Agricultural Sector Homologous Emission Inventory of Air Pollutants and Greenhouse Gases for China

Abstract

The agricultural sector is the largest source for air pollutants of ammonia (NH₃) and greenhouse gases (GHGs) of nitrous oxide (N₂O) and methane (CH₄). Establishing a unified and homologous emission inventory of air pollutants and GHGs is essential for synergistic abatement pathway studies of air pollution and climate change. However, current agriculture emission inventories of air pollutants and GHGs are unclear due to the separated source classification and inconsistent calculating methodologies. This study adopted a synergistic approach to develop a unified emission inventory for NH₃, N₂O, and CH₄ from the agricultural sector in China for 2021, based on crop and livestock types as the activity level data, and considered regional-specific species in emission factors. The results showed that China’s agricultural emissions in 2021 amounted to 7566.17 Gg of NH₃, 486.14 Gg of N₂O, and 14,979.71 Gg of CH₄. Rice, cattle, and pigs were the primary contributors of NH₃, N₂O, and CH₄. Hotspots of NH₃ and N₂O emissions were concentrated in the North China Plain and Sichuan Basin, whereas CH₄ emissions were predominantly located in southern China. This study provides a crop- and livestock-specific data foundation for making region-specific and priority-based integrated strategies to improve air quality, mitigate climate change, and promote sustainable agricultural development in China.

1. Introduction

Facing the dual environmental challenges of climate change and air pollution [1,2], the Chinese Government has committed to reach a carbon emission peak before 2030 and achieve carbon neutrality before 2060 [3,4]. In the past years, a series of policies have been implemented, including the Air Pollution Control and Prevention Action Plan (APPA) and Blue-Sky Defense War, resulting in a substantial reduction in emissions of primary pollutants and fine particulate matter (PM_2.5) pollution [5,6]. However, the synergetic abatement paths face bottlenecks in terms of carbon neutrality and air quality compliance. In 2022, the annual average concentration of PM_2.5 in China was 29 µg/m³, which is considerably higher than the target of 5 µg/m³ set by the World Health Organization (WHO) [7]. For greenhouse gas (GHG) abatement, most decarbonization strategies focus on industrial and transportation sectors [8,9], which together account for most CO₂ and pollutant emissions, while relatively less emphasis has been placed on addressing non-CO₂ GHG emissions. To resolve this bottleneck, a deeper understanding of the characteristics of air pollutant and GHG emissions in other key sectors is required.

Sustainable agricultural production is a crucial pillar of China’s economy, playing a fundamental role in ensuring food security, supporting rural livelihoods, and maintaining social stability [10,11]. At the same time, the agricultural sector is the largest source of emissions of air pollution of ammonia (NH₃) [12], and about 85% of total NH₃ emissions came from livestock farming and crops of synthetic nitrogen fertilizers in China [13,14,15]. As the precursors to the formation of secondary inorganic aerosols (SIA) and PM_2.5, the contribution of agriculture-related NH₃ emissions to SIA and PM_2.5 are 29% and 16%, respectively [16]. Moreover, agricultural activities significantly contribute to global warming through the emission of greenhouse gases (GHGs) methane (CH₄) and nitrous oxide (N₂O) [17], producing approximately 60% of CH₄ [18,19,20] and 40% of N₂O emissions in China [21,22,23]. Therefore, the agricultural sector is a crucial one with significant potential in the integrated mitigation of both air pollutants and GHGs.

Developing a unified and homologous air pollutant and GHGs emission inventories is the foundation for identifying key emission sources and making realistic policy-relevant synergetic abatement strategies [5,24,25]. In previous research, GHG or NH₃ emissions have usually been quantified separately, with a focus on a specific gas and/or emission source. For instance, the estimates of agricultural NH₃ emissions in China range from 8.0–10.36 Tg [15,26,27]. Similarly, agricultural N₂O emissions in China range from 0.57–2.08 Tg [22,28,29]. CH₄ emissions from livestock enteric fermentation (LEF) and manure management were estimated to be between 9.44 and 15.01 Tg, while those from rice cultivation ranged from 5.61–14.16 Tg [20,30,31]. These inventories have provided a comprehensive assessment of agricultural emissions at the national scale. However, there are still significant variations in emissions across provinces and regions due to differences in economic development levels and agricultural production structures. Existing research on agricultural emission sources has primarily concentrated on major economic zones such as the Yangtze River Delta (YRD) [32], Pearl River Delta (PRD) [33], and Henan province [34]. Consequently, the spatial heterogeneity of agricultural emissions at the regional and provincial level remains insufficiently understood.

Meanwhile, only a few global or Asian inventories, such as Emissions Database for Global Atmospheric Research (EDGAR v8.0) [35], Greenhouse Gas and Air Pollution Interactions and Synergies (GAINS-China) [36], and the Regional Emission Inventory in Asia (REAS v1.1) [37], have turned out to be a homologous one of both air pollutants and GHGs. Due to inconsistencies in the classification of emission sources, substantial gaps and uncertainties have emerged across those inventories. Specifically, NH₃ emissions from animal manure are typically classified under manure management, whereas N₂O emissions from manure application are categorized under fertilizer-related emissions. This inconsistency complicates inter-inventory contrast and mitigation policy alignment. Furthermore, in these inventories, European or globally unified emission factors (EFs) were applied, which might not accurately reflect region-specific agricultural practices and species variations across different provinces of China. In particular, cropland soil N₂O and NH₃ emissions are typically estimated based on nitrogen fertilizer consumption as activity data [14]. However, neglecting the changes in fertilizer application across crop species, such as the transition from ammonium bicarbonate (ABC) to urea and compound fertilizers in China, leads to overestimation of agricultural emissions [38]. To date, not a single coupled inventory involving crop- and livestock-specific emissions have been reported. This gap hinders the quantification of sector-specific contributions and the formulation of precise and synergetic mitigation strategies.

The objectives of this inventory are to (1) develop a unified emission inventory for NH₃, N₂O, and CH₄ in China’s agricultural sector for 2021; (2) analyze the source contribution characteristics of these emissions; (3) elucidate the spatial distribution of NH₃, N₂O, and CH₄ within Chinese agriculture; and (4) identify crop- and livestock-specific and regional discrepancies in agricultural emissions. This study provides a homologous crop- and livestock-specific dataset on agricultural sector emissions to support the formulation of precise mitigation strategies for improving air quality, mitigating climate change, and promoting ecological sustainability.

2. Methodology and Data

2.1. Study Domain and Source Categorization

2.1.1. Study Domain

This study encompassed 31 provinces, municipalities, and autonomous regions in mainland China; Taiwan, Hong Kong, and Macao were excluded due to data limitations. To identify the discrepancies in NH₃, N₂O, and CH₄ emissions in different agricultural regions across China, we classified the study area into six regions based on the distribution of agricultural product systems and codes for the administrative divisions of the Peoples Republic of China [39]. As shown in Figure 1, the agricultural regions in China are divided into six parts: Northeast China (NEC: Liaoning, Jilin, and Heilongjiang), North China (NC: Beijing, Tianjin, Shanxi, Hebei, and Inner Mongolia), East China (EC: Shanghai, Jiangsu, Zhejiang, Anhui, Fujian, Jiangxi, and Shandong), South Central China (SCC: Henan, Hubei, Hunan, Guangdong, Guangxi, and Hainan), Southwest China (SWC: Chongqing, Sichuan, Guizhou, Yunnan, and Tibet), and Northwest China (NWC: Shaanxi, Gansu, Qinghai, Ningxia, and Xinjiang). The base year for the emission inventory was 2021.

2.1.2. Sources Categorization

Based on existing regulations and guidelines for atmospheric NH₃ [40] and GHGs [41] emission inventory in China, we developed a unified classification system for the major emission sources of NH₃, CH_4, and N₂O in the agricultural sector. This classification was comprised of three hierarchical levels of emission sources. The first level encompassed five main categories: livestock manure management (LMM), cropland fertilizers (CF), LEF, paddy cultivation (PC), and aquaculture (Aq). The second levels divided these categories into sectors according to livestock, poultry, or crop species, and third levels further divided breeding method (intensive or free-range) and fertilizer type (urea, ABC and NPK). The official statistical data provide a relatively complete data set for livestock headcounts, crop sown area and the amount of fertilizer application based on available the statistical yearbook, Supplementary Material Texts S1 and S3. A detailed classification of the emission sources is provided in

Table 1. Source categorization for agricultural NH₃, CH_4, and N₂O emission inventories in China.

Level I	Level II	Level III	Gas	Activity Data	Spatial Surrogates
LEF	Cattle;	Intensive; Free-range	CH4	Livestock population a,b,c	Rural residential area e
	Pig;
	Sheep;
	Draft animals;
LMM	Cattle;	Intensive; Free-range	NH3; N2O; CH4	Livestock population a,b,c	Rural residential area e
	Pig;
	Sheep;
	Draft animals;
	Poultry;
CF	Rice;	Urea; ABC; NPK	NH3; N2O	Crop sown area a,b; Fertilization rate d	The cultivated land area includes upland and paddy areas f
	Maize;
	Vegetables;
	Fruits;
	Wheat;
	Others
PC	Early rice;		CH4	Crop sown area a,b;	The area of land cultivated for paddy f
	Single rice;
	Late-rice
Aq	Mariculture;		N2O; CH4	Aquaculture area b,c	Water bodies for rivers, lakes, and ponds f
	Freshwater

^a Data collected from China Statistical Yearbook 2022 [42]. ^b Data collected from the China Rural Statistical Yearbook 2022 [43]. ^c Data collected from the China Animal Husbandry and Veterinary Yearbook 2022 [44]. ^d Data collected from the National Cost and Profit of Agri-Products Materials Compilation 2022 [45]. ^e Data from the National Earth System Science Data Sharing Infrastructure [46] (http://www.geodata.cn). ^f Data from Resource and Environment Science and Data Center [47], Chinese Academy of Sciences (http://www.resdc.cn).

.

2.2. Calculation of NH₃, N₂O and CH₄ Emissions

We used a bottom-up approach to aggregate the 2021 annual emissions of China’s agricultural sector by accounting for the emissions from each sub-source; the emissions from each sub-source were calculated as follows:

$$E_{i,j,k,l}=E_{LEF,i,j,k,l}+E_{LMM,i,j,k,l}+E_{CF,i,j,k,l}+E_{PC,i,j,k,l}+E_{Aq,i,j,k,l}$$

(1)

where $E_{i,j,k,l}$ is the 2021-based annual emissions of the total agriculture emissions (tons/year); E_LEF,i,j,k,l, E_LMM,i,j,k,l, E_CF,i,j,k,l, E_PC,i,j,k,l, and E_Aq,i,j,k,l represent the NH₃, CH₄, and N₂O emissions from LEF, LMM, CF, PC and Aq (tons/year), respectively; i represents emission gas species (NH₃, CH₄ and N₂O); j represents the provinces, including the land area of 31 provinces in mainland China (excluding Hong Kong, Macao, and Taiwan); k represents the second-level source type; and l represents the third-level source type. The specific activity data compilation approach and emission factor determination methods are described by category in detail in the following chapters.

2.2.1. Livestock Enteric Fermentation

Livestock enteric fermentation emits only CH₄, which is calculated as follows [41]:

$$E_{LEF,i}={\sum }_{jk}{PL}_{j,k}\times {EF}_{LEF, i,k,l}$$

(2)

where $E_{LEF,i}$ (tons/year) denotes the emissions from LEF; i represents CH₄; PL_j,k,l (head) denotes the livestock population in j province, k source of livestock, and l type of breeding method. The breeding methods included intensive and free range. According to the collected intensive breeding rates from reports and references, using linear regression to fit and forecast rates, the intensive farming rates in 2021 were 71.91% for broilers and laying hens, 68.65% for cows, 40.33% for beef cattle, 53% for pigs, and 38.4% for sheep. Farming methods in the other categories were all considered free range in this study. The details of PL_j,k are provided in Supplementary Material Text S1. EF_LEF,i,k,l (kg/head) represents the emission factors (EFs) for CH₄ from the k-th livestock sub-source and l-th type of breeding method. Owing to the relatively minor impact of regional variations, the EFs were not disaggregated by province. The CH₄ EFs for enteric fermentation across different livestock categories, based on feeding practices, were sourced from provincial guidelines and are described in detail in Supplementary Tables S1 and S2.

2.2.2. Livestock Manure Management

Livestock manure management emits NH₃, N₂O, and CH₄ simultaneously. NH₃ emissions were calculated as follows:

$$E_{LMM,i}={\sum }_{jk}{PL}_{j,k}\times {EF}_{LMM,i,k,l,m}$$

(3)

where $E_{LMM,i}$ denotes the NH₃ emissions from LMM, i represents NH₃, PL_j,k,l (head) represents the population of livestock for j province and k source of livestock, and l type of breeding method. EF_i,j,k,l,m is the emission factors for i gas, k sub-source of livestock, l type of breeding method (kg/head), and t temperature. For the LMM EFs of NH₃, according to the relevant data provided by the technical guidelines for the preparation of the atmospheric ammonia source emission inventory (Trial) [40], we obtained pixel-level monthly EFs of manure management using monthly temperature data. Details of the methods and parameters are presented in Supplementary Material Text S1, Tables S1 and S3.

The LMM emissions of N₂O and CH₄ were calculated as follows:

$$E_{LMM,i}={\sum }_{jk}{PL}_{j,k}\times {EF}_{LMM,i,j,k,l}$$

(4)

where $E_{LMM,i}$ is the emissions from LMM, i represents CH₄ or N₂O; PL_j,k,l (head) represents the population of livestock for j province and k source of livestock, and l type of breeding method. EF_i,j,k,l is the emission factors for i gas, j province, k source of livestock, and l type of breeding method (kg/head). The EFs of CH₄ and N₂O were obtained by referring to the guidelines for the Emission Inventory of Greenhouse Gases in China [41]. A detailed description of the EFs of CH₄ and N₂O from LMM is presented in Supplementary Tables S4 and S5, respectively.

2.2.3. Cropland Fertilizer

Cropland fertilizer emissions primarily from the use of synthetic fertilizers. These emissions are nitrogen-rich pollutants, namely NH₃ and N₂O. These emissions were quantified using the methods described by Misselbrook et al. [48] and Yue et al. [49]. The formula used for this quantification is as follows:

$$E_{CF,i}=\sum C_{j,k}\times {ARSNF}_{k,l}\times {EF}_{CF,i,k,l}$$

(5)

where E_CF,i denotes the emissions from the CF sub-sector (tons/year), i represents NH₃ and N₂O; C_j,k (ha⁻¹) denotes the cultivation area within province j for crop type k; ARSNF_jk (kg /ha⁻¹) represents the application rate of synthetic fertilizer for crop type k and fertilizer type l (data from “National Cost and Profit of Agri-product Materials Compilation, 2022” [45], and Supplementary Material Tables S6 and S7); ${EF}_{CF,i,k,l}$ (kg/ha⁻¹) represents the emission factors corresponding to gas type i, crop type k, and fertilizer type l. The emission factors of N₂O were obtained from field measurements and localized experiments ([49]; Supplementary Table S8). The NH₃ emission factors for CF were modified by functions associated with other variables (e.g., soil pH and land use) to obtain EFs, which were calculated using the following equation from Misselbrook et al. [48]:

$${EF}_{CF,NH3}={EF}_{max}\times {RF}_{soilPH}\times {RF}_{landuse}\times {RF}_{rate}\times {RF}_{rainfall}\times {RF}_{temperature}$$

(6)

where ${EF}_{CF,NH3}$ (%) denotes the NH₃ emission factors from CF; EF_max is the maximum potential emission for each fertilizer type. RF_soilpH, RF_landuse, RF_rate, RF_rainfall, and RF_temperature are modified parameters of soil pH, land use, application rate, rainfall, and temperature, respectively. Details of the methods and parameters are presented in Supplementary Material Texts S3 and S4, Tables S6, S7, and S9.

2.2.4. Paddy Cultivation

Paddy cultivation (excluding fertilization) exclusively gives rise to CH₄ emissions. CH₄ emissions were calculated based on the guidelines for the emission inventory of GHGs in China [41] in the following formula:

$$E_{PC,i}=\sum C_{j,k}\times {EF}_{PC,j,k}$$

(7)

where E_PC,i is total CH₄ emissions from PC (tons/year), C_j,k (ha⁻¹) denotes the rice cultivation areas within province j for rice type k, and ${EF}_{PC,i,j,k}$ denotes the CH₄ emission factors corresponding to province j and paddy type k (kg/ha⁻¹) (Supplementary Tables S6 and S10).

2.2.5. Aquaculture

Aquaculture leads to both N₂O and CH₄ emissions, which within this sub-sector were quantified in the method described by Xu [50]. The formula is as follows:

$$E_{A,i}={\sum }_{ij}{AA}_{jk}\times {EF}_{i,k}$$

(8)

where $E_{A,i}$ is N₂O or CH₄ emissions for Aq (tons/year), ${AA}_{jk}$ is the Aq area to province j and aquaculture type k (ha⁻¹), and ${EF}_{i,k}$ is the N₂O or CH₄ emission factor for Aq type k (mg/ha⁻¹/h) (Supplementary Tables S11 and S12).

2.3. Spatial Allocation

Agricultural emissions spatial allocation of NH₃, N₂O, and CH₄ were mapped to 3 km × 3 km grid cells by means of the Geographic Information System (ArcGIS; version 10.2). The formula is as follows:

$$E_{g,c}=E_c\times L_{g,c}/L_c$$

(9)

where $E_{g,c}$ represents the gas emissions in area c of grid g; g represents the ID of each grid, automatically generated by ArcGIS when the grid was created; c represents different provinces in China; and L represents latitude and longitude.

Based on source characteristics, different emission sources were allocated using various spatial proxies. The detailed spatial proxies used for ARS emissions allocation are presented in Table 1. In summary, data from rural residential areas were used to spatially allocate emissions from LEF and LMM. Since CF includes both upland and paddy areas, emissions from this source were apportioned based on changes in Land-Use and Land-Cover (LUCC) data for arable lands, covering both upland and paddy areas within the study domain. CH₄ emissions from PC were allocated to paddy areas identified in LUCC data. Emissions from Aq, being relatively small, were allocated to LUCC-identified water bodies, such as rivers, lakes, and ponds. After completing the spatial allocation, aggregation was conducted to sum all emissions. Our previous research included detailed approaches for the spatial allocation of these emission sources [15,51,52].

2.4. Uncertainty Analysis

Uncertainty in emission inventories can be described qualitatively, semi-quantitatively, and quantitatively. In this study, a quantitative approach in the Monte Carlo Simulation with R (version 4.2.3) was applied to evaluate the uncertainty of all sources of agricultural emissions according to the method by Zheng et al. [51]. By establishing emission factor databases from the literature, the uncertainties for each source and total emissions were quantified and key sources leading to uncertainty in model outputs were identified in the sensitivity analysis approach. Detailed input parameters are provided in Supplementary Table S13.

3. Results and Discussion

3.1. Emission Inventory and Uncertainty

The unified emission inventory of air pollutants (NH₃) and GHGs (N₂O and CH₄) from China’s agricultural sector in 2021 is presented in

Table 2. Agricultural emission inventory of NH₃, N₂O, and CH₄ in China for 2021 (Gg).

Level I	Level II	NH3	N2O	CH4
LEF				5774.28
	Sheep			1625.75
	Cattle			2632.39
	Pig			456.07
	Draft animals			1060.07
LMM		3758.48	336.83	2576.16
	Cattle	1013.61	64.39	148.15
	Pig	280.99	167.45	2020.45
	Sheep	1104.69	41.99	52.26
	Draft animals	252.76	17.86	62.67
	Poultry	1106.44	45.14	292.63
CF		3798.37	146.24
	Rice	1127.91	24.79
	Vegetable	796.28	23.89
	Fruits	461.21	16.29
	Wheat	264.98	23.49
	Maize	508.25	21.25
	Other crops	639.73	36.53
PC				6607.65
	Single rice			4310.81
	Late rice			1270.59
	Early rice			1026.26
Aq			2.24	21.62
	Freshwater		2.2	21.51
	Mariculture		0.05	0.1
Total		7556.85	485.32	14,979.71

and Figure 2. The total emissions of NH₃, N₂O, and CH₄ were estimated to be 7556.85 Gg (−52–71%), 485.32 Gg (−63–120%), and 14,979.71 Gg (−44–56%), respectively.

Livestock manure management is the dominant source of agricultural emissions, contributing significantly to all three gases (3758.48 Gg for NH₃, 336.83 Gg for N₂O, and 2576.16 Gg for CH₄). Cropland Fertilizer is the primary contributor to NH₃ emissions, contributing 3798.37 Gg. Paddy Cultivation is the main source of CH₄ emissions, contributing 6607.65 Gg. The rapid development of China’s agricultural economy, combined with its large population, has significantly increased the demand for agricultural production, consequently leading to higher agriculture-related emissions [10,53]. Sector-based emission uncertainties and comparisons with previous studies are discussed in the following sections.

Figure 2 illustrates the uncertainty ranges for different agricultural emission sources of NH₃, N₂O, and CH₄. Overall, the NH₃ and N₂O emissions exhibited higher uncertainties compared to CH₄. The largest source of uncertainty was LMM, with total uncertainties of −71–113% for NH₃, −81–212% for N₂O, and –64–96% for CH₄ (95% confidence). Poultry NH₃ (−98–520%), cattle CH₄ (–89–293%), and sheep N₂O (–77–187%) emissions exhibited the greatest uncertainty, driven by substantial variations in emission factors in breeding periods and livestock management practices of regional differences. For CF, the uncertainties were lower, at −56–79% for NH₃ and −44–67% for N₂O. Among crops, fruit trees (−97–463%) and maize (−100–368%) showed higher NH₃ uncertainty than wheat (–48–75%) and rice (−72–84%), which were influenced by factors such as fertilizer type, soil properties, weather, and application methods. The uncertainties in CH₄ emissions from LEF were influenced by the species of livestock: sheep (−80–109%), pigs (−13–14%), cattle (–38–53%), and draft animals (86–137%). CH₄ emissions from PC exhibited lower uncertainties (–43–56%), with single rice, late rice, and early rice showing values of −71–82%, −59–85%, and 62–63%, respectively. Aquaculture emissions in both freshwater and mariculture showed high uncertainties for CH₄ (−99–536%) and N₂O (−96–568%) due to significant regional variations and limited data availability; thus, further investigation is required.

3.2. Comparison with Other Emission Inventories

We preliminarily compared China’s agricultural NH₃, N₂O, and CH₄ emissions with those reported by the Emissions Database for Global Atmospheric Research (EDGARv8.0), Greenhouse Gas and Air Pollution Interactions and Synergies (GAINS-China), Regional Emission Inventory in Asia (REAS v2.1), and other studies (

Table 3. Comparison of agricultural emissions of NH₃, N₂O, and CH₄ in China.

Emission Study	Base Year	Total			Livestock (LMM + LEF)			Crops (CF + PC)
		NH3	N2O	CH4	NH3	N2O	CH4	NH3	N2O	CH4
This study	2021	7.56	0.49	14.97	3.76	0.34	8.35	3.76	0.14	6.60
EDGAR 8.0	2022	7.46	0.74	22.38	2.51	0.21	8.58	4.94	0.53	13.78
GAINS-China	2015	14.32	1.33	17.34	6.56	1.18	11.47	7.76	0.15	5.87
REAS v2.1	2008	12.24	1.40	16.17	2.83	0.37	9.29	9.41	1.03	6.88
Li et al. [54]	2016	10.10			5.42			4.67
Fu et al. [55]	2016	11.99			5.09			6.9
Zhang et al. [56]	2008	10.36			5.31			5.05
Kang et al. [27]	2012	8.07			5.26			2.81
Liang et al. [28]	2020		0.57			0.16			0.41
Feng et al. [21]	2022		0.62			0.29			0.33
Luo et al. [22]	2015		2.08			0.36			1.72
Wang et al. [29]	2019		0.71			0.20			0.51
Zhou et al. [57]	2008		1.33			0.61			0.72
Duan et al. [18]	2020			23.39			13.59			9.79
Gong & Shi. [59]	2015			26.56			15.01			11.45
Huang et al. [58]	2015			19.98			10.21			9.77
Peng et al. [20]	2010			15.79			10.03			5.61

Note: Emission units are uniformly converted to teragrams (Tg).

). Notably, the comprehensive agricultural emission inventories for NH₃, N₂O, and CH₄ in China are available only from EDGAR [35], GAINS-China [36], REAS [37], and our study. There were significant variations in the emission estimation for NH₃, N₂O, and CH₄ among the EDGAR, GAINS-China, REAS, and other databases. Our total agricultural NH₃ emission estimate for 2021 was similar to EDGAR’s estimate of 7.46 Tg. However, it was 30–40% lower than those reported by GAINS-China, REAS v2.1, and other studies, such as those by Li et al. [54], Fu et al. [55], and Zhang et al. [56]. For N₂O, the total agricultural emissions was 0.49 Tg in 2021, which was consistent with those of Liang et al. [28] and Feng et al. [21], but lower than that observed by Zhou et al. [57] (1.03 Tg in 2008). The CH₄ emission in 2021 (14.97 Tg) was lower than those reported by EDGARv8.0 (22.38 Tg), Huang et al. [58] (19.98 Tg), Duan et al. [18] (23.39 Tg), and Gong & Shi [59] (26.56 Tg). These discrepancies were attributed to revisions in the source categories, emission factors, and activity data for the different base years.

Source-based comparisons provide deeper insights into the causes of emission discrepancies. For livestock-related sources, including livestock manure management (LMM) and Livestock enteric fermentation (LEF), our emission estimates were 10–20% higher than those reported by REAS and EDGAR but more than 30% lower than those reported by GAINS-China and most other studies [27,54,55,56]. These discrepancies can be attributed to two primary factors. First, our methods included time-varying temperature-dependent parameters, which differ from the constant parameters used in REAS and EDGAR. Second, although REAS and EDGAR include NH₃ emissions from animal manure as a subset of fertilizer emissions, our study did not follow this categorization.

For crop-related sources, including cropland fertilization (CF) and paddy cultivation (PC), our study revealed lower (10–20% NH₃, 25–30% N₂O, and 5–15% CH₄) emission estimates compared with most other studies. These differences can be primarily attributed to three key factors: First, emission factor discrepancies are substantial, as many previous studies relied on IPCC-derived N₂O emission factors or CH₄ emission factors for PC based on field measurements in Europe, which differ significantly from the region-specific emission factors used in this study; Second, croplands soil N₂O and NH₃ emissions are estimated by using the amount of nitrogen fertilizer consumption as activity data. Without accounting for shifts in fertilizer use across crop species, where ABC has been replaced by urea and compound fertilizer, agricultural emissions in China have been overestimated. Third, in order to keep the sources consistent for both N₂O and NH₃ emissions in cropland, our estimates do not include the return of straw to the field and indirect N₂O emissions.

3.3. Emission Contributions by Category

Figure 3 shows the contributions of various emission sources to NH₃, N₂O, and CH₄ emissions from China’s agricultural sector in 2021. Emissions from livestock-related sources, such as LMM and LEF, are crucial, constituting 49.7, 69.4, and 55.8% of total emissions for NH₃, N₂O, and CH₄, respectively. Within the livestock category, LEF accounted for 38.6% of CH₄ emissions, and the remainder (17.2%) was attributed to LMM. There was a notable variation among different livestock types; cattle and sheep were primary contributors to NH₃ (13.4% and 14.6%, respectively), N₂O (13.3% and 8.7%, respectively), and CH₄ (18.6% and 11.2%, respectively) emissions, with NH₃ and N₂O emissions predominantly controlled by LMM and CH₄ emissions influenced by LEF. Pigs, owing to their high consumption of traditional Chinese diets, significantly contribute to N₂O (34.5%) and CH₄ (16.5%) emissions. Poultry, including geese, chickens, ducks, and rabbits, contributed 14.6% of NH₃, 9.3% of N₂O, and 7.5% of CH₄ emissions. Meanwhile, draft animals contribute minimally, owing to high agricultural mechanization and fewer numbers of livestock in farms. The contributions of Aq to N₂O and CH₄ emissions were negligible at 0.5% and 0.2%, respectively.

Crop-related emissions (CF and PC) accounted for 49.7, 30.1, and 44.1% of NH₃, N₂O, and CH₄ emissions, respectively. CF is a significant source of NH₃ and N₂O, with rice, other crops, and vegetables emitting more gases than maize, wheat, and fruits because of their higher fertilizer requirements. Vegetables contributed 10.5% of NH₃ and 4.9% of N₂O emissions owing to higher temperatures and fertilization. Maize, primarily grown in northern China with higher summer temperatures, contributed 6.7% of NH₃ and 4.4% of N₂O. Wheat, despite its widespread cultivation, had lower emissions (3.5% NH₃ and 4.8% N₂O) due to cooler fertilization seasons. Other crops, including soybeans, tubers, oilseeds, cotton, tobacco, tea, and sugar, contributed a smaller proportion (8.5% NH₃, 7.5% N₂O) but exhibited concentrated emission patterns due to summer fertilization. Rice, a staple food in China, emitted 14.9% of NH₃ and 5.1% of N₂O during the fertilization process. The extensive rice cultivation in China, including single, late, and early rice, contributed 44.1% of total CH₄ emissions due to the anaerobic growth conditions.

3.4. Regional Distributions and Discrepancies in Emission Sources

Figure 4 presents the NH₃, N₂O, and CH₄ emission contributions and distributions by region (%) and province (Gg) in China for 2021. In general, the spatial distribution indicated that SCC and EC had higher emissions, reflecting significant agricultural intensity due to the extensive use of fertilizers and high-density livestock populations. SWC, NC, and NEC showed moderate emissions from agricultural activities but not at the levels observed in SCC and EC. NWC had lower agricultural emissions than the other regions, but the emissions were notable.

In detail, the largest agricultural emissions region was from SCC, constituting 26% of NH₃, 27% of N₂O, and 32% of CH₄ emissions. From a provincial perspective, Henan was the second-largest NH₃-emitting province, at 572 Gg, and the third-largest N₂O-emitting province, at 33 Gg. Additionally, it was the seventh-largest CH₄-emitting province, contributing 766 Gg. EC is another pivotal region in terms of agricultural emissions, constituting 23% of NH₃, 22% of N₂O, and 23% of CH₄ emissions. In particular, Shandong was one of the largest NH₃- and N₂O-emitters provinces. In 2021, NEC contributed 11%, 11%, and 10% of NH₃, N₂O, and CH₄ emissions, respectively. Despite having similar climate features and flat terrain, Heilongjiang contributed more CH₄ emissions than Liaoning and Jilin due to its larger area of PC. Conversely, NC, characterized by higher per capita grain yields, contributed 15% of NH₃, 14% of N₂O, and 9% of CH₄ emissions. Among the different provinces, emissions were lower in Beijing and Tianjin, whereas higher emissions were observed in Hebei. SWC contributed 14%, 18%, and 18% of China’s total agricultural NH₃, N₂O, and CH₄ emissions in 2021. Sichuan accounted for the largest share of agricultural emissions, being the second-largest source of CH₄, the largest source of N₂O, and the fourth-largest source of NH₃ in the country. Therefore, the high emissions and driving forces in Sichuan have a significant impact on agricultural emissions in SWC. NWC exhibited the lowest emissions, accounting for 12% of NH₃, 10% of N₂O, and 8% of CH₄. The region’s dry climate, low temperatures, and limited agricultural crop activity contribute to its lower emission levels. Xinjiang is a significant contributor to regional emissions, largely because of the prevalence of oasis and livestock farming in the province.

Designing effective regional and special priority development strategies requires a nuanced understanding of the structural characteristics of emission sources within different regions. To achieve this objective, the analysis focused on understanding the total amount of individual emissions sources and sub-sources to various gases in the six key regions (see Figure 5). Crops and livestock with the largest agricultural emissions were located in SCC and EC, which are the most dominant agricultural economic regions in China, where demand for pig and poultry products is growing alongside their rising living standard. The emission structures of the two regions were similar in that rice and vegetables were the dominant contributors to CF emission as well as poultry NH₃, pig N₂O, and cattle CH₄ emissions, the main LMM sub-sources. Meanwhile draft animals contributed 4.92% of N₂O emissions in SCC but only 0.79% in EC, and this disparity can be attributed to the higher level of agricultural mechanization in EC, where the use of draft animals has been largely replaced by agricultural machinery [42]. In SWC, LEF contributed the most of CH₄ emissions, followed by PC and LMM. As far as CH₄ emissions in the 6 regions of China are concerned, draft animals constituted the largest proportion in SWC. Consisting of the Sichuan Basin, the Yunnan-Guizhou Plateau, and southern part of the Qinghai-Tibet Plateau, SWC is characterized with rugged terrain that is not conducive to large-scale mechanized production. Consequently, agricultural production there predominantly relies on animal power. In NC, livestock farming in Inner Mongolia and planting in Hebei coexist, which explains the high levels of emissions caused by intense agricultural activities. In general, for NH₃, N₂O, and CH₄, sheep were the leading contributor, followed by cattle and pigs. Furthermore, in NC, maize significantly contributed to the NH₃ and N₂O emissions. The NEC region mainly grows structural grain crops. Rice and maize are the most important grain crops worldwide. Compound fertilizers are applied at a high rate, with low emissions and a low agricultural multiple cropping index. Compared to other regions, NWC had lower emissions of N₂O and CH₄. Sheep and cattle are the main sub-sources of N₂O emissions from LEF and CH₄ emissions from LMM, whereas crop planting contributes only marginally to these emissions. This is attributed to the region’s unique agricultural practices, including extensive grazing and limited oasis farming, which are adapted to the arid and semi-arid conditions.

3.5. Spatial Distribution

The spatial distributions with a high resolution of 3 km × 3 km for the agricultural sector emission sources in China are shown in Figure 6a–c. Agricultural sector emissions varied significantly across China, with a strong east-west gradient. In general, high NH₃, N₂O, and CH₄ emissions were observed in areas east of the Heihe–Tengchong line [60]. This region includes Beijing–Tianjin–Hebei region, the Yangtze River Delta, the Sichuan Basin, the Triangle of Central China, and the Pearl River Delta. High population density and intensive agricultural activities were the primary reasons for the high emissions. In contrast, the area west of the Heihe–Tengchong Line demonstrated remarkably low agricultural emissions. This region is predominantly a continental plateau and mountainous area, which leads to lower emissions because extensive utilization of agriculture is constrained. However, it is worth noting that northwest Xinjiang is generally higher than other regions, which may be linked to its distribution in oasis agricultural regions. Meanwhile, we found significant variability in the spatial distribution of NH₃, N₂O, and CH₄ in China’s agricultural sector, owing to the contributions of various sources across distinct regions. The spatial distributions of NH₃ and N₂O emissions were similar, primarily distributed in the North China Plain and the Sichuan Basin. In contrast, CH₄ is primarily concentrated in the southern regions. The primary factor contributing to this discrepancy was the divergence in cropping structures, with dryland-based cultivation in the northern regions and rice-based cultivation in the southern regions.

4. Conclusions and Recommendations

This study is the first integrated agricultural emissions inventory framework for China developed for the base year 2021. The developed inventory couples the air pollutant of NH₃, and GHGs of N₂O and CH₄, based on a unified crop- and livestock- specific emission source. This homologous agricultural inventory consists of three hierarchical levels classification system for emission sources, which helps identify regional and categorical discrepancies, provide air quality and climate change modeling input, and outlines precise mitigation strategies in China. In 2021, China’s agricultural emissions of NH₃, N₂O, and CH₄ were 7566.17 Gg, 486.14 Gg, and 14,979.71 Gg, respectively. In addition, the uncertainties of NH₃, N₂O, and CH₄ were (–52.1–71.15%), (–63.16–119.95%), and (–43.7–56.02%), respectively. Emissions from livestock-related sources is the largest primary source of emissions. Rice, vegetables, cattle, and pigs were the agricultural subsectors with the highest contribution to emissions. Furthermore, agricultural NH₃ and N₂O emission hotspots were mainly distributed in the North China Plain and the Sichuan Basin, whereas CH₄ was distributed in the southern region. Shandong had the highest NH₃ emissions, Sichuan had the highest N₂O emissions, and Hunan had the highest CH₄ emissions. Meanwhile, regional discrepancies were observed in the agricultural emissions from crop and livestock species across China. SCC and EC were primary contributors due to extensive fertilizer application in rice, vegetables, and other crops. SWC, NC, and NEC exhibited moderate emissions despite diverse agricultural systems. NWC had lower N₂O and CH₄ emissions, where sheep and cattle were the main sub-sources. Further improvements in the accuracy of NH₃, N₂O, and CH₄ emission estimates in agriculture can be achieved by conducting regional emission factor testing for N₂O and CH₄ from livestock sources and collecting detailed activity data across various crops species.