2.1 Overview of MIXv2

The key features of MIXv2 are summarized in Table 1. Anthropogenic sources, including power, industry, residential, transportation, and agriculture, along with open biomass burning and shipping, are included. Dust and aviation are not included in the current version of MIX. Monthly emissions between 2010–2017 are allocated to grids at 0.1 × 0.1°. Open biomass burning emissions are also available with daily resolution upon request. Emissions of 10 species, including CO₂ and air pollutants of NO_x, SO₂, CO, NMVOC, NH₃ (ammonia), PM₁₀ (particulate matter with diameter less than or equal to 10 µm), PM_2.5 (particulate matter with diameter less than or equal to 2.5 µm), BC (black carbon) and OC (organic carbon) are estimated. MIX can support most atmospheric models compatible with gas-phase chemical mechanisms of SAPRC99, SAPRC07 and CB05, and these can be mapped based on these three chemical mechanisms (e.g., GEOS-Chem, MOZART) (Li et al., 2014). MIXv2 stretches from Afghanistan in the west to Japan in the east and from Indonesia in the south to Mongolia in the north. The domain is consistent with the REASv3 gridded emissions product.

Table 1Key features of MIXv2 and the component emission inventories. Note that NA stands for not available.

^ahttps://csl.noaa.gov/groups/csl4/modeldata/data/Li2023/ (last access: October 2023). ^b https://www.nies.go.jp/REAS/index.html (for REASv3.2.1) (last access: April 2022). ^c http://meicmodel.org.cn/​​​​​​​ (last access: June 2022). ^d http://meicmodel.org.cn/?page_id=1772&lang=en (last access: June 2022). ^e Lu et al. (2011). ^f Chatani et al. (2018), Shibata and Morikawa (2021). ^g National Institute of Environmental Research Center. ^h http://meicmodel.org.cn/?page_id=1772&lang=en. ⁱ https://www.globalfiredata.org/ (last access: October 2022). ^j https://edgar.jrc.ec.europa.eu/ (last access: October 2022).

Download Print Version | Download XLSX

Table 2Sector and subsectors included in MIXv2^a.

^a Detailed source profiles assigned to sources within each subsector are summarized in Table S1. ^b Reference report: https://www.ipcc.ch/report/ar6/wg3/downloads/report/IPCC_AR6_WGIII_FOD_AnnexII.pdf (last access: May 2022).

Download Print Version | Download XLSX

Table 2 summarizes the subsectors for each sector in the development of MIX, along with the corresponding source codes used by IPCC.

2.2 Mosaic methodology

We follow a mosaic methodology similar to the development of MIXv1 (M. Li et al., 2017b), as shown in Fig. 1. In brief, nine regional and two global emission inventories were collected and integrated into a uniform format, including REAS version 3 for Asia (referred to as REASv3) (Kurokawa and Ohara, 2020); the Multi-resolution Emission Inventory for China version 2.0 (MEICv2, http://meicmodel.org.cn/, last access: June 2022) (M. Li et al., 2017a; Zheng et al., 2021, 2018); a process-based NH₃ emission inventory developed by Peking University (referred to as PKU-NH₃) (Kang et al., 2016); an official Japan emission inventory (referred to as JPN) (Chatani et al., 2018; Shibata and Morikawa, 2021); an Indian emission inventory for power plants from Argonne National Laboratory (referred to as ANL-India) (Lu and Streets, 2012; Lu et al., 2011); an open biomass burning emission inventory from Peking University (PKU-Biomass) (Yin et al., 2019); the official emissions from Clean Air Policy Support System (CAPSS) for the Republic of Korea (Lee et al., 2011); the fourth version of Global Fire Emissions Database with small fires (GFEDv4s) (van der Werf et al., 2017); and the Emissions Database for Global Atmospheric Research (EDGAR) version 6 for the shipping emissions (Janssens-Maenhout et al., 2019). Figure 1 shows the distribution of the components of emission inventories, which are mosaicked into the MIXv2.

Figure 1Components of regional emission inventories in the MIXv2 mosaic.

Download

We follow a similar hierarchy in MIXv2 as the previous version in the mosaic process. We use REASv3 as the default inventory for anthropogenic sources, and then further replace it with official emission inventories at a finer scale. MEIC, PKU-NH₃ and ANL-India are demonstrated as the best available inventories through inter-comparisons to represent local source distribution with advanced methodology and reliable data sources (M. Li et al., 2017b). Thus, MEIC overrides REAS for anthropogenic emissions over mainland China. PKU-NH₃, which was developed with a process-based model was further applied to replace the NH₃ emissions in MEIC. In Japan, JPN provides emissions for all air pollutants. The ratios of CO₂ to NO_x by sectors derived from REAS are combined with JPN to develop CO₂ gridded emissions. In detail, we firstly calculated the total CO₂ emissions of Japan by multiplying the CO₂ to NO_x ratios and JPN's NO_x emission estimates by sectors, then we developed the spatial proxies based on the NO_x gridded emissions of JPN. Lastly, the calculated CO₂ emissions were allocated to each grid month by month. We use ANL-India for SO₂, NO_x and CO₂ for power plants in India directly, and we complement emissions of other species and sectors with REASv3. Regarding Hong Kong, we use the updated REASv3 emissions.

The open biomass emissions of MIX are developed by combining GFEDv4s and PKU-Biomass inventories. GFED emissions over Asia are processed to the objective domain. We re-gridded the GFED emissions from 0.25 to 0.1° based on an area-weighted algorithm in a mass-balanced way. Wildfires of various vegetation types (including savanna, forest, peatland) and in-field agricultural waste burning are aggregated into the “open biomass burning” sector. PKU-Biomass overrides the emissions of GFED over China (including Hong Kong and Taiwan) on both a monthly and a daily basis.

The consistency of data is ensured in three aspects: source aggregation, spatial distribution and NMVOC speciation. Firstly, a consistent source definition system was applied in source aggregation from regional emission inventories to the final emission mosaic, as outlined in Table 2 and Table S1 in the Supplement. Secondly, the consistency of emissions spatial distribution during emissions mosaic between different inventories are ensured carefully. In India, we integrated the ANL-India emissions for NO_x, SO₂ and CO₂ for pointed power plants and emissions from REAS for other species. To keep the consistency in spatial distribution, we developed spatial proxies based on the CO₂ emissions from ANL-India and re-located REAS emissions for other species. Thirdly, a consistent NMVOC speciation framework was applied throughout all component emission inventories for both anthropogenic and open biomass burning sources, which is described in detail in Sect. 2.4.

2.3 Components of regional emission inventory

2.4 NMVOC speciation

NMVOC speciation has substantial impacts on the model-ready emissions accuracy and performance of chemical transport models (Li et al., 2014). Selection of profiles turns out to be the most important contributor to uncertainties in emissions of individual species. To reduce the uncertainties due to the inaccurate species mapping, Li et al. (2014) developed an explicit assignment approach based on multiple profiles and mechanism-dependent mapping tables.

We processed the gas-phase speciation for NMVOCs following the profile-based mapping procedure, as shown in Fig. 2. The speciation was conducted at a detailed source basis. Firstly, we developed the composite source profile database by combining the US EPA's SPECIATE database v4.5 (https://www.epa.gov/air-emissions-modeling/speciate-2, last access: June 2019​​​​​​​) (Simon et al., 2010) and available local measurements (e.g., Akagi et al., 2011; Mo et al., 2016; Xiao et al., 2018; Yuan et al., 2010). The complete list of source profiles used in this work is provided in Table S1. To diminish the uncertainties due to inappropriate sampling and analyses techniques regarding oxygenated volatile organic compounds (OVOCs), we applied the OVOC correction to those incomplete profiles. The detailed method can be found in previous studies (Li et al., 2014). The following sources in particular have significant OVOC emitted that should be addressed: coal combustion in residential stoves (31 %), residential wood and crop residue fuel use (23 %–33 %), and diesel engines (28 %–47 %). Secondly, we assigned the composite profile database to each component inventory to develop emissions of individual species. Lastly, individual species were lumped to three chemical mechanisms (SAPRC99, SAPRC07, CB05) based on the conversion factors derived from mechanism-dependent mapping tables (Carter, 2015).

Figure 2NMVOC speciation scheme used in MIXv2.

Download

For the Republic of Korea and Hong Kong, we applied the speciation factor for SAPRC99 and CB05 by sectors developed from the SMOKE-Asia model, which have been used for MIXv1 (Woo et al., 2012). Emissions of SAPRC07 were further developed from SAPRC99 based on Table S2, which applied MEIC speciated results.

2.5 Limitations and uncertainties

As a mosaic inventory, MIXv2 has several limitations when integrating various gridded products into a unified dataset. Firstly, inconsistencies could exist at country boundaries where different datasets were used for the adjacent countries (Janssens-Maenhout et al., 2015), for example, the border between China and India. But limited effects are anticipated here given the small area affected due to the high spatial resolution and the low population density at the border. Secondly, extracting emissions by country from gridded maps may introduce uncertainties, especially for power plants located near the coast. This issue is more important because gridded emissions are developed with higher spatial resolutions than earlier versions. Using an extended country map by assigning extra adjacent grids of “ocean” with the neighboring country can reduce this bias. We also acknowledge the general inconsistency of uncertainty levels between countries where different inventories are used following various data sources and approaches. For air quality simulation purposes, the lack of diurnal variations and vertical distribution is another limitation when applying MIXv2 data directly. Development of Asia-specific temporal and vertical profiles is important to improve the model simulation performance in the future.

Table 3Uncertainties in emission estimates for Asian regions (95 % confidence intervals if not noted; unit: %).

^∗ The 97 % confidence interval.

Download Print Version | Download XLSX

It is always difficult to quantify the uncertainties for a mosaic emission inventory such as MIXv2. The uncertainties for each of the component inventories are discussed in detail in corresponding studies (Kang et al., 2016; Kurokawa and Ohara, 2020; Yin et al., 2019; Zheng et al., 2018). Here we summarized the uncertainty estimation by Asian regions in previous studies in Table 3. The uncertainty ranges are quantified based on propagation of uncertainty (Kurokawa and Ohara, 2020; Lei et al., 2011; Zhang et al., 2009) or thousands of Monte Carlo simulations (Lu et al., 2011; Paliwal et al., 2016; Shan et al., 2020; Shi and Yamaguchi, 2014; Sun et al., 2018; Zhao et al., 2011, 2012, 2013; Zhou et al., 2017).

In regard to anthropogenic sectors, the precision of emission estimates for SO₂, NO_x and CO₂ is higher than that of other pollutants, owing to the minimal uncertainties associated with power plants and large industrial facilities. This is particularly notable in the case of MIXv2, where uncertainties are even lower due to the integration of unit-based power plant information for both China and India. While uncertainties for CO and NMVOC are comparable, they are higher than those for SO₂, NO_x and CO₂ because of substantial emission contributions from biofuel combustion. Emissions for particulate matter (especially BC and OC) tend to be more uncertain compared to trace gases, primarily due to the low data availability of accurate activity rates and emission factors related to residential biofuel combustion. The need for more detailed information at the technology or facility level in regions such as India, OSA and SEA is crucial to narrow down the overall uncertainties in Asia in the future. For open biomass burning, previous investigations have estimated low uncertainty ranges for species like CO, NMVOC and OC, while further analyses are urgently needed. In this work, we conducted uncertainty analyses qualitatively by comparing the MIXv2 estimates with other bottom-up inventories and those derived from satellite retrievals in Sect. 4 (Li et al., 2018). In summary, generally consistent emission estimates and trends over Asia are found based on bottom-up and top-down comparisons in Sect. 4. Discrepancies persist, especially in regions like South Asia and Southeast Asia and among species like BC and NMVOC.

3.1 Asian emissions in 2017

In 2017, MIXv2 estimated emissions of Asia as follows: 41.6 Tg NO_x, 33.2 Tg SO₂, 258.2 Tg CO, 61.8 Tg NMVOC, 28.3 Tg NH₃, 24.0 Tg PM₁₀, 16.7 Tg PM_2.5, 2.7 Tg BC, 5.3 Tg OC and 18.0 Pg CO₂ for all anthropogenic sources including power, industry, residential, transportation and agriculture. Emissions are summarized by Asian regions, including China, East Asia other than China (OEA), India, South Asia other than India (OSA) and Southeast Asia (SEA), as shown in Tables 4 and S3. China, India and SEA together account for > 90 % of the total Asian emissions. China dominates the emissions (> 50 %) of CO₂ (11.3 Pg, 63 %), NO_x (22.4 Tg, 54 %), and CO (137.0 Tg, 53 %) and contributes more than 30 % to all other species. The contributions of India are larger than 30 % for SO₂ (13.8 Tg, 42 %), NH₃ (9.8 Tg, 35 %), BC (0.86 Tg, 32 %), OC (1.7 Tg, 33 %), PM₁₀ (7.2 Tg, 30 %) and PM_2.5 (5.0 Tg, 30 %). SEA ranks third for all species, including CO₂ (1.8 Pg, 10 %), NO_x (6.1 Tg, 15 %), SO₂ (5.9 Tg, 18 %), NMVOC (11.9 Tg, 19 %), NH₃ (4.8 Tg, 17 %) and ∼ 15 % of aerosol species. OEA's share varies from 1 % (aerosol species) to 8 % (CO₂). Emission proportions of OSA are around 11 % for NH₃, OC and PM₁₀ and less than 10 % for others.

Table 4Anthropogenic/open biomass^a emissions of MIXv2 for Asian countries and regions in 2017.

^a Anthropogenic includes power, industry, residential, transportation and agriculture. Open biomass represents the “Open Biomass Burning” sector. ^b Tg yr⁻¹for CO₂; Gg yr⁻¹ for other species. ^c Bold values are with the following units: Pg yr⁻¹ for CO₂; Tg yr⁻¹ for other species. ^d Other East Asia represents East Asia other than China. ^e Other South Asia represents South Asia other than India. ^f Asia (year/2017) represents the ratio of emissions (year) to emissions (2017) for all of Asia within the MIXv2 domain. The detailed emissions for 2010–2017 by country and sector are summarized in Table S3.

Download Print Version | Download XLSX

Sectoral contribution varies among species in Asia according to our estimates. Power plants contribute significantly to SO₂ (first contributor, 38 %) and CO₂ (second contributor, 33 %). Industry dominates the emissions of CO₂ (41 %), NMVOC (44 %), PM_2.5 (39 %) and PM₁₀ (47 %). For NO_x, transportation accounts for 33 % of the total emissions, followed by industry (24 %) and inland and international shipping (18 %). The residential sector contributes > 38 % of emissions for PM_2.5, BC, CO and OC. NH₃ is dominated by agriculture (81 %), followed by 13 % from residential.

Figure 3Emission changes by countries/regions from 2010–2017. Shares of open biomass burning for each region are shown as shadowed blocks.

Download

Open biomass burning plays a key role in the SEA and OEA emissions budgets, and is a minor contributor for other regions, as shown in Table 4 and Fig. 3. Including the open biomass burning sector increases the emissions of OC, PM_2.5, PM₁₀, NMVOC, CO and CO₂, by 69 %, 53 %, 46 %, 37 %, 34 % and 15 %, respectively for SEA. Due to the active fire events, Southeast Asia is the largest emission contributor for OC, PM₁₀, PM_2.5 and NMVOC in 2014 and 2015. Additionally, OEA has a significant emission increment for NMVOC (30 %), OC (143 %), PM₁₀ (52 %) and PM_2.5 (81 %) when taking biomass burning into account. Given the large contributions to ozone and climate change from NMVOC, CO and aerosols, it is important to address the open biomass burning contributions in designing mitigation strategies for these areas.

In a global context, Asia shares out 43 %–56 % of the global anthropogenic emissions in 2017, including 44 % for NO_x, 43 % for SO₂, 49 % for CO₂, 51 % for NH₃, 55 % for CO, 50 % for NMVOC and over 50 % for all PM species. Emissions over Asia are derived from MIXv2, and emissions over the rest of the world are estimated by EDGARv6. Figure S1 depicts the emissions trend by Asian regions, the US and OECD-Europe from anthropogenic sources. Asia is playing a more and more important role in global climate change as its CO₂ emission fraction has increased by 7 % during 2010–2017. Especially with the general emission reductions in the US and OECD-Europe, India and Southeast Asia are catching up with the emissions of these two developed regions for NO_x, SO₂ and CO₂ and already surpassed their emissions for other species. As of now, the US and OECD-Europe emissions are in general comparable to those of OSA for most air pollutants.

3.2 Emissions evolution from 2010–2017

For anthropogenic sources, driven by stringent air pollution control measures implemented over China and OEA since 2010, Asian emissions have declined rapidly by 24.3 % for SO₂, 16.6 % for CO, 17.2 % for PM₁₀, 18.7 % for PM_2.5, 14.2 % for BC and 19.8 % for OC according to our estimates. On the contrary, CO₂, NMVOC and NH₃ still show emissions increasing continuously, with growth rates of 16.5 %, 12.6 % and 4.4 % during 2010–2017, respectively. The emissions for 2010–2017 are summarized in Table S3 and shown in Figs. 3 and 4. As demonstrated in Fig. 4, power, industry, residential and transportation contribute to the rapid emission changes. In contrast to the smoothly changing patterns for anthropogenic, open biomass burning emissions vary from year to year, peaking in 2015 as a result of El Niño (Field et al., 2016). Open-fire activities dominate the SEA emission changes for CO, OC, NMVOC, PM₁₀ and primary PM_2.5 (see Fig. 3). Marked reductions are estimated for China, with a concurrent increase over India and Southeast Asia for all species except CO₂, NMVOC and NH₃ (see Fig. 5). Consequently, air pollutants, including ozone and secondary aerosol precursors of NO_x, have shifted southward (Zhang et al., 2016). This changing spatial pattern has been confirmed from observations as described in previous studies (Samset et al., 2019). We illustrate the driving forces of emissions evolution for each species below. Shipping is not included in the following analyses.

Figure 4Emission changes by sectors in Asia from 2010–2017.

Download

Figure 5Emission changes for anthropogenic sources (power, industry, residential and transportation) by regions and sectors from 2010–2017 for (a) NO_x, (b) SO₂, (c) NMVOC and (d) PM_2.5. The pie sizes are scaled with the total anthropogenic emissions in Asia. The unit for the total emission values in the center is Tg per year. OEA denotes East Asia other than China, OSA represents South Asia other than India and SEA is Southeast Asia.

Download

NO_x. NO_x emissions show a increasing–decreasing–increasing trend for 2010–2017, with a peak in 2012. This trend is a combination of significant power plant emissions reduction (−22 % from 2010–2017), and emissions increase from industry (+4 %) and transportation (+6 %). As estimated, China's emissions for all anthropogenic sectors dropped by 4.6 Tg (−17 %) from 2010–2017, along with 2.6 Tg (+38 %) emissions growth from India and 0.9 Tg (+19 %) from SEA. As a result, China's contribution decreased from 63 % to 54 %, and the Indian share grew from 16 % to 22 % (anthropogenic, Fig. 5a). In China, power plant emissions dropped by 4.5 Tg (−51 %) because very stringent emission standards are implemented for power plants since 2003, which has continuous substantial impacts on SO₂, NO_x and particulate matters (Chinese National Standards GB 13223-2003 and 3223-2011) (Zheng et al., 2018). Furthermore, “ultra-low” emission standards were set up by the Chinese government in 2015 to further reduce emissions from coal-fired power plants by 60 % by 2020. OEA shows 21 % emissions decrease because of continuous control measures over industry (−16 %) and transportation (−32 %). Due to insufficient control strategies in India, OSA and SEA, NO_x anthropogenic emissions have grown by 38 %, 18 % and 19 %, respectively, mainly driven by power plants and vehicle growth (see Fig. 5a). Open biomass burning has limited effect (∼ 11 %) on NO_x emissions over SEA and is neglectable for other regions (< 3 %).

SO₂. SO₂ emissions rapidly declined from 43.8 to 33.1 Tg during 2010–2017, peaking in 2012. Significant reductions in industrial SO₂ emissions (−8.7 Tg, −39 %) lead to the marked total emissions decrease (see Fig. 4). China's emissions dropped by 62 %, partly offset by the concurrent emissions increase of India (+46 %) and Southeast Asia (+41 %). Stringent control measures shutting down small industrial boilers and cleaning larger ones in China are the primary driving forces (Zheng et al., 2018). New emissions standards were set up for coal-fired industrial boilers with tightened SO₂ limit values (Zheng et al., 2018). In addition, nationwide phasing out of outdated industrial capacity and small, polluting units has been carried out in China since 2013. Consequently, China's emission fraction decreased from 64 % to 32 %, ranking second in 2017. India's proportion grew from 22 % to 41 %, and nowadays it is the largest SO₂ emitter in Asia (anthropogenic; see Fig. 5b). Relatively small emission changes are estimated for OEA (−0.2 Tg) and OSA (+0.6 Tg). Significant emissions growth from power plants drives the total anthropogenic increase by 44 %, and a 41 % rise when considering additional open biomass burning for SEA.

CO. Anthropogenic CO shows moderate emissions reduction (−16 %) since 2010, driven by the clean-air actions implemented in China covering industry (−38 % changes), residential (−20 %) and transportation (−22 %) sources. Based on the index decomposition analysis, the improvements in combustion efficiency and oxygen blast furnace gas recycling in industrial boilers are the largest contributors to the emission reductions in China (Zheng et al., 2018). Replacing polluted fuel (biofuel, coal) with cleaner fuels (natural gas, electricity) is the primary driving force in the residential sector. Despite the rapid vehicle growth, which would typically yield a CO increase, pollution control measures reduced the net CO emission factors by fleet turnover with cleaner models replacing the older, more polluted vehicles in the market. OEA shows emission reductions in industry (−37 %), residential (−22 %) and transportation (−28 %) sources. Residential fuel combustion decreases by 20 % and 11 % for SEA and India, respectively, having a canceling effect on the total emissions growth for these two regions. Open biomass burning accounts for 25 %–77 % of CO emissions in SEA, which drives the total emissions reduction by 19 % in 2017 compared to 2010. Additionally, the climate anomaly due to El Niño in 2015 leads to the rapid CO emissions drop from 2015 to 2016 in SEA.

CO₂. Driven by economic and population growth, anthropogenic CO₂ emissions show a rapid increasing trend for China (15 %), India (32 %), OSA (32 %) and SEA (21 %). We found slight emission decreases for OEA (−2 %). Power, industry and transportation grew by 28 %, 12 % and 35 %, respectively, driving the total emissions increase continuously. In contrast, we estimate a 5 % CO₂ emission reduction from the residential sector, attributed to reduced fuel combustion. Notable emissions have increased for sectors apart from residential for India (increasing rates varying between 39 %–57 %), OSA (43 %–56 %) and SEA (9 %–48 %). Fractions by region are stable during the studied period, with 62 % contribution from China, 8 % from OEA, 16 % from India, 3 % from OSA and 11 % from SEA (see Fig. 3). Open biomass burning curbs the total emission growth in SEA from +21 % to +5 %.

NMVOC. Differing from the decreasing emission trend for NO_x, SO₂ and CO, NMVOC increases by 13 % for anthropogenic sources and 6 % for all sources with open biomass burning. In China, the industrial sector (+5.1 Tg, +35 %) is the major reason for the emissions growth, and industrial solvent use (e.g., architecture paint use, wood paint use) is the largest contributor. The share of solvent use rapidly rises from 28 % in 2010 to 42 % in 2017. In addition, oil production, distribution and refineries, and chemical production lead to a corresponding emissions increase by 44 % (M. Li et al., 2019). Due to fuel transfer in residential stoves and the effective pollution control measures for on-road vehicles, China shows 18 % and 22 % emissions decreases, respectively, slowing down the increasing trend. Industry and transportation drive the anthropogenic emissions in India and SEA growing by 18 % and 13 %, respectively (see Fig. 5c). In SEA, 64 % of the total emissions are contributed by open biomass burning in 2015. Compared to 2010, 2017 total emissions decreased by 12 % in SEA, attributed to biomass burning. NMVOCs are speciated into three chemical mechanisms following the source-profile-based methodology (see Sect. 2.4). We analyzed the speciation results in Sect. 3.5.

Figure 6Monthly variations of emissions in Asia by sectors for 2017. For agriculture, only NH₃ emissions are estimated.

Download

Figure 7Monthly variations in emissions in Asian regions in 2017 for (a) CO (solid lines) and SO₂ (short, dashed lines) and (b) NMVOC (solid lines) and PM_2.5 (short, dashed lines). For each month, the dominant sector is labeled by circle color. Circle area is scaled based on the emission fraction of the dominant sector for each month. Both anthropogenic and open biomass burning are included.

Download

Figure 8Spatial distribution of emissions in MIXv2 in 2017 for gaseous species.

NH₃. As estimated by MIXv2, NH₃ emissions are generally flat, with slight increases (+4 %) over Asia due to the lack of targeted control measures. Over 80 % of the total emissions are contributed by fertilizer application and livestock manure. Transportation emissions in 2017 are 2.8 times greater than those in 2010 due to vehicle growth, which can play a key role for urban air quality. India is the largest contributor (35 %), followed by China (32 %) and SEA (17 %) (see Fig. 3). India and OSA emissions show monotonic 7 % and 20 % increases, respectively. According to our estimates, China decreases by 8 %, reflecting the agricultural activity rate changes. Limited effects are estimated from open fires over NH₃ emissions budget, peaking at 19 % of the total in 2015 for SEA.

Particulate Matter (PM). PM emissions are estimated to have decreased in Asia: −4.9 Tg PM₁₀ (−17 %), −3.8 Tg PM_2.5 (−19 %), −0.45 Tg BC (−14 %) and −1.3 Tg OC (−20 %) for anthropogenic sources and −6.5 Tg PM₁₀ (−20 %), −5.0 Tg​​​​​​​ PM_2.5 (−21 %), −0.51 Tg BC (−15 %) and −1.9 Tg OC (−23 %) after including open biomass burning. Industrial and residential sectors are the primary driving forces of the emissions reduction. In China, the strengthened particulates standard for all emission-intensive industrial activities, including iron- and steel-making, cement, brick, coke, glass, chemicals, and coal boilers have driven the technology renewal and the phasing out of outdated, highly polluting small facilities (Zheng et al., 2018). Pollution control measures reduced power plant emissions by ∼ 30 % for PM₁₀, PM_2.5 and BC and counterbalanced the transportation emissions despite vehicle ownership increasing by 270 % in 7 years. Other East Asia shows significant anthropogenic emissions reduction for all PM species: −32 % for PM₁₀, −33 % for PM_2.5, −27 % for BC and −31 % for OC. Flat trends are estimated in India for all PM species (±8 %). Reduction in biofuel use led to the residential emission reduction of PM in India and SEA. Increasing industrial activities led to 24 %–35 % anthropogenic emissions growth for PM₁₀ and PM_2.5 in OSA. As a result, China's emission fractions are shrinking, with growing contributions from India and OSA (see Fig. 5d). Taking PM_2.5 as an example, the emissions shares among Asian regions have changed significantly between 2010 and 2017: from 58 % to 46 % for China, 23 % to 30 % for India, and 12 % to 14 % for SEA (anthropogenic, Fig. 5d). Open biomass burning dominates the SEA emissions changes between 2010–2017: −19 % for PM₁₀, −24 % for PM_2.5, −20 % for BC and −30 % for OC.

Figure 9Spatial distribution of emissions in MIXv2 in 2017 for PM species.

3.3 Seasonality

Monthly variations of emissions, which are highly sector dependent, are estimated in MIXv2. Within the same sector, similar monthly emission variations are found among different species as they are mainly driven by activity rates (see Fig. 6) (M. Li et al., 2017b). Figures 6 and 7 illustrate the emission fractions by sectors, and the dominant sector (classified by circle color) for each month for CO, SO₂, NMVOC and PM_2.5 by regions in 2017, including both anthropogenic and open biomass burning. Contribution of the “dominant sector” is scaled to the circle area. Large circles represent the significant role of the dominant sector and small ones (near 17 %) indicate the balanced contribution from six sectors. The monthly emissions patterns show large disparities varying with regions. Notable variations are estimated for emissions of China, OEA and SEA. The residential sector of China is the largest contributor in winter for CO and PM_2.5, leading to the “valley” curves. Industrial emissions show relatively high fractions in the second half of the year aiming to achieve the annual production goal, which dominate the seasonal patterns of SO₂, CO, NMVOC and PM_2.5 for most of the months. The emissions peaks in summer for OEA and March for SEA are attributed to significant in-field biomass burning activities. Indian and OSA emissions show relatively small monthly variations, compared to other Asian regions. This pattern is attributed to the predominant role of the residential sector in the monthly emissions for the investigated species (as illustrated in Fig. 7). The minimal seasonal variations in surface temperature within the tropical climate of India and OSA contribute to the overall stability in monthly residential emission patterns. Thus, it is important to take both anthropogenic and open biomass sectors into account in seasonality analyses given their dominant roles varying by months, such as model evaluations based on ground, satellite, and aircraft measurements.

3.4 Spatial distribution

Gridded emissions at 0.1 × 0.1° were developed in our inventory. Power plant emissions in China and India are developed on a unit basis and assigned with exact geophysical locations. For other sources, emissions are allocated to grids based on spatial proxies, such as road map, population and gross domestic product (GDP). The gridded MODIS fire product with high spatial resolution (up to 1 km) is the essential dataset for open biomass burning emission estimation. Figures 8 and 9 depict the spatial distribution of both CO₂ and air pollutants over Asia in 2017, showing the distinct patterns of point source, roads and city clusters. Emission intensities of hot spots over the Indo-Gangetic Plain, spanning northern Pakistan, northern India and Bangladesh are comparable to those of northern China and Indonesia, especially for NH₃, NMVOC, BC and OC. Clear shipping routes can be seen for NO_x, SO₂, CO₂ and PM species.

Emission reductions in East Asia highlight the importance of air pollution control in Southern Asia. We show the emission changes by latitude bands from 2010 to 2017 for ozone precursors (NO_x, NMVOC, CO) and primary PM_2.5 in Figs. 10 and S2 (featuring open biomass burning). Largest reductions are estimated between 35–40° N for NO_x (−25 %), CO (−32 %) and PM_2.5 (−35 %) because of China's effective emission control strategies. On the other hand, NO_x anthropogenic emissions have increased by 15 % over 10–0° S (Southeast Asia) and +27 % over 10–20° N (driven by India). Open biomass burning enlarged the emission amplitude for 15–0° S (Southeast Asia), while it had limited effect on the trends over other latitude bands (see Fig. S2). To conclude, NO_x has shifted southward in Asia. NMVOC emissions show a generally increasing trend over all latitude bands (+5 %–+38 %, anthropogenic). Differently, CO and primary PM_2.5 show general emissions reduction since 2010 except over 15–10° S and 10–20° N. These latitudinal shifts are of particular importance for the global tropospheric ozone budget as ozone precursors emitted at low latitudes are more efficient at producing ozone than if the same quantity of emissions is released at high latitudes (Zhang et al., 2016, 2021).

Figure 10Emissions in 2010 and 2017 (a) and anthropogenic emission changes from 2010 to 2017 (b) by latitude bands for NO_x, NMVOC, CO and PM_2.5. The changing patterns of total emissions by latitude (anthropogenic + open biomass burning) are shown in Fig. S2.

Download

3.5 Speciated NMVOC emissions

As one of the key precursors of ozone and secondary organic aerosols (SOAs), NMVOCs gain more and more attention because of the emission increase due to relatively loose targeted control measures. As estimated in MIXv2, Asian emissions have increased by 13 % for anthropogenic sources and 6 % with additional open biomass burning. We speciated the total NMVOC to three chemical mechanisms, SARPC99, SAPRC07 and CB05, following the profile-based approach (Sect. 2.5). Emission changes during 2010–2017 by chemical groups are shown in Fig. 11. Alkanes, alkenes and aromatics comprise 78 % of the total emissions on a mole basis in 2017. Driven by the growing activities in industry, emissions of alkanes and aromatics increased by ∼ 20 % within 7 years, according to our estimates. Alkenes and alkynes show a stable trend, reflecting the combined results of emission reduction in residential and growth in the industry and transportation sectors. For OVOCs, especially aldehydes, emissions decreased by 10 % since 2010 due to reduced residential fuel combustion. Open biomass burning plays a role over other OVOC (OVOCs other than aldehydes and ketones) emission changes. India, Southeast Asia and China are the largest contributors to the total Asian budget, with varying sector distributions and driving forces by chemical groups.

Figure 11Emissions in 2010 and 2017 (right columns) and the emission changing rates from 2010–2017 (left columns) of NMVOCs by chemical groups. For each country and region, the left column represents the emission changing rates (in %), while the right column shows the emissions by sector in 2010 and 2017. Chemical groups are lumped from the SAPRC07 species following Table S4. Open bb denotes open biomass burning.

Download

Industry, mainly industrial solvent use, is the primary driving sector for emissions increase of alkanes (+15 %) and aromatics (+21 %) in China. Moderate reductions are estimated for anthropogenic OVOCs (−33 % aldehydes, +20 % ketones, −22 % other OVOCs) attributed to fuel transfer in the residential sector. OEA emissions show generally decreasing trends from −13 % (ketones, alkenes) to +3 % (alkynes) for anthropogenic sectors and −10 % (ketones, aromatics, others) to +25 % (Other OVOCs) with additional open biomass burning. Industrial emissions have decreased over all chemical groups for OEA. Similar sectoral distributions across chemical species are found for India and OSA, dominated by the residential and transportation sectors. More than 29 % emissions growth is estimated for alkanes and aromatics, driven by industry, residential and transportation sectors in India and OSA. In SEA, ∼ 20 % increases are estimated for emissions of alkanes and aromatics, and minor changes are estimated for alkenes, alkynes, aldehydes and ketones (within 10 %) during 2010–2017. OVOCs emissions in 2017 are 25 % lower than the values in 2010, contributed by residential sources and open biomass burning.

Figure 12Emission comparisons for anthropogenic sources between MIXv2, REASv3.2, EDGARv6 and CEDS_GBD-MAPs for (a) NO_x, (b) NMVOC, (c) CO₂ and (d) BC during 2010–2017 for Asian regions.

Download