Background & Summary

Methane (CH4) is the second most potent greenhouse gas contributing to climate change, following carbon dioxide (CO2). Its global warming potential (GWP) is 28 times greater than that of CO2 over a 100-year period1. Over the past decade, atmospheric CH4 concentrations have been rising at an accelerating rate2,3,4. In 2023, the atmospheric CH4 concentration reached 1922.41 ppb, with an annual increase exceeding 10 ppb between 2020 and 2023 (https://gml.noaa.gov/ccgg/trends_CH4/). This persistent rise in atmospheric methane levels threatens to negate the climate benefits achieved through CO2 emission reductions3,5. Given CH4’s relatively short atmospheric lifetime of approximately ten years, prompt mitigation actions could offer significant climate benefits5. Reducing CH4 emissions is a cost-effective strategy for achieving the temperature targets set by the Paris Agreement. Additionally, CH4 contributes to the formation of tropospheric ozone (O3); thus, lowering CH4 emissions can help reduce ozone formation, improve air quality, and mitigate ozone-related health problems. According to the Global Methane Budget6, China is the largest contributor to global CH4 emissions, accounting for approximately 10% of the global total from 2010 to 2019. Therefore, understanding and addressing China’s CH4 emissions is crucial for achieving global climate targets and mitigating global warming potential.

Bottom-up approaches based on inventories from sectoral statistic data and top-down approaches based on inversions from the observation data of satellite or in-situ network are widely used to quantify the methane emissions7,8,9,10,11,12,13,14. The latest national greenhouse gas inventory submitted by China8 (bottom-up) to the UNFCCC (The People’s Republic of China Third Biennial Update Report on Climate Change) reported 64.1 Tg CH4 emissions in 2018, with contributions from coal mining (39%), livestock (22%), rice paddies (15%), solid waste (7%), wastewater management (5%), oil and gas systems (3%), and other sources (9%). Previous studies11,12,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29 have reported discrepancies in methane emission estimates for China, ranging from 44.9 to 67.6 Tg CH4 yr−1 across different inventories based on bottom-up approaches (Supplementary Table 1). Sectoral differences are more significant, with estimates for solid waste, wastewater, and oil and gas sectors varying by over 100% between studies (Supplementary Table 1). These discrepancies arise from differences in activity data scales12, emission factors11,30,31, and data quality32, additional contributing factors include differences in estimation methodologies9,31, spatial and temporal resolutions33, sectoral coverage5,30, approaches to uncertainty quantification5, and variations in the enforcement of regional policies9,15,33, such as differences in coal mine CH4 utilization efforts across provinces. Among these annual inventories (Supplementary Table 1), only the Emissions Database for Global Atmospheric Research15,21,22 (EDGAR, https://edgar.jrc.ec.europa.eu/dataset_ghg2024) and FAOSAT Database23 (https://www.fao.org/faostat/) are regularly updated. In addition, EDGAR has provided monthly CH4 emission inventories for anthropogenic sources since version 7.0. Given the strong seasonality of CH4 emissions from major sources, developing a monthly inventory for China is essential. Paddy fields and wetlands are highly sensitive to climatic factors such as temperature, precipitation, and humidity34,35,36, while human activities, including winter heating, rice cultivation, and livestock rearing, also exhibit significant seasonal variations37. Apart from EDGAR15,21,22, Gong and Shi38 has also estimated monthly CH4 emission inventories for China in 2015, yet long-term datasets remain scarce.

Observation data from Greenhouse gases observing satellite (GOSAT) and TROPOspheric Monitoring Instrument (TROPOMI) are widely applied in top-down methods to estimate CH4 emissions39. GOSAT inversions (43–61.5 Tg yr−1) show lower estimates for China’s CH4 emissions compared with TROPOMI inversion (65 Tg yr-1)5,7,40,41,42,43,44,45. In terms of sectoral emissions, the major discrepancies in CH4 emissions are in coal, livestock and rice sector between GOSAT inversions and TROPOMI inversions (Supplementary Table 2). Qu et al.40 reported that the differences between GOSAT and TROPOMI CH4emission estimates in China are mainly concentrated in the southeastern region, likely due to the spatial overlap of coal and rice emission sources and the seasonal cloud cover during peak emissions of rice planting. However, these differences could potentially be reduced by using higher spatiotemporal resolution prior inventory and applying additional filtering to TROPOMI data to address the cloud overcorrection7. Deng et al.41 pointed out that in situ inversions tend to estimate lower CH4 emissions for China compared to GOSAT and bottom-up inventories. It is worth noting that in situ inversions are often combined with satellite observation inversions to generate more reliable methane emission estimates46,47,48. These discrepancies highlight the need for a high-resolution, long-term methane inventory for China that can serve as a more reliable basis for both top-down and bottom-up estimates.

To address the lack of monthly methane emission inventory for China and the low temporal resolution of prior inventories in satellite observation data inversions, this study develops a long-term, monthly methane emissions dataset at the provincial level for mainland China using a bottom-up inventory approach. Our dataset includes seven anthropogenic sectors, coal mining, oil and gas systems, energy combustion, rice cultivation, livestock, solid waste, and wastewater, along with one natural source, wetlands. Our inventory also holds potential for the following applications: (1) Enhancing the accuracy of prior inventories in climate models to improve assessments of methane’s chemical reactions, transport processes, and interactions with the climate system49; (2) Capturing short-term emission events that are often averaged out in annual inventories, such as sudden industrial accidents or pipeline leaks50, and the impact of floods or droughts on wetland methane emissions51; (3) Facilitating integration with high spatiotemporal resolution observations from satellites and in situ networks52,53, enabling dynamic calibration and optimization of emission estimates; (4) Providing a clearer representation of temporal dynamics in sectoral emissions, as different emission sources vary significantly over time, allowing for better identification of multi-sector contributions and their driving factors54; (5) Supporting climate agreements and monitoring initiatives that require high temporal resolution data.

Workflow Overview

The Monthly Methane Emission Inventory for China’s Provinces (MMCP) dataset provides monthly CH4 emission inventories for 31 provinces of mainland China from January 2013 to December 2022, covering eight key sectors: coal mining, oil and gas systems, energy combustion, rice cultivation, livestock, solid waste, wastewater, and wetlands. This dataset was developed using bottom-up approaches following the 2006 IPCC Guidelines for National Greenhouse Gas Inventories (2006 IPCC Guidelines)31 and the 2019 Refinement to the 2006 IPCC Guidelines for National Greenhouse Gas Inventories (2019 Refinement)55. Sector-specific methodologies were adopted based on data availability and emission characteristics. For example, monthly or daily activity data (AD) were used directly to estimate emissions from fossil fuel-based energy combustion, oil and gas systems, and coal mining. In constract, sectors like biofuel combustion, rice cultivation, and solid waste relied on annual activity data, which were temporally downscaled using monthly proxy indicators (e.g. PM2.5 emissions, methane flux observations, temperature data, and monthly industrial production). Figure 1 illustrates the comprehensive workflow for constructing the MMCP dataset, showcasing the integration of subsectoral processes and proxy selection. This detailed framework ensures high temporal and sectoral resolution, enabling an accurate depiction of monthly methane emissions at the provincial level. Validation was conducted by comparing MMCP results with both bottom-up and top-down estimates from previous studies, and an uncertainty assessment was carried out to quantify the variability of key parameters across provinces and sectors.

Fig. 1
figure 1

Framework of the monthly methane emission inventory for China’s provinces (MMCP).

Full size image

Methods

Calculation scope

The Monthly Methane Emission Inventory for China’s Provinces (MMCP) dataset encompasses eight major sectors: (1) coal mining, (2) oil and gas Systems, (3) energy combustion, (4) rice cultivation, (5) livestock, (6) solid waste, (7) wastewater, and (8) wetlands. Each sector is further divided into subsectors and sub-subsectors to ensure comprehensive coverage and detailed emission estimation (Table 1). CH4 emissions from coal mining sector are estimated for underground mines, surface mines, and abandoned mines. For underground and surface mining activities, both mining and post-mining processes are considered. In the oil and gas systems sector, emissions are categorized into inshore and onshore oil wells as well as natural gas wells, with specific processes such as venting, flaring, exploration, production and upgrading, transport, refining/processing, transmission, and storage contributing to the total emissions. The energy combustion sector is divided into two subsectors: fossil fuel combustion (FFC) and biofuel combustion (BFC). FFC includes emissions from power generation, industrial activities, ground transportation, aviation, and residential energy use. While BFC accounts for CH4 released during the consumption of firewood and in-field straw. Methane emissions from rice cultivation are estimated separately for single-season and double-season rice systems, incorporating variations in growth periods, paddy field areas, and fertilizer and irrigation management practices. Livestock includes enteric fermentation subsector and manure management subsector. Enteric fermentation includes species such as beef cattle, dairy cows, goats, sheep, pigs, and camels. Manure management considers the same species while factoring in management practices and temperature variations. For solid waste, emissions are calculated from three primary waste treatment methods: landfill, incineration, and biological treatment. Meanwhile, the wastewater sector includes industrial, domestic, and agricultural wastewater, with emissions varying depending on the treatment or disposal method. Finally, methane emissions from the wetlands sector are estimated for both inland and coastal wetlands.

Table 1 Calculation scope of MMCP.
Full size table

Coal mining

The CH4 emissions from coal mining include fugitive emissions from underground coal mining and post-mining processes, surface coal mining and post-mining processes, as well as abandoned mines. The CH4 emissions from underground and surface mines are estimated using Eq. (1):

$${{SE}}_{i,p,m}={{AD}}_{i,p,m}\times {{EF}}_{i,p,y}-{R}_{i,p,m}$$
(1)

Where SE represents the total CH4 emissions from coal mining sectors. i, p, m, and y indicate the indices of subsector, province, month, and year, respectively. AD represents the activity data. The monthly coal production from underground and surface mines is used as activity data for these respective subsectors. The provincial-level monthly coal production is calculated based on the total monthly coal production obtained from the National Bureau of Statistics (https://data.stats.gov.cn)56 and the proportion of underground and surface mine production at provincial level (Table 5)57.

EF represents the emission factor, specifically accounting for processes such as coal mining and post-mining. The CH4 EFs for underground coal mining processes (Supplementary Table 3) are derived from four groups of provincial data from Sheng et al.12, Wang et al.58, Zhu et al.24 and Gao et al.30, For surface coal mines, a specific EF value of 2 m3 tons−1 was adopted based on the Guidelines for the Provincial Greenhouse Gas Inventories in 2011 (hereinafter referred to as GCPGGI)59. For post-mining processes, the CH4 EFs is (0.9–4.0) m3 tons−1 for underground coal from Ma et al.60 and 0.5 (0–0.5) m3 tons−1 for surface coal from GCPGGI59, respectively.

R represents the utilization amount of CH4 from underground coal mining. The provincial utilization amounts of CH4 from underground coal mining are obtained from the previous studies and social media (Supplementary Table 4). The data sources of monthly CH4 emissions from coal mining sector are listed in Table 2.

Table 2 Data source of methane emssion estimates from coal mining sector.
Full size table

The total CH4 emissions from coal mining in province p during month m, denoted as CEp,m, are calculated using Eq. (2).

$${{CE}}_{p,m}={{SE}}_{u,p,m}+{{SE}}_{{pu},p,m}+{{SE}}_{s,p,m}+{{SE}}_{{ps},p,m}+{E}_{a,p,m}$$
(2)

Where SEu,p,m and SEpu,p,m represent CH4 emissions from underground mining and associated post-mining processes, respectively. SEs,p,m and SEps,p,m represent CH4 emissions from surface mining and its post-mining processes, respectively. Ea,p,m denotes emissions from abandoned coal mines. Emissions from abandoned mines are estimated as 1% of the total CH4 emissions from coal mining, following previous studies25,31.

Oil and gas system

For the oil and gas system, we also use the Eq. (1) and Eq. (2) to estimate monthly CH4 emissions at the provincial level. For the onshore sector, activity data consist of monthly crude oil and natural gas production at the provincial level, obtained from the National Bureau of Statistics (https://data.stats.gov.cn)56. For the offshore sector, activity data were estimated using Eq. (3):

$${{MADoff}}_{i,p,m}={A{ADoff}}_{i,p,y}\times {{Vm}}_{i,p,m}$$
(3)

Where MADoffi,p,m represents the activity data for monthly offshore crude oil and natural gas methane emissions. i denotes the two sectors: crude oil and natural gas. p represents the six coastal provinces of China, Tianjin, Hebei, Liaoning, Shanghai, Shandong, and Guangdong, which are the primary contributors to output of offshore crude oil and natural gas. m and y indicate the month and year, respectively. AADoffi,p,y represents the annual output of offshore crude oil or natural gas in province p for year y, with data for with data for 2013–2018 sourced from the China Marine Statistical Yearbook61. Due to the absence of available provincial-level monthly production for offshore crude oil and natural gas after 2018, we collected national-level offshore production data for 2010–2021 (with missing data for 2020) from online sources (Supplementary Table 5). To estimate the missing national-level offshore production, interpolation was applied for 2020 using data from 2019 and 2021, and linear extrapolation was performed for 2022 based on observed trends from 2010 to 2021. The estimated 2019–2022 national-level offshore production data were then allocated to individual provinces based on their 2018 provincial production proportions, allowing for the derivation of provincial-level offshore crude oil and natural gas production. Vmi,p,m represents the monthly variation factor of crude oil or natural gas production in province p, derived from the monthly production variations reported by the National Bureau of Statistics (https://data.stats.gov.cn)56. The EFs of CH4 emissions from the processes including venting, flaring, exploration, production and upgrading, transport, refining/processing, transmission, and storage are collected from the 2019 Refinement55 and Yang et al.62. The data source of CH4 emssion estimates for the oil and gas system sector are shown in Table 3.

Table 3 Data source of methane emssion estimates from oil and gas system sector.
Full size table

Energy combustion

The energy combustion sector includes emissions from fossil fuel combustion and bio-fuel combustion. For the fossil fuel combustion, we calculated the monthly emissions from power combustion, industry, transport, other sectors (residential) at the provincial level with the following Eq. (4).

$${{ECE}(C{H}_{4})}_{p,m,i}=\sum _{m}{{CM}(C{O}_{2})}_{p,d,i}\times \frac{{{EF}(C{H}_{4})}_{i}}{{{EF}(C{O}_{2})}_{i}}$$
(4)

Where ECE(CH4)p,m,i represents CH4 emissions from sub-sectors i in province p during month m, CM(CO2)p,d,i represents the daily CO2 emissions from sub-sectors i in province p on day d, including contributions from power combustion, industry, aviation ground transport, and residential sectors. These data are obtained from CM-China35,63,64 (https://cn.carbonmonitor.org).

EF(CH4)i and EF(CO2)i are the CH4 and CO2 emission factors (Supplementary Table 7), respectively, for fuel type i, source from the 2006 IPCC Guidelines31.

For bio-fuel combustion, provincial-level monthly CH4 emissions were also calculated using Eq. (1). However, firewood and in-field straw consumption were used as the activity data, along with CH4 emission factors specific to firewood and in-field straw combustion. Since no province-level firewood and in-field straw consumption data are available after 2010 (except for Zhejiang province), we estimated provincial values using national-level data and provincial shares. National consumption data for 2012, 2014, and 2018 were collected from Tian et al.65, Tang and Li66 and Cong et al.67. Missing values for 2013, 2015, 2016, and 2017 were estimated through linear interpolation between available data points. For 2019–2022, we applied linear extrapolation based on trends observed in 2012, 2014, and 2018 to extend the dataset forward. To downscale national estimates to the province level, we used the most recent available province-level share from 201068. We assumed this share remained constant over time, as interannual variations in the provincial share of firewood and in-field straw consumption between 2000 and 2008 were less than 5%, based on data from the China Energy Statistical Yearbook (2000–2008)69. For Zhejiang province, firewood and in-field straw consumption data for 2010–2018 were collected from the Zhejiang Natural Resources and Environment Statistical Yearbook70, while values for 2019–2022 were extrapolated based on the observed trend. This dataset was used to replace the estimated values for Zhejiang province that were previously derived using the fixed provincial share approach. We will continue to collect and update data to further improve the quality of our dataset in future work.

The EFs of CH4 from firewood and in-field straw consumption is taken as 3.17 g kg−1 and 4.85 g kg−1 from Yue et al.71. The province-level of monthly proxy data corresponding to emissions from firewood and in-field straw consumption is taken as the regional monthly variation of PM2.5 emissions from in-field straw and Firewood estimated from Wu et al.72 (Supplementary Table 7).

Rice cultivation

We classify the rice cultivation sector based on the cropping cycle into single-season rice and double-season rice, with the latter further divided into early rice and late rice. Monthly CH4 emissions from rice cultivation in province p during month m and year y, denoted as REy,p,m, are calculated using the following Eq. (5).

$${{RE}}_{y,p,m}=\sum _{i}{({AD}}_{y,p,i}\times {{EF}}_{p,i}\times {t}_{p,i,m})\times {{Or}}_{p,m}$$
(5)

Where ADy,p,i is the annual rice planting area for rice type i in province p during year y. EFp,i is the CH4 emission factor for rice type i in province p. tp,i,m represents the growing period for rice type i in month m. Orp,m is monthly CH4 flux observations from 21 sites in 16 provinces as monthly proxy data for annual emissions (Table 4). The sub-category i includes single-season rice and double-season rice. We collected four groups of EFs from rice cultivation, considering temperature, rice growing period, yield, area, transplanting, sowing and harvesting dates, organic fertilizer use, irrigation management, rice variety, and soil sand content (Supplementary Table 8). Due to incomplete observation coverage across all provinces and years (2013–2022), missing values in Orp,m were substituted with the average CH4 flux from other provinces within the same rice-planting region.

Table 4 Average monthly methane fluxes (unit in mg m−2 h−1) or observed monthly emissions (unit in g m−2, represented by ‘*’) at 21 sites.
Full size table

Livestock

Annual CH4 emissions from livestock are estimated as Eq. (6).

$${LE}_{i,m,p}=\sum _{t}{L{EF}}_{i,t,p}\times (\frac{(a+b)\times {{time}}_{t}}{365\times 12})$$
(6)

Where LEi,m,p is the estimated CH4 emissions from livestock sector i in province p during month m. LEFi,m,p is the emission factor (kg CH4 head−1 m−1) for species t in sector i and province p. Sub-sector i includes enteric fermentation and manure management. a and b represent annual livestock stock and annual slaughter (both 10000 heads per year) of species t, respectively. time means the growth period of species t (unit in days). The slaughter numbers for cattle and pigs (2019–2022) and sheep (2021–2022) were obtained from the China Rural Statistical Yearbook73, while the year-end stock and slaughter numbers for other species were sourced from the National Bureau of Statistics (https://data.stats.gov.cn)56.

For enteric fermentation sector, the CH4 emission factor for species is from Huang et al.74 and Zhou et al.75. For manure management system, CH4 emissions are influenced by temperature, as stated in the 2006 IPCC Guidelines31. We adopted the Tier 2 method recommended in the 2019 Refinement55 to estimate the monthly CH4 emission factors for manure management systems, denoted as MEFt,p, which are calculated using Eq. (7):

$${{MEF}}_{t,p}=\frac{{{VS}}_{t}}{12}\times \left({B0}_{t}\times 0.67\times \sum {{MCF}}_{t,p,m,s}\times {{AWMS}}_{s}\right)$$
(7)

Where VSt is the annual volatile solid (VS) excreted for livestock category t (kg dry matter animal−1 yr−1). B0t is the represents maximum methane producing capacity (m3 CH4 kg−1) for species t. We collected B0t from Table 10.16 A in the 2019 Refinement55. MCFt,p,m,s represents the methane conversion factors (%) for livestock category t under manure management system s in province p during month m. AWMSs (dimensionless) is the fraction of livestock category t’s manure handled using animal waste management system s.

The value of VSt is calculated using Eq. (8):

$${{VS}}_{t}={{VS}}_{{rate}(t)}\times \frac{{{TAM}}_{t}}{1000}\times 365$$
(8)

Where VSrate is the default VS excretion rate (kg VS day−1) for livestock category t, which were collected them from Table 10.13a in the 2019 Refinement55. TAMt is typical animal mass (kg animal−1) for livestock category t, obtained from Table 10 A.5 in the 2006 IPCC Guidelines31.

For MCF estimation, we applied the Spreadsheet example for the calculation of a country or region-specific MCF provided in the 2019 Refinement55, requiring monthly inputs of temperature, B0 and VSt. Monthly average temperature data for each province p in China from January 2013 to December 2022 was obtained from the ERA5 dataset76.

Wastewater

CH4 emissions from wastewater are estimated as the total emissions from industrial wastewater, agricultural wastewater, and domestic wastewater. The methodological framework for estimating monthly provincial-level CH4 emissions from wastewater is illustrated in Fig. 2.

Fig. 2
figure 2

Workflow for monthly methane emission from wastewater at provincial level in mainland China.

Full size image

Monthly CH4 emissions from the three wastewater subsectors, denoted as WWEs,i,p,m, are estimated using Eq. (9):

$${{WWE}}_{s,i,p,m}={{AD}}_{s,i,p,m}\times {B0}_{s,i}\times {{MCF}}_{s,i}$$
(9)

Where ADs,i,p,m represents the activity data for subsector s using wastewater treatment method i in province p and month m. B0s,i denotes the maximum CH4-producing capacity of wastewater for subsector s using wastewater treatment method i. MCFs,i is the methane correction factor, accounting for variations in treatment efficiency. i represents the wastewater treatment method, which includes direct discharge and centralized treatment and other methods. s, p, and m indicate the wastewater subsector, province, and month, respectively.

Industrial wastewater is either directly discharged or treated through centralized treatment. For industrial wastewater with direct discharge treatment, the activity data is the chemical oxygen demand (COD) discharge in province p, denoted as ADinddp, which is estimated using Eq. (10).

$${{AD}{ind}d}_{p}={{DCODI}}_{p}\times (1-{{WWId}}_{p}/{{WWIe}}_{p})$$
(10)

Where DCODIp represents the COD discharge of industrial wastewater in province p. WWIdp and WWIep represent the amount of industrial wastewater treatment (tons) and industrial wastewater discharge (tons) in province p, respectively. Because no WWIdp data are available after 2015, a linear relationship was identified between WWIdp and the treatment capacity of industrial wastewater facilities (y = 196.38x - 3604.5, R² = 0.96). Based on this relationship, the treatment capacity of industrial wastewater facilities was used as a proxy to extrapolate WWId values beyond 2015. We collected the annual DCODIp, WWIdp, WWIep and the treatment capacity of industrial wastewater facilities from the China Environment Statistical Yearbook77.

For industrial wastewater treated through centralized treatment, the activity data is the COD removed from industrial wastewater at provincial level (RCODp) is calculated using Eqs. (11) and (12):

$${{RCOD}}_{p}={{WWId}}_{p}\times ({{PreCcod}}_{p}-{{AftCcod}}_{p})$$
(11)
$${{AftCcod}}_{p}=\,\frac{{{DCODI}}_{p}}{{{WWIe}}_{p}}$$
(12)

Where PreCcodp and AftCcodp represent the average COD concentration before and after treatment, respectively. Based on historical data from 2004 to 2010 in the China Environment Statistical Yearbook77, the trend in PreCcodp remain stable. Therefore, it was assumed that variations in PreCcod after 2010 were minimal, and the average value from 2004–2010 was used as the reference for PreCcodp beyond 2010.The value of DCODIp, WWIdp and WWIep for province p are colllected from the China Environment Statistical Yearbook77.

Monthly variation in CH4 emissions from industrial wastewater with centralized treatment is captured by a proxy RWWIei,p,m,y, as expressed in Eq. (13):

$${{RWWIe}}_{i,p,{my}}=\frac{{{IP}}_{i,p,m,y}}{\sum _{p}{{IP}}_{i,m,y}}\times \frac{{{WWIe}s}_{i,y}}{\sum _{i}{{WWIe}s}_{y}}$$
(13)

Where IPi,p,m,y refers to the industrial product output of industrial sector i in province p during month m of year y. WWIesi,y represents the wastewater discharge by industrial sectors i of year y. Supplementary Table 9 lists the seven major industrial sectors responsible for wastewater discharge along with their corresponding industrial products. The monthly industrial product output for each province (IPi,p,m,y) was obtained from the National Bureau of Statistics (https://data.stats.gov.cn)56 and wastewater discharge by industrial sectors (WWIesi,y) were sourced from the China Environment Statistical Yearbook77.

Agricultural wastewater is primarily discharged directly into rivers, lakes, or seas; therefore, direct discharge is the predominant treatment method. The activity data (DCODAp) for CH4 emissions from agricultural wastewater correspond to the COD of agricultural wastewater at the provincial level, sourced from the the China Environmental Statistics Yearbook77.

Domestic wastewater is classified into rural and urban wastewater based on its generation area. In terms of treatment methods, domestic wastewater is managed through direct discharge, centralized treatment, or other treatment, with the latter primarily referring to toilet wastewater treatment. For rural and urban wastewater treated by direct discharge, the activity data required for CH₄ emission estimation is the biochemical oxygen demand (BOD), which is estimated using Eq. (14).

$${{BODd}}_{p}=({{DCO}{Dd}}_{p}\times {(1-{Ru}}_{p})+{{DCO}{Dd}}_{p}\times {{Ru}}_{p}\times (1-{R{dom}}_{p}))\times {{Ccod}}_{p}$$
(14)

Where BODdp represents the BOD of domestic wastewater treated though direct discharge in province p. Since officially published BOD data is unavailable, BOD is estimated using COD and the conversion ratio between BOD and COD. DCODdp represent the COD discharge of domestic wastewater in province p which are obtained from the China Environment Statistical Yearbook77. Ccodp represent the the ratio of BOD and COD at provincial level and the data source are collected from Song et al.78 (Supplementary Table 10). Rup represents the provincial urbanization rate, obtained from the China Household Survey Data79. Rdomp represents the non-direct discharge treatment rate of urban domestic wastewater in province p, obtained from the China Environment Statistical Yearbook77.

For urban domestic wastewater with centralized treatment, the activity data is the removal of chemical oxygen demand (RCODdu). Since no official public data of RCODdu is available, it is estimated using Eq. (15) and Eq. (16), based on the definition provided in the China Environment Statistical Yearbook77.

$${{RCODdu}}_{p}={{PCODdu}}_{p}-{{DCODd}}_{p}\times {R{dom}}_{p}$$
(15)
$${{PCODdu}}_{p}={{ICOD}}_{p}\times {{Ru}}_{p}\times {{POP}}_{p}\times 365$$
(16)

Where RCODdup represents the removal of chemical oxygen demand in urban domestic wastewater for province p. PCODdup denotes the total COD in urban domestic wastewater discharge for province p. Rup and Rdomp have the same definations as in Eq. (14). ICODp represents per capita daily COD content in urban domestic wastewater, with a national average of 75 grams per person per day, sourced from the China Environment Statistical Yearbook77. POPp is the provincial-level resident population, obtained from the National Bureau of Statistics (https://data.stats.gov.cn)56.

For CH4 emissions from urban domestic toilet wastewater, the activity data are calculated using Eq. (17), and EFs are calculated using Eq. (18) and Eq. (19).

$${{BODo}}_{p}={{DCODd}}_{p}\times {{Ru}}_{p}\times ({{Rdom}}_{p}-{{Rdomc}}_{p})\times {{Ccod}}_{p}$$
(17)

Where BODop represents the BOD of domestic toilet wastewater in province p. DCODdp, Rup, Rdomp, and Ccodp are consistent with the definitions mentioned in Eq. (14) and Eq. (15). Rdomcp represents the rate of centralized treatment for urban domestic wastewater in province p, collected from the China Environment Statistical Yearbook77.

$${{doEFo}}_{p}=\,B0\times ({MCF}({ft})\times \frac{{{Num}({ft})}_{p}}{{{Num}({tt})}_{p}}+{MCF}({pt})\times \frac{{{Num}({pt})}_{p}}{{{Num}({tt})}_{p}})$$
(18)
$${{Num}({ft})}_{p}=\frac{{{Ru}}_{p}\times {{POP}}_{p}}{3}$$
(19)

Where doEFop represents CH4 emission factors from domestic toilet wastewater in province p. MCF(ft) and MCF(pt) represent the methane correction factor for household toilet wastewater and public toilet wastewater, respectively. B0 represents the Maximum CH4-producing capacity. Table 5 shows the value and data source for B0, MCF(ft) and MCF(fpt), respectively. Num(ft)p, Num(pt)p and Num(tt)p represents the number of household toilets, public toilets and total toilets in province p, respectively. We assume that each household has one toilet, allowing us to estimate the number of household toilets using the urban population and household size. The urban population is derived from the urbanization rate (Rup) and the resident population (POPp). The data source of Rup and POPp are shown in Table 6. According to the China Development Report80, the household size is set to 3. Given the numerous equations and parameters involved in the wastewater sector, we have summarized the data sources for these parameters in Table 6.

Table 5 B0 and MCF values under different wastewater treatment modes.
Full size table
Table 6 Data source of methane emssion estimates from wastewater sector.
Full size table

Solid waste

For the solid waste sector, we estimated the CH4 emissions from the following three subsectors, landfills, waste incineration, and biological treatment. The CH4 emissions from the landfill subsector is estimated by first-order decay (FOD) method according to the 2019 Refinement55 as Eq. (20).

$${{LE}}_{p}={({DOCdy}}_{p}+{{DOC}(a-1)}_{p}\times (1-{e}^{-k}))\ast \frac{16}{12}\ast F\ast (1-{O}_{x})$$
(20)

Where DOCdyp is the mass of decomposable DOC in the disposal year for province p. DOC(a-1)p represents the accumulated mass of decomposable DOC in the last year for province p. F represents the proportion of CH₄ in the generated landfill gas and is set to 0.5 in this study based on Peng et al.11 and Gao et al.32. Ox is the oxidation factor, set to 0.1, also sourced from Peng et al.11 and Gao et al.32. k is the reaction constant, set to 0.3 according to the People’s Republic of China National Greenhouse Gas Inventory81,82. DOCdyp and DDOCac are calculated using Eqs. (21) and (22).

$${{DOCdy}}_{p}={{DOCdep}}_{p}\times \left({1-e}^{(-k(\frac{13-M}{12}))}\right)$$
(21)
$${{DOC}(a)}_{p}={{DOCun}}_{p}+{{DOC}(a-1)}_{p}\times {e}^{-k}$$
(22)

Where DOCdepp and DOCunp represent the mass of eposited decomposable DOC and unreacted decomposable DOC in the disposal year, respectively. m represents the month when the reaction begins, which in this study is set to November. DOCdepp is calculated by Eq. (23).

$${{DOCdep}}_{p}=W\times {DOC}\times {DOCf}\times {MC}{F}_{p}$$
(23)

Where W is the mass of deposited waste. DOC represents the fraction of degradable organic carbon in municipal solid waste (MSW) at the time of deposition. Based on previous studies11,27,83,84,85, we set DOC to 6.5%, which reflects the average organic carbon content in China’s MSW composition. DOCf represents the ratio of divided waste. Based on previous studies11,27,83,84,85, the DOCf ranges from 0.5 to 0.77. Referring to Peng et al.11 and Du et al.83, we adopted a value of 0.6 in this study. MCFp represents the methane correction factor for aerobic decomposition in province p and is obtained from Du et al.84 (Supplementary Table 11). According to the 2019 Refinement55, the FOD method requires 50 years of activity data, covering landfill amounts from 1967 to 2022. Provincial-level W data for landfill are obtained from the China Statistical Yearbook86 for 2001–2002 and the Urban Construction Statistical Yearbook87 for 2003–2020. Since no W data are available before 2000, the missing values (W1966–2000) are calculated using Eq. (24).

$${W}_{1966-2000}=pW\times {{POP}}_{p}\times {RL}$$
(24)

Where pW represents the per capita waste collection amount, which is treated as a constant value of 0.025 tons per person per year, based on the China Greenhouse Gas Inventory Research (2005)81. Provincial population data (POPp) were obtained from he National Bureau of Statistics (https://data.stats.gov.cn)56, with missing values interpolated for 1986, Hainan before 1988, Tianjin before 1967, and Chongqing before 1997. RL denotes the proportion of landfilled waste to total waste. Since national RL data are unavailable for certain years between 1967 and 2000, values were assumed to be 10% for 1967–1978 and 20% for 1978–1990. Missing RL data for some provinces were interpolated between available years.

We calculated CH4 emissions from the waste incineration subsector (IWEp) following the method outlined in the 2006 IPCC Guidelines31 as shown in Eq. (25):

$${{IWE}}_{p}=({{IW}}_{p}\times {{EF}}_{{ie}})\cdot {10}^{-6}$$
(25)

Where IWp (Gg yr−¹) represents the amount of waste incinerated in province p, and EFie (kg CH4/Gg of waste) denotes the aggregate CH4 emission factor. In this study, the yearly amounts of waste incinerated at the provincial level were collected from the China Statistical Yearbook86. For regions such as Tibet, Shaanxi, and Qinghai, where data for certain years were unavailable, the missing values were estimated using linear interpolation based on data from adjacent years. Regarding EFie, we adopted the default value of 6500 g CH₄ per ton of municipal solid waste (wet weight), as suggested by the 2006 IPCC Guidelines31. The factor 10−6 was applied to convert values from kilograms to gigagrams.

The CH4 emissions from the biological treatment subsector (BWEp) are calculated using Eq. (26).

$${BWE}_{p}={M}_{{ow}}\times {{EF}}_{b}\times {10}^{-3}$$
(26)

Where M represents the mass of organic waste subjected to biological treatment. EFb is the CH4 emission factor for biological treatment, set to 4 g CH4 per kg of treated waste based on the 2019 Refinement55. Given the relatively stable monthly CH₄ emission pattern in the solid waste sector, annual emissions are distributed across months in proportion to the number of days in each month. The allocation is calculated using Eq. (27):

$${E}_{s,m}={E}_{s,y}\times \frac{{{Day}}_{m}}{{{Day}}_{y}}$$
(27)

Where Es,m represents CH₄ emissions from sector s in month m, Es,y denotes CH₄ emissions from sector s in year y, and Daym is the number of days in month m.

Wetland

CH4 emissions from wetland, including coastal wetlands and inland wetlands, are estimated using Eq. (4). The activity data for this sector correspond to the wetland area (WA). The Second National Wetland Resources Survey (2009–2013)88 provides provincial-level wetland area for 2013. For subsequent years, wetland areas are estimated using Eq. (28).

$${{WA}}_{y,p,i}={({WA}}_{2013}+{AR}\times (y-2013))\times \frac{{{AD}}_{2013,p,i}}{{{AD}}_{2013}}$$
(28)

Where WAy,p,i represents the wetland area in year y for province p and wetland type i. WA2013 is the national wetland area in 2013 (excluding rivers and lakes), totaling 3445.67 (10000 hectares). AR denotes the annual wetland area increase, set at 179.7 (10000 hectares), which are estimated based on Meng et al.89 and public reports90. Since province-specific wetland area data are unavailable for 2014–2022, national-scale wetland area growth was estimated using Eq. (28). The provincial distribution of wetland area was allocated using the 2013 provincial wetland area proportions, under the assumption that relative proportions remained stable during this period. To estimate CH4 emissions, regional climate-specific EFs were applied at the provincial level based on climate zone divisions (Supplementary Table 12)91. Monthly CH4 flux observations from 37 wetland sites in 16 provinces were used as proxies for distributing annual emissions at the monthly scale (Supplementary Table 13). For provinces with missing monthly CH4 flux observations, two approaches were applied to estimate emissions during the non-growing season. For Inner Mongolia, Heilongjiang, Liaoning, and Hebei, non-growing season emissions were assumed to contribute 5% of the annual total92. For Tibet, Sichuan, and Jiangxi, missing values were estimated based on the average monthly CH4 fluxes from provinces within the same climatic zone. For provinces without CH₄ flux observations, the mean flux values from the corresponding climatic zone were applied. The monthly distribution of CH₄ emissions for all 31 provinces is provided in Supplementary Table 14.

Data Records

The Monthly Methane Emission Inventory for China’s Provinces (MMCP) provides monthly CH4 emission data for 31 provinces in mainland China, spanning from January 2013 to December 2022. The dataset covers eight key sectors: (1) Coal Mining, (2) Oil and Gas Systems, (3) Energy Combustion, (4) Rice Cultivation, (5) Livestock, (6) Solid Waste, (7) Wastewater, and (8) Wetlands.

For each province, the data are presented in a consistent tabular format, with 1,080 rows representing sectoral and total monthly emissions. Table 7 lists the dataset structure and variable definitions. Figure 3 shows the temporal evolution of sectoral CH4 emissions across the 31 provinces. Model validation is conducted by comparing total anthropogenic emissions with those from the EDGARv8.0 dataset (red lines). Table 8 summarizes the 10-year average monthly methane emissions. The complete dataset is publicly available on Figshare93 (https://doi.org/10.6084/m9.figshare.26806522.v1). Users should note that all emissions are expressed in kilotons of CH4 per month. The dataset has been updated through December 2022, and future updates will be released as needed.

Table 7 Data attributes.
Full size table
Fig. 3
figure 3

Monthly methane emissions by sector across 31 provinces in mainland China from Jan. 2013 to Dec. 2022.

Full size image
Table 8 A ten-year average monthly CH4 emission inventory of China’s provinces, unit in kt CH4 per month.
Full size table

Technical Validation

Validation MMCP against other monthly datasets

In addition to this study, monthly methane emission inventories for China have been provided by EDGARv6.015 and Gong and Shi38 based on a bottom-up method. We compared the monthly emission results from these three datasets at the national (Fig. 4) and provincial (Fig. 3) levels. At the national level, the difference between our results and those of Gong and Shi38 ranges from -27.29% (August) to 5.30% (April). Our estimates are closest in spring (Feb-May) with an average difference of only 3.56%, 10.67% higher in summer (May-Jul), and 19.02% higher in winter (Nov-Jan). However, the largest discrepancies occur in autumn, with our estimates being nearly 30% higher in August and about 20% lower in October. Compared to EDGARv8.022 (red line in Fig. 4a), our estimates are closest in August, being nearly 8% (±1.17%) higher, but in other months our estimates are lower, with underestimates ranging from 10.70% (±1.91%) to 45.08% (±3.29%). Months with differences over 40% were February (43.96% ± 5.9%), April (41.95% ± 3.71%), and May (45.08% ± 3.29%).

Fig. 4
figure 4

Comparison of total monthly methane emissions (a) and sectoral monthly methane emissions (b–i) at national level between MMCP, EDGARv8.096 and Gong and Shi38.

Full size image

The observed differences primarily arise from the use of different monthly proxies to estimate annual methane emissions from the rice cultivation sector. This study employed methane flux data observed at paddy flux towers, EDGARv6.015 used planting area data derived from a single double-season rice site, and Gong and Shi38 used the normalized difference vegetation index (NDVI). As illustrated in Fig. 5a, these monthly proxies exhibit distinct seasonal variation patterns. NDVI shows relatively stable monthly trends, particularly from July to September, resulting in estimated methane emissions displaying a prolonged summer peak during these months (purple line in Fig. 5b). In contrast, the planting area used by EDGARv6.015 shows two distinct peaks in June and August (green line in Fig. 5a), corresponding to methane emissions with a bimodal distribution during the rice growing seasons (green line in Fig. 5b). Methane flux observations, however, reveal a gradual increase from April to June, a sharper rise in July and August, and a rapid decline after September, leading to a different seasonal trend in methane emissions (orange line in Fig. 5b). The significant differences in the seasonal dynamics of these proxies, particularly from June to October, account for the largest discrepancies in the monthly methane emission estimates between this study, EDGARv6.015, and Gong and Shi38. Specifically, our study’s estimates differ most notably from EDGARv6.0 in June and from Gong and Shi38 in October. It is worth noting that Gong and Shi38 did not provide the specific NDVI data used in their study. To better compare the proxies, we collected NDVI monthly variation data from Mao et al.94 and Laborte et al.95 to illustrate the differences in proxy selection.

Fig. 5
figure 5

Monthly variation in NDVI, planting area and observed methane flux used as monthly proxies for Gong and Shi38, EDGARv6.015, and This study, respectively (a) and comparison of monthly methane emissions from rice cultivation sector in 2015 among Gong and Shi38, EDGARv6.015, and This study (b).

Full size image

Apart from the rice cultivation sector, this study, EDGARv6.015, and Gong and Shi38 exhibit notable differences in the selection of monthly proxies for other sectors. In the energy combustion sector (Fig. 4h), we used daily CO2 emissions from fossil fuel combustion at the provincial level and monthly regional PM2.5 emissions as proxies for fossil fuel combustion and biofuel combustion, respectively. Meanwhile, Gong and Shi38 applied monthly fire emissions as a proxy for biofuel combustion, whereas EDGARv6.015 utilized rural population data from the Global Human Settlement Layer (GHSL) database as both temporal and spatial proxies96. For the manure management sector (Fig. 4c), both this study and Gong and Shi38 employed surface temperature as a monthly proxy. However, the two studies differed in data sources: we used surface temperature data from ERA576, while Gong and Shi38 relied on MOD11C3-MODIS. In the wastewater sector (Fig. 4f), we incorporated monthly industrial production data for major industries with substantial wastewater discharge to estimate industrial wastewater emissions. Whereas, EDGARv6.015 and Gong and Shi38 assumed wastewater emissions remained constant throughout the year without considering monthly variations. Regarding the wetland sector (Fig. 4i), this study utilized methane flux observations from wetland monitoring sites, whereas Gong and Shi38 employed monthly rainfall and surface temperature as proxies. Since EDGARv6.015 is an anthropogenic emission inventory, it does not include methane emissions from wetlands.

Validation MMCP against other yearly datasets

We compared our results with other bottom-up inventories on China’s annual emissions (Fig. 6a). In this study, the average annual emission from 2013 to 2022 is 56.35 Mt yr−1, with an average annual growth rate of 0.35 Mt yr−1. Differences between our results and those of EDGARv8.022, FAO23, NCCC16,17,18,19,20, Huang et al.74, and Gong and Shi38 are 20.61%, 4.13%, 3.14%, 10.51%, 9.53% and 9.64%, respectively. Our upward trends of China’s annual CH4 emissions are slightly slower compared to EDGARv8.022 (0.77 Mt yr−1) and FAO23 (0.55 Mt yr−1). Compared to top-down inversions using GOSAT satellite observations, our results are 9.37% and 2.17% lower than Miller et al.42 and Qu et al.40, but higher than other GOSAT-based inversions, with overestimates ranging from 2.09%41 to 22.01%45. Our estimates are 16.51% lower than the TROPOMI-based inversion of Chen et al.7.

Fig. 6
figure 6

Comparison in China’s annual methane emissions from 2013 to 2022 (a), and sectoral methane emissions by sectors in 2015 and 2019 (b) with other studies. “Waste” represents the sum of emissions from the solid waste and wastewater sectors. “Energy” represents the combined emissions from coal mining, oil and gas system, and energy combustion.

Full size image

Figure 6b presents a comparison of our results with other studies on China’s CH4 emissions by sector for 2015 and 2019. In the coal sector, the discrepancy between our results and the emissions from Gong and Shi38 and Chen et al.7 based on TROPOMI observations are 7.59% and 9.00%, respectively. However, compared to emission estimates from EDGARv8.022, FAO23, and Miller et al.42 based on GOSAT observations, the differences exceed 20%. Our lower CH4 emission estimates from the coal mining sector are partly due to the adoption of provincial EFs and recovery rates. Previous studies12,24,30 indicate that CH4 emission calculations from coal mining in China using provincial EFs and recovery rates are significantly lower than those based on national EFs and recovery rates. We compared our study with Sheng et al.12, which is based on provincial EFs and recovery rates, and found that the average annual difference in CH4 emissions from the coal mining sector is only 1.87%.

In the livestock sector, our results show the smallest difference of less than 2% compared to Gong and Shi38. Our estimates are 20.7% lower than those of Chen et al.7, which are based on TROPOMI observations. However, our estimates are higher than those from other inventories and GOSAT-based inversions, with differences ranging from 8.4% (NCCC) to 48.19% (FAO). For the rice sector, our estimates are over 30% higher than the FAO23 estimate but lower than other studies, with underestimates ranging from 14.05% to 81.6%. In the wastewater sector, our estimate is 7.82% higher than NCCC (UNFCCC)17 results and 13.81% and 18.51% lower than Gong and Shi38 and Huang et al.74, respectively. The EDGARv6.0 estimate for CH₄ emissions from the wastewater sector in China is approximately three times higher than our estimate. For the solid waste sector, our results are slightly less than 5% higher than those of Huang et al.74 and EDGARv8.022, but 16.27% and 46.46% lower than NCCC (UNFCCC)17 and Gong and Shi38, respectively. Comparing our overall emission estimates from the wastewater and solid waste sectors with Miller et al.42 based on GOSAT observations and Chen et al.7 based on TROPOMI observations, the differences are close to 4%. However, our emission estimates from total waste sector are significantly lower than those from FAO estimates, by approximately 90%.

Validation MMCP against in-situ observations

We compared the monthly inventory emission results for the Beijing-Tianjin-Hebei region and Qinghai Province with monthly atmospheric CH4 concentrations observed at the Shangdianzi and Waliguan atmospheric background stations (Fig. 7). Atmospheric CH4 concentration data for Waliguan (https://gml.noaa.gov/aftp/data/trace_gases/ch4/flask/surface/txt/ch4_wlg_surface-flask_1_ccgg_month.txt) and Shangdianzi (https://gml.noaa.gov/aftp/data/trace_gases/ch4/flask/surface/txt/ch4_sdz_surface-flask_1_ccgg_month.txt) were obtained from the Global Monitoring Laboratory (GML) of the National Oceanic and Atmospheric Administration (NOAA).

Fig. 7
figure 7

Comparison of atmospheric methane concentrations between MMCP estimates and observations from background monitoring stations at Shangdianzi (a) and Waliguan (b).

Full size image

In the Beijing-Tianjin-Hebei region, the monthly fluctuation of CH4 emissions aligns well with atmospheric methane concentrations at Shangdianzi, particularly in peak occurrence and duration. In Qinghai province, the monthly emissions in the inventory closely match atmospheric methane concentrations, particularly regarding the timing and duration of peak emissions. However, discrepancies between this study and observed atmospheric CH4 concentrations may arise due to the varying distances between the Waliguan station and different CH4 emission sources. These discrepancies are influenced by atmospheric transmission effects, leading to differences between emission data and atmospheric concentration trends. Additionally, limitations in observation data, such as those from wetlands, also contribute to these differences.

Uncertainty analysis

Both the Monte Carlo simulation and the error propagation method are widely used for uncertainty estimation in greenhouse gas emission inventories. Recent studies favor Monte Carlo simulations for their ability to model uncertainty probabilistically. However, this approach requires well-defined probability distribution functions for activity data and emission factors, which are often unavailable or highly uncertain for regional-scale methane emissions97. Given these challenges, the error propagation method remains a widely recommended approach in the IPCC’s Good Practice Guidance98 and has been extensively applied in national greenhouse gas inventories. Considering consistency and feasibility, we have adopted the error propagation method for uncertainty estimation in this study. We use the error propagation method to quantify uncertainty of total CH4 emissions as Eq. (29).

$${U}_{t}={(\sum _{i}{{U}_{i}}^{2})}^{\frac{1}{2}}$$
(29)

Where U represents the uncertainty of CH4 emissions from sector i at the provincial level. In emission inventories, uncertainty arises from multiple sources. For methane emissions across different sectors, the primary contributors to uncertainty include emission factors, activity data, and other estimation parameters, as defined by the sector-specific calculation equations. The uncertainty of CH4 emissions is estimated from the range of variability in activity data, EFs and other parameters.

Uncertainty of sectoral emissions

The uncertainty in CH4 emissions from coal mining is primarily influenced by the variability in emission factors (EFs) for underground coal mining and methane utilization rates. Provincial disparities in EFs are significant (Supplementary Table 3), for example, Zheng et al.99 reported an average EF of 5.97 m³/ton for Inner Mongolia, whereas Zhu et al.24 reported only 0.82 m³/ton. Furthermore, EFs from underground coal mines also are affected by gas content. For example, differences in emission factors are observed due to inconsistencies in coalbed methane content classification standards between Sheng et al.12 and Wang et al.58. Due to inconsistent data on methane utilization across provinces, the uncertainty in monthly CH4 emissions from the coal sector in this study ranges between 9.77% and 30.17% (Fig. 4b).

For livestock, it is derived from the ranges of EFs from enteric fermentation11,31,76. The uncertainty for monthly methane emissions from the livestock sector spans from 9.64% to 11.8% (Fig. 4c). The uncertainty of CH4 emissions is derived from the range of EFs from rice cultivation25,59,100,101. We obtained four sets of provincial-level EFs for single-season and double-season rice (Early rice and Late rice) from GCPGGI59, Wang et al.25, Fu and Yu100, and Yan et al.101 (Supplementary Table 5). GCPGGI59 and Wang et al.25 provided EFs based on the CH4MOD model and DNDC model, respectively, whereas Fu and Yu100 and Yan et al.101 derived EFs from observations of methane flux during the rice growth cycle. Among these sources, Wang et al.25 reported the highest EFs for single-season rice in most provinces, whereas Yan et al.101 reported the lowest EFs. For early rice, the EFs from Fu and Yu100 are comparatively lower than those from the other studies, while the EFs provided by GCPGGI59 for late rice are generally lower. The uncertainty for monthly methane emissions from the rice sector ranges from 15.92% to 25.12% (Fig. 4d).

The uncertainty for solid waste is calculated from the term of the content of DOC32,102. The uncertainty of CH4 emissions from wastewater is based on the range of MCF and B0. For industrial wastewater, The B0 range is from 0.45828 to 0.531, and for domestic wastewater is from 0.1 to 0.331. The MCF range (0.456–0.587 kg CH4 / kg BOD) is used to calculate the uncertainty of CH4 emissions from wastewater28. The uncertainty for monthly CH4 emissions from the solid waste and wastewater sectors ranges from 41.55% to 50.31% (Fig. 4e) and from 12.47% to 17.56% (Fig. 4f), respectively.

For oil and gas system, the uncertainty from EFs for onshore oil production (2.91–3.43 t per 1000 m3), transportation (0.0054–0.025 t per 1000 m3), onshore gas production (2.54–4.09 t per million m3), treatment (0.57–1.65 t per million m3), transportation (1.29–3.36 t per million m3), storage (0.29–0.67 t per million m3), and distribution (0.62–2.92 t per million m3)31,62. The annual methane emissions from the oil and gas sector exhibit the uncertainty ranging from 36.42% to 45.16% (Fig. 4g). The uncertainty of CH₄ emissions from energy combustion is calculated using the range of EFs for firewood (2.4–3.17 g CH4/ kg firewood) and in-field straw (2.8–5.2 g CH4 /kg straw)71. The uncertainty for annual methane emissions from the energy combustion sector ranges between 11.42% and 25.81% (Fig. 4h). Given the relatively low emissions and limited availability of data from the wetland sector, this study focuses solely on quantifying the uncertainties associated with emissions from anthropogenic sources.

Uncertainty of provincial emissions

We calculated the uncertainty in monthly methane emissions for each province using the error propagation equation, as illustrated in Fig. 8. The analysis reveals that Qinghai has the lowest monthly average emission uncertainty at 6.70%, whereas Anhui has the highest at 28.18%. Provinces with monthly average emission uncertainties exceeding 20% include Anhui, Beijing, Guizhou, Shaanxi, and Shanxi. In contrast, Gansu, Hebei, Jilin, Shandong, Tianjin, and Tibet have monthly average emission uncertainties below 10%. The remaining provinces exhibit uncertainties ranging from 10% to 20%. Furthermore, variations in monthly methane emission uncertainty are observed due to differing seasonal emission trends across sectors and the varying emission structures among provinces.

Fig. 8
figure 8

Uncertainty in monthly CH4 emissions for 31 provinces in mainland China.

Full size image

Limitations

Our datasets have the following limitations: (1) In the rice sector, we used methane flux observation data to allocate monthly emissions. However, due to gaps in temporal continuity and regional coverage, we interpolated data from nearby regions and time periods. Future efforts will focus on collecting additional observations to improve the accuracy of monthly emission estimates. (2) In the coal mining, oil and gas system, and fossil fuel combustion (FFC) sectors, the monthly variation in CH4 emissions primarily depends on monthly changes in activity data, while emission factors (EFs) are assumed to remain constant24,30,103. For the coal mining sector, CH4 emission factors are influenced by coal seam gas content, mining depth, and advanced mining technologies24,103. While these factors may change over time due to mining operations, they are minimally affected by seasonal variations such as temperature and precipitation. Additionally, quantifying the temporal variability of these factors remains challenging due to the lack of direct observational data. As a result, this study assumes a fixed EF for coal mining emissions. Future work will incorporate monthly EF variations as relevant datasets become available. Additionally, large-scale leakage events in coal mining and oil and gas systems can cause significant short-term fluctuations in emissions. However, our current inventory does not explicitly account for such events. Future research may improve the estimation of these episodic emissions by integrating remote sensing data, such as TROPOMI satellite observations or hyperspectral satellite imagery104,105,106,107. (3) In the solid waste sector, methane emissions were estimated based primarily on monthly activity data variations. However, research suggests that methane emission factors from landfills may also be influenced by temperature and precipitation108,109. Future studies will explore this relationship to improve the accuracy of the sector’s emissions. For wetlands, the estimation of provincial wetland areas relies on the assumption that the provincial proportions of wetlands in 2013 remain unchanged over time. Given that wetland restoration efforts vary across provinces, this assumption may introduce uncertainties. Future studies will incorporate province-specific wetland area changes to refine the estimates.