==== Front Carbon Balance Manag Carbon Balance Manag Carbon Balance and Management 1750-0680 Springer International Publishing Cham 163 10.1186/s13021-020-00163-2 Research A city-level comparison of fossil-fuel and industry processes-induced CO2 emissions over the Beijing-Tianjin-Hebei region from eight emission inventories Han Pengfei pfhan@mail.iap.ac.cn 1 Zeng Ning 2 Oda Tomohiro 234 Zhang Wen zhw@mail.iap.ac.cn 5 Lin Xiaohui 5 Liu Di 1 Cai Qixiang 1 Ma Xiaolin 6 Meng Wenjun 7 Wang Guocheng 5 Wang Rong 8 Zheng Bo 9 1 grid.9227.e0000000119573309State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China 2 grid.164295.d0000 0001 0941 7177Department of Atmospheric and Oceanic Science, and Earth System Science Interdisciplinary Center, University of Maryland, College Park, MD USA 3 grid.410493.b0000 0000 8634 1877Goddard Earth Sciences Research and Technology, Universities Space Research Association, Columbia, MD USA 4 grid.133275.10000 0004 0637 6666Global Modeling and Assimilation Office, NASA Goddard Space Flight Center, Greenbelt, MD USA 5 grid.9227.e0000000119573309State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China 6 grid.41156.370000 0001 2314 964XState Key Laboratory of Pollution Control and Resource Reuse, School of the Environment, Nanjing University, Nanjing, China 7 grid.11135.370000 0001 2256 9319Laboratory for Earth Surface Processes, College of Urban and Environmental Sciences, Peking University, Beijing, China 8 grid.8547.e0000 0001 0125 2443Department of Environmental Science and Engineering, Fudan University, Shanghai, China 9 grid.457340.10000 0001 0584 9722Laboratoire Des Sciences du Climat et de L’Environnement, CEA-CNRS-UVSQ, UMR8212, Gif-sur-Yvette, France 3 12 2020 3 12 2020 12 2020 15 2511 7 2020 24 11 2020 © The Author(s) 2020Open AccessThis article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.Background Quantifying CO2 emissions from cities is of great importance because cities contribute more than 70% of the global total CO2 emissions. As the largest urbanized megalopolis region in northern China, the Beijing-Tianjin-Hebei (Jing-Jin-Ji, JJJ) region (population: 112.7 million) is under considerable pressure to reduce carbon emissions. Despite the several emission inventories covering the JJJ region, a comprehensive evaluation of the CO2 emissions at the prefectural city scale in JJJ is still limited, and this information is crucial to implementing mitigation strategies. Results Here, we collected and analyzed 8 published emission inventories to assess the emissions and uncertainty at the JJJ city level. The results showed that a large discrepancy existed in the JJJ emissions among downscaled country-level emission inventories, with total emissions ranging from 657 to 1132 Mt CO2 (or 849 ± 214 for mean ± standard deviation (SD)) in 2012, while emission estimates based on provincial-level data estimated emissions to be 1038 and 1056 Mt. Compared to the mean emissions of city-data-based inventories (989 Mt), provincial-data-based inventories were 6% higher, and national-data-based inventories were 14% lower. Emissions from national-data-based inventories were 53–75% lower in the high-emitting industrial cities of Tangshan and Handan, while they were 47–160% higher in Beijing and Tianjin than those from city-data-based inventories. Spatially, the emissions pattern was consistent with the distribution of urban areas, and urban emissions in Beijing contributed 50–70% of the total emissions. Higher emissions from Beijing and Tianjin resulted in lower estimates of prefectural cities in Hebei for some national inventories. Conclusions National-level data-based emission inventories produce large differences in JJJ prefectural city-level emission estimates. The city-level statistics data-based inventories produced more consistent estimates. The consistent spatial distribution patterns recognized by these inventories (such as high emissions in southern Beijing, central Tianjin and Tangshan) potentially indicate areas with robust emission estimates. This result could be useful in the efficient deployment of monitoring instruments, and if proven by such measurements, it will increase our confidence in inventories and provide support for policy makers trying to reduce emissions in key regions. Keywords City-level fossil fuel CO2Industry processesMultiple inventoriesPolicy makingCO2 monitoringhttp://dx.doi.org/10.13039/501100002855Ministry of Science and Technology of the People's Republic of China2017YFB0504000Han Pengfei National Aeronautics and Space Administration (US)NNX14AM76GOda Tomohiro issue-copyright-statement© The Author(s) 2020 ==== Body Background Cities play a significant role in global greenhouse gas emissions, especially in urban areas, which are responsible for 67–76% of the global CO2 emissions and energy consumption [1]. Cities have become the critical and basic units for implementing emissions mitigation policies [2–5]. However, city-level mitigation actions remain daunting challenges [1, 6, 7]. City carbon emissions are influenced by the physical environment, economic development, urbanized density, industry structure, and energy use patterns specific to each city [1, 8]. Cities with a heavy industry, high traffic load, and high population density more easily have high emissions [3]. Discrepancies in the emissions and emission-socioeconomic characteristics among different cities require the development of corresponding policies [4]. Moreover, how to deploy observational instruments to form an efficient network is rather challenging [9–11], especially when there is no robust understanding of emissions. With emissions patterns identified by multiple inventories, such deployments would have more scientific guidelines. Furthermore, city strategies that reduce carbon emissions are expected to achieve emissions mitigation and meet a city’s economic growth goals [12]. Therefore, an accurate understanding of city-level CO2 emissions is of great importance in developing and implementing mitigation strategies. With rapid economic development and urbanization in China, cities account for 85% of China’s CO2 emissions [13]. Emission error could be much larger at subnational levels [14]. However, most of the existing studies focus on national [15–19], provincial [4, 16, 20, 21], or sectoral CO2 emissions inventories in China [22–25]. For example, using nine inventories, Han et al. [19] estimated that the national total fossil fuel and industrial process-related CO2 emissions were 9.8 (9.2–10.4) Gt CO2 in 2016, and the emissions estimated from provincial-data-based inventories were more consistent than those from spatial disaggregation of national energy statistics. A few efforts have been made to estimate city-level emissions, but these efforts have mainly focused on megacities or provincial capital cities due to the limited energy data [26–29]. To date, the studies by Shan et al. [12] and Cai et al. [30] included 182 and 305 Chinese cities, respectively, and still had gaps in city coverage. Zheng et al. [31] estimated all the cities’ CO2 emissions intensities in mainland China, yet the latest year included in the study was 2013, and the study lacked temporal dynamics. Thus, a more comprehensive assessment of city-level CO2 emissions in China is critical for understanding the role of cities in carbon emissions. The Beijing-Tianjin-Hebei (Jing-Jin-Ji, JJJ) region is the largest urbanized megalopolis region in northern China, covering an area of 218,000 km2 and home to 112.70 million people [32–34]. Cities in the JJJ region include the municipalities Beijing and Tianjin and eleven prefecture cities in Hebei Province. In 2018, the total energy consumption in the JJJ region accounted for more than 10% of China’s total energy consumption [34]. Moreover, coal is the primary energy source in this region [33, 34]. The JJJ region is under considerable pressure to reduce CO2 as well as air pollutant emissions [31, 35]. Beijing and Tianjin have committed to peak their CO2 emissions by 2020 and approximately 2025, respectively, in the 13th Five-Year Plan [36]. Hebei Province is experiencing rapid industrial and urban development [37], contributing greatly to the national CO2 emissions [4]. However, there is a wide range of CO2 emission estimates in the JJJ region, especially in the areas with high emissions. The emissions from Beijing and Tianjin estimated by Cai et al. [30] and Wang [38] were 41–57% higher than those estimated by Shan et al. [16]. In addition, based on the results of Mi et al. [39] and Cai et al. [30], Tangshan’s carbon emissions differed by 50%. To our knowledge, there is not a comprehensive assessment on prefectural city level CO2 emissions in the JJJ region. Cai et al. [33] reported the provincial emissions, but not covered the prefectural cities in JJJ. Therefore, an accurate estimation of the CO2 emissions in the JJJ region is of great significance in terms of providing accurate information for developing mitigation policies. The assessment of existing CO2 emissions inventories is urgently needed, yet direct observations of CO2 emissions at the city scale is limited, especially in developing countries [10, 11]. Here, we conducted a comprehensive analysis of 8 state-of-the-art inventories and presented the temporal dynamics, spatial distributions, and urban and non-urban fractions of 13 cities. We recognized the similarities and differences in emissions and thus improved the understanding of current inventories; this research provides useful information for policy making related to reducing city emissions and monitoring CO2. Data and methods Data We used annual CO2 emissions data from eight emission inventories, including the China High Resolution Emission Database (CHRED); China Emission Accounts and Datasets (CEADs); Multi-resolution Emission Inventory for China (MEIC), version 1.3; the Nanjing University CO2 emission inventory (NJU); the Peking University CO2 emission inventory (PKU), version 2 (PKU-CO2-V2); the Open-source Data Inventory for Anthropogenic CO2, version 2018 (ODIAC2018); the Emissions Database for Global Atmospheric Research, version 5.0 (EDGARv5.0); and the Fossil Fuel Data Assimilation System, version 2.2 (FFDAS v2.2). Below, the eight inventories were categorized into two based on the emission calculation methodologies (Table 1).Table 1 Summary of the emission inventories ( modified from Han et al. [19]) Data ODIAC EDGAR PKU-CO2 FFDAS CHRED MEIC NJU CEADs Domain Global Global Global Global China China China China Temporal coverage 2000–2016 1970–2012 1960–2014 1997–2015 2007, 2012 2000–2016 2000–2015 1997–2015 Temporal resolution Monthly Annual Monthly Hourly/Annual Biennially or triennially Monthly Annual Annual Spatial resolution 1 km 0.1 degree 0.1 degree 0.1 degree/ 1 km 10 km 0.25 degree 0.25 degree N/A Emission estimates Global/National Global/ National Global/National Global/National National/Provincial National/Provincial National/Provincial Prefectural/National/ Provincial Emission factor for raw coal (tC per t of coal) 0.746 0.713 0.518 – 0.518 0.491 0.518 0.499 National uncertainty 17.5% (95% CI)  ± 15%  ± 19% (95% CI) 5- 15%  ± 8%  ± 15% 7–10% (90% CI) −15%–25% (95% CI) Point source CARMA2.0 CARMA3.0 CARMA2.0 CARMA2.0 FCPSC CPED CEC;ACC;CCTEN N/A Line source N/A The OpenStreetMap and OpenRailwayMap, Int. aviation and bunker N/A Transport networks The national road, railway, navigation network, and traffic flows Transport networks N/A N/A Area source Nighttime light Population density, nighttime light Vegetation and population density, nighttime light Nighttime light Population density, land use, human activity Population density, land use Population density, GDP N/A Version name ODIAC2018 EDGARv50_ PKU-CO2-v2 FFDAS v2.2 CHRED MEIC v.1.3 NJU-CO2v2017 CEADs Year published/updated 2018 2019 2016 2014 2017 2018 2017 2017 Data sources http://db.cger.nies.go.jp/dataset/ODIAC/ https://edgar.jrc.ec.europa.eu/overview.php?v=50_GHG http://inventory.pku.edu.cn/download/download.html http://ffdas.rc.nau.edu/Data.html Data developer Data developer Data developer http://www.ceads.net/ References Oda [40] Janssens-Maenhout [41] Wang et al. [51] Asefi‐Najafabady et al. [42] Cai et al. [43]; Wang et al. [44] Zheng [45]; Liu et al. [46] Liu [47] Shan et al. [48]; Guan et al. [17] Emission inventories based on city-level data We used two inventories that are based on city-level data. CHRED was constructed by enterprise-level point-source data and China’s city and provincial statistics, and then carbon emissions were allocated to 10 km resolution using proxy data in 2007 and 2012 [15, 30]. The CEADs inventory provides total CO2 emission estimates at the provincial and city levels from 2000 to 2016 based on apparent energy consumption data and local optimized emission factors [4, 17, 49]. Emission inventories based on provincial or national level data We used two inventories that are based on province-level data. China’s carbon emissions from MEIC are developed by a technology-based methodology based on provincial energy consumption, combustion/industrial/control technologies and emission factor databases covering 2000 to 2016 at 0.25, 0.5, and 1 degree spatial resolutions by Tsinghua University [18, 31, 50]. NJU calculated China’s CO2 emissions using provincial energy statistics and spatially distributed emissions based on the location of large point sources (power plants and cement plants) and various proxy data, i.e., using GDP for industry-related emissions and population for transportation and other emissions, at 0.25 degree resolution from 2000 to 2016 [5, 18]. We used 4 inventories that are based on national-level emission estimates. These subnational emission distributions are largely achieved by emission downscaling. PKU distributes national or subnational fuel data with various proxies (e.g., power plants as point sources, night-time light to distribute national gas flaring and population for others) based on the subnational disaggregation method at 0.1 degree resolution from 1960 to 2014 [51]. The year 2018 version of the ODIAC emissions data product (ODIAC2018) distributes national emissions into 1 km and 1 degree grids from 2000 to 2017 based on spatial proxies, such as geographical locations of power plant emissions, satellite observations of nightlights to distribute nonpoint emissions, and aircraft and ship fleet tracks [52, 53]. EDGAR provides carbon emissions on the 0.1 degree grid from 1970 to 2018 based on national emissions by a variety of spatial proxy data, including power plants from CARMA3.0 for point sources, road network and different weighting factors for line sources and population for residential and commercial emissions [54, 55]. FFDAS quantifies carbon emissions using a data assimilation technique incorporating remote-sensing and national fuel accounts and power plants at a 0.1 degree resolution from 1997 to 2015 [42, 56]. All data sets used here have yearly data. For more details, please refer to Han et al. [19]. Methods These inventories were first extracted by a JJJ mask (in shapefile format) from the National Geomatics Center of China using ArcGIS 10.02 software (ESRI, 2012). The emissions from urban and non-urban areas were separated by using an urban mask from the European Space Agency (ESA) Climate Change Initiative (CCI) land cover maps with a 300 m resolution (https://www.esa-landcover-cci.org), and the urban area here mainly refers to impermeable surfaces, with high coherence (stability) and bright reflections, maintained in time and under varying angles detected by satellite. This means there is a density of human structures such as houses, commercial buildings, roads, bridges, and railways, while a city refers to prefectural level zones with a legal definition that defines a physical geographic boundary. ArcGIS was used to obtain prefectural cities’ total emissions in urban and non-urban areas using city mask data from the National Geomatics Center of China. The data pre-process procedures are as follows: (1) Convert the grid data to polygon data, which will keep the original value of the grid; (2) Split the polygon data using the prefectural city boundaries, and this step produces the real areas of a polygon within the city boundary; (3) Calculate the actual areas of each polygon using “Calculate geometry” and multiply the areas with emissions intensity (e.g. kg CO2/km2); (4) Finally, sum all the values within the boundary of a city. Emission intensity was calculated as the CO2 emissions divided by the GDP, data which were derived from the National Bureau of Statistics of the People's Republic of China (NBS) and websites of prefectural cities’ statistics bureaus in Hebei. Linear regressions were conducted using the Python scipy package between inventories with CEADs or CHRED. Results City-level CO2 emissions In 2012, the emissions of 13 cities varied widely from 13 to 282 Mt CO2 (or 72 ± 45, mean ± SD) (Fig. 1). Tangshan, Tianjin, Handan and Beijing are high-emission areas. The total emissions estimated from provincial-data-based inventories (i.e., MEIC and NJU) were 6% higher than those from city-data-based inventories (i.e., CHRED and CEADs) but were 14% lower from downscaled national-level emissions (i.e., PKU, ODIAC, EDGAR, and FFDAS) in 2012. There was a great discrepancy in national-data-based inventories with a range from 657 to 1132 Mt CO2 (or 849 ± 214, mean ± SD). The CO2 emissions from EDGAR and PKU were 28% and 34% lower than the average from city-data-based inventories. The 13 cities have substantial differences in natural resources and socioeconomic conditions. A city with high productivity, rapid economic growth, and a large population tends to have high carbon emissions. Emissions from metropolises (e.g., Beijing and Tianjin) with advanced economics and high urbanization rates greatly contributed to the JJJ total emissions (23–43%, Fig. 1). In Hebei Province, high emissions were also located in the provincial capital and in industrial cities, such as Shijiazhuang, Tangshan, Handan, and Baoding, which accounted for 57%–68% of the total emissions in Hebei. The 7 remaining cities accounted for 32–43% of the total emissions in Hebei.Fig. 1 City-level CO2 emissions from CHRED, CEADs, MEIC, NJU, PKU, ODIAC, EDGAR, and FFDAS in the Beijing-Tianjin-Hebei (JJJ) region in 2012 Since CHRED and CEADs use city-level statistical data and these two datasets are more consistent with each other than with the other datasets, they are used as references for national and provincial data-based inventories. The emissions of Beijing from EDGAR, MEIC, and NJU are comparable to those from CHRED, with differences ranging from 1 to 8 (1–7%) Mt CO2 in 2012. However, compared to the values estimated by CHRED, the emissions were 115% and 32% higher when estimated by ODIAC and FFDAS, respectively, and 31% lower when estimated by PKU. These trends were also found in Tianjin, and the emissions from ODIAC were equal to 227 Mt CO2, a value that was 47% higher than that from CHRED but 53% lower than that from PKU. Emissions in prefectural cities with coal mines and heavy-intensity industries also showed large differences, such as Tangshan and Handan. Tangshan’s emissions varied largely across these datasets. CHRED and CEADs both produced larger estimates (182 and 282 Mt), while estimates from PKU, EDGAR, and FFDAS were only 69–87 Mt, or 52–62% lower, than that by CHRED. Handan’s emissions were up to 121 and 130 Mt based on CHRED and CEADs, respectively, but ranged from 61 to 89 Mt (26–53% lower than CHRED and CEADs) for the other inventories. It should also be noted that CHRED and CEADs produced large differences in some cities, such as Tangshan, Shijiazhuang, and Cangzhou, and these differences highlighted future directions for city-level inventories in JJJ. According to the plan of “Collaborative Development of Beijing, Tianjin and Hebei Province”, Zhangjiakou and Chengde belong to ecological conservation areas; thus, heavy industry is not recommended, and more efforts should be concentrated in high-emission cities such as Tangshan, Shijiazhuang and Handan. Spatial pattern of CO2 emissions The uneven spatial distribution of CO2 emissions reflects the highly diverse conditions in population, economic development, and natural environment in the JJJ region. The spatial distributions showed reasonably good agreement in patterns such as the high emissions areas of Beijing-Tianjin-Tangshan, although they varied in detail (Fig. 2). Low emissions (e.g., < 100 ton CO2 km−2) from different datasets were mainly located in the northwestern part of the region. This result is because cities in the north, such as Zhangjiakou and Chengde, have a low population density and have lower economic development [39]. High-emission areas (> 10,000 t CO2 km−2) are clustered in urban centers in the south and east. Hotspots of CO2 emissions (> 50,000 t CO2 km−2) recognized by most inventories are located in the urban areas of Beijing, Tianjin, and Tangshan.Fig. 2 Spatial distribution of CO2 emissions (a) from CHRED, MEIC, NJU, PKU, ODIAC, EDGAR, and FFDAS in the JJJ region in 2012. Horizontal bars represent emissions fractions from each category of total emissions (b). Urban, cropland and tree and shrub cover from ESA_CCI_LC Specifically, the spatial distributions of CO2 emissions from EDGAR and FFDAS are consistent with the pattern from CHRED. However, the total amount of emissions in the region from EDGAR is 23% lower than that from CHRED. This difference is caused by the discrepancies estimated in Handan, Tianjin, and Tangshan from EDGAR, which is altogether approximately 204 Mt CO2 lower than that from CHRED. Although the difference in total emissions between FFDAS and CHRED is less than 3%, the emissions in Tangshan and Beijing from FFDAS are 62% lower and 32% higher than those from CHRED. ODIAC distributes the non-power plant portion of national emissions based on satellite nighttime light [52, 53]. The pattern of carbon emissions in ODIAC is highly consistent with the spatial distribution of urban areas. Low emissions ranging from 50 to 500 t CO2 km−2 are not found in ODIAC. This result is probably due to the low emissions mostly located in the non-urban areas, which are mainly covered by vegetation and do not emit strong night light (Fig. 2). ODIAC and FFDAS have similar patterns, shown in Fig. 2, which can be because they both used night-time light to distribute nonpoint emissions. Compared with CHRED, the carbon emissions for the remaining inventories all showed higher emissions in the southeastern areas, which are mainly covered by croplands, with the results of MEIC, NJU, PKU and ODIAC being more notable and the results of EDGAR and FFDAS being relatively weaker. For the spatial allocation of emissions, CHRED, EDGAR and FFDAS give more emissions to high emitting grids, with > 50,000 t CO2/km2 grids contributing more than 50% of the total emissions (Fig. 2c), while for NJU and PKU, the same level of high emitting grids contributed less than 10%. It should be noted that the urban area contributed largely to the total emissions because of its large energy consumption and population [57]. Here, according to the urban extent from the ESA CCI land cover maps with a 300 m resolution (https://www.esa-landcover-cci.org) (Fig. 2), we extracted urban emissions from the seven gridded inventories, namely, CHRED, NJU, MEIC, PKU, ODIAC, EDGAR and FFDAS. As illustrated in Fig. 3, the proportion of urban emissions shows obvious differences among datasets, ranging from 17% (NJU) to 50% (ODIAC). As the largest contributors, the proportions of urban emissions in Beijing and Tianjin varied between 37% (NJU) –73% (ODIAC) and 23% (NJU) –63% (ODIAC), respectively. ODIAC tended to overestimate emissions in urban areas, resulting in the strongest urban-non-urban emissions gradients, especially in Tianjin (63% from urban) and Hengshui (79% from urban). This result is due mainly to the use of the nightlight proxy [14, 58]. The urban emissions from FFDAS also show a similar pattern with those from ODIAC, except for Zhangjiakou and Hengshui. In addition, in high-emitting cities in Hebei, such as Tangshan, all inventories identified urban emissions accounting for a smaller fraction (20%–40%) than those in Beijing and Tianjin.Fig. 3 Spatial distribution of urban and non-urban CO2 emissions from CHRED, MEIC, NJU, PKU, ODIAC, EDGAR, and FFDAS in the JJJ region in 2012 for subplot (a). The area of the pie chart represents the amount of CO2 emissions per city. The number near each pie chart represents the urban emission fraction. And the urban extent from ESA_CCI_LC for subplot (b) Temporal variation in city-level CO2 emissions and emissions intensity The interannual variations in carbon emissions from existing inventories all showed an increasing trend during 2000–2012, ranging from 355 ± 58 Mt to 915 ± 178 Mt CO2, and then tended to level off or showed a slight downward trend afterwards (Fig. 4). Total emissions of the JJJ region from ODIAC, CEADs and EDGAR increased faster than others, with average growth rates of 9.7%, 9.2% and 9.1%, respectively, during 2000–2012. However, the regional total emissions from FFDAS, PKU, MEIC, and NJU showed a relatively small trend, with an annual growth rate of 6.0–8.4%. PKU and EDGAR tended to underestimate emissions compared to CHRED and CEADs, especially in high-emitting cities. The interannual variation at the city scale from MEIC was consistent with that from NJU, both of which were based on provincial statistical data.Fig. 4 Temporal variation in the annual CO2 emission amounts and emissions intensity from CHRED, CEADs, MEIC, NJU, PKU, ODIAC, EDGAR, and FFDAS in the JJJ region. Solid lines denote CO2 emissions, and dashed lines denote emissions intensity. Note that the scales are different across cities to show their variations Specifically, for Beijing, the CO2 emissions from MEIC, NJU, CEADs, PKU, and CHRED tended to be stable since 2007, which was consistent with the studies by Li et al. [59] and Shan et al. [60]. This result may be because Beijing decreased coal use by 43% and increased natural gas consumption by 144% from 2007 to 2014 [4, 16]. However, the annual emissions in Beijing from ODIAC fluctuated greatly, ranging from 73 Mt CO2 in 2000 to 263 Mt CO2 in 2017. PKU is a downscaled inventory; however, it uses provincial consumption fractions to rescale the IEA (International Energy Agency) total fossil fuel consumption when distributing emissions to grids, and thus, it can capture the Beijing decreasing trend after 2007, while other national downscaled inventories cannot. The emissions in Tianjin and Tangshan experienced rapid growth and then outpaced the emissions of other cities in recent years. The time series of emissions in Tianjin from ODIAC, MEIC, NJU, CEADs, and CHRED all showed an apparent growing trend during 2009–2014, which mainly resulted from the increment of coal use (22%) and crude oil (90%) [61]. The decline between 2005 and 2010 for CEADs was because of the change in statistical methods and dimensions, and blast furnace gas was lacking from the energy balance sheet during this period. For other periods, high emissions and large fluctuations in Tangshan’s emissions from CEADs were partially caused by its heavy industrial system [12, 39], whose iron and steel production accounted for 60–70% of Hebei’s total production. The emissions intensity (CO2 emissions per unit of GDP) in the JJJ region showed a decreasing trend since 2000 (Fig. 4). Interannual changes in emissions intensity among datasets were consistent with the total emissions. Moreover, the decoupled relationship between an increase in GDP and a decrease in emissions intensity indicated that a reduction in carbon emissions intensity could be achieved while also maintaining economic growth. For example, the GDP in Beijing and Tianjin significantly increased by average annual rates of 14% and 16%, respectively, during 2000–2016, but emissions intensity declined by 8–11% annually from PKU, MEIC, NJU, and CEADs and by 5–7% from ODIAC, EDGAR, and FFDAS. Comparison of city-level CO2 emissions from different inventories to CHRED and CEADs CHRED and CEADs were used as references to evaluate other datasets because they are both based on city-level energy statistics and consistent with each other. Furthermore, CHRED included unique comprehensive point sources (over 1.5 million enterprises) [33], and CEADs used measured local emission factors from 602 coal samples and 4243 coal mines [49]. As expected, the city-level CO2 emissions showed the best correlation between CEADs and CHRED, with a correlation coefficient (R) as high as 0.9 (Figs. 5a and 6a) and with a slope close to 1 and a smaller intercept than the others. The emissions from MEIC and NJU were both highly correlated with those from CEADs and CHRED, with R values ranging from 0.8 to 0.9 and slopes ranging from 0.5 to 0.7. These results were probably due to MEIC and NJU using provincial energy statistics as input data. CHRED and CEADs also used provincial inventories to calculate some city-level CO2 emissions because of the lack of consistent and accurate energy statistics in certain cities [12, 30].Fig. 5 Correlation of city-level emissions from CEADs and from CHRED (a), MEIC (b), NJU (c), PKU (d), ODIAC (e), EDGAR (f), and FFDAS (g) in the JJJ region Fig. 6 Correlation of city-level emissions from CHRED and from CEADs (a), MEIC (b), NJU (c), PKU (d), ODIAC (e), EDGAR (f), and FFDAS (g) in the JJJ region PKU was strongly correlated with CEADs and CHRED, with R values both equal to 0.8. The slopes were only 0.2 and 0.4 for the correlation relationships of PKU and CEADs and of PKU and CHRED, respectively (Figs. 5 and 6). The lower emissions estimated from PKU were mainly due to the rescaled energy data for China’s total by IEA statistics. Emissions from EDGAR had a relatively weak relationship with those of CEADs and CHRED, with R values of 0.6 and 0.8, respectively. ODIAC and FFDAS showed the lowest relationships with CEADs and CHRED, with R values of 0.5 and 0.7, respectively. These results were probably due to ODIAC, EDGAR and FFDAS all disaggregating emissions based on national energy statistics. Discussions Differences in prefectural-city-, provincial- and national-statistical data-based inventories The differences between provincial- and national-statistical data-based emissions have been well discussed in previous studies [19, 62, 63]. However, the differences between provincial and prefectural city statistical data-based emissions are poorly understood. Thus, it is difficult to conclude which is closer to the truth. Of these inventories, CEADs provided both prefectural-city- and provincial-statistical data-based estimates. We found that prefectural-city-statistical data-based estimates were 60% higher than provincial-statistical data-based estimates (Fig. 7) from 2000–2010 and gradually decreased to approximately zero after 2012, i.e., the sum of prefectural-statistical data became more consistent with provincial data in recent years due to the wider coverage of large-volume industries in local statistical authorities, indicating the improvement of statistical data between these two levels.Fig. 7 Percentage differences between prefectural-city-based and provincial-based estimates for CEADs. For the years without data, values were interpolated using data from nearby years At the prefectural city level for JJJ and other cities, the estimated differences generally increased from less than 10% to more than 300% when more inventories were included (Table 2) [5, 64–66]. Generally, the prefectural-city-statistical data-based inventories of CHRED and CEADs were more consistent than the others. For Hebei prefectural cities (such as Tangshan and Handan), the spread was relatively large among the three methods (i.e., the national, provincial and prefectural statistics data-based inventories), with differences ranging from 69 to 282 Mt CO2 for Tangshan and from 61 to 130 Mt CO2 for Handan in 2012 (Table 2). For Tangshan, the national inventories (ODIAC, EDGAR, PKU and FFDAS) were 69–117 Mt CO2, which were all smaller than the prefectural-city level and the provincial estimates of 138–282 Mt CO2 (MEIC, CHRED, MEIC and NJU). CEADs estimated Tangshan’s emissions to be much higher than the other inventories due to the large consumption of coke, e.g., the amount for 2015 was 31,986,000 tons, which was 41.4% that of Hebei Province [61]. The estimates for provincial-statistic data-based inventories were lower in Tangshan and Handan and thus tended to calculate higher emissions for other cities such as Baoding, Langfang and Hengshui due to the constraint of provincial total emissions. Moreover, the estimates for Beijing and Tianjin were more consistent for provincial estimates than the national estimates.Table 2 CO2 emissions estimates and differences at prefectural cities in JJJ and other cities City Number of inventories Year of emissions Estimates (MtCO2) (Max–min)/min (%) References Beijing 2 2006 115.2–160.7 39.5 Chen et al. [64] 2 2012 112.8–113.5 0.6 Cai et al. [33] 3 2012 79.7–134.0 68.1 Wang et al. [67] 8 2012 82.0–254.9 210.8 This study 8 2015 80.9–260.7 222.4 This study Tianjin 2 2012 167.4–180.0 7.1 Cai et al. [33] 3 2012 73.2–198.6 171.4 Wang et al. [67] 8 2012 72.1–227.0 214.7 This study 8 2015 80.0–231.8 189.7 This study Shijiazhuang 4 2007 52.5–123.4 134.8 Wang et al. [67] 8 2012 70.5–128.9 82.9 This study 8 2015 72.0–120.3 67.1 This study Tangshan 4 2010 47.8–188.1 293.7 Wang et al. [67] 8 2012 69.1–281.8 307.9 This study 8 2015 72.7–264.7 264.1 This study Handan 5 2008 26.1–86.0 229.1 Wang et al. [67] 2 2014 61.3–104.9 71.0 Wang et al. [67] 8 2012 60.9–130.3 114.0 This study 8 2015 64.5–153.1 137.4 This study Shanghai 2 2006 179.9–189.1 5.1 Chen et al. [64] 4 2010 112.6–420.9 273.6 Wang et al. [67] Paris 2 2005 44.7–50.3 12.5 Chen et al. [64] London 2 2003 28.1–32.4 15.1 Chen et al. [64] Los Angeles 2 2000 64.5–77.6 20.4 Chen et al. [64] Manhattan 2 2005 2.8–7.4 161.8 Chen et al. [64] Salt Lake City 2 2011 3.2–3.8 20.8 Gurney et al. [65] Indianapolis 2 2011 3.5–4.0 12.5 Gurney et al. [65] Northeastern U.S. cities 4 2011 NA 50–250 Gately and Hutyra [66] Urban and non-urban CO2 emissions and implications for carbon monitoring instrument distribution The urban extent of the ESACCI-LC product was 17%, 14%, and 5% of the land area for Beijing, Tianjin and Hebei, respectively, and the corresponding direct CO2 emissions were 64%, 37%, and 20%, respectively, of the total emissions for Beijing, Tianjin and Hebei for the mean of all inventories, which indicates that high emissions are spatially located more often in the urban areas of Beijing and Tianjin and are more diffusive in non-urban areas in Hebei. Similarly, an urban CO2 emissions study conducted by Cai et al. [33] showed that the urban extents were 17%, 17%, and 4% of land area in Beijing, Tianjin, and Hebei, which contributed 84%, 60%, and 41% of direct CO2 emissions, respectively. The differences between these two studies were mainly due to the inventories and urban land masks used. The ESA 300 m data showed more urban details, while Cai et al. [33] used a homemade dataset based mainly on the county/district and town/township GIS data, which are more continuous in space, to depict the urban areas. However, our study and Cai et al. [33] are largely consistent in their urban extent estimates for Beijing, Tianjin and Hebei and were approximately 20% lower than the CO2 emissions for urban areas calculated by Cai et al. [33]. Specifically, at the prefectural city level, emissions from national-data-based inventories were 53–75% lower in the industrial cities of Tangshan and Handan and 47–160% higher in Beijing and Tianjin than those from city-data-based inventories, and this difference was consistent with Gately and Hutyra [66]. The implication to the community and society is that national-based inventories have more biases in prefectural city levels than in local data-based inventories, and thus, we should be cautious when using national-based inventories for city-level use, such as in emissions evaluation, modeling research and policy making. The high-emission areas recognized by most of the present inventories have significant implications for monitoring instrument deployment [7, 10, 11, 67], although this type of comparison does not allow us to further discuss the accuracy of emission estimates. Thus, this information needs to be supplemented by objective physical measurements to validate the accuracy (e.g., [14, 53]). The areas with good agreement among inventories can be key areas for observation deployment, and in return, we can use these measurements to validate the accuracy of inventories. For example, third-party monitoring of CO2 emissions using high-density low-cost sensor networks is becoming possible due to the development of nondispersive infrared (NDIR) technology. Martin et al. [68] investigated a low-cost NDIR sensor and compared it with the standard instrument Los Gatos and found that the accuracy could reach 2–5 ppm after environmental factor corrections. More than 300 of such CO2 sensors were deployed in a network in Switzerland, and they were able to resolve CO2 changes and differences with magnitudes larger than ~ 20 ppm [11]. Bao et al. [69] proved such sensors to be promising in high-emission areas near Shijiazhuang, Hebei Province. However, how to deploy such nodes to form an efficient network has been a challenge [9]. With emissions patterns identified by most inventories, such deployment would have more scientific references. For example, in Beijing and Tangshan, emissions from urban areas consisted of 50–70% and 20–30%, respectively, which indicated that more sensors should be deployed in the urban areas of Beijing, while more nodes should be deployed in the non-urban areas of Tangshan. Moreover, atmospheric CO2 measurements need to consider the combined effects of biospheric and anthropogenic signals since the JJJ region is surrounded by vegetation, especially in the western and northern parts, and atmospheric transport also plays a significant role in the measurements. The identified high-emission areas might also indicate potential target areas for emission reductions for policy-makers if such high-emission areas are also confirmed by instruments. Beijing and Tianjin have committed to peak their CO2 emissions by 2020 and approximately 2025, respectively, in the 13th Five-Year Plan [36]. More specifically, local governments of Beijing have proposed a clear peak time of total and per capita CO2 emissions in 2020 in its "13th Five-Year Plan" for energy conservation and consumption reduction and climate change; thus, the city must promote the revolution in energy production and consumption, improve energy efficiency, and accelerate the construction of low-carbon cities [70]. High-emission areas (e.g., > 50,000 ton emissions in Fig. 2) need to allocate high priority to emissions control. The joint emissions control of JJJ also needs to identify the high-emission areas in Fig. 2. Point source contributions Point emissions consist of a large proportion of total emissions [15, 33]. For the JJJ regional total, CHRED estimated the highest point emissions proportion of 78.1%, including 12,991 industrial key emission sources and 187 industrial process sites [33]. Other inventories produced much smaller proportions ranging from 19% for NJU to 43% for EDGAR (Fig. 8), and MEIC and EDGAR both had more than 100 large emission grids associated with power plants. Point sources for CHRED consisted of power plants, industries and industrial processes, while other inventories (e.g., PKU, ODIAC) mostly used the CARMA dataset, which included only power plants. NJU included power plants ranking with the top 80% in electricity production and cement production that exceeded 1 Mt yr−1 [47], and there were only 42 power plants and 23 cement plants and thus fewer high-emission grids (N = 37, Fig. 8), which may be the reason why NJU point emission fractions were much smaller than those of other inventories. Since the gridded maps are mostly produced based on point sources, line sources (transportation emissions) and finally area sources, and area sources are distributed using the total estimate minus the point and line sources and proxies (e.g., night light, population, GDP), point source numbers, geolocations and emission magnitudes determined a very large degree of the gridded products (e.g., [53]). To improve the accuracy in mapping emissions and mitigate the errors in emission estimates, it is important to include reliable information, preferably reported information, on large point sources as much as possible. Thus, it is preferable for the community to share the point source information and for the information to be accurate to improve the understanding of the point source emissions (e.g., [14]). This is also important for emission monitoring purposes.Fig. 8 Characteristics of large emission grids and the corresponding emissions fractions and grid numbers (N) Conclusions Here, we conducted a comprehensive analysis of city-level fossil fuel and industrial process-related CO2 emissions from cities in the Beijing-Tianjin-Hebei region. We showed their temporal dynamics, spatial distributions, and urban and non-urban emissions fractions. We recognized the similarities and differences in emissions and thus improved the understanding of current inventories and provided useful information for policy making in terms of reducing city emissions and monitoring CO2. The results showed that compared to city-data-based inventories, provincial-data-based inventories were 6% higher, and national-data-based inventories were 14% lower in 2012. Compared with city-data-based inventories, the lower estimates (53–75%) in the industrial cities of Tangshan and Handan resulted in higher estimates (47–160%) in Beijing and Tianjin for national-data-based inventories. Due to the more complete data of industrial enterprises above the state designated scale, the differences between city statistical data-based estimates and provincial statistical data-based estimates decreased from 60% to approximately zero from 2000 to 2012, indicating the improvement in local statistical authorities. Spatially, all datasets agreed with high emissions in the triangular spatial distribution pattern of Beijing-Tianjin-Tangshan and low emissions in the northern parts of Zhangjiakou and Chengde. The implications of the consistent spatial distribution patterns recognized by these inventories provide useful information for the efficient deployment of monitoring instruments, and in return, the independent measurements from these areas will increase our confidence in inventories and thus provide support for policy makers in joint emissions reductions. Abbreviations JJJJJJ, Beijing-Tianjin-Hebei ODIACOpen-source Data Inventory for Anthropogenic CO2 EDGAREmissions Database for Global Atmospheric Research PKUPeking University CO2 emission inventory FFDASFossil Fuel Data Assimilation System CEADsChina Emission Accounts and Datasets CHREDChina High Resolution Emission Database MEICMulti-resolution Emission Inventory for China NJUNanjing University CO2 emission inventory SDStandard deviation CARMACARbon Monitoring for Action CIConfidence interval FCPSCThe First China Pollution Source Census CPEDChina Power Emissions Database CECCommission for Environmental Cooperation ACCChina Cement Almanac CCTENChina Cement Industry Enterprise Indirectory GDPGross domestic product ESAEuropean Space Agency CCIClimate Change Initiative NBSNational Bureau of Statistics NDIRNon-dispersive infrared Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Acknowledgements This work was supported by the National Key R&D Program of China (No. 2017YFB0504000). Tomohiro Oda is supported by the NASA Carbon Cycle Science program (grant no. NNX14AM76G). We thank Dr. Dabo Gun’s team, Dr. Qiang Zhang, Dr. Shu Tao and Dr. Bofeng Cai for kindly providing CEADs, MEIC, PKU-CO2 and CHRED data. Authors are listed alphabetically after Qixiang Cai. Authors' contributions PFH and WZ conceived and designed the study. PFH and XHL collected and analyzed the data sets. PFH, NZ, XHL, and WZ led the paper writing with contributions from all coauthors. TO, DL and QXC helped with the data plots and improved the discussion section. Data developers for each inventory, i.e., MC and GJM for EDGAR, TO for ODIAC, XLM and HKW for NJU, WJM and RW for PKU, BZ for MEIC, contributed to the descriptions and discussions of their data sets. All authors made suggestions and contributed to the revision of the final manuscript. All authors read and approved the final manuscript. Funding This research was supported by the National Key R&D Program of China (No. 2017YFB0504000) and the NASA Carbon Cycle Science program (No. NNX14AM76G). Availability of data and materials The data sets of ODIAC, EDGAR, PKU, CEADs and FFDAS are freely available from http://db.cger.nies.go.jp/dataset/ODIAC/DL_odiac2018.html, https://edgar.jrc.ec.europa.eu/overview.php?v=50_GHG, http://inventory.pku.edu.cn/download/download.htmlhttp://www.ceads.net/, and http://ffdas.rc.nau.edu/Data.html, respectively. CHRED, MEIC and NJU are available from the data developers upon request. Competing interests The authors declare that they have no conflicts of interest. ==== Refs References 1. IPCC. IPCC, 2014: Summary for Policymakers. In: Climate Change 2014: Mitigation of Climate Change. Contribution of Working Group III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Edenhofer O, Pichs-Madruga R, Sokona Y, Farahani E, Kadner S, Seyboth K, Adler A, Baum I, Brunner S, Eickemeier P, Kriemann B, Savolainen J, Schlömer S, von Stechow C, Zwickel T, Minx JC (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York. 2014. 2. Li X Yang L Zheng H Shan Y Zhang Z Song M City-level water-energy nexus in Beijing-Tianjin-Hebei region Appl Energy. 2019 235 827 834 10.1016/j.apenergy.2018.10.097 3. Cai B Lu J Wang J Dong H Liu X Chen Y A benchmark city-level carbon dioxide emission inventory for China in 2005 Appl Energy. 2019 233–234 659 673 10.1016/j.apenergy.2018.10.016 4. Shan Y Liu J Liu Z Xu X Shao S Wang P New provincial CO2 emission inventories in China based on apparent energy consumption data and updated emission factors Appl Energy 2016 184 742 750 10.1016/j.apenergy.2016.03.073 5. Wang H Lu X Deng Y Sun Y Nielsen CP Liu Y China’s CO2 peak before 2030 implied from characteristics and growth of cities Nat Sustain. 2019 2 8 748 754 10.1038/s41893-019-0339-6 6. Kort EA, Frankenberg C, Miller CE, Oda T. Space-based observations of megacity carbon dioxide. Geophys Res Lett. 2012;39(17). 7. Zheng B Chevallier F Ciais P Broquet G Wang Y Lian J Observing carbon dioxide emissions over China's cities with the Orbiting Carbon Observatory-2 Atmos Chem Phys Discuss 2020 2020 1 17 8. Kennedy CA Ibrahim N Hoornweg D Low-carbon infrastructure strategies for cities Nat Climate Change. 2014 4 5 343 346 10.1038/nclimate2160 9. Wu L Broquet G Ciais P Bellassen V Vogel F Chevallier F What would dense atmospheric observation networks bring to the quantification of city CO2 emissions? Atmos Chem Phys 2016 16 12 7743 7771 10.5194/acp-16-7743-2016 10. Shusterman AA Teige VE Turner AJ Newman C Kim J Cohen RC The BErkeley atmospheric CO2 observation network: initial evaluation Atmos Chem Phys 2016 16 21 13449 13463 10.5194/acp-16-13449-2016 11. Mueller M Graf P Meyer J Pentina A Dominik B Perez-Cruz F Integration and calibration of NDIR CO2 low-cost sensors, and their operation in a sensor network covering Switzerland Atmos Meas Tech Discuss 2019 2019 1 25 12. Shan Y Guan D Hubacek K Zheng B Davis SJ Jia L City-level climate change mitigation in China Sci Adv. 2018 4 6 0390 13. Shan Y Guan D Liu J Mi Z Liu Z Liu J Methodology and applications of city level CO2 emission accounts in China J Cleaner Prod. 2017 161 1215 1225 10.1016/j.jclepro.2017.06.075 14. Oda T Bun R Kinakh V Topylko P Halushchak M Marland G Errors and uncertainties in a gridded carbon dioxide emissions inventory Mitigation Adaptation Strategies Global Change. 2019 24 6 1007 1050 10.1007/s11027-019-09877-2 15. Wang J Cai B Zhang L Cao D Liu L Zhou Y High resolution carbon dioxide emission gridded data for China derived from point sources Environ Sci Technol 2014 48 12 7085 7093 10.1021/es405369r 24840164 16. Shan Y Guan D Zheng H Ou J Li Y Meng J Data Descriptor: China CO2 emission accounts 1997–2015 Sci Data. 2017 5 170201 10.1038/sdata.2017.201 17. Guan D Meng J Reiner DM Zhang N Shan Y Mi Z Structural decline in China’s CO 2 emissions through transitions in industry and energy systems Nat Geosci 2018 11 8 551 555 10.1038/s41561-018-0161-1 18. Liu M Wang H Oda T Zhao Y Yang X Zang R Refined estimate of China's CO 2 emissions in spatiotemporal distributions Atmos Chem Phys 2013 13 21 10873 10882 10.5194/acp-13-10873-2013 19. Han P Zeng N Oda T Lin X Crippa M Guan D Evaluating China's fossil-fuel CO2 emissions from a comprehensive dataset of nine inventories Atmos Chem Phys Discuss 2020 2020 1 21 20. Wang K Zhang X Wei Y-M Yu S Regional allocation of CO2 emissions allowance over provinces in China by 2020 Energy Policy. 2013 54 214 229 10.1016/j.enpol.2012.11.030 21. Zheng B Zhang Q Tong D Chen C Hong C Li M Resolution dependence of uncertainties in gridded emission inventories: a case study in Hebei, China Atmos Chem Phys 2017 17 2 921 933 10.5194/acp-17-921-2017 22. Liu F Zhang Q Tong D Zheng B Li M Huo H High-resolution inventory of technologies, activities, and emissions of coal-fired power plants in China from 1990 to 2010 Atmos Chem Phys 2015 15 23 13299 13317 10.5194/acp-15-13299-2015 23. Zheng B, Huo H, Zhang Q, Yao Z, Wang X, Yang X, et al. High-resolution mapping of vehicle emissions in China in 2008. Atmospheric Chemistry & Physics. 2014;14(18). 24. Shan Y Liu Z Guan D CO2 emissions from China’s lime industry Appl Energy 2016 166 245 252 10.1016/j.apenergy.2015.04.091 25. Guo B Geng Y Franke B Hao H Liu Y Chiu A Uncovering China’s transport CO2 emission patterns at the regional level Energy Policy 2014 74 134 146 10.1016/j.enpol.2014.08.005 26. Dhakal S Urban energy use and carbon emissions from cities in China and policy implications Energy Policy. 2009 37 11 4208 4219 10.1016/j.enpol.2009.05.020 27. Fang C Wang S Li G Changing urban forms and carbon dioxide emissions in China: a case study of 30 provincial capital cities Appl Energy. 2015 158 519 531 10.1016/j.apenergy.2015.08.095 28. Mi Z Zhang Y Guan D Shan Y Liu Z Cong R Consumption-based emission accounting for Chinese cities Appl Energy. 2016 184 1073 1081 10.1016/j.apenergy.2016.06.094 29. Tan X Dong L Chen D Gu B Zeng Y China’s regional CO2 emissions reduction potential: a study of Chongqing city Appl Energy. 2016 162 1345 1354 10.1016/j.apenergy.2015.06.071 30. Cai B Cui C Zhang D Cao L Wu P Pang L China city-level greenhouse gas emissions inventory in 2015 and uncertainty analysis Appl Energy. 2019 253 113579 10.1016/j.apenergy.2019.113579 31. Zheng B Tong D Li M Liu F Hong C Geng G Trends in China's anthropogenic emissions since 2010 as the consequence of clean air actions Atmos Chem Phys 2018 18 14095 14111 10.5194/acp-18-2018 32. Chen W Wu S Lei Y Li S Interprovincial transfer of embodied energy between the Jing-Jin-Ji area and other provinces in China: a quantification using interprovincial input-output model Sci Total Environ. 2017 584 990 1003 10.1016/j.scitotenv.2017.01.152 28169031 33. Cai B Li W Dhakal S Wang J Source data supported high resolution carbon emissions inventory for urban areas of the Beijing-Tianjin-Hebei region: spatial patterns, decomposition and policy implications J Environ Manag. 2018 206 786 799 10.1016/j.jenvman.2017.11.038 34. NBS NBoSotPsRoC. China Statistical Yearbook 2019. China Statistics Press. 2019. 35. Meng W Zhong Q Chen Y Shen H Yun X Smith KR Energy and air pollution benefits of household fuel policies in northern China Proc Natl Acad Sci 2019 116 34 16773 10.1073/pnas.1904182116 31383761 36. NDRC. The 13th five-year plan for energy saving and emissions reduction of the People's Repubic of China (2016–2020). 2016. http://www.ndrc.gov.cn/zcfb/zcfbqt/201701/t20170105_834500.html. 37. Wang Z Yang L Delinking indicators on regional industry development and carbon emissions: Beijing–Tianjin–Hebei economic band case Ecol Indicators. 2015 48 41 48 10.1016/j.ecolind.2014.07.035 38. Wang H Zhang R Liu M Bi J The carbon emissions of Chinese cities Atmos Chem Phys 2012 12 6197 6206 10.5194/acp-12-6197-2012 39. Mi Z Zheng J Meng J Zheng H Li X Coffman DM Carbon emissions of cities from a consumption-based perspective Appl Energy. 2019 235 509 518 10.1016/j.apenergy.2018.10.137 40. Oda T Maksyutov S Andres RJ The Open-source Data Inventory for Anthropogenic CO2, version 2016 (ODIAC2016): a global monthly fossil fuel CO2 gridded emissions data product for tracer transport simulations and surface flux inversions Earth Syst Sci Data 2018 10 87 107 10.5194/essd-10-87-2018 31662803 41. Janssens-Maenhout G Crippa M Guizzardi D Muntean M Schaaf E Dentener F Bergamaschi P Pagliari V Olivier JGJ Peters JAHW van Aardenne JA Monni S Doering U Petrescu AMR EDGAR v4.3.2 Global Atlas of the three major Greenhouse Gas Emissions for the period 1970–2012 Earth Syst Sci Data Discuss. 2017 10.5194/essd-2017-79 42. Asefi-Najafabady S Rayner P Gurney K McRobert A Song Y Coltin K A multiyear, global gridded fossil fuel CO2 emission data product: evaluation and analysis of results J Geophys Res Atmospheres. 2014 119 17 10 213 10.1002/2013JD021296 43. Cai B Liang S Zhou J Wang J Cao L Qu S China high resolution emission database (CHRED) with point emission sources, gridded emission data, and supplementary socioeconomic data Resources Conserv Recycling. 2018 129 232 239 10.1016/j.resconrec.2017.10.036 44. Wang J Bofeng C Lixiao Z Dong C Lancui L Ying Z High resolution carbon dioxide emission gridded data for China derived from point sources Environ Sci Technol 2014 48 12 7085 7093 10.1021/es405369r 24840164 45. Zheng B Tong D Li M Liu F Hong C Geng G Li H Li X Peng L Qi J Yan L Zhang Y Zhao H Zheng Y He K Zhang Q Trends in China's anthropogenic emissions since 2010 as the consequence of clean air actions Atmos Chem Phys 2018 18 14095 14111 10.5194/acp-18-2018 46. Liu Z Guan D Wei W Davis SJ Ciais P Bai J Reduced carbon emission estimates from fossil fuel combustion and cement production in China Nature 2015 524 335 10.1038/nature14677 26289204 47. Liu M Wang H Wang H Oda T Zhao Y Yang X Zang R Zang B Bi J Chen J Refined estimate of China's CO2 emissions in spatiotemporal distributions Atmos Chem Phys 2013 13 10873 10882 10.5194/acp-13-2013 48. Shan Y Guan D Zheng H Ou J Li Y Meng J China CO2 emission accounts 1997–2015 Sci Data 2018 5 170201 10.1038/sdata.2017.201 29337312 49. Liu Z Guan D Wei W Davis SJ Ciais P Bai J Reduced carbon emission estimates from fossil fuel combustion and cement production in China Nature 2015 524 7565 335 346 10.1038/nature14677 26289204 50. Li M Zhang Q Kurokawa J-I Woo J-H He K Lu Z MIX: a mosaic Asian anthropogenic emission inventory under the international collaboration framework of the MICS-Asia and HTAP Atmospheric Chem Phys. 2017 17 935 963 10.5194/acp-17-935-2017 51. Wang R Tao S Ciais P Shen H Huang Y Chen H High-resolution mapping of combustion processes and implications for CO 2 emissions Atmos Chem Phys 2013 13 10 5189 5203 10.5194/acp-13-5189-2013 52. Oda T Maksyutov S A very high-resolution (1 km× 1 km) global fossil fuel CO 2 emission inventory derived using a point source database and satellite observations of nighttime lights Atmos Chem Phys 2011 11 2 543 556 10.5194/acp-11-543-2011 53. Oda T Maksyutov S Andres RJ The Open-source Data Inventory for Anthropogenic CO 2, version 2016 (ODIAC2016): a global monthly fossil fuel CO 2 gridded emissions data product for tracer transport simulations and surface flux inversions Earth Syst Sci Data 2018 10 1 87 107 10.5194/essd-10-87-2018 31662803 54. Crippa M, Oreggioni G, Guizzardi D, Muntean M, Schaaf E, Lo Vullo E, et al. Fossil CO2 and GHG emissions of all world countries - 2019 Report, EUR 29849 EN, Publications Office of the European Union, Luxembourg, ISBN 978-92-76-11100-9. 2019. 10.2760/687800, JRC117610. 55. Janssens-Maenhout G Crippa M Guizzardi D Muntean M Schaaf E Dentener F EDGAR v4.3.2 Global Atlas of the three major Greenhouse Gas Emissions for the period 1970–2012 Earth Syst Sci Data. 2019 11 3 959 1002 10.5194/essd-11-959-2019 56. Rayner P Raupach M Paget M Peylin P Koffi E A new global gridded data set of CO2 emissions from fossil fuel combustion: methodology and evaluation J Geophys Res Atmospheres. 2010 115 D19306 10.1029/2009JD013439 57. Ramaswami A Tong K Fang A Lal RM Nagpure AS Li Y Urban cross-sector actions for carbon mitigation with local health co-benefits in China Nat Climate Change. 2017 7 10 736 742 10.1038/nclimate3373 58. Gurney KR Liang J Patarasuk R Song Y Huang J Roest G The vulcan version 3.0 high-resolution fossil fuel Co2 emissions for the United States Earth Syst Sci Data Discuss. 2019 2019 1 42 59. Li JS Zhou HW Meng J Yang Q Chen B Zhang YY Carbon emissions and their drivers for a typical urban economy from multiple perspectives: A case analysis for Beijing city Appl Energy. 2018 226 1076 1086 10.1016/j.apenergy.2018.06.004 60. Shan Y Guan D Liu J Liu Z Liu J Schroeder H CO2 emissions inventory of Chinese cities Atmos Chem Phys Discuss 2016 2016 1 26 61. NBS NBoSotPsRoC. China Statistical Yearbook 2016. China Statistics Press. 2016. 62. Guan D Liu Z Geng Y Lindner S Hubacek K The gigatonne gap in China’s carbon dioxide inventories Nat Climate Change 2012 2 9 672 676 10.1038/nclimate1560 63. Hong C Zhang Q He K Guan D Li M Liu F Variations of China's emission estimates: response to uncertainties in energy statistics Atmos Chem Phys 2017 17 2 1227 1239 10.5194/acp-17-1227-2017 64. Chen J Zhao F Zeng N Oda T Comparing a global high-resolution downscaled fossil fuel CO2 emission dataset to local inventory-based estimates over 14 global cities Carbon Balance Manag. 2019 15 1 9 10.1186/s13021-020-00146-3 65. Gurney KR Liang J O’Keeffe D Patarasuk R Hutchins M Huang J Comparison of global downscaled versus bottom-up fossil fuel CO2 emissions at the urban scale in four US urban areas J Geophys Res Atmospheres. 2019 10.1029/2018JD028859 66. Gately CK Hutyra LR Large uncertainties in urban-scale carbon emissions J Geophys Res Atmospheres. 2017 122 20 11 242 10.1002/2017JD027359 67. Wang Y Ciais P Broquet G Bréon FM Oda T Lespinas F A global map of emission clumps for future monitoring of fossil fuel CO2 emissions from space Earth Syst Sci Data 2019 11 2 687 703 10.5194/essd-11-687-2019 68. Martin CR Ning Z Karion A Dickerson RR Weber KJ Evaluation and environmental correction of ambient CO2 measurements from a low-cost NDIR sensor Atmospheric Measurement Tech 2017 10 7 2383 2395 10.5194/amt-10-2383-2017 69. Bao Z Han P Zeng N Liu D Wang Y Tang G Observation and modeling of vertical carbon dioxide distribution in a heavily polluted suburban environment Atmospheric Oceanic Sci Lett. 2020 10.1080/16742834.2020.1746627 70. The People's Government of Beijing Municipality P. Beijing's "13th Five-Year Plan" for energy conservation and consumption reduction and climate change. 2016. http://www.beijing.gov.cn/zfxxgk/110001/szfwj/2016-08/07/content_c7607556c0e74fe58c1c85a5d25183b6.shtml.