China’s provincial process CO2 emissions from cement production during 1993–2019

Carbon dioxide (CO2) emissions from China’s cement production process have increased rapidly in recent decades, comprising the second-largest source of CO2 emissions in the country, next only to fossil fuel combustion. However, there used to lack high-quality data to estimate provincial process-related CO2 emissions from the cement industry of China. It has been recognised that many previous publications have adopted cement-based accounting methods or national average emission factors to estimate them. This study assembles fundamental provincial clinker production data and provincial clinker emission factors from multiple official statistics sources, following the Intergovernmental Panel on Climate Change (IPCC) methodology (A clinker-based estimation methodology), to develop a high-quality panel dataset of China’s provincial process-related cement emissions during 1993–2019. In 2019, the gross cement process emissions of China amounted to 818.2 Mt CO2, and the cumulative emissions between 1993–2019 were estimated to be approximately 12.5 Gt CO2. There are significant differences in provincial process-related CO2 emissions. The dataset is crucial to provincial cement process emission characterisation and emissions reduction policy-making in China. Measurement(s) carbon dioxide emission process Technology Type(s) computational modeling technique Factor Type(s) cement production • clinker production • the emission factor of clinker Sample Characteristic - Organism cement • clinker • Nonmetallic Minerals Sample Characteristic - Environment manufacturing process Sample Characteristic - Location Province (China) Measurement(s) carbon dioxide emission process Technology Type(s) computational modeling technique Factor Type(s) cement production • clinker production • the emission factor of clinker Sample Characteristic - Organism cement • clinker • Nonmetallic Minerals Sample Characteristic - Environment manufacturing process Sample Characteristic - Location Province (China)


Background & Summary
As a common building material, cement is widely used in housing and road construction. In general, the global cement industry accounts for about 5-7% of the world's CO 2 emissions 1 . Since 1990, global cement production has increased nearly fourfold, the growth rate of which is substantially faster than energy production over the same time period 2 . Global cement production in 2019 is estimated at approximately 4.1 Gt 3 . According to China National Bureau of Statistics, in 2019, China produced 2.3 Gt cement 4 , accounting for 56.8% of the world's total cement production, and in recent years its process-related emissions from cement production accounted for more than 50% of the global process emissions from cement production 2 . Hence, China's cement industry is a crucial sector for reducing industrial process CO 2 emissions in the world. In May 2021, seven ministries of China, including the State Administration for Market Regulation, the Ministry of Industry and Information Technology, the National Development and Reform Commission, and the Ministry of Ecology and Environment, jointly issued the Opinions on Improving the Quality of Cement Products and Regulating the Cement Market, which stated that the carbon emissions of the cement industry must reach its peak before 2030 5 . Although it is very important to reduce CO 2 emissions from the direct combustion of fossil fuels and the use of fossil energy-based electricity in the cement industry, reducing process-related emissions from cement production is also a crucial part of achieving the carbon peaking and neutrality goal of China's cement industry.
To quantify CO 2 emissions from the cement production process is the basis of managing process-related CO 2 emissions from China's cement production plants. Currently, there lacks of consistency in the reported CO 2 emissions from China's cement production process. There are also no official cement process emissions data in consecutive years' order, with only national-level data for a few years 1994 6 , 2005 7 , 2010 8 , 2012 9 , 2014 10 being publicly available. Although some studies have estimated process CO 2 emissions from China's cement industry, these works rely on the national-level data, such as Emissions Database for Global Atmospheric Research (EDGAR), Carbon Dioxide Information Analysis Center (CDIAC) and Global Carbon Budget (GCB) [11][12][13][14] ( Table 1). The more precise provincial-level emission data were seldomly counted. On the other hand, many of these studies calculated the cement-related process emissions based on the cement output 12,[15][16][17] (that is, using an average cement emission factor to directly multiply with cement output). With the average cement emission factor (0.2906 tonne CO 2 per tonne cement), Carbon Emission Accounts and Datasets (CEADs) calculated provincial process cement emissions with provincial cement production data during 1997-2019 16,18,19 . However, according to the 2006 IPCC Guidelines for National Greenhouse Gas Inventories 20 , the cement-based method doesn't reflect the actual process of emission occurrence in the cement production. Those calculations are generally of low accuracy since it does not consider regional differences in cement manufacturing process and cement-clinker ratios across China 21 . The clinker production provides the best activity data for the process emissions calculation of cement industry 20,22 . Only a few studies have recently used the clinker production method to estimate China's provincial carbon emissions of the cement production process 21 . For example, CEADs used a national average emission factor of 0.4964 tonne CO 2 per tonne clinker to calculate provincial process cement emissions during 1996-2016 21 . However, in terms of emission factors, those studies use a single average clinker emission factor to calculate the cement emissions of different provinces, which fail to conform to the best practices recommended by the IPCC guidelines 20 . This is because the emission factors of cement industries in different provinces in China are quite different 23 .
To provide high-quality data to estimate China's process-related emissions from cement production, this study developed a panel dataset for the cement industries of China's 31 provinces during the time interval of 1993-2019. A clinker-based estimation methodology was adopted. The dataset was assembled from provincial-level data of cement production, clinker production and cement-clinker ratios, as well as other basic data and the official provincial emission factors. The dataset provides a robust scientific support for further analyses of China's greenhouse gas emission issues and emission management strategies. Figure 1 shows the overall structural design of the cement emission estimation methodology in this study, and the dataset is available online at https://doi.org/10.11922/sciencedb.00024.

Methods
The process-related CO 2 emissions from cement production in the dataset were estimated according to the IPCC territorial-based accounting scope. It means that these emissions 'taking place within national (including administered) territories and offshore areas over which the country has jurisdiction' 24 . The administrative territorial accounting method can be used to estimate the human-induced emissions directly within one region's boundaries 25 . Research scope. Carbon dioxide emissions in the cement production process mainly occur in the clinker production process. Clinker is an intermediate product in the production of cement. In the production of  Table 1. Representative studies of China process-related CO 2 emissions from cement production.

Fig. 1
Construction flows of China's provincial CO 2 emission inventory from the cement production process.
clinker, limestone containing calcium carbonate and magnesium carbonate is heated over 1000 °C to calcine the limestone, and carbon dioxide is released as a by-product. In general, there are two primary sources of CO 2 emission in the cement manufacturing process. The first source of the CO 2 emission is from the chemical reaction of limestone calcination process (largely CaCO 3 in limestone) 26 , i.e. the process depicted by Eq. (1). The second source is from the direct usage of fossil fuels and electricity that power the cement manufacturing process. In this study, we only focus on CO 2 emissions from the limestone calcination process. This is because, in CO 2 emission inventories, fossil fuel and electricity emissions consumed by cement production are normally attributed to the emissions of the energy sector 27 .
Calculate process-related CO 2 emissions of cement manufacture. According to the 2006 IPCC Guidelines for National Greenhouse Gas Inventories, there are three primary methods to measure CO 2 emissions from cement production process 20 .The first method is to multiply the estimated clinker production by the emission factor. Clinker production estimates are inferred from the output and clinker content ratio of cement by types; exports and imports of clinker are also considered. The second method is to multiply the actual clinker output by the emission factor and the cement kiln dust correction factor. Compared with the first method, the emission accounting method that directly uses clinker production data has lower uncertainty. The third method is to calculate CO 2 emissions from the cement production process based on the weight and composition of carbonates in raw materials and fuel sources, the emission factor of carbonate, and the proportion of calcination achieved. The third method generally covers a wide range of data and is more accurate. But in fact, it is often difficult to collect plant-level supporting data. Based on the collected and estimated data of clinker production by province, this paper uses the second method in order to obtain the provincial cement production process emission data from 1993 to 2019. Clinker or cement production data are used, depending on data availability, to estimate the CO 2 emissions. If there are clinker production data of 31 provinces in the corresponding year, we calculate the CO 2 emissions of cement production process directly. However, if the clinker production data of year t is missing, then the annual clinker data can be estimated with the cement production data by Eq. (4) as below.
Where M cl,t represents the estimated cement clinker output of the missing year t, M c,t the cement production of year t without cement clinker data, t1 and t2 are the latest years with data before and after the data-missing year respectively. emission coefficient. Various organisations have suggested default emission factors for clinker. IPCC Tier 2 and Cement Sustainability Initiative (CSI), respectively, provided emission factors of 0.510 tonne CO 2 per tonne clinker (not including a correction for CKD) and 0.525 tonne CO 2 per tonne clinker (including a correction for MgCO 3 ) 20,30 . Using the cement-based method, EDGAR, CDIAC and CEADs suggested emission factors of 0.390 tonne CO 2 per tonne cement, 0.499 tonne CO 2 per tonne cement and 0.2906 tonne CO 2 per tonne cement, respectively 31 . 29 suggests that the average clinker emission factors of China in 1994 is 0.5277 tonne CO 2 per tonne clinker. The book also provided the clinker emission factors of various provinces in China in 1994. According to the data, with the exception of a few provinces, the numerical deviations of the clinker emission factors in most of Chinese provinces were relatively small. Shanxi, Liaoning and Hunan have higher clinker emission factors due to higher content of MgO in the produced clinker; Ningxia has a low content of CaO and moderate content of MgO in the clinker; therefore, its clinker emission factor is the lowest. For the calculation of emissions from the cement production process in various provinces from 1993 to 1999, this study uses the clinker emission factors of each province in 1994.
According to the 2005 China national greenhouse gas inventory Study 28 , the clinker emission factors of 31 provinces in China ranged between 0.5232-0.5556 tonne CO 2 per tonne clinker, which is higher than 0.4964 tonne CO 2 per tonne clinker production used by Shan et al. 21 . Moreover, provincial differences of clinker emission factors were ignored by these studies, which simply use a national average emission factor to calculate emissions from the cement production process in different provinces. It would cause inaccurate accounting of cement process CO 2 emissions at the provincial level.
If assuming that the calcined CKD in the system is not lost, then the CKD correction factor will be 1 (Vol.3 ch.2 p.2.12) 20 . According to the 2005 China National Greenhouse Gas Inventory research group's field survey and expert experience, Chinese companies directly screen raw material ores, maximise the use of low-grade ores and rocks, and install dust removal devices, so that the amount of unrecovered clinker dust is close to zero (p. 105) 28 . As a result, the correction factor for cement kiln dust would be 1 28 . The uncertainty caused by this way will be considered in the uncertainty analysis later. www.nature.com/scientificdata www.nature.com/scientificdata/ Data source. Up-to-date statistics and analyses of raw materials inputs and clinker production and country-specific emission factors are preferred for estimating the process CO 2 emissions from cement production 32 . Following this suggestion, our dataset mainly includes two groups of raw data: provincial clinker production data and provincial cement production data. The China National Bureau of Statistics had provided national and provincial cement production data from 1993 to 2019 4 . The provincial clinker production data have different sources in various years. The provincial cement clinker output data in 1993, 1994, and 1997

Data Records
Our dataset is an Excel file containing six sheets. The six sheets involve the raw cement production, the raw clinker production, the calculated clinker-cement ratio, the clinker emission factor, the final cement process emission inventory data and the uncertainty analysis. The China's provincial cement production process emissions dataset includes 833 data records (31 × 27 − 4 = 833). The Chongqing city was separated from the Sichuan Province in 1997 and became a municipality directly under the Central Government. Therefore, there is no data for Chongqing during 1993-1996. The annual cement production data for all other provinces of China have been provided in the dataset. Our dataset is stored on the website of Science Data Bank website 77     Notably, the ratios of clinker production to cement production in the data set vary by province. For example, it is relatively high (0.9614 in 2019) in Anhui, while the clinker-cement ratio in Jiangsu is relatively low (0.3522 in 2019). Anhui has a large clinker and cement production capacity, and lots of clinker are produced in the cities of Anhui province, such as Wuhu, Tongling, Fanchang, Zongyang and other places along the Yangtze River. Hence, a large part of Anhui's clinker production is transported to coastal areas downstream the Yangtze River, such as in Jiangsu and Zhejiang, for grinding and processing to produce cement. Clinker trade is one of the important reasons for the difference in the ratio of clinker production to cement production in various provinces.
technical Validation Uncertainty analysis. In general, there are two primary sources of uncertainty for cement-related emissions. One comes from the uncertainty of activity data; the other one pertains to the uncertainty of emission factors 78 . The uncertainty of the activity data is further comprised of uncertainties from the clinker output data of the cement industry, the clinker output data of enterprises, and the under-counted amount of kiln dust. The clinker output may be under-reported by local cement industry authorities. The clinker output provided by enterprises may contain technical errors of clinker weighing. The under-counted amount of kiln dust refers to the kiln ash lost during limestone burning, which contains calcium oxide after the decomposition of calcium carbonate. The uncertainty of emission factors comes from the sampling error of MgO and CaO contents in clinker and the error of chemical analysis of clinker conducted by enterprises. It may also come from the error of MgO and CaO contents in raw materials brought into the final calculation 20,28 .
The IPCC Good Practice Guide suggests that the uncertainty of plant-level clinker data is typically around 1-2% 79 .This study uses the median value of 1.5%. The industry statistics of clinker output are generally completed by the China Building Materials Quantitative Economic Supervision Committee, whose uncertainties are unclear. In the 2005 China Greenhouse Gas Emissions Inventory Study, the level of uncertainty was set at 5%, and the uncertainty caused by the correction coefficient of the kiln dust is 0.3% 28 . Based on this, the combined uncertainty of raw clinker production data is 6.8%. For those estimated clinker production data, this study temporarily set a higher uncertainty at 10%. The uncertainty in our clinker production data for all years are determined in this way.
The IPCC Good Practice Guide sets the uncertainty of chemical analysis at 1-2% 79 . This study uses the upper limit value of the range as the 2005 uncertainty of chemical analysis, i.e. 2%. The sampling error is about 0.2% 29 . Hence, the combined uncertainty of actual clinker emission factor of Chinese provinces is 2.2% in 2005 and 1994. The estimated clinker emission factor of the province by the regional average value is likely to have a higher uncertainty, and we set the uncertainty level at 4% in 2005. The 2005 clinker emission factors, which were derived from the 2005 China Greenhouse Gas Emission Inventory study, are used in the calculation of   This study uses the error propagation method (Approach 1) recommended by the IPCC to determine uncertainties of provincial process-related CO 2 emissions of cement production. According to the multiplication relationship of clinker production data and clinker emission factor, the combined uncertainty of each province's carbon dioxide emissions is 7.1-7.9% in 2005. During 1993-2019, the combined uncertainties of provincial carbon dioxide emissions range from 7.1% to 11.7%. The max and min of them by years are shown in Fig. 6.
For the combined uncertainty of the process carbon dioxide emissions from China's total cement production over the years, this study use two methods, including the error propagation method and the Monte Carlo method. First, the combined uncertainty of the national clinker production is calculated with the average uncertainty of provincial clinker production with the weight of provincial clinker production. The combined uncertainty of national average clinker emission factor is calculated with the uncertainty of provincial clinker emission factor with the weight of provincial clinker production. Then the combined uncertainty of national process-related cement emission is around 7.1-11.0% with the error propagation method during 1993-2019, as shown in Fig. 6. Second, under the normal distribution for the national clinker production and clinker emission factor, we employ the Monte Carlo method to calculate the combined uncertainty for the national process emission of cement industry. Based on determining the combined uncertainty of the national clinker production and the average clinker factor, we respectively calculated the standard deviation of the normal distribution patterns. Then we conduct random sampling on both the activity data and emission factors for 200,000 times and get 200,000 estimations on the national process-related CO 2 emissions in the cement industry. The uncertainty range, therefore, was 95% confidential intervals of the estimations. The Monte Carlo simulation was conducted  www.nature.com/scientificdata www.nature.com/scientificdata/   21 . The Coefficient of Variation (CV, the standard deviation divided by the mean) by these institutions is 3.2%. In the 1994 China Greenhouse Gas Emission Inventory study, the combined uncertainty of national process-related CO 2 emission in cement industry is about 6.6-8.3%. The calculated uncertainties of this study by the two methods are 7.1%, which are close to the calculated uncertainty of the 1994 China Greenhouse Gas Emission Inventory study. In addition to the above uncertainties of clinker emission factors and activity data, some other uncertainties should also be considered when using the datasets. For example, the carbon absorption effect in the use of cement has been neglected in the calculation, which also creates a certain degree of uncertainty. There may be differences between the China Cement Association and the National Bureau of Statistics in the statistical methods of cement production and clinker production. This difference in data sources may also increase the uncertainty of the calculation results.
Comparison with previous cement emission estimates. To verify the cement emissions data given in this study, we compared our results with estimates of other professional institutes or research groups, as shown in Fig. 7. It illustrates that our estimates for cement-related emissions of China are lying in the middle range, close to the official estimates, with less than 1% difference.
According to the first and second national communications submitted by Chinese officials to the United Nations Framework Convention on Climate Change (UNFCCC) 2 , process emissions from the cement production process in Mainland China in 1994 and 2005 were 157.8 Mt CO 2 6 and 411.7 Mt CO 2 7 , respectively. In the third national communication, the first and the second Biennial Update Report, the process emissions of non-metallic mineral products and the clinker production were reported, but the process emissions of cement production were not directly reported. In 2010, 2012 and 2014, China's cement clinker production was 1.18875 Gt clinker 10 , 1.3392 Gt clinker 9 and 1.4865 Gt clinker 8 , respectively. The 2005 National Greenhouse Gas Inventory reported that the national average carbon emission factor for clinker was 0.5383. Like Andrew (2019), we use this clinker emission coefficient to estimate the total cement process carbon emissions in Mainland China 2 . In 2010, 2012 and 2014, its process emission were be roughly 639.9 Mt CO 2 , 701.9 Mt CO 2 and 758.3 Mt CO 2 , respectively.
Our estimation results suggest that China process-related cement CO 2 emissions have been likely overestimated by the old version data of CDIAC and EDGAR. For example, in the year 2012, the old version of CDIAC overestimated 55% of the emissions and EDGAR overestimated by 22%. However, due to the lower emission factors of cement and clinker, the cement-based results and clinker-based results of CEADs have underestimated China's cement-related CO 2 emissions by 10% and 7%, respectively. The new version data of CDIAC by Gilfillan et al. 12 show much lower cement process emission in China than its old version data of CDIAC by Gilfillan et al. 12,80 . In 2010, 2012 and 2014, the gap between CDIAC's new version of China's cement process data and the results of China's greenhouse gas inventory was −8.5%, 1.6%, and 2.5%, respectively. Since 2013, EDGAR 6.0 14 www.nature.com/scientificdata www.nature.com/scientificdata/ has shown much lower values than EDGAR v5.0 13 for China's process-related cement CO 2 emissions because of an adjustment acknowledging the lower clinker ratio. Based on the comparisons (Fig. 7), only the estimated values of GCB 11 and the new version data of CDIAC (Gilfillan et al.) 12 are close to the results of this study, and the differences between GCB and this study are generally within 2.5% in the last ten years.
In previous studies, only CEADs have provided provincial-level CO 2 emission estimates for the cement industry with both the cement-based method and clinker-based method. The process emission difference between this study and CEADs by provinces in 2014 are shown in Fig. 8. The difference between the results of this article and the two methods of CEADs varies from province to province. In Jiangsu, the cement-based and clinker-based estimates by CEADs is 26.9 Mt CO 2 and 6.3 Mt CO 2 higher than this study's estimates, respectively. In Anhui, the cement-based and clinker-based estimates by CEADs are 33.4 Mt CO 2 and 9.4 Mt CO 2 lower than this study's estimates, respectively. In Henan, the cement-based estimate is 6.2 Mt CO 2 higher than this study's estimates, but the clinker-based estimate is 3.8 Mt CO 2 lower than this study's estimates. Our provincial-level clinker emission factor estimation provides the latest and longest-term emission inventory of China and its provinces, and is an important supplement to existing emission estimates and official emission inventories. Also, the two methods of CEADs adopted the national average emission factor, which cannot reflect provincial differences in the cement manufacturing process and cement-clinker ratios across China.
Limitations. Our dataset has the following limitations: first, despite we applied the clinker emission factors of various provinces in China, the emission factor of the year 1994 and 2005 was assumed to be representative for the time interval of 1993-1999 and 2000-2019 respectively. Considering the differences in clinker production technology in different years, variations in annual emission factors are expected. In future studies, the provincial clinker emission factors of different years, if possible, should be applied to achieve a more accurate emission inventory estimation of the cement production process. Second, due to the lack of clinker data for the years 1995-1996, 1998-2001 and 2003-2004, we assumed that the clinker-to-cement ratio is constant, so that we used the averages of the clinker-cement ratio of the previous and next years, and the annual cement data of the previous years to estimate the clinker data. Third, this study uses the standard territorial principle in estimating the emissions of each province. We use the perspective of administrative boundaries to calculate the CO 2 emissions of the cement production process, without considering the impact of the transfer of cement and cement clinker between different provinces from the perspective of final consumption.

Code availability
The data file with XLSX format are accessible on ScienceDB. No additional code is used during the calculation of provincial process CO 2 emissions from China cement production. The installation of Microsoft Office or WPS Office is recommended to manage the data and reproduce the study results.