The Second National Communication (SNC) on Climate Change of the People’s Republic of China was submitted to the Secretariat of the United Nations Framework Convention on Climate Change (UNFCCC, hereinafter referred to as the Convention) in November 2012[①] and was released to the international community. The second part of the communication was about China’s GHG inventory in 2005. This was the third time China officially released its GHG data after the release of GHG data of 1994 in Initial National Communication on Climate Change of the People’s Republic of China (INC) in 2004[②] and data of 2004 in National Climate Change Programme (NCCP) in 2007[③]. And this was also the first time China officially released the base year data after its 2020 CO2 mitigation target was announced in 2009 before the Copenhagen Conference, attracting wide attention from the international community.

However, there are some differences between the data released by China and those by internationally renowned research agencies or databases. These agencies are far surpassing China in terms of the frequency and timeliness of their data release, often resulting in the international community’s preconceived impression of China’s GHG emissions. This paper analyzes and compares the major international CO2 emissions data with China’s official data from the perspectives of original data source, calculation scope and methodology.[④] The purpose of the study is, on the one hand, to clarify the causes of differences through comparative analysis; and on the other hand, to identify the weaknesses in China’s energy statistics.

I. Comparison between CO2 Emissions Data of Chinese Government and Other Sources

2.1 China’s Official CO2 Emissions Data and Estimated Data, 2006－2011

Table 1 lists China’s official CO2 emissions data. Both INC and SNC adopt TIER2 method provided by the Revised 1996 IPCC Guidelines for National Greenhouse Gas Inventories (hereinafter referred to as IPCC1996) for inventory development, and the reference approach is also used for comparison.[⑤] The main sources of CO2 emissions covered by SNC are fossil fuel combustion, industrial processes (the production of cement, steel, lime, and calcium carbide and utilization of limestone and dolomite) as well as the combustion of a very small amount of non-botanic waste. CO2 emissions from fossil fuel combustion do not cover international bunker (listed separately) or non-energy use.[⑥] Other energy activities may also produce significant CO2 emissions, such as coal mining, venting and flaring in oil and gas system, but they are not estimated in SNC.

The paper estimates CO2 emissions from energy combustion in China in 2006－2011 based on the implied emission factors of primary fossil fuels in 2005, namely, CO2 emissions per unit of coal, oil and gas (t CO2 (TJ)－1). Meanwhile, according to CO2 emission factor of clinker production and proportion of clinker in cement, CO2 emissions from China’s cement production in 2006－2011 are also estimated[⑦], as shown in Figure 1.



The above estimations are all based on China’s official energy consumption and industrial production statistic data. China has gradually established its energy statistics system since the early 1980s, and a relatively complete system of statistical indicators and statistical agencies has been set up on both national and local levels to meet the needs of the time. Nevertheless, China’s energy statistics still has certain limitations, such as a shortage of high quality data, unsound energy statistical scopes and statistical indicators, and weak energy statistics foundations, etc.[⑧] The marketization of energy industries has further weakened the original energy statistic system.[⑨] Therefore, there are certainly deviations in the official energy data released by China. Using these data as the basis of comparison with other data, by no means, denies the potential uncertainties in the data. Nevertheless, the comparison with other data will help further identify the shortcomings of China’s energy statistics and propose concrete recommendations for improvement.

2.2 Data Comparison between International and Domestic Researches

The research agencies and databases regularly publishing national CO2 emissions include the Secretariat of the Convention[⑩], International Energy Agency (IEA), British Petroleum (BP), Emission Database for Global Atmospheric Research (EDGAR) jointly developed by Netherlands Environmental Assessment Agency (PBL) and the European Commission’s Joint Research Centre (JRC), Oak Ridge National Laboratory Carbon Dioxide Information Analysis Center (CDIAC), U.S. Energy Information Administration (EIA) and the Climate Analysis Indicators Tool (CAIT) of U.S. World Resources Institute (WRI).

Figure 2 shows the emissions data of China in 2005－2011 published by these research agencies and their comparison with the data released by China and estimated by this study. Upon examining the change of these series in 2005－2011, one can see that: 1) as for CO2 emissions from energy combustion, the estimate data have basically formed the pattern of BP>EDGAR>EIA>IEA≈CAIT>CDIAC, and some agencies’ data are less than or closer to the data released by China, but the rest are much higher, especially the data after 2008 (Fig. 2a and Fig. 1); 2) as for CO2 emissions from cement production process, the data in 2005－2006 from EDGAR are close to that of SNC and the estimate of this paper and then demonstrates significant deviation, and CDIAC estimate is far higher than that of this paper (Fig. 2b and Fig. 1).



II. Scope, Methodology and Sources of Underlying Data of CO2 Estimation in Selected International Research Agencies

3.1 IEA data

IEA data has a relatively broad coverage (131 countries) and time series are quite complete (OECD countries: 1960－2010; other countries: 1971－2010). The limitation lies in its lack of timeliness, and the data generally lags by one and a half years (for example, the data in 2010 was released in mid-2012).

IEA uses the reference approach in IPCC1996 and the TIER1 method in sectoral approach for fossil fuel CO2 estimation.[11] The scope of the sectoral approach matches that of fuel combustion specified in IPCC1996, that is, excluding non-energy use emissions, international bunkers emissions, CO2 fugitive emissions from energy production process, and flaring emissions. Therefore, the energy combustion CO2 emissions data of IEA shows good comparability with that in China’s SNC in terms of estimation scope and method.

The activity data is from IEA energy statistics, and emission factors adopt the default emission factors in IPCC1996.[12] According to the cooperation agreement between China and IEA, the National Bureau of Statistics of China (NBSC) provides China’s energy production and consumption data annually to IEA which will reprocess the data according to its own information and format available. Thus, the China’s energy data in IEA publications is generally consistent with China’s official data.

However, there are significant differences between the CO2 emissions from fossil fuels in 2005 estimated by IEA and the data in China’s communications, and between IEA’s data of 2006－2007 and the estimates of this paper. This research has carried out a special study on this issue. Table 2 lists the data on China’s fossil fuel production and consumption in 2005 released by NBSC and IEA. It can be seen that the data from IEA is smaller than that of NBSC, especially in coal consumption. The reason lies in that in 2009－2010, NBSC adjusted China’s energy data in 1996 and onwards according to the second economic census,[13] and the coal data witnessed the most significant adjustment. The adjusted data in 2005 for coal consumption showed an increase of 150 Mt, equivalent to about 300 Mt of additional CO2 emissions. Similarly, China also adjusted its coal data in 2006 and 2007, but there was no corresponding update in IEA data, so its results were significantly lower.



3.2 BP Data

The 2012 world energy review published by BP in June 2012 included the energy-related CO2 emissions data of 72 countries and regions, and covered the time span of 1965－2011.[14] The scope and method of BP’s CO2 estimation is the simplest, most straightforward and transparent. The emissions data are obtained by multiplying coal, oil and natural gas consumption with their average emission factors, not excluding international bunkers emissions or any potential carbon stored. It can be concluded that what BP estimates is the maximum potential CO2 emissions from fossil energy use in various countries, bearing no comparability with the data from other sources and in particular that from national inventories.

BP energy statistics has its own system, and its official release time is earlier than IEA. This study compares BP energy statistics and NBSC energy statistics (Fig. 3). The observed difference shows BP data generally overestimates China’s oil and coal consumption while underestimating natural gas consumption. The difference in oil data is relatively small, not exceeding 2% (but there is an increasing trend in recent years). The difference in coal consumption did not exceed 2.5% in 2002－2008, but from 2009 (2009 included), the difference in coal consumption data between these two systems soared and reached 10.4% in 2011, almost 350 Mt of raw coal (NBSC announced China’s coal consumption was 2380 Mtce in 2011, while BP data showed 2627 Mtce). This is the major reason for BP’s much higher estimates of China’s energy-related CO2 emissions, and also the reason why the CO2 estimates of other databases based on BP energy data are higher.


3.3 EDGAR and PBL/JRC

3.3.1 Scope and Methodology

EDGAR and PBL/JRC have currently published the CO2 emissions data in 1970－2011, covering over 214 countries and regions.[15] The time series is inferior to that of IEA, but with better timeliness. Another difference between IEA and EDGAR is that the scope of the latter is much broader: it not only contains energy-related CO2 emissions, but also includes CO2 emissions from some industrial processes as well as CO2 emissions from non-energy uses, and even emissions from some fuel spontaneous combustion.[16] Energy-related CO2 emissions cover not only CO2 emissions from fossil fuel combustion, but also CO2 emissions from gas flaring which is not included in either INC or SNC of China; industrial processes cover cement production, lime production/utilization and sodium carbonate (soda) production/utilization processes, in which cement production is the major source. It can be seen that there are many differences between EDGAR’s calculation scope and that of China.

EDGAR uses TIER 1 in 2006 IPCC Guidelines for National Greenhouse Gas Inventories (hereafter referred to as IPCC2006) to estimate CO2 emissions, but its emission source classification still sticks to what is required in IPCC1996 to ensure the comparability with other data sources.[17]

3.3.2 Underlying Data Sources

With regard to the latest data in 1970－2011 released by PBL/JRC in June 2012, the basic data of different emission categories came from various sources. The energy consumption data before 2008 (2008 included) were mainly from IEA, and the energy consumption data in 2009－2011 came from BP; gas flaring data were from the analysis of satellite monitoring data provided by the National Oceanic and Atmospheric Administration (NOAA) and the Global Gas Flaring Reduction Partnership (GGFR) organized by World Bank; cement production data came mainly from the data released by U.S. Geological Survey (USGS), and the data on China came from NBSC. Meanwhile, PBL/JRC also utilizes the clinker ratio data in the report released by World Business Council for Sustainable Development (WBCSD); the basic data on non-energy uses were mainly from USGS and the World Steel Association (WSA).

3.3.3 Comparison and Analysis

In general EDGAR emissions data have a broader coverage and a relatively poor comparability with China and IEA data. Here we will make a rough comparison between IEA emissions series and EDGAR energy-related emissions data (Fig. 4). It can be seen, the gap between the two remained relatively stable at about 200－400 Mt before 2008, but since 2008, the gap began to jump to 700 Mt and reached 900 Mt in 2010. As the energy data in 2009－2011 in current EDGAR database were from BP, it seems to be inevitable that the gap between the calculated CO2 emissions and IEA data is widening.[18]



This paper also carries out a simple analysis on CO2 emissions from China’s cement production released by EDGAR. It is found that EDGAR’s CO2 emission factor per unit of produced cement remained unchanged in 2005－2011 (0.39 t CO2 per ton cement). In fact, the proportion of clinker to cement in China has been decreasing in recent years. It decreased from 70% in 2005 to 61% in 2011,[19] and correspondingly the intensity of emissions from producing one ton of cement also decreased significantly. Given 0.54 t of CO2 emissions for one ton of clinker (data in 2005), CO2 emissions from China’s cement production were around 700 Mt in 2011, while EDGAR’s estimates were 17% higher.

Trends in Global CO2 Emissions published by PBL/JRC has a specific chapter dedicated to presenting the uncertainty in China’s estimation.[20] It recognizes that the 2008 data is the highest in the available emissions data concerning China, and the uncertainty may be about 10%, but this uncertainty is asymmetrical, with higher possibility of overestimation. At the same time, it considers that there is great uncertainty in NBSC coal consumption data, and the deviation may not be fully adjusted by NBSC’s self-revision in 2005 and 2009. It also finds the support from relevant materials[21] and BP’s estimate/correction on China’s coal production.

3.4 CDIAC Data

By January 2013, CDIAC had collected, estimated and officially published the CO2 emissions data in 1751－2009 of 224 countries/regions;[22] in the pursuit of timeliness, CDIAC also informally released in January 2013 the selected countries’ CO2 emissions data in 2010－2011.[23] CDIAC’s data includes both energy-related and industrial process CO2 emissions. The former includes not only energy combustion but also natural gas flaring and non-energy use emissions, while the latter refers especially to CO2 emissions from cement production. In hence, CDIAC’s data scope shares some similarities with that for EDGAR. However, the two have differences in dealing with international bunkers: CDIAC includes international bunkers emissions of a country into its total emissions, while EDGAR removes it from the country’s total emissions and includes it into total global bunkers emissions. Due to inclusion of gas flaring and international bunker emissions, CDIAC energy-related CO2 emissions data is relatively less comparable with the official data of China.

CDIAC’s estimates are based on the methods of Marland and Rotty[24] similar to the reference approach in IPCC1996. That is, it uses the apparent consumption of varieties of energies to calculate corresponding CO2 emissions.

The basic data in 1950－2008 in CDIAC database are primarily from the United Nations Statistical Office (UNSO), with appropriate reference to the official statistical publications of each country, of which UNSO data largely reflect the results of the questionnaire distributed by IEA.[25] Cement production data are mainly from the data released by the USGS, and the flaring data come basically from the United Nations and are supplemented by the U.S. EIA data. The estimation in 2009－2010 uses BP energy data.

In theory, CDIAC’s and EDGAR’s energy-related emissions data, owing to similar calculation scopes and underlying data sources, should be relatively close. But the reality is that CDIAC data is closer to IEA data instead of EDGAR data (Fig. 2a).

Although CDIAC shares similar basic data with EDGAR, its estimation of China’s cement production is far higher (Fig. 2b). The previous analysis indicates that EDGAR estimates are relatively high, so CDIAC data proves even much higher.

3.5 EIA Data

By 2012, EIA had collected, estimated and officially published the CO2 emissions data in 1980-2010 for 217 countries/regions.[26] These data covers only energy-related emissions, including not only energy combustion but also natural gas flaring as well as some non-energy use emissions, and the emissions from international bunkers are included in national total emissions.[27] As a result, the scope of EIA’s CO2 emissions data is quite comparable with that of CDIAC’s energy-related CO2 emissions data. However, the comparability between EIA’s CO2 emissions data and China’s official data is also relatively poor due to the differences in dealing with gas flaring and international bunkers emissions. Similar to CDIAC, EIA also uses the apparent consumption method to calculate CO2 emissions from fossil fuel combustion. But EIA has its own energy statistics channels,[28] and it also uses the internal carbon content and oxidation rate data.[29]

Despite the consistency in the method and scope adopted by EIA and CDIAC to estimate energy-related CO2 emissions, their results show great differences. As can be seen from Figure 2a, the estimation of the two shows great differences except in 2006 and 2007 when the two are close, and EIA data in 2010 were nearly 100 Mt higher than CDIAC.

In order to explore the reason for EIA’s slightly higher CO2 emissions estimation, this paper analyzes China’s fossil energy consumption data in 2007－2010 released by EIA and makes a comparison with NBSC data (Fig. 5). The results show that EIA’s natural gas consumption data are slightly lower than the statistics of China, but its coal and oil consumption data are significantly larger than NBSC data, of which its coal consumption data in 2010 were about 437 Mtce higher and crude oil consumption data in 2009 were 80 Mtce higher than China’s statistics. Therefore, it is not difficult to understand the high EIA CO2 emissions data.



In addition, it is also found that the net calorific value of coal adopted by EIA (0.7994 kgce (kg coal)－1) is 11.9 % higher than the value commonly used in China (0.7143 kgce (kg coal) －1). Therefore, from the perspective of physical quantity, EIA statistics on coal consumption in 2010 was about 230 Mt higher than NBSC statistics, but it grew up to more than 400 Mtce after equivalent conversion. The differences both in physical quantity and calorific value conversion factor lead to much higher EIA energy-related CO2 emissions data than IEA data and CDIAC data in recent years.

3.6 CAIT Database of WRI

The emissions data in CAIT database, developed by WRI, were chosen from selected database instead of conducting original research. The CO2 emissions data provided by CAIT cover three sources of emissions: energy activities, industrial process (cement production) and land uses & land changes.[30] Currently, the database covers the data in 1850－2008 for 185 countries.

CAIT’s energy-related CO2 emissions data are from a variety of databases. Based on the criteria of completeness (geographic and temporal completeness) and accuracy, CAIT has selected IEA, CDIAC and EIA as the sources of basic data, and the combination is shown in Table 3. CAIT gives the highest priority to IEA data, followed by CDIAC, and finally EIA. Accordingly, China’s energy-related CO2 emissions data also follows this pattern. The data in 1971－2008 are from IEA, and the data in 1899－1970 from CDIAC.



CAIT data on CO2 emissions from cement production are from CDIAC, covering the period of 1928－2008. Based on the above analysis, CDIAC data for CO2 emissions from cement production are noticeably higher, and they therefore affect the accuracy of the CAIT data.

3.7 Overall Assessments of CO2 Data from Various Databases

This paper compares the CO2 emissions data from selected databases from the perspective of methodology, scope, geographic/time coverage, basic data source and comparability with China’s official data. The results are summarized in Table 4.

Since IEA’s energy-related CO2 emissions calculation shares the same scope and similar methodology with China’s inventory, and its energy statistics is close to China’s official data, so IEA emissions data has good comparability with China’s

Source of documents： Global Review