, N., Jamsranjav, B., Fukuda, M., and Troxler, T.: 2013 Supplement to the 2006 IPCC Guidelines for National Greenhouse Gas Inventories: Wetlands, IPCC, Switzerland, available at: https://www.ipcc-nggip.iges.or.jp/public/wetlands/ (last access: 14 March 2019), 2013.Hoesly, R. M., Smith, S. J., Feng, L., Klimont, Z., Janssens-Maenhout, G., Pitkanen, T., Seibert, J. J., Vu, L., Andres, R. J., Bolt, R. M., Bond, T. C., Dawidowski, L., Kholod, N., Kurokawa, J.-I., Li, M., Liu, L., Lu, Z., Moura, M. C. P., O'Rourke, P. R., and Zhang, Q.: Historical (1750–2014) anthropogenic emissions of reactive gases and aerosols from the Community Emissions Data System (CEDS), Geosci. Model Dev., 11, 369–408, 10.5194/gmd-11-369-2018, 2018.
Högbom, A.: Om sannolikheten för sekulära förändringar i
atmosfärens kolsyrehalt (On the probability of secular changes in the
level of atmospheric carbon dioxide), Svensk Kemisk Tidskrift, VI, 169–176,
1894.Höhne, N., Blum, H., Fuglestvedt, J., Skeie, R. B., Kurosawa, A., Hu,
G., Lowe, J., Gohar, L., Matthews, B., Nioac de Salles, A. C., and
Ellermann, C. J. C. C.: Contributions of individual countries' emissions to
climate change and their uncertainty, Clim. Change, 106, 359–391,
10.1007/s10584-010-9930-6, 2011.Houghton, J. T., Meira Filho, L. G., Lim, B., Treanton, K., Mamaty, I., Bonduki, Y., Griggs, D. J., and Callender, B. A.: Revised 1996 IPCC Guidelines for National Greenhouse Gas Inventories, UK Meteorological Office, Bracknell, UK, available at: http://www.ipcc-nggip.iges.or.jp/public/gl/invs1.html (last access: 17 October 2014), 1996.I4CE: Chiffres clés du climat France, Europe et Monde, Institute for
Climate Economics, Commissariat général au développement
durable, available at: https://www.statistiques.developpement-durable.gouv.fr/sites/default/files/2018-12/datalab-46-chiffres-cles-du-climat-edition-2019-novembre2018_1.pdf (last access: 14 December 2019), 2018.IEA: CO2 emissions from fuel combustion 2018, International Energy
Agency, Paris, France, available at: https://webstore.iea.org/co2-emissions-from-fuel-combustion-2018-highlights (last
access: 8 February 2019), 2018a.IEA: World Energy Balances (2018 Edition), International Energy Agency,
Paris, available at: https://webstore.iea.org/world-energy-balances-2018 (last access: 19 February 2019), 2018b.IEA: CO2 Emissions From Fuel Combustion: Database Documentation,
International Energy Agency, Paris, France, available at: http://wds.iea.org/wds/pdf/Worldco2_Documentation.pdf (last
access: 8 February 2019), 2018c.IEA: CO2 Emissions From Fuel Combustion: Database Documentation,
International Energy Agency, Paris, France, available at: http://wds.iea.org/wds/pdf/Worldco2_Documentation.pdf, last
access: 16 December 2019a.IEA: CO2 Emissions from Fuel Combustion 2019 Highlights, International
Energy Agency Paris, available at: https://webstore.iea.org/co2-emissions-from-fuel-combustion-2019-highlights, last access: 28 November 2019b.IEA: CO2 emissions from fuel combustion 2019, International Energy
Agency, Paris, France, available at: https://webstore.iea.org/co2-emissions-from-fuel-combustion-2019 (last
access: 5 February 2020), 2019c.IEA: World Energy Statistics 2019 Edition: Database Documentation,
International Energy Agency, Paris, available at: https://webstore.iea.org/world-energy-statistics-2019, last
access: 29 August 2019d.IEA: World Energy Balances 2019 Edition, International Energy Agency, Paris, available at: https://webstore.iea.org/world-energy-balances-2019, last access: 29 August 2019e.IEA: History, available at: https://www.iea.org/about/history/, last
access: 2 August 2019f.IEA: Statistics Resources: Questionnaires, available at: https://www.iea.org/statistics/resources/questionnaires/, last access: 2 August 2019g.
IPCC: Climate Change 2014: Synthesis Report, Contribution of Working Groups
I, II and III to the Fifth Assessment Report of the Intergovernmental Panel
on Climate Change, IPCC, Geneva, Switzerland, 2014.
IPCC: 2019 Refinement to the 2006 IPCC Guidelines for National Greenhouse
Gas Inventories, 2019.Janssens-Maenhout, G., Pagliari, V., Guizzardi, D., and Muntean, M.: Global
emission inventories in the Emission Database for Global Atmospheric
Research (EDGAR) – Manual (I): Gridding: EDGAR emissions distribution on
global gridmaps, Joint Research Centre, Report EUR 25785 EN, available at:
http://publications.jrc.ec.europa.eu/repository/bitstream/JRC78261/edgarv4_manual_ i_gridding_pubsy_final.pdf (last access: 28 March 2019), 2012.Janssens-Maenhout, G., Crippa, M., Guizzardi, D., Muntean, M., Schaaf, E., Dentener, F., Bergamaschi, P., Pagliari, V., Olivier, J. G. J., Peters, J. A. H. W., van Aardenne, J. A., Monni, S., Doering, U., Petrescu, A. M. R., Solazzo, E., and Oreggioni, G. D.: EDGAR v4.3.2 Global Atlas of the three major greenhouse gas emissions for the period 1970–2012, Earth Syst. Sci. Data, 11, 959–1002, 10.5194/essd-11-959-2019, 2019.Jeffery, M. L., Gütschow, J., Gieseke, R., and Gebel, R.: PRIMAP-crf: UNFCCC CRF data in IPCC 2006 categories, Earth Syst. Sci. Data, 10, 1427–1438, 10.5194/essd-10-1427-2018, 2018.Ke, J., McNeil, M., Price, L., Khanna, N. Z., and Zhou, N.: Estimation of
CO2 emissions from China's cement production: Methodologies and
uncertainties, Energ. Policy, 57, 172–181, 10.1016/j.enpol.2013.01.028, 2013.Keeling, C. D.: Industrial production of carbon dioxide from fossil fuels
and limestone, Tellus, 25, 174–198, 10.3402/tellusa.v25i2.9652,
1973.
Klein Goldewijk, C. and Battjes, J.: The IMAGE 2 hundred year (1890–1900)
data base of the global environment (HYDE), RIVM Report 481507008, 1995.
Klein Goldewijk, C. and Battjes, J.: A Hundred Year (1890–1900) Database
for Integrated Environmental Assessments (HYDE, version 1.1), RIVM Report 422514002,
1997.Korsbakken, J. I., Peters, G. P., and Andrew, R. M.: Uncertainties around
reductions in China's coal use and CO2 emissions, Nature Climate
Change, 6, 687–690, 10.1038/nclimate2963, 2016.Lei, S.: Key opportunities to reduce emissions in China's cement industry,
Joint workshop by MOST & IEA on carbon capture and storage: opportunities
in energy-intensive industry, Beijing, China, available at: https://www.iea.org/media/workshops/2013/most/1_3ShenLei_en.pdf (last access: 5 May 2015), 2012.Li, M., Zhang, Q., Kurokawa, J.-I., Woo, J.-H., He, K., Lu, Z., Ohara, T., Song, Y., Streets, D. G., Carmichael, G. R., Cheng, Y., Hong, C., Huo, H., Jiang, X., Kang, S., Liu, F., Su, H., and Zheng, B.: MIX: a mosaic Asian anthropogenic emission inventory under the international collaboration framework of the MICS-Asia and HTAP, Atmos. Chem. Phys., 17, 935–963, 10.5194/acp-17-935-2017, 2017.Liu, Z., Guan, D., Wei, W., Davis, S. J., Ciais, P., Bai, J., Peng, S.,
Zhang, Q., Hubacek, K., Marland, G., Andres, R. J., Crawford-Brown, D., Lin,
J., Zhao, H., Hong, C., Boden, T. A., Feng, K., Peters, G. P., Xi, F., Liu,
J., Li, Y., Zhao, Y., Zeng, N., and He, K.: Reduced carbon emission
estimates from fossil fuel combustion and cement production in China,
Nature, 524, 335–338, 10.1038/nature14677, 2015.Macknick, J.: Energy and CO2 emission data uncertainties, Carbon Manag.,
2, 189–205, 10.4155/cmt.11.10, 2011.
Marland, G. and Rotty, R. M.: Carbon dioxide emissions from fossil fuels: a
procedure for estimation and results for 1950–1982, Tellus B, 36, 232–261,
1984.
Marland, G. and Boden, T.: The Magnitude and Distribution of
Fossil-Fuel-Related Carbon Releases, in: The Global Carbon Cycle, Berlin,
Heidelberg, 117–138, 1993.Marland, G., Andres, R. J., and Boden, T. A.: Global, regional, and national
CO2 emissions, in: Trends '93: A compendium of data on global change,
edited by: Boden, T. A., Kaiser, D. P., Sepanski, R. J., Stoss, F. W., and
Logsdon, G. M., Oak Ridge National Lab., TN (United States), 505–584,
https://www.osti.gov/servlets/purl/10106351 (last access: 11
February 2019), 1994.Marland, G., Brenkert, A., and Olivier, J.: CO2 from fossil fuel burning: a
comparison of ORNL and EDGAR estimates of national emissions, Environ.
Sci. Pol., 2, 265–273, 10.1016/S1462-9011(99)00018-0, 1999.MATCH: Ad hoc group for the modelling and assessment of contributions of
climate change (MATCH), DEFRA and ECOFYS,, available at: http://www.match-info.net/, last access: 4 September 2019.Matthews, H. D., Gillett, N. P., Stott, P. A., and Zickfeld, K.: The
proportionality of global warming to cumulative carbon emissions, Nature,
459, 829–832, 10.1038/nature08047, 2009.Miljødirektoratet: Greenhouse Gas Emissions 1990–2017, National
Inventory Report, The Norwegian Environment Agency, Oslo, available at: https://unfccc.int/documents/195137 (last access: 29 January 2020), 2019.Myhre, G., Alterskjær, K., and Lowe, D.: A fast method for updating
global fossil fuel carbon dioxide emissions, Environ. Res. Lett.,
4, 034012, 10.1088/1748-9326/4/3/034012, 2009.Nauels, A., Rosen, D., Mauritsen, T., Maycock, A., McKenna, C., Rogelj, J., Schleussner, C.-F., Smith, E., Smith, C., and Forster, P.: ZERO IN ON the remaining carbon budget and decadal warming rates,
The CONSTRAIN Project Annual Report 2019, 10.5518/100/20, last access: 7 December 2019.Oda, T. and Maksyutov, S.: A very high-resolution (1 km × 1 km) global fossil fuel CO2 emission inventory derived using a point source database and satellite observations of nighttime lights, Atmos. Chem. Phys., 11, 543–556, 10.5194/acp-11-543-2011, 2011.Oda, T., Maksyutov, S., and Andres, R. J.: The Open-source Data Inventory for Anthropogenic CO2, version 2016 (ODIAC2016): a global monthly fossil fuel CO2 gridded emissions data product for tracer transport simulations and surface flux inversions, Earth Syst. Sci. Data, 10, 87–107, 10.5194/essd-10-87-2018, 2018.Oda, T., Bun, R., Kinakh, V., Topylko, P., Halushchak, M., Marland, G.,
Lauvaux, T., Jonas, M., Maksyutov, S., Nahorski, Z., Lesiv, M., Danylo, O.,
and Horabik-Pyzel, J.: Errors and uncertainties in a gridded carbon dioxide
emissions inventory, Mitigation and Adaptation Strategies for Global Change,
24, 1007–1050, 10.1007/s11027-019-09877-2, 2019.Olivier, J. G. J.: Scenarios for global emissions from air traffic, National
Institutde of Public Health and Environmental Protection, Bilthoven, The
Netherlands, Report no. 773002003, available at: https://www.rivm.nl/bibliotheek/rapporten/773002003.html (last access: 8 February 2019), 1995.Olivier, J. G. J., Bouwman, A. F., van der Maas, C. W. M., Berdowski, J. J.
M., Veldt, C., Bloos, J. P. J., Visschedijk, A. J. H., Zandveld, P. Y. J.,
and Haverlag, J. L.: Description of EDGAR Version 2.0: A set of global
emission inventories of greenhouse gases and ozone-depleting substances for
all anthropogenic and most natural sources on a per country basis and on 1
degree × 1 degree grid, National Institute of Public Health and the
Environment, Bilthoven, the Netherlands, available at: https://www.rivm.nl/bibliotheek/rapporten/771060002.html (last access: 8
February 2019), 1996.Olivier, J. G. J., Bouwman, A. F., Berdowski, J. J. M., Veldt, C., Bloos, J.
P. J., Visschedijk, A. J. H., van der Maas, C. W. M., and Zandveld, P. Y.
J.: Sectoral emission inventories of greenhouse gases for 1990 on a per
country basis as well as on 1∘× 1∘,
Environ. Sci. Pol., 2, 241–263, 10.1016/S1462-9011(99)00027-1, 1999.Olsson, A.: Reporting and review of GHG inventories under the Convention and
the Kyoto Protocol, Conference on Climate Change and Official Statistics,
Oslo, Norway, 14–16 April 2008, available at: https://unstats.un.org/unsd/climate_change/docs/presentations/CCPresentation_UNFCCC.pdf (last
access: 30 January 2020), 2008.Penman, J., Gytarsky, M., Hiraishi, T., Irving, W., and Krug, T.: Overview,
in: 2006 IPCC Guidelines for National Greenhouse Gas Inventories, edited by:
Eggleston, S., Buendia, L., Miwa, K., Ngara, T., and Tanabe, K., IGES,
Japan, available at: http://www.ipcc-nggip.iges.or.jp/public/2006gl/index.html (last access: 10 October 2018), 2006.Plass, G. N.: The Carbon Dioxide Theory of Climatic Change, Tellus A, 8,
140–154, 10.3402/tellusa.v8i2.8969, 1956.Revelle, R. and Suess, H. E.: Carbon Dioxide Exchange Between Atmosphere
and Ocean and the Question of an Increase of Atmospheric CO2 during the
Past Decades, Tellus B, 9, 18–27, 10.1111/j.2153-3490.1957.tb01849.x,
1957.
Revelle, R., Broecker, W., Craig, H., Keeling, C. D., and Smagorinsky, J.:
Atmospheric Carbon Dioxide, in: Restoring the Quality of Our Environment,
edited by: The Environmental Pollution Panel, President's Science Advisory Committee, 111–133,
1965.Rogelj, J., Schaeffer, M., Friedlingstein, P., Gillett, N. P., van Vuuren,
D. P., Riahi, K., Allen, M., and Knutti, R.: Differences between carbon
budget estimates unravelled, Nat. Clim. Change, 6, 245–252, 10.1038/nclimate2868, 2016.Rotty, R. M.: Commentary on and extension of calculative procedure for
CO2 production, Tellus B, 25, 508–517, 10.1111/j.2153-3490.1973.tb00635.x, 1973.Rotty, R. M.: First estimates of global flaring of natural gas, Atmos.
Environ., 8, 681–686, 10.1016/0004-6981(74)90159-0, 1974.Rotty, R. M.: Distribution of and changes in industrial carbon dioxide
production, J. Geophys. Res.-Oceans, 88, 1301–1308,
10.1029/JC088iC02p01301, 1983.Scott, R.: Major Policies and Actions, in: The History of the International
Energy Agency, International Energy Agency, Paris, available at: https://www.iea.org/media/about/2ieahistory.pdf (last access: 2 August
2019), 1994.
Smith, S. J., Zhou, Y., Kyle, P., Wang, H., and Yu, H.: A Community
Emissions Data System (CEDS): Emissions for CMIP6 and Beyond, Proceedings of
the 2015 International Emission Inventory Conference, San Diego, CA, USA,
14–16 April 2015, 12–16, 2015.SOeS: the carbon footprint of French consumption: evolution between 1990 and
2007, Service de l'observation et des statistiques, Commissariat
Général au Développement Durable, available at: https://www.statistiques.developpement-durable.gouv.fr/sites/default/files/2018-10/LPS 114_Pointcarbone_GB_BD_v2.pdf
(last access: 14 Decemeber 2019), 2012.Stocker, T. F., Qin, D., Plattner, G.-K., Alexander, L. V., Allen, S. K.,
Bindoff, N. L., Bréon, F.-M., Church, J. A., Cubasch, U., Emori, S.,
Forster, P., Friedlingstein, P., Gillett, N., Gregory, J. M., Hartmann, D.
L., Jansen, E., Kirtman, B., Knutti, R., Kumar, K. K., Lemke, P., Marotzke,
J., Masson-Delmotte, V., Meehl, G. A., Mokhov, I. I., Piao, S., Ramaswamy,
V., Randall, D., Rhein, M., Rojas, M., Sabine, C., Shindell, D., Talley, L.
D., Vaughan, D. G., and Xie, S.-P.: Technical Summary, in: 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, UK,
New York, NY, USA, available at: https://www.ipcc.ch/assessment-report/ar5/ (last access:
10 February 2020),
2013.Temertekov, T., Grabert, J., Ishmukhametov, R., and Lalieu, V.: Greenhouse
Gas Inventory Information System, 12th International Emission Inventory
Conference: “Emission Inventories – Applying New Technologies”, San Diego,
29 April–1 May 2003, available at: https://www3.epa.gov/ttnchie1/conference/ei12/ (last access: 11 May 2020), 2003.Treanton, K., Ibitoye, F., Kainou, K., Olivier, J. G. J., Pretel, J.,
Simmons, T., and Yang, H.: Reference Approach, in: 2006 IPCC Guidelines for
National Greenhouse Gas Inventories, edited by: Eggleston, S., Buendia, L.,
Miwa, K., Ngara, T., and Tanabe, K., IGES, Japan, available at: http://www.ipcc-nggip.iges.or.jp/public/2006gl/index.html (last access: 21 May 2017), 2006a.Treanton, K., Ibitoye, F., Kainou, K., Olivier, J. G. J., Pretel, J.,
Simmons, T., Yang, H., and Quadrelli, R.: Reference Approach, in: 2006 IPCC
Guidelines for National Greenhouse Gas Inventories, edited by: Eggleston,
S., Buendia, L., Miwa, K., Ngara, T., and Tanabe, K., IGES, Japan, available at:
http://www.ipcc-nggip.iges.or.jp/public/2006gl/index.html (last
access: 21 May 2017), 2006b.UN: Paris Agreement, available at: http://unfccc.int/paris_agreement/items/9485.php (last
access: 23 June 2020), 2015.UNFCCC: Report of the Conference of the Parties on its first session, held
at Berlin from 28 March to 7 April 1995: Addendum, Part two: Action taken by
the Conference of the Parties at its first session, United Nations Framework
Convention on Climate Change, FCCC/CP/1995/7/Add.1, available at: https://unfccc.int/resource/docs/cop1/07a01.pdf (last access: 30 January
2020), 1995.UNFCCC: Review of the Implementation of Commitments and of Other Provisions
of the Convention, United Nations Framework Convention on Climate Change,
FCCC/CP/1999/7, available at: https://unfccc.int/resource/docs/cop5/07.pdf (last access: 30 January
2020), 2000.UNFCCC: Copenhagen Accord, United Nations Framework Convention on Climate
Change, available at: http://unfccc.int/meetings/cop_15/copenhagen_accord/items/5262.php (last access: May 2011), 2009.UNFCCC: Report of the Conference of the Parties on its nineteenth session,
held in Warsaw from 11 to 23 November 2013: Addendum, Part two: Action taken
by the Conference of the Parties at its nineteenth session, United Nations
Framework Convention on Climate Change, FCCC/CP/2013/10/Add.3, available at:
https://unfccc.int/resource/docs/2013/cop19/eng/10a03.pdf (last
access: 8 February 2019), 2014.UNFCCC: Report on the individual review of the inventory submission of
Norway submitted in 2018, CC/ERT/ARR/2019/18, available at: https://unfccc.int/process-and-meetings/transparency-and-reporting/reporting-and-review-under-the-convention/greenhouse-gas-inventories-annex-i-parties/inventory-review-reports/inventory-review-reports-2018
(last access: 30 January 2020), 2019a.UNFCCC: Report of the Conference of the Parties serving as the meeting of
the Parties to the Paris Agreement on the third part of its first session,
held in Katowice from 2 to 15 December2018: Addendum, Part two: Action taken
by the Conference of the Parties serving as the meeting of the Parties to
the Paris Agreement, United Nations Framework Convention on Climate Change,
FCCC/PA/CMA/2018/3/Add.2, available at: https://unfccc.int/sites/default/files/resource/CMA2018_03a02E.pdf, last access: 10 February 2019b.UNFCCC: National Inventory Submissions 2019, United Nations Framework
Convention on Climate Change, available at: https://unfccc.int/process-and-meetings/transparency-and-reporting/reporting-and-review-under-the-convention/greenhouse-gas-inventories-annex-i-parties/national-inventory-submissions-2019, last access: 27 August 2019c.UNFCCC: Emissions from fuels used for international aviation and maritime
transport, United Nations Framework Convention on Climate Change, available at:
https://unfccc.int/topics/mitigation/workstreams/emissions-from-international-transport-bunker-fuels, last access: 6 February 2020.UNFCCC: Time Series – Annex I, United Nations Framework Convention on
Climate Change, available at: https://di.unfccc.int/time_series (last access: 2 August 2019d).
United Nations: World Energy Requirements in 1975 and 2000, International
Conference on the Peaceful Uses of Atomic Energy, Geneva, 8–20 August 1955, 1956.UNSD: International Recommendationsfor Energy Statistics(IRES), United
Nations Department of Economic and Social Affairs Statistics Division,
ST/ESA/STAT/SER.M/93, available at: https://unstats.un.org/unsd/energystats/methodology/documents/IRES-web.pdf
(last access: 6 February 2020), 2018.
USGS: Cement, in: 2016 Minerals Yearbook – Metals and Minerals, available at:
https://www.usgs.gov/centers/nmic/cement-statistics-and-information,last
access: 26 May 2020.WRI: CAIT Country Greenhouse Gas Emissions: Sources & Methods, World
Resources Institute, available at: http://cait.wri.org/docs/CAIT2.0_CountryGHG_Methods.pdf (last access: 8 February 2019), 2015.