Atmospheric observations suggest methane emissions in north-eastern China growing with natural gas use

The dramatic increase of natural gas use in China, as a substitute for coal, helps to reduce CO2 emissions and air pollution, but the climate mitigation benefit can be offset by methane leakage into the atmosphere. We estimate methane emissions from 2010 to 2018 in four regions of China using the GOSAT satellite data and in-situ observations with a high-resolution (0.1° × 0.1°) inverse model and analyze interannual changes of emissions by source sectors. We find that estimated methane emission over the north-eastern China region contributes the largest part (0.77 Tg CH4 yr−1) of the methane emission growth rate of China (0.87 Tg CH4 yr−1) and is largely attributable to the growth in natural gas use. The results provide evidence of a detectable impact on atmospheric methane observations by the increasing natural gas use in China and call for methane emission reductions throughout the gas supply chain and promotion of low emission end-use facilities.

Over the last decade, natural gas (NG) has become the fastest-growing fossil energy in China as a result of coalto-gas switch action to reduce air pollution and carbon dioxide (CO 2 ) emissions. The NG consumption increased dramatically from 108.5 billion standard cubic meters (bcm) in 2010 (4% of primary energy consumption) to a record level of 306.4 bcm in 2019 (8.1% of primary energy consumption), and it will keep increasing according to China's energy plan, and the share of gas in the energy mix is expected to reach 15% by 2030, while coal and oil consumption will decline 1 . Domestic production of natural gas has increased approximately twofold from 94.8 to 176.2 bcm, and the imported NG also increased dramatically. Methane (CH 4 ) is the primary component of NG and the second most important anthropogenic greenhouse gas after CO 2 with an estimated 20-year global warming potential 84-86 times greater than CO 2 2 . Oil and natural gas production is one of the major sources of CH 4 in the atmosphere 3,4 . The CH 4 leakage rate from NG upstream (extraction and gathering, processing, transmission and storage, distribution) and end-use combustion to the atmosphere is the key factor determining climatic advantage of the coal-to-gas shift 5,6 . Atmospheric measurements studies found that a large amount of methane emissions from oil and gas production are unaccounted for the bottom-up inventories 7,8 . Chan et al. 9 reported eight-year estimates of methane emissions from oil and gas operations in western Canada and found that they are nearly twice of those from inventories. Zhang et al. 10 estimated a leakage equivalent to 3.7% (~ 60% higher than the national average leakage rate) of the gross gas extracted from the largest oil-producing basin in the United States (US) using high-resolution satellite observations. Moreover, basin-wide estimates of emissions using in situ airborne data reported an inverse relationship between the basin-level leakage rate and gas production 11 .
Emissions from NG distribution network were found to be the major CH 4 contributor (56%) accounting for the detectable CH 4 emissions in Paris, as evidenced from CH 4 and its isotopic composition by mobile measurement on the ground 12 . Leaks from the NG pipelines were identified as the main source of CH 4 in emissions in London 13 and several US cities and the leak rates vary in a large range (from 0.004 leaks/km to 0.63 leaks/km) 14 www.nature.com/scientificreports/ Advanced mobile leak detection (AMLD) platform combined with GIS information of utility pipeline is used to estimate CH 4 leakage from pipelines from the local distribution systems in the United States. It is found that the leakage from those pipelines is approximately 5 times greater than inventories based on self-reported utility leakage data 17 . It was also found that the chances of leakage increase with the aging of the pipeline infrastructure irrespective of the material types. From 2010 to 2019, the length of the gas supply pipelines in the urban areas of China has increased approximately threefold from 298.6 to 935.6 million meters including 82% in the city and 18% in the county seat 18 . The CH 4 leakage from those pipelines is not actively monitored, which might be a potential threat to the net carbon reduction of China's energy switch strategy to reach the carbon-neutral goal in 2060 19 . China is the biggest methane emitting country 3 and many studies report increases in methane emissions from China in the past decade considering China as one region (eg. Zhang et al. 20 , Jackson et al. 21 , Sheng et al. 22 ). But there is limited data publicly available on upstream emissions and local distribution of natural gas emissions in China among different subregions. To overcome the limited access to the proprietary data, the combination of surface observations and satellite observations of column-averaged dry-mol fractions of methane provide an opportunity to discover the leaks and to estimate emissions through top-down approaches. Here we use nine years of observations by the GOSAT satellite and the WDCGG (World Data Centre for Greenhouse Gases) surface stations, for the first time to estimate methane emissions in subregions of China from 2010 to 2018 using a high-resolution inverse model and find an impact of increasing natural gas use in China on CH 4 emissions, signaling a need for effective mitigation strategies.

Results and discussions
Regional inversion of CH 4 emission. The NIES-TM-FLEXPART-VAR (NTFVAR) global inverse model 23,24 is used to estimate the CH 4 emissions constrained by GOSAT and surface observations from 2010 to 2018. Here we focus on the analysis of inverse model results over China and its subregions. There are several ways of regional division of China. In this study, we use a four-region division, based on different geographical features, that is, the north-eastern China (NE), the south-eastern China (SE), the north-western China (NW), and the Qinghai-Tibetan Plateau (TP) areas (Fig. 1). These regions differ in climate, agriculture type, also differ in the major economic activities and CH 4 emission sources. The NE and SE regions are in the Eastern monsoon area and are divided by the Qingling Mountains-Huai River, which is also the dividing line of 800 mm mean annual precipitation, with the SE region experiencing more precipitation. Daxinganling-Yinshan-Helan mountain is the physical geographic boundary of North and Northwest, which is also the dividing line of 400 mm mean annual precipitation. The Northwest region is a non-monsoon area with mean annual precipitation of less than 400 mm, including Xinjiang and Inner Mongolia where the main agriculture is animal husbandry 25 . TP is a region at an average elevation over 4000 m, with the Kunlunshan range, Qilianshan range, and Hengduan mountain chain as the division to other three regions.
Our optimized estimate of the average Chinese emissions is 58 Tg CH 4 yr −1 during 2010-2018 (with model uncertainty of 8.6 Tg CH 4 yr −1 ), around 12% lower than the average prior emission of 65.6 Tg CH 4 yr −1 , and our estimate is consistent with Sheng's 22 top-down study with 57.6 Tg CH 4 Table 1). Figure 32 . The rapid increases in the latter half of 2018 in all regions may be due to using 11-month running mean instead of signal from the inversion and the variation in NW need to be further investigated.

Interannual changes in CH 4 emissions.
Statistically significant increase trends in 2010-2018 are detected in NE for both prior and posterior flux anomalies of total CH 4 emissions (95% confidence P ≤ 0.05 by Mann-Kendall approach 33 ). The increasing trend in SE is weaker but still significant (95% confidence). No trends in TP and NW are found. We estimate a yearly increase rate of 0.87 Tg CH 4 yr −1 for the whole of China during 2010-2018, which is consistent with Zhang's 20 simulation of 0.72 Tg CH 4 yr −1 for China during 2010-2016, but lower than the estimate of Miller's 28 with 1.1 Tg CH 4 yr −1 over 2010-2015. NE contributes the most to the growth rate (0.77 Tg CH 4 yr −1 ) followed by SE (0.13 Tg CH 4 yr −1 ), which is stronger compared to the increase of the prior fluxes (0.35 Tg CH 4 yr −1 ). Trends in regional emission sectors. Anthropogenic CH 4 emission is about 90% of the total CH 4 emission in China 31,34 . Figure 3 shows the relative contributions of major CH 4 emission sources in China and related productions percentage in the four regions analysed in this study using data from EDGAR v5 and China National Statistic Yearbook 18 . The major anthropogenic emission sources are solid fuel (31%), rice production (24%), waste (18%), and livestock (15%). NE is the most energy production region producing 55, 77, and 40% of Chinese national coal, oil, and natural gas each year, followed by NW which produces 29, 14, and 23% of Chinese national coal, oil and natural gas each year. 77% of rice is produced in SE and 21% in NE. The total volume of collected municipal solid waste and wastewater discharged are mainly in SE (60%) and NE (35%) due to its large population in SE (58% of national population) and NE (36% of national population). Ruminant population spreads in the four regions with 33% in NE, 31% in NW, 26% in SE, and 10% in TP.
Coal production in China peaked in 2012 and declined since then from 41 to 37 Gt yr −1 until 2018 with a significant decrease in SE (decrease from 7.5 to 4 Gt yr −1 ). 18 Rice production, ruminant population, and crude oil production remain relatively stable (Supplementary Fig. S1). Waste (volume of discharged sewage and domestic removed solid waste) and natural gas production show a dramatic increase during 2010-2018, which might be the major contributors to CH 4 emission changes. The increase of CH 4 emissions from waste in China is 0.40 ± 0.08 Tg CH 4 yr −1 during 2010-2018 (according to EDGAR v6.0 3 ), with 60% occurring in SE counterweighing the decrease in emissions caused by coal production. The increase in total CH 4 emissions in SE is not as significant as it is in NE since CH 4 emission in SE from coal production decrease significantly and counterweighs the emissions from increasing waste and NG. The estimated CH 4 emission from coal production in SE is 0.36 Tg CH 4 yr −1 with a national average emission factor of 9.3 m 3 t −1 , while previous studies suggested that the coal

Contribution of natural gas emission in north-eastern China. Estimation of CH 4 emissions from
NG includes leakage from energy extraction, processing, transport, and leakage at end-use applicant according to site measurements and province-level data on pipeline distribution leakage provided by NG suppliers (see Methods). The left panel in Fig. 4 shows the upper and lower bounds of the estimates of CH 4 emission from NG and the total CH 4 emission trend with proportioned uncertainty during 2010-2018 in NE, both of which depict statistically significant increasing trends. The variation of total CH 4 emission increase closely follows the changes of CH 4 emissions from NG ( Fig. 4 right panel), indicating NG leakage of CH 4 emissions is a notable driver of CH 4 emission increase in NE. Removing the waste sector increment (0.14 Tg CH 4 yr −1 ) estimated by EDGAR v6.0 from the total increase in NE, we estimate an average CH 4 emission growth of 0.63 Tg CH 4 yr −1 in NE. The estimated NG leakage contributes 0.14 ~ 0.23 Tg CH 4 yr −1 , and the changes of pipeline leakage dominate the variation of CH 4 emissions from NG which is accounted from province-level loss amount from gas supply pipeline. Natural gas pipeline leaks (estimated by the difference between the amount of gas purchased and the amount of gas sold) is 3.4% in NE and 2.7% in China during 2010-2018, which is higher than the estimate of ~ 1.4% in Russia 37 and within the range of British estimate between 1.9 and 10.8% 38 . Taking 2018 year as example, in NE region the NG production is 63 bcm, and the NG consumption is 101.5 bcm 18 . The estimated total NG emission is 5.2 ~ 8.6% of the regional NG production or 3.2 ~ 5.3% of the regional NG consumption. A previous study 39 has found that the self-accounting by NG supplies potentially can be lower than the actual leakage. The discrepancy between top-down and bottom-up estimations of CH 4 emissions implies that a significant amount of CH 4 leaks are not accounted for. Our estimates of the growing trend are higher than it in the inventory, and some other studies suggest that CH 4 emissions from NG use can be underestimated by the bottom-up approach in China 40 . The analysis shows a strong correlation between the trend in NG use and the increase of the CH 4 concentration over NE China, translated into emission changes by the inverse model.

Implications for natural gas emission mitigation. Our analysis shows that the top-down approach
based on inversion modeling can support an observation-driven assessment of methane emissions from methane leakage, especially GOSAT data providing a long-term trend. Results highlight the relevance of NG use and pipeline expansion to methane emissions. Such an increase of leaking methane from NG production and use chain will cause potential danger to diverse stakeholders despite introducing a net carbon reduction. Given the large NG distribution pipelines 935.6 million meters in China, NG leakage can be a significant waste of energy and money. It can also accelerates ozone formation in urban areas 41 , especially in North China where surface ozone is already a severe air pollution problem 42,43 . Increase the monitoring of NG CH 4 emissions through a number of methods from facility-level measurements 44 to city-scale surveys 39,45,46 is a pressing task to successful mitigation strategy. Advanced leak reduction technologies in the NG end-use sector can also bring economic, environmental, and health benefits 17,46 .

Methods
Methane observations. Atmospheric CH 4 observations from satellite, surface, aircraft, and ship platforms are used in this study. Greenhouse Gases Observing Satellite (GOSAT) is the first satellite dedicated to observing greenhouse gases from space launched in January 2009 47 . The orbit overpasses at around 12:49 (local time) every three days, and the diameter of the footprint in nadir is approximately 10.5 km. The Thermal and Nearinfrared Sensor for carbon Observation-Fourier Transform Spectrometer (TANSO-FTS) is the main instrument of GOSAT, measuring short-wavelength infrared (SWIR) radiance reflected from the Earth's surface and atmosphere. We use column-averaged dry-air mole fraction of methane (XCH 4 ) data from the NIES GOSAT Level 2 retrievals (v. 02.81) 48 . The GOSAT XCH 4 data is further corrected by subtracting the monthly mean difference for each 5° latitude band between GOSAT observations and the model simulated XCH 4 optimized with inversion that uses only surface observations 31 . Ground-based atmospheric CH 4 observation data are obtained from WDCGG, and aircraft and ships observations are from NIES (map shown in Supplementary Fig. S2). Weekly   www.nature.com/scientificreports/ isentropic vertical levels 51 and FLEXPART model v.8.0 52 . The transport model is driven by the meteorological data from the Japanese Meteorological Agency (JMA) Climate Data Assimilation System (JCDAS) 53,54 . In this study, variational inversion scheme is combined with the high-resolution variant of the transport model and its adjoint described by Maksyutov et al. 24 , and applied to inverse modelling of methane emissions in a number of studies 4,21,31 . The inverse modeling problem is formulated and solved to find the optimal value of corrections to prior fluxes considering mismatches of observations and modelled concentrations. Variational optimization is applied to obtain flux corrections as anthropogenic and wetland scaling factors to vary prior uncertainty fields on a monthly basis at a 0.1° × 0.1° resolution separately for anthropogenic and natural wetland emissions with bi-weekly time steps. The inverse model operates at the resolution of coupled transport model of 0.1° × 0.1° and applies spatial flux covariance length of 500 km. Uncertainty tests for the inverse model have been performed using randomly perturbed observations and perturbed fluxes for different regions. We perturbed five sets of observations consistently with the observation uncertainty at each site and produced five sets of perturbed monthly Emissions Database for Global Atmospheric Research (EDGAR) and VISIT fluxes with a random scaling factor applied separately for regions and each month. We then performed an inversion using the perturbed pseudo-observations as measurement data and the perturbed fluxes (perturbed EDGAR and VISIT combined with the non-perturbed soil sink, biomass burning, and other natural emissions from the ocean, geological sources, and termites) as the prior fluxes and compared the inversion results to get the standard deviation of the estimated emissions. The uncertainty of inverse simulation in China is 16.5% 23,24 . Independent evaluation of the inverse model was made by observations at Dongsha Island (DSI) (20.7 N, 116.7E) and Novosibirsk (NOV) (55 N, 83E). DSI is located to the south of China and NOV is to the north of China (shown in Supplementary  Fig. S1) Estimation of natural gas emissions. As a comparison, fugitive CH 4 emissions from NG systems are estimated on the basis of the 2006 IPCC Guidelines for Greenhouse Gas Inventories, and recent measurements. The province-level annual natural gas production data is collected from the China Statistical Yearbook 18 (2010-2018) (see Supplementary Fig. S3 for more details). The fugitive emissions rates (FERs) for NG systems upstream (energy extraction, processing, transport, and distribution) in China is set as constant at 1.8% (0.35kt CH 4 PJ −1 ) for the 2010-2018 period taken from Schwietzke's 63 study. There exist only limited measurements of gas leakage in China. The conventional gas methane leakage rates reported in the latest U.S. field studies are applied to China. Alvarez et al. 8 found agreement between site-level results and top-down results, with the best estimate of supply chain emissions. This estimate of oil/NG CH 4 emissions can also be expressed as a production-normalized emission rate of 2.3% (+ 0.4%/− 0.3%) by normalizing annual gross natural gas production, which is mainly from production, gathering, and processing sources. Zhang et al. 10 estimated natural gas production emission rate (or methane leakage rate) of 3.7 ± 0.7% by a high-resolution satellite data-based atmospheric inversion framework which is ~ 60% higher than the national average of 2.3 ± 0.3% in the largest oil and gas production basin in the US. The leakage rate is even higher for the rapidly developing Delaware sub-basin (4.1%). The emission distribution is based on province-level loss amount from gas supply pipeline, defined as the difference between the amount of gas purchased (e.g., what enters the gateway to a province) and the amount of gas sold (e.g., what is metered to consumers), obtained from the China City Statistical Yearbook (2010-2018) 64 (see Supplementary Fig. S3 for more details). End-use (power generation, residential cooking, and industrial boilers) processes included in the estimation is 0.4-0.9% following by Lebel's 65 estimation of appliance level leakage. Taking the upstream and end-use EFs into account, the CH 4 emission is estimated as the sum of low FERs (1.8%) and high FERs (4.4%) for production, 0.4-0.9% of end-user combustion, and the accounted province-level loss.

Data availability
The inverse model and forward transport model code can be made available to potential research collaborators upon reasonable request. The XCH 4 retrievals (NIES Level 2 retrievals) are available at: https:// data2. gosat. nies. go. jp. Furthermore, in-situ data are archived on the WDCGG Global Network: https:// gaw. kishou. go. jp (more details on data file information and references see Supplementary S4). Other data products used in the study like the EDGAR emissions inventory and GFAS Database are available for download at http:// edgar. jrc. ec. europa. eu/ and https:// www. ecmwf. int/ en/ forec asts/ datas et/ global-fire-assim ilati on-system. Wetland emission VISIT are available at https:// www. nies. go. jp/ doi/ 10. 17595/ 20210 521. 001-e. html, and the NIES airborne and JR STATION (Japan-Russia Siberian Tall Tower Inland Observation network) data are available at https:// db. cger. nies. go. jp/ portal/ geds/ atmos pheri cAndO ceani cMoni toring. Ship data are available by request to NIES observation group. www.nature.com/scientificreports/ Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This 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:// creat iveco mmons. org/ licen ses/ by/4. 0/.