Global carbon budget 2014

. Accurate assessment of anthropogenic carbon dioxide (CO 2 ) emissions and their redistribution among the atmosphere, ocean, and terrestrial biosphere is important to better understand the global carbon cycle, support the development of climate policies, and project future climate change. Here we describe data sets and a methodology to quantify all major components of the global carbon budget, including their uncertainties, based on the combination of a range of data, algorithms, statistics, and model estimates and their interpretation by a broad scientiﬁc community. We discuss changes compared to previous estimates, consistency within and among components, alongside methodology and data limitations. CO 2 emissions from fossil G ATM was high in 2013, reﬂecting a steady increase in E FF and smaller and opposite changes between S OCEAN and S LAND compared to the past decade (2004–2013). The global atmospheric CO 2 concentration reached 395.31 ± 0.10 ppm averaged over 2013. We estimate that E FF will increase by 2.5 % (1.3– 3.5 %) to 10.1 ± 0.6 GtC in 2014 (37.0 ± 2.2 GtCO 2 yr − 1 ), 65 % above emissions in 1990, based on projections of world gross domestic product and recent changes in the carbon intensity of the global economy. From this projection of E FF and assumed constant E LUC for 2014, cumulative emissions of CO 2 will reach about 545 ± 55 GtC (2000 ± 200 GtCO 2 ) for 1870–2014, about 75 % from E FF and 25 % from E LUC . This paper documents changes in the methods and data sets used in this new carbon budget compared with previous publications of this living data set (Le Quéré et al., 2013, 2014). All observations presented here can be downloaded from the Carbon Dioxide Information Analysis Center (doi:10.3334/CDIAC/GCP_2014).


Introduction
The concentration of carbon dioxide (CO 2 ) in the atmosphere has increased from approximately 277 parts per million (ppm) in 1750 (Joos and Spahni, 2008), the beginning of the Industrial Era, to 395.31 ppm in 2013 (Dlugokencky and. Daily averages went above 400 ppm for the first time at Mauna Loa station in May 2013 (Scripps, 2013). This station holds the longest running record of direct measurements of atmospheric CO 2 concentration (Tans and Keeling, 2014; Fig. 1). The atmospheric CO 2 increase above preindustrial levels was initially, primarily, caused by the release of carbon to the atmosphere from deforestation and other land-use-change activities . While emissions from fossil fuel combustion started before the Industrial Era, they only became the dominant source of anthropogenic emissions to the atmosphere from around 1920 and their relative share has continued to increase until present. Anthropogenic emissions occur on top of an active natural carbon cycle that circulates carbon between the atmosphere, ocean, and terrestrial biosphere reservoirs on timescales from days to millennia, while exchanges with geologic reservoirs occur at longer timescales (Archer et al., 2009).
The global carbon budget presented here refers to the mean, variations, and trends in the perturbation of CO 2 in the atmosphere, referenced to the beginning of the Industrial Era. It quantifies the input of CO 2 to the atmosphere by emissions from human activities, the growth of CO 2 in the atmosphere, and the resulting changes in the storage of carbon in the land and ocean reservoirs in response to increasing atmospheric CO 2 levels, climate and climate variability, and other anthropogenic and natural changes (Fig. 2). An understanding of this perturbation budget over time and the underlying variability and trends of the natural carbon cycle are necessary to understand the response of natural sinks to changes in climate, CO 2 and land-use-change drivers, and the permissible emissions for a given climate stabilisation target.
The components of the CO 2 budget that are reported annually in this paper include separate estimates for (1) the CO 2 emissions from fossil fuel combustion and cement production (E FF ; GtC yr −1 ), (2) the CO 2 emissions resulting from deliberate human activities on land leading to land-use change (LUC; E LUC ; GtC yr −1 ), (3) the growth rate of CO 2 in the atmosphere (G ATM ; GtC yr −1 ), and the uptake of CO 2 by the "CO 2 sinks" in (4) the ocean (S OCEAN ; GtC yr −1 ) and (5) on land (S LAND ; GtC yr −1 ). The CO 2 sinks as defined here include the response of the land and ocean to elevated CO 2 and changes in climate and other environmental conditions. The global emissions and their partitioning among the atmosphere, ocean, and land are in balance: Atmospheric CO 2 concentration (ppm) NOAA/ESRL (Dlugokencky & Tans, 2014) Scripps Institution of Oceanography (Keeling et al., 1976) Figure 1. Surface average atmospheric CO 2 concentration, deseasonalised (ppm). The 1980-2014 monthly data are from NOAA/ESRL (Dlugokencky and Tans, 2014). The 1980-2014 estimate is an average of direct atmospheric CO 2 measurements from multiple stations in the marine boundary layer (Masarie and Tans, 1995). The 1958The -1979 monthly data are from the Scripps Institution of Oceanography, based on an average of direct atmospheric CO 2 measurements from the Mauna Loa and South Pole stations (Keeling et al., 1976). To take into account the difference of mean CO 2 between the NOAA/ESRL and the Scripps station networks used here, the Scripps surface average (from two stations) was harmonised to match the NOAA/ESRL surface average (from multiple stations) by adding the mean difference of 0.542 ppm, calculated here from overlapping data during 1980-2012. The mean seasonal cycle was removed from both data sets. (1) G ATM is usually reported in ppm yr −1 , which we convert to units of carbon mass, GtC yr −1 , using 1 ppm = 2.120 GtC (Prather et al., 2012; Table 1). We also include a quantification of E FF by country, computed with both territorial and consumption based accounting (see Methods). Equation (1) partly omits two kinds of processes. The first is the net input of CO 2 to the atmosphere from the chemical oxidation of reactive carbon-containing gases from sources other than fossil fuels (e.g. fugitive anthropogenic CH 4 emissions, industrial processes, and changes of biogenic emissions from changes in vegetation, fires, wetlands), primarily methane (CH 4 ), carbon monoxide (CO), and volatile organic compounds such as isoprene and terpene. CO emissions are currently implicit in E FF , while anthropogenic CH 4 emissions are not and thus their inclusion would result in a small increase in E FF . The second is the anthropogenic perturbation to carbon cycling in terrestrial freshwaters, estuaries, and coastal areas, which modifies lateral fluxes from land ecosystems to the open ocean, the evasion CO 2 flux from rivers, lakes and estuaries to the atmosphere, and the net airsea anthropogenic CO 2 flux of coastal areas (Regnier et al.,Figure 2. Schematic representation of the overall perturbation of the global carbon cycle caused by anthropogenic activities, averaged globally for the decade 2004-2013. The arrows represent emission from fossil fuel burning and cement production (E FF ), emissions from deforestation and other land-use change (E LUC ), the growth of carbon in the atmosphere (G ATM ), and the uptake of carbon by the "sinks" in the ocean (S OCEAN ) and land (S LAND ) reservoirs. All fluxes are in units of GtC yr −1 , with uncertainties reported as ±1σ (68 % confidence that the real value lies within the given interval) as described in the text. This figure is an update of one prepared by the International Geosphere-Biosphere Programme for the GCP, first presented in Le  2013). The inclusion of freshwater fluxes of anthropogenic CO 2 would affect the estimates of, and partitioning between, S LAND and S OCEAN in Eq. (1) in complementary ways, but it would not affect the other terms. These flows are omitted in absence of annual information on the natural versus anthropogenic perturbation terms of these loops of the carbon cycle, and they are discussed in Sect. 2.7.
The CO 2 budget has been assessed by the Intergovernmental Panel on Climate Change (IPCC) in all assessment reports Denman et al., 2007;Prentice et al., 2001;Schimel et al., 1995;Watson et al., 1990), as well as by others (e.g. Ballantyne et al., 2012). These assessments included budget estimates for the decades of the 1980s and 1990s (Denman et al., 2007) and, most recently, the period 2002-2011 . The IPCC methodology has been adapted and used by the Global Carbon Project (GCP, www.globalcarbonproject.org), which has coordinated a cooperative community effort for the annual publication of global carbon budgets up to year 2005  including fossil emissions only), 2006 GCP, 2007), 2008 (Friedlingstein et al., 2010, year 2010 Consumption-based fossil fuel and cement emissions (E FF ) by country (consumption) Peters et al. (2011b) updated as described in this paper Land-use-change emissions (E LUC ) Houghton et al. (2012) combined with Giglio et al. (2013) Atmospheric CO 2 growth rate (G ATM ) Dlugokencky and Tans (2014; NOAA/ESRL: www.esrl.noaa.gov/gmd/ccgg/trends/) Ocean and land CO 2 sinks (S OCEAN and S LAND ) This paper for S OCEAN and S LAND and references in Table 6 for individual models. (Peters et al., 2012b), 2012 (Le , and, most recently, 2013 (Le , where the carbon budget year refers to the initial year of publication. Each of these papers updated previous estimates with the latest available information for the entire time series. From 2008, these publications projected fossil fuel emissions for one additional year using the projected world gross domestic product (GDP) and estimated improvements in the carbon intensity of the global economy. We adopt a range of ±1 standard deviation (σ ) to report the uncertainties in our estimates, representing a likelihood of 68 % that the true value will be within the provided range if the errors have a Gaussian distribution. This choice reflects the difficulty of characterising the uncertainty in the CO 2 fluxes between the atmosphere and the ocean and land reservoirs individually, particularly on an annual basis, as well as the difficulty of updating the CO 2 emissions from LUC. A likelihood of 68 % provides an indication of our current capability to quantify each term and its uncertainty given the available information. For comparison, the Fifth Assessment Report of the IPCC (AR5) generally reported a likelihood of 90 % for large data sets whose uncertainty is well characterised, or for long time intervals less affected by year-to-year variability. Our 68 % uncertainty value is near the 66 % which the IPCC characterises as "likely" for values falling into the ±1σ interval. The uncertainties reported here combine statistical analysis of the underlying data and expert judgement of the likelihood of results lying outside this range. The limitations of current information are discussed in the paper.
All quantities are presented in units of gigatonnes of carbon (GtC, 10 15 gC), which is the same as petagrams of carbon (PgC; Table 1). Units of gigatonnes of CO 2 (or billion tonnes of CO 2 ) used in policy are equal to 3.664 multiplied by the value in units of GtC.
This paper provides a detailed description of the data sets and methodology used to compute the global carbon budget estimates for the period pre-industrial (1750) to 2013 and in more detail for the period 1959 to 2013. We also provide decadal averages starting in 1960 and including the last decade (2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013), results for the year 2013, and a projection of E FF for year 2014. Finally, we provide the total or cumulative emissions from fossil fuels and land-use change since the year 1750; the pre-industrial period; and since year 1870, the reference year for the cumulative carbon estimate used by the IPCC (AR5) based on the availability of global temperature data (Stocker et al., 2013b). This paper will be updated every year using the format of "living data" so as to keep a record of budget versions and the changes in new data, revision of data, and changes in methodology that lead to changes in estimates of the carbon budget. Additional materials associated with the release of each new version will be posted on the Global Carbon Project (GCP) website (http://www.globalcarbonproject.org/carbonbudget). Data associated with this release are also available through the Global Carbon Atlas (http://www.globalcarbonatlas.org). With this approach, we aim to provide the highest trans-parency and traceability in the reporting of key indicators and drivers of climate change.

Methods
Multiple organisations and research groups around the world generated the original measurements and data used to complete the global carbon budget. The effort presented here is thus mainly one of synthesis, where results from individual groups are collated, analysed, and evaluated for consistency. We facilitate access to original data with the understanding that primary data sets will be referenced in future work (see Table 2 for how to cite the data sets). Descriptions of the measurements, models, and methodologies follow below, and in-depth descriptions of each component are described elsewhere (e.g. Andres et al., 2012;Houghton et al., 2012). This is the ninth version of the "global carbon budget" (see Introduction for details) and the third revised version of the "global carbon budget living data paper". It is an update of Le , including data to year 2013 (inclusive) and a projection for fossil fuel emissions for year 2014. The main changes from Le  are as follows: (1) we use 3 years of BP energy consumption growth rates (coal, oil, gas) to estimate E FF compared to 2 years in the previous version (Sect. 2.1), (2) we updated S OCEAN estimates from observations to 2013 extending the Surface Ocean CO 2 Atlas (SOCAT) v2 database Sect. 2.4) with additional new cruises, and (3) we introduced results from three atmospheric inverse methods using atmospheric measurements from a global network of surface stations through 2013 that provide a latitudinal breakdown of the combined land and ocean fluxes (Sect. 2.6). The main methodological differences between annual carbon budgets are summarised in Table 3. 2.1 CO 2 emissions from fossil fuel combustion and cement production (E FF )

Fossil fuel and cement emissions and their uncertainty
The calculation of global and national CO 2 emissions from fossil fuel combustion, including gas flaring and cement production (E FF ), relies primarily on energy consumption data, specifically data on hydrocarbon fuels, collated and archived by several organisations (Andres et al., 2012). These include the Carbon Dioxide Information Analysis Center (CDIAC), the International Energy Agency (IEA), the United Nations (UN), the United States Department of Energy (DoE) Energy Information Administration (EIA), and more recently also the Planbureau voor de Leefomgeving (PBL) of the Netherlands Environmental Assessment Agency. We use the emissions estimated by the CDIAC . The CDIAC emission estimates constitute the only data set that extends back in time to 1751 with consistent and well-documented emissions from fossil fuel combustion, cement production, and gas flaring for all countries and their uncertainty (Andres et al., 1999(Andres et al., , 2012; this makes the data set a unique resource for research of the carbon cycle during the fossil fuel era. During the period 1959-2010, the emissions from fossil fuel consumption are based primarily on energy data provided by the UN Statistics Division (Table 4; UN, 2013a, b). When necessary, fuel masses/volumes are converted to fuel energy content using coefficients provided by the UN and then to CO 2 emissions using conversion factors that take into account the relationship between carbon content and energy (heat) content of the different fuel types (coal, oil, gas, gas flaring) and the combustion efficiency (to account, for example, for soot left in the combustor or fuel otherwise lost or discharged without oxidation). Most data on energy consumption and fuel quality (carbon content and heat content) are available at the country level (UN, 2013a). In general, CO 2 emissions for equivalent primary energy consumption are about 30 % higher for coal compared to oil, and 70 % higher for coal compared to natural gas . All estimated fossil fuel emissions are based on the mass flows of carbon and assume that the fossil carbon emitted as CO or CH 4 will soon be oxidised to CO 2 in the atmosphere and can be accounted for with CO 2 emissions (see Sect. 2.7).
For the three most recent years (2011, 2012, and 2013) when the UN statistics are not yet available, we generated preliminary estimates based on the BP annual energy review by applying the growth rates of energy consumption (coal, oil, gas) for -2013(BP, 2014 to the CDIAC emissions in 2010. BP's sources for energy statistics overlap with those of the UN data but are compiled more rapidly from about 70 countries covering about 96 % of global emissions. We use the BP values only for the year-to-year rate of change, because the rates of change are less uncertain than the absolute values and to avoid discontinuities in the time series when linking the UN-based energy data (up to 2010) with the BP energy data (2011)(2012)(2013). These preliminary estimates are replaced with the more complete CDIAC data based on UN statistics when they become available. Past experience and work by others  shows that projections based on the BP rate of change are within the uncertainty provided (see Sect. 3.2 and Supplement from Peters et al., 2013).
Emissions from cement production are based on cement production data from the U.S. Geological Survey up to year 2012 (van Oss, 2013), and up to 2013 for the top 18 countries (representing 85 % of global production; USGS, 2014). For countries without data in 2013 we use the 2012 values (zero growth). Some fraction of the CaO and MgO in cement is returned to the carbonate form during cement weathering, but this is generally regarded to be small and is ignored here.
Emission estimates from gas flaring are calculated in a similar manner as those from solid, liquid, and gaseous fuels, and rely on the UN Energy Statistics to supply the amount Table 3. Main methodological changes in the global carbon budget since first publication. Unless specified below, the methodology was identical to that described in the current paper. Furthermore, methodological changes introduced in one year are kept for the following years unless noted. Empty cells mean there were no methodological changes introduced that year. of flared or vented fuel. For emission years 2011-2013, flaring is assumed constant from 2010 (emission year) UN-based data. The basic data on gas flaring report atmospheric losses during petroleum production and processing that have large uncertainty and do not distinguish between gas that is flared as CO 2 or vented as CH 4 . Fugitive emissions of CH 4 from the so-called upstream sector (e.g. coal mining and natural gas distribution) are not included in the accounts of CO 2 emissions except to the extent that they are captured in the UN energy data and counted as gas "flared or lost". The published CDIAC data set has 250 countries and regions included. This expanded list includes countries/regions that no longer exist, such as the USSR and East Pakistan. For the budget, we reduce the list to 216 countries by reallocating emissions to the currently defined territories. This involved both aggregation and disaggregation, and does not change global emissions. Examples of aggregation include merging East and West Germany to the currently defined Germany. Examples of disaggregation include reallocating the emissions from the former USSR to the resulting independent countries. For disaggregation, we use the emission shares when the current territory first appeared. For the most recent years, 2011-2013, the BP statistics are more aggregated, but we retain the detail of CDIAC by applying the growth rates of each aggregated region in the BP data set to its constituent individual countries in CDIAC.
Estimates of CO 2 emissions show that the global total of emissions is not equal to the sum of emissions from all countries. This is largely attributable to emissions that occur in international territory, in particular the combustion of fuels used in international shipping and aviation (bunker fuels), where the emissions are included in the global totals but are not attributed to individual countries. In practice, the emissions from international bunker fuels are calculated based on where the fuels were loaded, but they are not included with national emissions estimates. Other differences occur because globally the sum of imports in all countries is not equal to the sum of exports and because of differing treatment of oxidation of non-fuel uses of hydrocarbons (e.g. as solvents, lubricants, feedstocks), and changes in stock (Andres et al., 2012).
The uncertainty in the annual fossil fuel and cement emissions for the globe has been estimated at ±5 % (scaled down from the published ±10 % at ±2σ to the use of ±1σ bounds reported here; Andres et al., 2012). This is consistent with a more detailed recent analysis of uncertainty of ±8.4 % at ±2σ . This includes an assessment of uncertainties in the amounts of fuel consumed, the carbon and heat contents of fuels, and the combustion efficiency. While in the budget we consider a fixed uncertainty of ±5 % for all years, in reality the uncertainty, as a percentage of the emissions, is growing with time because of the larger share of global emissions from non-Annex B countries (emerging economies and developing countries) with less precise statistical systems . For example, the uncertainty in Chinese emissions has been estimated at around ±10 % (for ±1σ ; Gregg et al., 2008). Generally, emissions from mature economies with good statistical bases have an uncertainty of only a few percent (Marland, 2008). Further research is needed before we can quantify the time evolution of the uncertainty, and its temporal error correlation structure. We note that, even if they are presented as 1σ estimates, uncertainties of emissions are likely to be mainly country-specific systematic errors related to underlying biases of energy statistics and to the accounting method used by each country. We assign a medium confidence to the results presented here because they are based on indirect estimates of emissions using energy data (Durant et al., 2010). There is only limited and indirect evidence for emissions, although there is a high agreement among the available estimates within the given uncertainty (Andres et al., 2012, and emission estimates are consistent with a range of other observations , even though their re-gional and national partitioning is more uncertain (Francey et al., 2013).

Emissions embodied in goods and services
National emission inventories take a territorial (production) perspective and "include greenhouse gas emissions and removals taking place within national territory and offshore areas over which the country has jurisdiction" (Rypdal et al., 2006). That is, emissions are allocated to the country where and when the emissions actually occur. The territorial emission inventory of an individual country does not include the emissions from the production of goods and services produced in other countries (e.g. food and clothes) that are used for consumption. Consumption-based emission inventories for an individual country is another attribution point of view that allocates global emissions to products that are consumed within a country, and are conceptually calculated as the territorial emissions minus the "embedded" territorial emissions to produce exported products plus the emissions in other countries to produce imported products (consumption = territorial − exports + imports). The difference between the territorial-and consumption-based emission inventories is the net transfer (exports minus imports) of emissions from the production of internationally traded products. Consumption-based emission attribution results (e.g. Davis and Caldeira, 2010) provide additional information on territorial-based emissions that can be used to understand emission drivers (Hertwich and , quantify emission (virtual) transfers by the trade of products between countries (Peters et al., 2011b), and potentially design more effective and efficient climate policy (Peters and Hertwich, 2008). We estimate consumption-based emissions by enumerating the global supply chain using a global model of the economic relationships between economic sectors within and between every country Peters et al., 2011a). Due to availability of the input data, detailed estimates are made for the years 1997, 2001, 2004, and 2007 (using  The consumption-based emission inventories in this carbon budget incorporate several improvements over previous versions (Le Peters et al., 2011bPeters et al., , 2012b. The detailed estimates for 2004 and 2007 and time series approximation from 1990 to 2012 are based on an updated version of the GTAP database (Narayanan et al., 2013). We estimate the sector level CO 2 emissions using our own calculations based on the GTAP data and methodology, include flaring and cement emissions from CDIAC, and then scale the national totals (excluding bunker fuels) to match the CDIAC estimates from the most recent carbon budget. We do not include international transportation in our estimates of national totals, but we do include them in the global total. The time series of trade data provided by GTAP covers the period 1995-2009 and our methodology uses the trade shares as this data set. We do not provide an uncertainty estimate for these emissions, but based on model comparisons and sensitivity analysis, they are unlikely to be larger than for the territorial emission estimates (Peters et al., 2012a). Uncertainty is expected to increase for more detailed results and decrease with aggregation (Peters et al., 2011b; e.g. the results for Annex B countries will be more accurate than the sector results for an individual country).
The consumption-based emissions attribution method considers the CO 2 emitted to the atmosphere in the production of products, but not the trade in fossil fuels (coal, oil, gas). It is also possible to account for the carbon trade in fossil fuels (Davis et al., 2011), but we do not present those data here. Peters et al. (2012a) additionally considered trade in biomass.
The consumption data do not modify the global average terms in Eq. (1), but they are relevant to the anthropogenic carbon cycle as they reflect the trade-driven movement of emissions across the Earth's surface in response to human activities. Furthermore, if national and international climate policies continue to develop in an unharmonised way, then the trends reflected in these data will need to be accommodated by those developing policies.

Growth rate in emissions
We report the annual growth rate in emissions for adjacent years (in percent per year) by calculating the difference between the 2 years and then comparing to the emissions in the first year: × % yr −1 . This is the simplest method to characterise a 1-year growth compared to the previous year and is widely used. We apply a leap-year adjustment to ensure valid interpretations of annual growth rates. This affects the growth rate by about 0.3 % yr −1 ( 1 365 ) and causes growth rates to go up approximately 0.3 % if the first year is a leap year and down 0.3 % if the second year is a leap year.
The relative growth rate of E FF over time periods of greater than 1 year can be re-written using its logarithm equivalent as follows: (2) Here we calculate relative growth rates in emissions for multi-year periods (e.g. a decade) by fitting a linear trend to ln(E FF ) in Eq.
(2), reported in percent per year. We fit the logarithm of E FF rather than E FF directly because this method ensures that computed growth rates satisfy Eq. (6). This method differs from previous papers Raupach et al., 2007) that computed the fit to E FF and divided by average E FF directly, but the difference is very small (< 0.05 %) in the case of E FF .

Emissions projections using GDP projections
Energy statistics are normally available around June for the previous year. We use the close relationship between the growth in world GDP and the growth in global emissions  to project emissions for the current year. This is based on the so-called Kaya identity (also called IPAT identity, the acronym standing for human impact (I) on the environment, which is equal to the product of population (P), affluence (A), and technology (T), whereby E FF (GtC yr −1 ) is decomposed by the product of GDP (USD yr −1 ) and the fossil fuel carbon intensity of the economy (I FF ; GtC USD −1 ) as follows: Such product-rule decomposition identities imply that the relative growth rates of the multiplied quantities are additive. Taking a time derivative of Eq. (3) gives and, applying the rules of calculus, finally, dividing Eq. (5) by Eq. (3) gives where the left-hand term is the relative growth rate of E FF , and the right-hand terms are the relative growth rates of GDP and I FF , respectively, which can simply be added linearly to give overall growth rate. The growth rates are reported in percent by multiplying each term by 100. As preliminary estimates of annual change in GDP are made well before the end of a calendar year, making assumptions on the growth rate of I FF allows us to make projections of the annual change in CO 2 emissions well before the end of a calendar year.
2.2 CO 2 emissions from land use, land-use change, and forestry (E LUC ) LUC emissions reported in the 2014 carbon budget (E LUC ) include CO 2 fluxes from deforestation, afforestation, logging (forest degradation and harvest activity), shifting cultivation (cycle of cutting forest for agriculture and then abandoning), and regrowth of forests following wood harvest or abandonment of agriculture. Only some land management activities (Table 5) are included in our LUC emissions estimates (e.g. emissions or sinks related to management and management changes in established pasture and croplands are not included). Some of these activities lead to emissions of CO 2 to the atmosphere, while others lead to CO 2 sinks. E LUC is the net sum of all anthropogenic activities considered. Our annual estimate for 1959-2010 is from a bookkeeping method (Sect. 2.2.1) primarily based on net forest area change and biomass data from the Forest Resource Assessment (FRA) of the Food and Agriculture Organization (FAO), which is only available at intervals of 5 years and ends in 2010 . Interannual variability in emissions due to deforestation and degradation have been coarsely estimated from satellite-based fire activity in tropical forest areas (Sect. 2.2.2; Giglio et al., 2013;. The bookkeeping method is used to quantify the E LUC over the time period of the available data, and the satellite-based deforestation fire information to incorporate interannual variability (E LUC flux annual anomalies) from tropical deforestation fires. The satellite-based deforestation and degradation fire emissions estimates are available for years 1997-2013. We calculate the global annual anomaly in deforestation and degradation fire emissions in tropical forest regions for each year, compared to the 1997-2010 period, and add this annual flux anomaly to the E LUC estimated using the bookkeeping method that is available up to 2010 only and assumed constant at the 2010 value during the period 2011-2013. We thus assume that all land management activities apart from deforestation and degradation do not vary significantly on a year-to-year basis. Other sources of interannual variability (e.g. the impact of climate variability on regrowth fluxes) are accounted for in S LAND . In addition, we use results from dynamic global vegetation models (see Sect. 2.2.3 and Table 6) that calculate net LUC CO 2 emissions in response to land-cover-change reconstructions prescribed to each model in order to help quantify the uncertainty in E LUC and to explore the consistency of our understanding. The three methods are described below, and differences are discussed in Sect. 3.2.

Bookkeeping method
LUC CO 2 emissions are calculated by a bookkeeping method approach (Houghton, 2003) that keeps track of the carbon stored in vegetation and soils before deforestation or other land-use change, and the changes in forest age classes, or cohorts, of disturbed lands after land-use change including possible forest regrowth after deforestation. It tracks the CO 2 emitted to the atmosphere immediately during deforestation, and over time due to the follow-up decay of soil and vegetation carbon in different pools, including wood product pools after logging and deforestation. It also tracks the regrowth of vegetation and associated build-up of soil carbon pools after LUC. It considers transitions between forests, pastures, and cropland; shifting cultivation; degradation of forests where a fraction of the trees is removed; abandonment of agricultural land; and forest management such as wood harvest and, in the USA, fire management. In addition to tracking logging debris on the forest floor, the bookkeeping method tracks the fate of carbon contained in harvested wood products that is eventually emitted back to the atmosphere as CO 2 , although a detailed treatment of the lifetime in each product pool is not performed (Earles et al., 2012). Harvested wood products are partitioned into three pools with different turnover times. All fuel wood is assumed burned in the year of harvest (1.0 yr −1 ). Pulp and paper products are oxidised at a rate of 0.1 yr −1 , timber is assumed to be oxidised at a rate of 0.01 yr −1 , and elemental carbon decays at 0.001 yr −1 . The general assumptions about partitioning wood products among these pools are based on national harvest data (Houghton, 2003). The primary land-cover-change and biomass data for the bookkeeping method analysis are from the Forest Resource Table 5. Comparison of the processes included in the E LUC of the global carbon budget and the DGVMs. See Table 6 for model references. All models include deforestation and forest regrowth after abandonment of agriculture (or from afforestation activities on agricultural land). Assessment of the FAO, which provides statistics on forestcover change and management at intervals of 5 years (FAO, 2010). The data are based on countries' self-reporting, some of which includes satellite data in more recent assessments (Table 4). Changes in land cover other than forest are based on annual, national changes in cropland and pasture areas reported by the FAO Statistics Division (FAOSTAT, 2010). LUC country data are aggregated by regions. The carbon stocks on land (biomass and soils), and their response functions subsequent to LUC, are based on FAO data averages per land-cover type, biome, and region. Similar results were obtained using forest biomass carbon density based on satellite data (Baccini et al., 2012). The bookkeeping method does not include land ecosystems' transient response to changes in climate, atmospheric CO 2 , and other environmental factors, but the growth/decay curves are based on contemporary data that will implicitly reflect the effects of CO 2 and climate at that time. Results from the bookkeeping method are available from 1850 to 2010.

Fire-based method
LUC-associated CO 2 emissions calculated from satellitebased fire activity in tropical forest areas  provide information on emissions due to tropical deforestation and degradation that are complementary to the bookkeeping approach. They do not provide a direct estimate of E LUC as they do not include non-combustion processes such as respiration, wood harvest, wood products, and forest regrowth. Legacy emissions such as decomposition from onground debris and soils are not included in this method either. However, fire estimates provide some insight into the year-to- year variations in the sub-component of the total E LUC flux that result from immediate CO 2 emissions during deforestation caused, for example, by the interactions between climate and human activity (e.g. there is more burning and clearing of forests in dry years) that are not represented by other methods. The "deforestation fire emissions" assume an important role of fire in removing biomass in the deforestation process, and thus can be used to infer gross instantaneous CO 2 emissions from deforestation using satellite-derived data on fire activity in regions with active deforestation. The method requires information on the fraction of total area burned associated with deforestation versus other types of fires, and this information can be merged with information on biomass stocks and the fraction of the biomass lost in a deforestation fire to estimate CO 2 emissions. The satellite-based deforestation fire emissions are limited to the tropics, where fires result mainly from human activities. Tropical deforestation is the largest and most variable single contributor to E LUC . Fire emissions associated with deforestation and tropical peat burning are based on the Global Fire Emissions Database (GFED) described in  but with updated burned area (Giglio et al., 2013) as well as burned area from relatively small fires that are detected by satellite as thermal anomalies but not mapped by the burnedarea approach (Randerson et al., 2012). The burned-area information is used as input data in a modified version of the satellite-driven Carnegie-Ames-Stanford Approach (CASA) biogeochemical model to estimate carbon emissions associated with fires, keeping track of what fraction of fire emissions was due to deforestation (see ). The CASA model uses different assumptions to compute    to include full carbon, nitrogen, and phosphorus cycle (Wang et al., 2010) and land cover and land-cover change. CLM4.5BGC a Oleson et al. (2013) Updated model from CLM4.0CN to CLM4.5BGC. Major changes include revised photosynthesis, slower turnover times for decomposition of litter and SOM, vertically resolved soil biogeochemistry, revised soil denitrification and nitrification, new fire model, and revised frozen-soil hydrology. As shown in Koven et al. (2013), these changes collectively bring model into better agreement with 20th century C budget. ISAM Jain et al. (2013) Ito and Inatomi (2012). h With no nutrient restoring below the mixed layer depth. i Referred to as LSCE in previous carbon budgets. j With updates to the physical model as described in Tjiputra et al. (2013). k Further information (e.g. physical evaluation) for CNRM model can be found in Danabasoglu et al. (2014). l Using winds from Atlas et al. (2011). m Updated version "s81_v3.6gcp". n The MACC v13.1 CO 2 inversion system, initially described by Chevallier et al. (2005), relies on the global tracer transport model LMDZ (Hourdin et al., 2006; see also Supplement to Peylin et al., 2013).
decay functions compared to the bookkeeping method, and does not include historical emissions or regrowth from landuse change prior to the availability of satellite data. Comparing coincident CO emissions and their atmospheric fate with satellite-derived CO concentrations allows for some validation of this approach (e.g. van der Werf et al., 2008). Results from the fire-based method to estimate LUC emissions anomalies added to the bookkeeping mean E LUC estimate are available from 1997 to 2013. Our combination of LUC CO 2 emissions where the variability of annual CO 2 deforestation emissions is diagnosed from fires assumes that year-to-year variability is dominated by variability in deforestation.

Dynamic global vegetation models (DGVMs)
LUC CO 2 emissions have been estimated using an ensemble of seven DGVMs. New model experiments up to year 2013 have been coordinated by the project "Trends and drivers of the regional-scale sources and sinks of carbon dioxide" (TRENDY; http://dgvm.ceh.ac.uk/node/9). We use only models that have estimated LUC CO 2 emissions and the terrestrial residual sink following the TRENDY protocol (see Sect. 2.5.2), thus providing better consistency in the assessment of the causes of carbon fluxes on land. Models use their latest configurations, summarised in Tables 5 and 6. The DGVMs were forced with historical changes in landcover distribution, climate, atmospheric CO 2 concentration, and N deposition. As further described below, each historical DGVM simulation was repeated with a time-invariant pre-industrial land-cover distribution, allowing for estimation of, by difference with the first simulation, the dynamic evolution of biomass and soil carbon pools in response to prescribed land-cover change. All DGVMs represent deforestation and (to some extent) regrowth, the most important components of E LUC , but they do not represent all processes resulting directly from human activities on land (Table 5). DGVMs represent processes of vegetation growth and mortality, as well as decomposition of dead organic matter associated with natural cycles, and include the vegetation and soil carbon response to increasing atmospheric CO 2 levels and to climate variability and change. In addition, four models explicitly simulate the coupling of C and N cycles and account for atmospheric N deposition ( Table 5). The DGVMs are independent of the other budget terms except for their use of atmospheric CO 2 concentration to calculate the fertilisation effect of CO 2 on primary production.
The DGVMs used a consistent land-use-change data set (Hurtt et al., 2011), which provided annual, half-degree, fractional data on cropland, pasture, and primary and secondary vegetation, as well as all underlying transitions between landuse states, including wood harvest and shifting cultivation. This data set used the HYDE  spatially gridded maps of cropland, pasture, and ice/water fractions of each grid cell as an input. The HYDE data are based on annual FAO statistics of change in agricultural area (FAOSTAT, 2010). For the years 2011, 2012, and 2013, the HYDE data set was extrapolated by country for pastures and cropland separately based on the trend in agricultural area over the previous 5 years. The HYDE data set is independent of the data set used in the bookkeeping method (Houghton, 2003, and updates), which is based primarily on forest area change statistics (FAO, 2010). Although the Hurtt land-usechange data set indicates whether land-use changes occur on forested or non-forested land, typically only the changes in agricultural areas are used by the models and are implemented differently within each model (e.g. an increased cropland fraction in a grid cell can be at the expense of either grassland or forest, the latter resulting in deforestation; land-cover fractions of the non-agricultural land differ between models). Thus the DGVM forest area and forest area change over time is not consistent with the Forest Resource Assessment of the FAO forest area data used for the bookkeeping model to calculate E LUC . Similarly, model-specific assumptions are applied to convert deforested biomass or deforested area, and other forest product pools, into carbon in some models (Table 5).
The DGVM model runs were forced by either 6-hourly CRU-NCEP or monthly temperature, precipitation, and cloud cover fields (transformed into incoming surface radiation) based on observations and provided on a 0.5 • × 0.5 • grid and updated to 2013 (CRU TS3.22; Harris et al., 2014). The forcing data include both gridded observations of climate and global atmospheric CO 2 , which change over time (Dlugokencky and Tans, 2014), and N deposition (as used in 4 models, Table 5; Lamarque et al., 2010). E LUC is diagnosed in each model by the difference between a model simulation with prescribed historical land-cover change and a simulation with constant, pre-industrial land-cover distribution. Both simulations were driven by changing atmospheric CO 2 , climate, and, in some models, N deposition over the period 1860-2013. Using the difference between these two DGVM simulations to diagnose E LUC is not consistent with the definition of E LUC in the bookkeeping method (Gasser and Ciais, 2013;Pongratz et al., 2014). The DGVM approach to diagnose land-use-change CO 2 emissions would be expected to produce systematically higher E LUC emissions than the bookkeeping approach if all the parameters of the two approaches were the same (which is not the case). Here, given the different input data of DGVMs and the bookkeeping approach, this systematic difference cannot be quantified.

Uncertainty assessment for E LUC
Differences between the bookkeeping, the addition of firebased interannual variability to the bookkeeping, and DGVM methods originate from three main sources: the land-coverchange data set, the different approaches used in models, and the different processes represented (Table 5). We examine the results from the seven DGVM models and of the bookkeeping method to assess the uncertainty in E LUC . Table 7. Comparison of results from the bookkeeping method and budget residuals with results from the DGVMs and inverse estimates for the periods 1960-1969, 1970-1979, 1980-1989, 1990-1999, 2000-2009, last decade, and last year available. All values are in GtC yr −1 . The DGVM uncertainties represents ±1σ of results from the nine individual models; for the inverse models all three results are given where available.
The uncertainties in the annual E LUC estimates are examined using the standard deviation across models, which ranged from 0.3 to 1.1 GtC yr −1 , with an average of 0.7 GtC yr −1 from 1959 to 2013 (Table 7). The mean of the multi-model E LUC estimates is the same as the mean of the bookkeeping estimate from the budget (Eq. 1) at 1.3 GtC for 1959 to 2010. The multi-model mean and bookkeeping method differ by less than 0.5 GtC yr −1 over 90 % of the time. Based on this comparison, we assess that an uncertainty of ±0.5 GtC yr −1 provides a semi-quantitative measure of uncertainty for annual emissions and reflects our best value judgment that there is at least 68 % chance (±1σ ) that the true LUC emission lies within the given range, for the range of processes considered here. This is consistent with the uncertainty analysis of Houghton et al. (2012), which partly reflects improvements in data on forest area change using data, and partly more complete understanding and representation of processes in models. The uncertainties in the decadal mean estimates from the DGVM ensemble are likely correlated between decades, and thus we apply the annual uncertainty as a measure of the decadal uncertainty. The correlations between decades come from (1) common biases in system boundaries (e.g. not counting forest degradation in some models); (2) common definition for the calculation of E LUC from the difference of simulations with and without LUC (a source of bias vs. the unknown truth); and (3) common and uncertain land-cover-change input data which also cause a bias (though if a different input data set is used each decade, decadal fluxes from DGVMs may be partly decorrelated); and (4) model structural errors (e.g. systematic errors in biomass stocks). In addition, errors arising from uncertain DGVM parameter values would be random, but they are not accounted for in this study, since no DGVM provided an ensemble of runs with perturbed parameters.
Prior to 1959, the uncertainty in E LUC is taken as ±33 %, which is the ratio of uncertainty to mean from the 1960s (Table 7), the first decade available. This ratio is consistent with the mean standard deviation of DGMVs' LUC emissions over 1870-1958 over the multi-model mean (0.94 GtC).

Atmospheric CO 2 growth rate (G ATM )
Global atmospheric CO 2 growth rate estimates The atmospheric CO 2 growth rate is provided by the US National Oceanic and Atmospheric Administration Earth System Research Laboratory (NOAA/ESRL; Dlugokencky and Tans, 2014), which is updated from Ballantyne et al. (2012). For the 1959-1980 period, the global growth rate is based on measurements of atmospheric CO 2 concentration averaged from the Mauna Loa and South Pole stations, as observed by the CO 2 Program at Scripps Institution of Oceanography (Keeling et al., 1976). For the 1980-2012 time period, the global growth rate is based on the average of multiple stations selected from the marine boundary layer sites with well-mixed background air (Ballantyne et al., 2012), after fitting each station with a smoothed curve as a function of time, and averaging by latitude band (Masarie and Tans, 1995). The annual growth rate is estimated by Dlugokencky and Tans (2014) from atmospheric CO 2 concentration by taking the average of the most recent December-January months corrected for the average seasonal cycle and subtracting this same average 1 year earlier. The growth rate in units of ppm yr −1 is converted to units of GtC yr −1 by mul-tiplying by a factor of 2.120 GtC ppm −1 (Prather et al., 2012) for consistency with the other components.
The uncertainty around the annual growth rate based on the multiple stations data set ranges between 0.11 and 0.72 GtC yr −1 , with a mean of 0.60 GtC yr −1 for 1959-1980 and 0.19 GtC yr −1 for 1980-2013, when a larger set of stations were available (Dlugokencky and Tans, 2014). It is based on the number of available stations, and thus takes into account both the measurement errors and data gaps at each station. This uncertainty is larger than the uncertainty of ±0.1 GtC yr −1 reported for decadal mean growth rate by the IPCC because errors in annual growth rate are strongly anti-correlated in consecutive years, leading to smaller errors for longer timescales. The decadal change is computed from the difference in concentration 10 years apart based on a measurement error of 0.35 ppm. This error is based on offsets between NOAA/ESRL measurements and those of the World Meteorological Organization World Data Center for Greenhouse Gases (NOAA/ESRL, 2014) for the start and end points (the decadal change uncertainty is the 2(0.35 ppm) 2 (10 yr) −1 assuming that each yearly measurement error is independent). This uncertainty is also used in Table 8.
The contribution of anthropogenic CO and CH 4 is neglected from the global carbon budget (see Sect. 2.7.1). We assign a high confidence to the annual estimates of G ATM because they are based on direct measurements from multiple and consistent instruments and stations distributed around the world (Ballantyne et al., 2012).
In order to estimate the total carbon accumulated in the atmosphere since 1750 or 1870, we use an atmospheric CO 2 concentration of 277 ± 3 or 288 ± 3 ppm, respectively, based on a cubic spline fit to ice core data (Joos and Spahni, 2008). The uncertainty of ±3 ppm (converted to ±1σ ) is taken directly from the IPCC's assessment . Typical uncertainties in the atmospheric growth rate from ice core data are ±1-1.5 GtC decade −1 as evaluated from the Law Dome data (Etheridge et al., 1996) for individual 20-year intervals over the period from 1870 to 1960 (Bruno and Joos, 1997).

Ocean CO 2 sink
Estimates of the global ocean CO 2 sink are based on a combination of a mean CO 2 sink estimate for the 1990s from observations and a trend and variability in the ocean CO 2 sink for 1959-2013 from seven global ocean biogeochemistry models. We use three observation-based estimated of S OCEAN available for the recent decade(s) to provide a qualitative assessment of confidence in the reported results.

Observation-based estimates
A mean ocean CO 2 sink of 2.2 ± 0.4 GtC yr −1 for the 1990s was estimated by the IPCC (Denman et al., 2007) based on indirect observations and their spread: ocean-land CO 2 sink partitioning from observed atmospheric O 2 / N 2 concentration trends (Keeling et al., 2011;Manning and Keeling, 2006), an oceanic inversion method constrained by ocean biogeochemistry data (Mikaloff Fletcher et al., 2006), and a method based on penetration time scale for CFCs (McNeil et al., 2003). This is comparable with the sink of 2.0 ± 0.5 GtC yr −1 estimated by Khatiwala et al. (2013) for the 1990s, and with the sink of 1.9 to 2.5 estimated from a range of methods for the period 1990period -2009period (Wanninkhof et al., 2013, with uncertainties ranging from ±0.3 to ±0.7 GtC yr −1 . The most direct way to estimate the observation-based ocean sink is from the product of (seaair pCO 2 difference) × (gas transfer coefficient). Estimates based on sea-air pCO 2 are fully consistent with indirect observations (Zeng et al., 2005), but their uncertainty is larger mainly due to difficulty in capturing complex turbulent processes in the gas transfer coefficient .
Two of the three observation-based estimates computed the interannual variability in the ocean CO 2 sink using interpolated measurements of surface ocean fugacity of CO 2 (pCO 2 corrected for the non-ideal behaviour of the gas; Pfeil et al., 2013). The measurements were from the Surface Ocean CO 2 Atlas (SOCAT v2; Bakker et al., 2014), which contains data to the end of 2011. This was extended with 2.4 million additional measurements from 2012 and 2013 from all basins (see data attribution table in Appendix A), submitted to SOCAT but not yet fully quality-controlled following standard SOCAT procedures. Revisions and corrections to measurements from before 2012 were also included where they were available. All new data were subjected to an automated quality control system to detect and remove the most obvious errors (e.g. incorrect reporting of metadata such as position, wrong units, clearly unrealistic data). The combined SOCAT v2 and preliminary 2012-2013 data were implemented in an inversion method ) and a combined self-organising map and feed-forward neural network . The observationbased estimates were corrected to remove a background (not part of the anthropogenic ocean flux) ocean source of CO 2 to the atmosphere of 0.45 GtC yr −1 from river input to the ocean  so as to make them comparable to S OCEAN , which only represents the annual uptake of anthropogenic CO 2 by the ocean.
We also compare the results with those of Park et al. (2010) based on regional correlations between surface temperature and pCO 2 , changes in surface temperature observed by satellite, and wind speed estimates also from satellite data for 1990-2009 (Atlas et al., 2011). The product of Park et al. (2010) provides a data-based assessment of the interannual variability combined with a model-based assessment of the trend and mean in S OCEAN . Several other data-based products are in preparation (e.g. Zeng et al., 2014) and the comparison with data products should help constrain the ocean CO 2 sink in the future. Table 8. Decadal mean in the five components of the anthropogenic CO 2 budget for the periods 1960-1969, 1970-1979, 1980-1989, 1990-1999, 2000-2009, last decade, and last year available. All values are in GtC yr −1 . All uncertainties are reported as ±1σ . A data set containing data for each year during 1959-2013 is available at http://cdiac.ornl.gov/GCP/carbonbudget/2014/. Please follow the terms of use and cite the original data sources as specified on the data set.
We use the data-based product of Khatiwala et al. (2009), updated by Khatiwala et al. (2013), to estimate the anthropogenic carbon accumulated in the ocean during 1765-1958 (60.2 GtC) and 1870-1958, and assume an oceanic uptake of 0.4 GtC for 1750-1765, for which time no data are available, based on the mean uptake during 1765-1770. The estimate of Khatiwala et al. (2009) is based on regional disequilibrium between surface pCO 2 and atmospheric CO 2 , and a Green's function utilising transient ocean tracers like CFCs and 14 C to ascribe changes through time. It does not include changes associated with changes in ocean circulation, temperature and climate, but these are thought to be small over the time period considered here . The uncertainty in cumulative uptake of ±20 GtC (converted to ±1σ ) is taken directly from the IPCC's review of the literature (Rhein et al., 2013), or about ±30 % for the annual values (Khatiwala et al., 2009).

Global ocean biogeochemistry models
The trend in the ocean CO 2 sink for 1959-2013 is computed using a combination of seven global ocean biogeochemistry models ( Table 6). The models represent the physical, chemical and biological processes that influence the surface ocean concentration of CO 2 and thus the air-sea CO 2 flux. The models are forced by meteorological reanalysis and atmospheric CO 2 concentration data available for the entire time period. Models do not include the effects of anthropogenic changes in nutrient supply. They compute the air-sea flux of CO 2 over grid boxes of 1 to 4 • in latitude and longitude. The ocean CO 2 sink for each model is normalised to the observa-tions by dividing the annual model values by their observed average over 1990-1999 (obtained from Keeling et al., 2011;Manning and Keeling, 2006;McNeil et al., 2003;Mikaloff Fletcher et al., 2006) and multiplying this by the observationbased estimate of 2.2 GtC yr −1 . The ocean CO 2 sink for each year (t) is therefore where n is the number of models. This normalisation ensures that the ocean CO 2 sink for the global carbon budget is based on observations, whereas the trends and annual values in CO 2 sinks are from model estimates. The normalisation based on a ratio assumes that if models over-or underestimate the sink in the 1990s, it is primarily due to the process of diffusion, which depends on the gradient of CO 2 . Thus a ratio is more appropriate than an offset as it takes into account the time dependence of CO 2 gradients in the ocean. The mean uncorrected ocean CO 2 sink from the seven models for 1990-1999 ranges between 1.5 and 2.6 GtC yr −1 , with a multi-model mean of 1.9 GtC yr −1 .

Uncertainty assessment for S OCEAN
The uncertainty around the mean ocean sink of anthropogenic CO 2 was quantified by Denman et al. (2007) for the 1990s (see Sect. 2.4.1). To quantify the uncertainty around annual values, we examine the standard deviation of the normalised model ensemble. We use further information from the three data-based products to assess the confidence level. The average standard deviation of the ocean model ensemble is 0.15 GtC yr −1 during 1980-2010 (with a maximum of 0.22), but it increases as the model ensemble goes back in time, with a standard deviation of 0.28 GtC yr −1 across models in the 1960s. We estimate that the uncertainty in the annual ocean CO 2 sink is about ±0.5 GtC yr −1 from the fractional uncertainty in the data uncertainty of ±0.4 GtC yr −1 and standard deviation across models of up to ±0.28 GtC yr −1 , reflecting both the uncertainty in the mean sink from observations during the 1990s (Denman et al., 2007;Sect. 2.4.1) and in the interannual variability as assessed by models. We examine the consistency between the variability of the model-based and the data-based products to assess confidence in S OCEAN . The interannual variability of the ocean fluxes (quantified as the standard deviation) of the three data-based estimates for 1990-2009 (when they overlap) is  2010), respectively (simple linear regression), but their mutual correlation ranges between 0.24 and 0.31 only. The use of annual data for the correlation may reduce the strength of the relationship because the dominant source of variability associated with El Niño events is less than 1 year. We assess a medium confidence level to the annual ocean CO 2 sink and its uncertainty because they are based on multiple lines of evidence, and the results are consistent in that the interannual variability in the model and data-based estimates are all generally small compared to the variability in atmospheric CO 2 growth rate. Nevertheless the various results do not show agreement in interannual variability on the global scale or for the relative roles of the annual and decadal variability compared to the trend.

Terrestrial CO 2 sink
The difference between the fossil fuel (E FF ) and LUC net emissions (E LUC ), the growth rate in atmospheric CO 2 concentration (G ATM ), and the ocean CO 2 sink (S OCEAN ) is attributable to the net sink of CO 2 in terrestrial vegetation and soils (S LAND ), within the given uncertainties. Thus, this sink can be estimated as the residual of the other terms in the mass balance budget, as well as directly calculated using DGVMs; alternatively, it can be estimated from inverse models that close a spatio-temporally explicit form of the mass balance in Eq. (1). The residual land sink (S LAND ) is thought to be in part because of the fertilising effect of rising atmospheric CO 2 on plant growth, N deposition, and effects of climate change such as the lengthening of the growing season in northern temperate and boreal areas. S LAND does not include gross land sinks directly resulting from LUC (e.g. regrowth of vegetation) as these are estimated to be part of the net landuse flux (E LUC ). System boundaries make it difficult to exactly attribute CO 2 fluxes on land between S LAND and E LUC (Erb et al., 2013), and by design most of the uncertainties in our method are allocated to S LAND for those processes that are poorly known or represented in models.

Residual of the budget
For 1959-2013, the terrestrial carbon sink was estimated from the residual of the other budget terms by rearranging Eq. (1): The uncertainty in S LAND is estimated annually from the root sum of squares of the uncertainty in the right-hand terms assuming the errors are not correlated. The uncertainty averages to ±0.8 GtC yr −1 over 1959-2013 (Table 7). S LAND estimated from the residual of the budget includes, by definition, all the missing processes and potential biases in the other components of Eq. (8).

DGVMs
A comparison of the residual calculation of S LAND in Eq. (8) with estimates from DGVMs as used to estimate E LUC in Sect. 2.2.3, but here excluding the effects of changes in land cover (using a constant pre-industrial land-cover distribution), provides an independent estimate of the consistency of S LAND with our understanding of the functioning of the terrestrial vegetation in response to CO 2 and climate variability ( Table 7). As described in Sect. 2.2.3, the DGVM runs that exclude the effects of changes in land cover include all climate variability and CO 2 effects over land but do not include reductions in CO 2 sink capacity associated with human activity directly affecting changes in vegetation cover and management, which by design is allocated to E LUC . This effect has been estimated to have led to a reduction in the terrestrial sink by 0.5 GtC yr −1 since 1750 (Gitz and Ciais, 2003). The models in this configuration estimate the mean and variability of S LAND based on atmospheric CO 2 and climate, and thus both terms can be compared to the budget residual.
The multi-DGVM mean of 2.5 ± 1.0 GtC yr −1 for the period 2004-2013 agrees well with the value computed from the budget residual (Table 7). The standard deviation of the annual CO 2 sink across the nine DGVMs ranges from ±0.4 to ±1.4 GtC yr −1 , with a mean standard deviation of ±0.9 GtC yr −1 for the period 1959 to 2013. The model mean, over different decades, correlates with the budget residual with r = 0.71, compared to correlations of r = 0.46 to r = 0.70 (median of 0.61) between individual models. The standard deviation is similar to that of the five model ensembles presented in Le , but the correlation is improved compared to r = 0.54 obtained in the earlier study. The DGVM results suggest that the sum of our knowledge on annual CO 2 emissions and their partitioning is plausible (see Discussion), and provide insight into the underlying processes and regional breakdown. However as the standard deviation across the DGVMs (of ±0.9 GtC yr −1 ) is of the same magnitude as the combined uncertainty due to the other components (E FF , E LUC , G ATM , S OCEAN ; Table 7), the DGVMs do not provide further reduction of uncertainty on the terrestrial CO 2 sink compared to the residual of the budget (Eq. 8). Yet, DGVM results are largely independent of the residual of the budget, and it is worth noting that the residual method and ensemble mean DGVM results are consistent within their respective uncertainties. We assess a medium confidence level to the annual land CO 2 sink and its uncertainty because the estimates from the residual budget and averaged DGVMs match well within their respective uncertainties, and the estimates based on the residual budget are primarily dependent on E FF and G ATM , both of which are well constrained.

The atmospheric perspective
The worldwide network of atmospheric measurements can be used with atmospheric inversion methods to constrain the location of the combined total surface CO 2 fluxes from all sources, including fossil and LUC emissions and land and ocean CO 2 fluxes. As the geographical distribution of fossil fuel emissions is already known, it can be subtracted from the total surface CO 2 flux to provide CO 2 fluxes over land and over the ocean. Here we used preliminary atmospheric CO 2 data to the end of 2013, and three atmospheric CO 2 inversions (Table 6) to infer the total CO 2 flux over land regions, and the distribution of the total land and ocean CO 2 fluxes for the mid-to high-latitude Northern Hemisphere (30-90 • N), tropics (30 • S-30 • N) and mid-to high-latitude region of the Southern Hemisphere (30-90 • S). We focus here on the largest and most consistent sources of information, and use these estimates to comment on the consistency across various data streams and process-based estimates.

Atmospheric inversions
The three inversion systems used in this release (Chevallier et al., 2005;Peters et al., 2010;Rödenbeck, 2005) are based on the same Bayesian inversion principles that interpret the same, for the most part, observed time series (or subsets thereof), but they use different methodologies that represent some of the many approaches used in the field. This mainly concerns the time resolution of the estimates (i.e. weekly or monthly), spatial breakdown (i.e. grid size), assumed cor-relation structures, and mathematical approach. The details of these approaches are documented extensively in the references provided. Each system had used a different transport model, which was demonstrated to be a driving factor behind differences in atmospheric-based flux estimates, and specifically their global distribution (Stephens et al., 2007). Most inverse models use estimates for the ocean and land biosphere, which can be very similar to those described in Sects. 2.4.1 and 2.5.1 to assign prior fluxes. They do not estimate E FF separately but assign E FF using similar data sources to those used described in Sects. 2.1.1 and 2.2.2. Finally atmospheric inversions include CO 2 fluxes from rivers (which need to be taken into account to allow comparison to other sources) and chemical oxidation of reactive carboncontaining gases (which are neglected here). These inverse estimates are not truly independent of the other estimates presented here as the atmospheric observations include a set of observations used to estimate the global atmospheric growth rate (Sect. 2.3). However they provide new information on the regional distribution of fluxes.
In this first application of inverse methods to the carbon budget we focus the analysis on two known strengths of the inverse approach: the derivation of the year-toyear changes in total land (E LUC + S LAND ) fluxes consistent with the whole network of atmospheric observations, and the spatial breakdown of land and ocean fluxes (E LUC + S LAND + S OCEAN ) across large regions of the global. The total land flux correlates well with those estimated from the budget residual (Eq. 1) with corrections for the annual time series ranging from r = 0.84 to 0.93, and with the DGVM multi-model mean with correlations for the annual time series ranging from r = 0.71 to 0.84 (r = 0.37 to 0.82 for individual DGVMs and inversions). The spatial breakdown is discussed in Sect. 3.1.3.

Contribution of anthropogenic CO and CH 4 to the global carbon budget
Anthropogenic emissions of CO and CH 4 to the atmosphere are eventually oxidised to CO 2 and are thus part of the global carbon budget. These contributions are omitted in Eq. (1), but an attempt is made in this section to estimate their magnitude and identify the sources of uncertainty. Anthropogenic CO emissions are from incomplete fossil fuel and biofuel burning and deforestation fires. The main anthropogenic emissions of fossil CH 4 that matter for the global carbon budget are the fugitive emissions of coal, oil, and gas upstream sectors (see below). These emissions of CO and CH 4 contribute a net addition of fossil carbon to the atmosphere. In our estimate of E FF we assumed (Sect. 2.1.1) that all the fuel burned is emitted as CO 2 ; thus CO anthropogenic emissions and their atmospheric oxidation into CO 2 within a few months are already counted implicitly in E FF and should not be counted twice (same for E LUC and anthropogenic CO emissions by deforestation fires). Anthropogenic emissions of fossil CH 4 are not included in E FF , because these fugitive emissions are not included in the fuel inventories. Yet they contribute to the annual CO 2 growth rate after CH 4 gets oxidised into CO 2 . Anthropogenic emissions of fossil CH 4 represent 15 % of total CH 4 emissions (Kirschke et al., 2013), which is 0.061 GtC yr −1 for the past decade. Assuming steady state, these emissions are all converted to CO 2 by OH oxidation, thus explaining 0.06 GtC yr −1 of the global CO 2 growth rate in the past decade.
Other anthropogenic changes in the sources of CO and CH 4 from wildfires, biomass, wetlands, ruminants, or permafrost changes are similarly assumed to have a small effect on the CO 2 growth rate.

Anthropogenic carbon fluxes in the land-to-ocean continuum
The approach used to determine the global carbon budget considers only anthropogenic CO 2 emissions and their partitioning among the atmosphere, ocean, and land. In this analysis, the land and ocean reservoirs that take up anthropogenic CO 2 from the atmosphere are conceived as independent carbon storage repositories. This approach thus omits that carbon is continuously displaced along the land-ocean aquatic continuum (LOAC) comprising freshwaters, estuaries, and coastal areas (Bauer et al., 2013;Regnier et al., 2013). A significant fraction of this lateral carbon flux is entirely "natural" and is thus a steady-state component of the pre-industrial carbon cycle that can be ignored in the current analysis. The remaining fraction is anthropogenic carbon entrained into the lateral transport loop of the LOAC, a perturbation that is relevant for the global carbon budget presented here. The results of the analysis of Regnier et al. (2013) can be summarised in three points of relevance to the anthropogenic CO 2 budget. First, the anthropogenic carbon input from land to hydrosphere, F LH , estimated at 1 ± 0.5 PgC is significant compared to the other terms of Eq. (1) ( Table 8), and implies that only a portion of the anthropogenic CO 2 taken up by land ecosystems remains sequestered in soil and biomass pools. Second, some of the exported anthropogenic carbon is stored in the LOAC ( C LOAC , 0.55 ± 0.3 GtC yr −1 ) and some is released back to the atmosphere as CO 2 (E LOAC , 0.35 ± 0.2 GtC yr −1 ), the magnitude of these fluxes resulting from the combined effects of freshwaters, estuaries, and coastal seas. Third, a small fraction of anthropogenic carbon displaced by the LOAC is transferred to the open ocean, where it accumulates (F HO , 0.1 ± > 0.05 GtC yr −1 ). The anthropogenic perturbation of the carbon fluxes from land to ocean does not contradict the method used in Sect. 2.5 to define the ocean sink and residual land sink. However, it does point to the need to account for the fate of anthropogenic carbon once it is removed from the atmosphere by land ecosystems (summarised in Fig. 2). In theory, direct estimates of changes of the ocean inorganic carbon inventory over time would see the land flux of anthropogenic carbon and would thus have a bias relative to air-sea flux estimates and tracerbased reconstructions. However, currently the value is small enough to be not noticeable relative to the errors in the individual techniques.
More importantly the residual land sink calculated in a budget which accounts for the LOAC (S LAND+LOAC = 3.25 ± 0.9 GtC yr −1 ) is larger than the residual land sink (S LAND ) value of 2.9 ± 0.85 GtC yr −1 (2004-2013) calculated according to Eq. (8). This is because this flux is partially offset by the net source of CO 2 to the atmosphere of 0.35 ± 0.3 GtC yr −1 from rivers, estuaries, and coastal seas (E LOAC ): In addition, because a fraction of anthropogenic CO 2 taken up by land ecosystems is exported to the LOAC (F LH ), the annual terrestrial ecosystems carbon storage change comprising the land vegetation, litter, and soil ( C TE , 1.35 GtC yr −1 ) is notably smaller than what would be calculated in a traditional budget that ignores the LOAC. In this case, the carbon storage change for the period 2004-2013 (Table 8) is estimated to be 2 Gt C yr −1 from the difference between S LAND (2.9 GtC yr −1 ) and E LUC (0.9 GtC yr −1 ). With the LOAC included, we now have A significant fraction of the anthropogenic carbon displaced from land ecosystems to LOAC is stored in freshwater and coastal sediments ( C LOAC ) and, to a lesser extent, in the open ocean (F HO ), while the rest is re-emitted to the atmosphere by freshwaters (E LOAC ). The annual ocean carbon storage change with LOAC included ( C OCEAN ) is therefore equal to C OCEAN = S OCEAN + F HO .
All estimates of LOAC are given with low confidence, because they originate from a single source. The carbon budget presented here implicitly incorporates the fluxes from the LOAC with S LAND . We do not attempt to separate these fluxes, because the uncertainties in either estimate are too large and there is insufficient information available to estimate the LOAC fluxes on an annual basis.

Global carbon budget averaged over decades and its variability
The global carbon budget averaged over the last decade (2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013) is shown in Fig. 2. For this time period, 91 % of the total emissions (E FF + E LUC ) were caused by fossil fuel combustion and cement production, and 9 % by landuse change. The total emissions were partitioned among the atmosphere (44 %), ocean (26 %), and land (29 %). All components except land-use-change emissions have grown since 1959 (Figs. 3 and 4), with important interannual variability in the atmospheric growth rate and in the land CO 2 sink (Fig. 4), as well as some decadal variability in all terms (Table 8).
Both ocean and land CO 2 sinks increased roughly in line with the atmospheric increase, but with significant decadal variability on land (Table 8). The ocean CO 2 sink increased from 1.1 ± 0.5 GtC yr −1 in the 1960s to 2.6 ± 0.5 GtC yr −1  (Chevallier et al., 2005) in red; Rödenbeck et al. (2003) in orange; Peters et al. (2010) in purple; see Table 6), and the carbon balance from Eq. (1) (black). In (c) the inversions were corrected for the pre-industrial land sink of CO 2 from river input by adding a sink of 0.45 GtC yr −1 . This correction does not take into account the anthropogenic contribution to river fluxes (see Sect. 2.7.2).
during 2004-2013, with interannual variations of the order of a few tenths of GtC yr −1 generally showing an increased ocean sink during El Niño events (i.e. 1982-1983, 1991-1993, 1997-1998) (Fig. 7; Rödenbeck et al., 2014). Although there is some coherence between the ocean models and data products and among data products, their mutual correlation is weak and highlights disagreement on the exact amplitude of the interannual variability, as well as on the relative importance of the trend versus the variability (Sect. 2.4.3 and Fig. 7). Most estimates produce a mean CO 2 sink for the 1990s that is below the mean assessed by the IPCC from indirect (but arguably more reliable) observations (Denman et al., 2007;Sect. 2.4.1). This could reflect issues with the vertical diffusion in ocean models, although as the data products also support a lower mean CO 2 sink, this discrepancy suggests we may need to reassess estimates of the mean ocean carbon sinks.
The land CO 2 sink increased from 1.8 ± 0.7 GtC yr −1 in the 1960s to 2.9 ± 0.8 GtC yr −1 during 2004-2013, with important interannual variations of up to 2 GtC yr −1 generally showing a decreased land sink during El Niño events, overcompensating for the increased in ocean sink and accounting for the enhanced atmospheric growth rate during El Niño events . The high-uptake anomaly around year 1991 is thought to be caused by the effect of the volcanic eruption of Mount Pinatubo on climate (Achard et al., 2014;Fig. 6c) and is not generally reproduced by the DGVMs but assigned to S LAND by the two inverse systems that include this period (Fig. 6). The larger land CO 2 sink during 2004-2013 compared to the 1960s is reproduced by all the DGVMs in response to combined atmospheric CO 2 increase, climate, and variability (average change of 1.4 GtC yr −1 ; eight models ranging between 0.8 and 2.3 GtC yr −1 with one model at 0.1 GtC yr −1 ), consistent with the budget residual and reflecting a common knowledge of the processes ( Table 7). The decadal change is also consistent with the results from the atmospheric inversions, which estimate a trend of 0.84 and 0.62 GtC yr −1 per decade for the inversions of Chevallier et al. (2005) and Rödenbeck et al. (2003), respectively.
The total CO 2 fluxes on land (E LUC + S LAND ) constrained by the atmospheric inversions show in general very good agreement with the global budget estimate, as expected given the strong constrains of G ATM and the small relative uncertainty typically assumed on S OCEAN and E FF by inversions. The total sink of similar magnitude for the decadal average, with estimates for 2004-2013 from the inversions of 1.7, 2.0, and 3.1 GtC yr −1 compared to 2.0 ± −0.7 GtC yr −1 for the budget residual (Table 7). The inversions' total land sink would be 1.2, 1.5, and 2.6 GtC yr −1 when including a mean river flux correction of 0.45 GtC yr −1 , though the exact correction would be smaller when taking into account the anthropogenic contribution to river fluxes (Sect. 2.7.2). The interannual variability of the inversions also matched the residual-based S LAND closely (Fig. 6). The multi-model mean from the DGVM ensemble that preformed the LUC simulations also compares well with the estimate from the residual budget and atmospheric inversions, with a decadal mean of 1.4 ± −1.0 GtC yr −1 (Table 7; Figure 7. Comparison of the atmosphere-ocean CO 2 flux shows the budget values of S OCEAN (black line), individual ocean models before normalisation (blue lines), and the three ocean databased products  in orange, Landschützer et al. (2014) in red, andPark et al. (2010) in purple; see Table 6). All data-based products were corrected for the pre-industrial ocean source of CO 2 from river input to the ocean, which is not present in the models, by adding a sink of 0.45 GtC yr −1  in order to make them comparable to S OCEAN . This correction does not take into account the anthropogenic contribution to river fluxes (see Sect. 2.7.2). though individual models differ by several gigatonnes of carbon for some years (Fig. 6).

Distribution
The total surface CO 2 fluxes on land and ocean including LUC (E LUC + S LAND + S OCEAN ) estimated from process models and atmospheric inversions can provide information on the regional distribution of those fluxes by latitude band (Fig. 8). The global mean CO 2 fluxes from process models for 2004-2013 is 2.8 GtC yr −1 , an underestimate compared to the fluxes of 4.5 GtC yr −1 inferred from the remainder of the carbon budget (E FF − G ATM in Eq. 1; Table 8). In contrast, the total CO 2 fluxes from the three inversions range between 4.1 and 4.7 GtC yr −1 , consistent with the carbon budget as expected from the constraints on the inversions.
In the south (south of 30 • S), the atmospheric inversions and combined models all suggest a CO 2 sink for 2004-2013 of between 1.3 and 1.6 GtC yr −1 (Fig. 8), although the details of the interannual variability are not fully consistent across methods. The interannual variability in the south is low because of the dominance of ocean area with low variability compared to land areas. In the tropics (30 • S-30 • N), both the atmospheric inversions and combined models suggest the carbon balance in this region is close to neutral over the past decade, with fluxes for 2004-2013 ranging between  (Chevallier et al., 2005) in red; Rödenbeck et al. (2003) in orange; Peters et al. (2010) in purple; Table 6).
with the ensemble mean of the process models suggesting a smaller total Northern Hemisphere sink for 2004-2013 of 2.0 ± 0.8 GtC yr −1 , while the inversions estimate a sink of between 2.4 and 3.6 GtC yr −1 , though some agreement exists in the interannual variability. The mean difference can only partly be explained by the influence of river fluxes, as this flux in the Northern Hemisphere would be less than 0.45 GtC yr −1 , particularly when the anthropogenic contribution to river fluxes are accounted for. This analysis thus suggests that the global underestimate of process models originates in the north.
In 2013, global CO 2 emissions were dominated by emissions from China (28 %), the USA (14 %), the EU (28 member states; 10 %), and India (7 %) compared to the global total including bunker fuels. These four regions account for 58 % of global emissions. Growth rates for these countries from 2012 to 2013 were 4.2 % (China), 2.9 % (USA), −1.8 % (EU28), and 5.1 % (India). The countries contributing most to the 2013 change in emissions were China (58 % of the increase), the USA (20 % of the increase), India (17 % of the increase), and EU28 (11 % of the decrease). The per Table 9. Actual CO 2 emissions from fossil fuel combustion and cement production (E FF ) compared to projections made the previous year based on world GDP (IMF October 2013) and the fossil fuel intensity of GDP (I FF ) based on subtracting the CO 2 and GDP growth rates. The "Actual" values are the latest estimate available, and the "Projected" value for 2013 refers to those estimates presented in this paper. A correction for leap years is applied (Sect.   capita CO 2 emissions in 2013 were 1.4 tC person −1 yr −1 for the globe, and were 4.5 (USA), 2.0 (China), 1.9 (EU28) and 0.5 tC person −1 yr −1 (India) for the four highest emitting countries (Fig. 5e). Territorial emissions in Annex B countries have remained stable from 1990 to 2012, while consumption emissions grew at 0.5 % yr −1 (Fig. 5c). In non-Annex B countries, territorial emissions have grown at 4.4 % yr −1 , while consumption emissions have grown at 4.1 % yr −1 . In 1990, 62 % of global territorial emissions were emitted in Annex B countries (34 % in non-Annex B, and 4 % in bunker fuels used for international shipping and aviation), while in 2012 this had reduced to 37 % (58 % in non-Annex B, and 6 % in bunker fuels). In terms of consumption emissions this split was 63 % in 1990 and 43 % in 2012 (33 to 51 % in non-Annex B). The difference between territorial and consumption emissions (the net emission transfer via international trade) from non-Annex B to Annex B countries has increased from 0.05 GtC yr −1 in 1990 to 0.46 GtC yr −1 in 2012 (Fig. 5), with an average annual growth rate of 11 % yr −1 . The increase in net emission transfers of 0.41 GtC yr −1 from 1990 to 2012 compares with the emission reduction of 0.27 GtC yr −1 in Annex B countries. These results clearly show a growing net emission transfer via international trade from non-Annex B to Annex B countries. In 2012, the biggest emitters from a consumption perspective were China (23 % of the global total), the USA (16 %), EU28 (13 %), and India (6 %).
Based on DGVMs only, the global CO 2 emissions from land-use-change activities are estimated to be 0.9 ± 0.6 GtC in 2013, slightly below the 2004-2013 average of 1.0 ± 0.7 GtC yr −1 . However, although the decadal mean generally agreed, the estimated annual variability was not consistent between the LUC emissions estimated based on the combined bookkeeping method and fire-based estimate and the DGVMs, except that they are small relative to the variability from the residual land sink (Fig. 6a). This could be partly due to the design of the DGVM experiments, which use flux differences between simulations with and without land-cover change, and thus may overestimate variability due to, for example, fires in forest regions where the contemporary forest cover is smaller than pre-industrial cover used in the runs without land-cover change. The extrapolated landcover input data for 2010-2013 may also explain part of the discrepancy, though it would not account for the larger variability in the DGVMs.
The ocean CO 2 sink was 2.9 ± 0.5 GtC yr −1 in 2013, an increase of 0.1 GtC yr −1 over 2012 according to ocean models. Five of the seven ocean models produce an increase in the ocean CO 2 sink in 2013 compared to 2012. However the two data products available over that period produce a decrease of −0.1 GtC yr −1 . All estimates suggest relatively small change in the ocean CO 2 sink, consistent with El Niño neutral conditions observed in 2013. All estimates suggest an ocean CO 2 sink for 2013 that is larger than the 2004-2013 average of 2.6 ± 0.5 GtC yr −1 . The terrestrial CO 2 sink calculated as the residual from the carbon budget was 2.5 ± 0.9 GtC in 2013, just below the 2.7 ± 0.9 GtC in 2012 and the 2004-2013 average of 2.9 ± 0.8 GtC yr −1 (Fig. 4), and also consistent with El Niño neutral conditions. The DGVM model mean suggests the same terrestrial CO 2 sink in 2013 of 2.4 ± 1.2 GtC (Table 7), but results cover a range among models.
Cumulative emissions for 1870-2013 were 390 ± 20 GtC for E FF and 145 ± 50 GtC for E LUC based on the bookkeeping method of Houghton et al. (2012) for 1870-1996 and a combination with fire-based emissions for 1997-2013 as described in Sect. 2.2 (Table 10). The cumulative emissions are rounded to the nearest 5 GtC. The total cumulative emissions for 1870-2013 are 535 ± 55 GtC. These emissions were partitioned among the atmosphere (225 ± 5 GtC based on atmospheric measurements in ice cores of 288 ppm (Sect. "Global atmospheric CO 2 growth rate estimates"; Joos and Spahni, 2008) and recent direct measurements of 395.31 ppm (Dlugokencky and Tans, 2014)), ocean (150 ± 20 GtC using Khatiwala et al., 2013, prior to 1959 and Table 8 otherwise), and land (155 ± 60 GtC by the difference).
Cumulative emissions for the early period 1750-1869 were 3 GtC for E FF and about 45 GtC for E LUC (rounded to nearest 5), of which 10 GtC was emitted in the period 1850-1870  and 30 GtC was emitted in the period 1750-1850 based on the average of four publications (22 GtC by Pongratz et al., 2009;15 GtC by van Minnen et al., 2009;64 GtC by Shevliakova et al., 2009;and 24 GtC by Zaehle et al., 2011). The growth in atmospheric CO 2 during that time was about 25 GtC, and the ocean uptake about 20 GtC, implying a land uptake of 5 GtC. These numbers have large relative uncertainties but balance within the limits of our understanding.
Cumulative emissions through to year 2014 can be estimated based on the 2014 projections of E FF (Sect. 3.2), the largest contributor, and assuming a constant E LUC of 0.9 GtC. For 1870-2014, these are 545 ± 55 GtC (2000 ± 200 GtCO 2 ) for total emissions, with about 75 % contribution from E FF (400 ± 20 GtC) and about 25 % contribution from E LUC (145 ± 50 GtC). Cumulative emissions since year 1870 are higher than the emissions of 515 [445 to 585] GtC reported in the IPCC (Stocker et al., 2013b) because they include an additional 32 GtC from emissions in 2012-2014 (mostly from E FF ). The uncertainty presented here (±1σ ) is smaller than the range of 90 % used by the IPCC, but both estimates overlap within their uncertainty ranges.

Discussion
Each year when the global carbon budget is published, each component for all previous years is updated to take into account corrections that are due to further scrutiny and verification of the underlying data in the primary input data sets. The updates have generally been relatively small and focused on the most recent years, except for LUC, where they are more significant but still generally within the provided uncertainty range (Fig. 9). The difficulty in accessing land-cover-change data to estimate E LUC is the key problem to providing continuous records of emissions in this sector. Current FAO estimates are based on statistics reported at the country level and are not spatially explicit. Advances in satellite recovery of land-cover change could help to keep track of LUC through time (Achard et al., 2014;Harris et al., 2014). Revisions in E LUC for the 2008/2009 budget were the result of the release of FAO 2010, which contained a major update to forest-cover change for the period 2000-2005 and provided the data for the following 5 years to 2010 (Fig. 9b). The differences in this year could be attributable to both the different data and the different methods. Updates to values for any give year in each component of the global carbon budget were highest at 0.34 GtC yr −1 for the atmospheric growth rate, 0.19 GtC yr −1 for the fossil fuel and cement emissions, and 0.1 GtC yr −1 for the ocean CO 2 sink, all within the reported uncertainty. The update for the residual land CO 2 sink was also large (Fig. 9e), with a maximum value of 0.71 GtC yr −1 , directly reflecting revisions in other terms of the budget, but still within the reported uncertainty.  1750-2013, 1870-2013, and 1870-2014 in gigatonnes of carbon. All uncertainties are reported as ±1σ . All values are rounded to the nearest 5 GtC as in Stocker et al. (2013b), reflecting the limits of our capacity to constrain cumulative estimates. Thus some columns will not exactly balance because of rounding errors. Our capacity to separate the carbon budget components can be evaluated by comparing the land CO 2 sink estimated through three approaches: (1) the budget residual (S LAND ), which includes errors and biases from all components; (2) the land CO 2 sink estimate by the DGVM ensemble, which are based on our understanding of processes of how the land responds to increasing CO 2 , climate, and variability; and (3) the inverse model estimates which formally merge observational constraints with process-based models to close the global budget. These estimates are generally close (Fig. 6), both for the mean and for the interannual variability. The DGVM mean over 1959 to 2013 correlates with the budget residual with r = 0.71 (Sect. 2.5.2; Fig. 6). The DGVMs produce a decadal mean and standard deviation across models of 2.6 ± 0.9 GtC yr −1 for the period 2000-2009, nearly the same as the estimate produced with the budget residual (Table 7). New insights from the comparison with the atmospheric inversions and their regional breakdown already provide a semi-independent way to validate the results. The comparison shows a first-order consistency but a lot of discrepancies, particularly for the allocation of the mean land sink between the tropics and the Northern Hemisphere. Understanding these discrepancies and further analysis of regional carbon budgets would provide additional information to quantify and improve our estimates, as has been undertaken by the project REgional Carbon Cycle Assessment and Processes (RECAPP; Canadell et al., 2012).
Annual estimates of each component of the global carbon budgets have their limitations, some of which could be improved with better data and/or better understanding of carbon dynamics. The primary limitations involve resolving fluxes on annual timescales and providing updated estimates for recent years for which data-based estimates are not yet available or only beginning to emerge. Of the various terms in the global budget, only the burning of fossil fuels and at-mospheric growth rate terms are based primarily on empirical inputs supporting annual estimates in this carbon budget. The data on fossil fuel consumption and cement production are based on survey data in all countries. The other terms can be provided on an annual basis only through the use of models. While these models represent the current state of the art, they provide only simulated changes in primary carbon budget components. For example, the decadal trends in global ocean uptake and the interannual variations associated with El Niño-Southern Oscillation (ENSO) are not directly constrained by observations, although many of the processes controlling these trends are sufficiently well known that the model-based trends still have value as benchmarks for further validation. Data-based products for the ocean CO 2 sink provide new ways to evaluate the model results, and could be used directly as data become more rapidly available and methods for creating such products improve. Estimates of land-use emissions and their year-to-year variability have even larger uncertainty, and much of the underlying data are not available as an annual update. Efforts are underway to work with annually available satellite area change data or FAO reported data in combination with fire data and modelling to provide annual updates for future budgets. The best resolved changes are in atmospheric growth (G ATM ), fossil fuel emissions (E FF ), and, by difference, the change in the sum of the remaining terms (S OCEAN + S LAND − E LUC ). The variations from year to year in these remaining terms are largely model-based at this time. Further efforts to increase the availability and use of annual data for estimating the remaining terms with annual to decadal resolution are especially needed.
Our approach also depends on the reliability of the energy and land-cover-change statistics provided at the country level, and are thus potentially subject to biases. Thus it is critical to develop multiple ways to estimate the carbon balance at the global and regional level, including estimates from the inversion of atmospheric CO 2 concentration used here for the first time, the use of other oceanic and atmospheric tracers, and the compilation of emissions using alternative statistics (e.g. sectors). It is also important to challenge the consistency of information across observational streams, for example to contrast the coherence of temperature trends with those of CO 2 sink trends. Multiple approaches ranging from global to regional scale would greatly help increase confidence and reduce uncertainty in CO 2 emissions and their fate.

Conclusions
The estimation of global CO 2 emissions and sinks is a major effort by the carbon cycle research community that requires a combination of measurements and compilation of statistical estimates and results from models. The delivery of an annual carbon budget serves two purposes. First, there is a large demand for up-to-date information on the state of the anthropogenic perturbation of the climate system and its underpinning causes. A broad stakeholder community relies on the data sets associated with the annual carbon budget, including scientists, policy makers, businesses, journalists, and the broader society increasingly engaged in adapting to and mitigating human-driven climate change. Second, over the last decade we have seen unprecedented changes in the human and biophysical environments (e.g. increase in the growth of fossil fuel emissions, ocean temperatures, and strength of the land sink), which call for more frequent assessments of the state of the planet, and by implications a better understanding of the future evolution of the carbon cycle, and the requirements for climate change mitigation and adaptation. Both the ocean and the land surface presently remove a large fraction of anthropogenic emissions. Any significant change in the function of carbon sinks is of great importance to climate policymaking, as they affect the excess carbon dioxide remaining in the atmosphere and therefore the compatible emissions for any climate stabilisation target. Better constraints of carbon cycle models against contemporary data sets raises the capacity for the models to become more accurate at future projections.
This all requires more frequent, robust, and transparent data sets and methods that can be scrutinised and replicated. After nine annual releases from the GCP, the effort is growing and the traceability of the methods has become increasingly complex. Here, we have documented in detail the data sets and methods used to compile the annual updates of the global carbon budget, explained the rationale for the choices made and the limitations of the information, and finally highlighted the need for additional information where gaps exist. This paper, via "living data", will help to keep track of new budget updates. The evolution over time of the carbon budget is now a key indicator of the anthropogenic perturbation of the climate system, and its annual delivery joins a set of other climate indicators to monitor the evolution of humaninduced climate change, such as the annual updates on the global surface temperature, sea level rise, minimum Arctic sea ice extent, and others.

Data access
The data presented here are made available in the belief that their wide dissemination will lead to greater understanding and new scientific insights into how the carbon cycle works, how humans are altering it, and how we can mitigate the resulting human-driven climate change. The free availability of these data does not constitute permission for publication of the data. For research projects, if the data are essential to the work, or if an important result or conclusion depends on the data, co-authorship may need to be considered. Full contact details and information on how to cite the data are given at the top of each page in the accompanying database and summarised in Table 2.
The accompanying database includes an Excel file organised in the following spreadsheets (accessible with the free viewer http://www.microsoft.com/en-us/download/ details.aspx?id=10): Appendix A: Additional ocean data combined with SOCATv2 and used by the ocean data products Table A1. Attribution of f CO 2 measurements for years 2012-2013 used in addition to SOCAT v2  to inform ocean data products.