Major components of China’s anthropogenic primary particulate emissions

This paper presents the first comprehensive estimates of particulate emissions in China by size distribution and major components. Using a technology-based emission inventory approach, we are able to classify particulate emissions into three size ranges, TSP, PM10 and PM2.5, and identify the contributions of black carbon (BC), organic carbon (OC), Ca and Mg. Total particulate emissions are estimated to be 27.4 Tg for the year 2001, of which 17.8 Tg are PM10 and 12.7 Tg are PM2.5. Industrial processes are the major sources of particles over all three size ranges, but residential biofuel use and transportation sources become increasingly important for PM10 and PM2.5. The industrialized coastal provinces, such as Shandong, Jiangsu and Hebei, are the major sources of particulate emissions. The industrialized and developing regions show different characteristic emission ratios of PM2.5/TSP, (BC+OC)/PM2.5 and (Ca+Mg)/TSP. In the future, we can expect significant reductions in primary particulate emissions and major changes in the patterns of size and species.


Introduction
Particulate matter (PM) impacts local, regional and global environments in many ways. Fine PM with diameters less than or equal to 2.5 μm (PM 2.5 ) has negative effects on human health (Pope et al 1995). Particles are able to travel across national boundaries and impact regional air quality and hemispheric backgrounds (Heald et al 2006). Carbonaceous particles have also been linked with global and regional climate change (Andreae 2001, Jacobson 2001, Menon et al 2002, Hansen and Nazarenko 2004. In China, PM 10 is the dominant air pollutant in cities, and PM 10 concentrations often 4 Author to whom any correspondence should be addressed. exceed China's national air quality standard in major urban areas (He et al 2002).
Although there are several inventories that address China's carbonaceous aerosol emissions as national inventories or as part of regional/global inventories (Streets et al 2001, 2003a, Bond et al 2004, Cao et al 2006, a detailed approach that considers the main components of aerosols and their size distribution with a consistent methodology is missing. Recently, we revisited China's inventory and updated CO and NO x emissions with an improved methodology (Streets et al 2006, Zhang et al 2007. The purpose of this paper is to present China's emissions of total suspended particulates (TSP), PM 10 , PM 2.5 , BC, OC, Ca and Mg in a consistent way, based on our improved methodology.

Methodology
We have implemented an improved, technology-based methodology to estimate China's anthropogenic primary aerosol emissions.
Emissions are calculated from the combination of activity rate, technology distributions, emission factors in raw gas and the penetration of emission control equipment. The general approach used in this work is similar to that of Klimont et al (2002); that is, emissions are estimated for three diameter ranges: large particles (>PM 10 ), coarse fraction (PM 2.5−10 ) and fine fraction (PM 2.5 ). Thereby, PM 10 is calculated as the sum of fine and coarse fractions and TSP as the sum of fine, coarse and >PM 10 fractions. The size-fractioned PM emissions are calculated by the following equation: k,m E i, j,k,m,y = j,k,m A i, j,k,m ef i, j,k,m,y . (1) For a given j , k and m, the final emission factor of diameter range y is estimated by the following equation: where i represents province; y represents diameter range of PM; j represents the sector; k represents fuel or product type; m represents the combustion or production technology; n represents PM control technology or equipment; E i,y means the emission amount of PM in diameter y in province i ; A means activity level, such as fuel consumption, production, etc; ef means emission factor; EF TSP means raw emission factor of TSP; F y means mass proportion of PM in diameter y out of total PM; C n means the percentage of control technology n; η n,y means the removal efficiency of control technology n for PM in diameter y.
In addition to total aerosol emissions, we also estimate the emissions of several key chemical components in aerosols: BC, OC, Ca and Mg. Emission factors of BC and OC are determined by the mass ratio of BC and OC to PM 2.5 emission factors, which is similar to the approach described in Bond et al (2004) and Kupiainen and Klimont (2007); similarly, emission factors of Ca and Mg are determined by their fraction in TSP emissions, following Zhu et al (2004).
We classify emission sources into four groups: stationary combustion, industrial process, mobile sources and biomass open burning. The stationary combustion sources involve three sectors (power plants, industry and residential) and eight types of fuel (coal, diesel, kerosene, fuel oil, liquid petroleum gas, coal gas, natural gas and biomass). The industrial process sources include 11 products in metallurgical industries, mineral products industries and chemical industries. For each industrial process, end-of-pipe and fugitive PM emission are estimated separately. The mobile emission sources include seven types of on-road mobile sources: light-duty gasoline vehicles, light-duty gasoline trucks, mid-duty gasoline trucks, light-duty diesel vehicles, heavy-duty gasoline vehicles, heavyduty diesel vehicles and motorcycles; and six types of off-road mobile sources: rural vehicles, tractors, building equipment, farm equipment, locomotives and vessels. Biomass burning emissions include forest fires, grassland fires and the open burning of crop residues.
According to equations (1) and (2), the necessary data for emissions calculations include: (a) fuel consumption and industrial production; (b) distribution of emission control technologies; (c) PM emission factor for each type of emission source; (d) share of PM emissions in each diameter range out of total PM emissions; (e) mass ratio of emissions of specific components (BC, OC, Ca and Mg) in total PM emissions; and (f) removal efficiency of each PM control technology in each diameter range. The activity data (a) and (b) and emission factors ((c)-(f)) were developed from a wide variety of sources.
The general approach used in this work to develop detailed activity data has been described in Streets et al (2006) and Zhang et al (2007), in connection with the development of improved CO and NO x emission inventories for China. The activity data at sector level rely mainly on statistics published by different government agencies of China. Fuel consumption by sector and by province is derived from the China Energy Statistical Yearbook 2001 (NBS 2004). Industrial production by sector and by province and the vehicle population data are derived from a variety of government statistics (ACT 2002, AISIC 2002, NBS 2002a, 2002b. Technology distributions within each sector are obtained from a wide variety of Chinese technology reports and other publications. Vehicle population data and fuel consumption by vehicle type are taken from He et al (2005), using a transportation fuel consumption model. Table 1 summarizes the main activity data used in this study. Readers are referred to Streets et al (2006) and Zhang et al (2007) for further details of sources of activity data. We have collected appropriate emission factors based on available measurements in China or used estimations based on the actual technology level and practice, e.g. Zhang et al (2000), Chen et al (2005Chen et al ( , 2006 and Yi et al (2006). SEPA (1996) was widely consulted for TSP emission factors. In some cases, where local information is lacking, we use emission factors for similar activities from the US AP-42 database (USEPA 1996) and the RAINS-PM model (Klimont et al 2002). Table 2 summarizes China's anthropogenic primary PM emissions in 2001 by size and by sector. We estimate that anthropogenic TSP emissions in China in 2001 were 27.4 Tg, and PM 10 emissions were 17.8 Tg, accounting for 65% of TSP emissions. PM 2.5 emissions, which are the most harmful part of the aerosols, comprise a large proportion of primary aerosol emissions. National PM 2.5 emissions were 12.7 Tg in 2001, accounting for 45% of TSP emissions and 70% of PM 10 emissions. Industrial processes are the largest contributor to TSP emissions, with a contribution of 58% to total TSP emissions, followed by residential biofuel and power plants, which contributed 12% and 9% of total TSP emissions, respectively. For PM 2.5 , industrial processes are still the largest contributor, but their contribution drops to 46%, while the contribution of residential biofuel emissions to national total PM 2.5 emissions increases to 24%. The sector differences in  (2005) a Coal consumption in cement kilns, brick kilns and lime kilns was subtracted from total industry coal combustion to avoid double counting. which have low PM removal efficiencies, especially for fine particles. Residential combustion is another important source of PM emissions, especially for fine particles. In 2001, PM 2.5 emissions from residential combustion were 3.85 Tg, with 80% of them from biofuel combustion. Residential combustion is also identified as a major contributor of carbonaceous aerosol emissions in China (Streets et al 2001, Streets andAunan 2005), which will be discussed later.

Size distribution
It is surprising that PM emissions from industrial processes are higher than the sum of emissions from all other stationary combustion sources. In 2001, industrial processes emitted 15.9 Tg TSP and 5.87 Tg PM 2.5 . Production of cement, brick, lime, coke, and iron and steel are identified as the main contributors of PM emissions. China's cement industry produced 660 million tons of cement in 2001 and released 7.5 Tg TSP and 4.35 Tg PM 2.5 , representing the largest individual source group. TSP and PM 2.5 emissions from China's brick production were 3.63 Tg and 0.37 Tg, respectively. This estimate is thought to be quite uncertain, because specific PM emission factors for brick production are not available, and we assumed that brick industrial PM emissions are similar to those of stokers in the industrial sector. We estimate that TSP emissions from China's coke production and iron and steel production in 2001 were 0.98 Tg and 0.84 Tg, respectively, and their PM 2.5 emissions were 0.50 Tg and 0.21 Tg, respectively. This estimation is also quite uncertain, partly due to the lack of local emission factors and partly due to the fact that a significant part of the PM emission is fugitive, which is very difficult to measure. We also observe that most PM emissions in coke production come from fugitive emissions in indigenous coking processes, which emitted 0.36 Tg. China's recent regulation on closing indigenous coking plants could be very positive for PM emission reduction.
Most particles emitted from vehicles are fine particles. PM 2.5 emissions in the transportation sector were 0.28 Tg in China in 2001, about 76% of PM 2.5 emissions of the transportation sector in the USA in 2002 (USEPA 2007), but only accounting for 2% of China's total anthropogenic PM 2.5 emissions. We also calculated aerosol emissions from open biomass burning by using the methodology described in Streets et al (2003b) and the emission factors from Andreae and Merlet (2001). We estimate that TSP emissions from open biomass burning in China were 2.24 Tg in 2001, where 1.02 Tg were PM 2.5 .

Carbonaceous aerosols
In previous studies, we estimated that China's BC emissions were 1.34 Tg in 1995 and 1.05 Tg in 2000, respectively (Streets et al 2001(Streets et al , 2003a. Our previous estimate of OC emissions for 2000 was 3.38 Tg (Streets et al 2003a). The decreasing trend in BC emissions is due to the decrease in solid fuel consumption in the residential sector. Bond et al (2004) estimated that China's BC and OC emissions in 1996 were 1.49 Tg and 2.82 Tg, respectively, as a part of their global inventory. They obtained higher estimates of BC emissions by identifying coke production and brick production as two important BC sources for China, while their OC estimates were lower than previous studies due to the lower emission factors used. Cao et al (2006) presented 1.50 Tg BC for China for 2000 and attributed the increase from previous studies to coal combustion in rural industry and rural residences. However, they did not report which industries were responsible for the increase.
We estimate that BC and OC emissions from anthropogenic sources in China in 2001 were 1.71 Tg and 3.58 Tg, respectively. The residential sector is the largest contributor to carbonaceous aerosols, emitting 0.87 Tg BC and 2.25 Tg OC in 2001, which is quite comparable with previous studies. We restate our previous estimates (Streets et al 2003b) for open biomass burning: 0.11 Tg BC and 0.75 Tg. By using the same emission factors as in Bond et al (2004), we estimate that industrial processes contributed 0.46 Tg BC and 0.47 Tg OC in 2001, mostly from coke production and brick production. However, we are not very confident about these numbers, because the emission factors are determined by very few measurements and local measurements are not available.
Estimates of carbonaceous aerosol emissions are thought to be of high uncertainty, especially due to the lack of local measurements in China. In recent studies, Chen et al (2005Chen et al ( , 2006 measured small coal combustion devices in China and then calculated carbonaceous aerosol emissions for residential coal combustion. Their BC estimates were lower than other previous studies by a factor of two or three, indicating that there are still huge uncertainties in carbonaceous aerosol inventories and additional comprehensive investigations are needed.

Ca and Mg emissions
Alkaline Ca and Mg are important components of particulate emissions, because of their ability to neutralize acidic species in the atmosphere such as sulfuric and nitric acids. Ca and Mg emissions come mainly from large combustion devices and industrial processes. China's anthropogenic Ca and Mg emissions from anthropogenic sources in 2001 were 4.75 Tg combined. About 94% of Ca and 76% of Mg emissions were emitted from industrial processes, e.g. cement production, lime production, and iron and steel production. The remainder was mainly from combustion, especially from large combustion devices such as power plants and large industrial boilers. Ca and Mg emissions from residential stoves are very small.
Cement production and lime production are the largest contributors of Ca and Mg emissions. Cement production emitted 3.0 Tg Ca and 0.07 Tg Mg in China in 2001, accounting for 66% of total Ca emissions and 30% of total Mg emissions. Ca and Mg emissions from lime production were 1.0 Tg and 0.02 Tg, respectively. The huge Ca and Mg emissions from cement production can be attributed to the following reasons. First, the proportion of Ca in TSP from the process is very high. The unabated TSP emission factor for cement production is about 100-150 g kg −1 cement, where about 40% is Ca. Second, under the lenient PM emission standard of the cement industry in 2001, many old cement plants were only equipped with wet scrubbers and cyclones, which had low removal efficiencies and resulted in high final PM emissions. However, it should be noted that China implemented a new strict emission standard for the cement industry in 2005, which required that TSP emissions from all cement plants could not exceed 100 mg per Nm 3 flow gas, corresponding to 1.5 g kg −1 cement (SEPA 2004). If all factories are able to meet this requirement, the final TSP emission factor of China's cement industry will be reduced by 70%. This will reduce China's TSP emissions markedly in the future and also reduce Ca and Mg emissions that affect the chemical composition of the atmosphere. This could have important implications for acid/base reactions in the atmosphere, e.g. acid rain formation.    use carbonaceous aerosols and base cation aerosols as tracers for residential emissions and industrial emissions, respectively. In remote and developing regions, residential emissions are the dominant fraction of PM emissions, with the effect that PM 2.5 /TSP and (BC + OC)/PM 2.5 ratios in those provinces are higher, but the (Ca + Mg)/TSP ratio is low. In the industrialized provinces, the industrial sector contributes a large share of PM emissions, where the (Ca + Mg)/TSP ratio is high but the PM 2.5 /TSP and (BC + OC)/PM 2.5 ratios are lower. These regional differences in PM emissions in China are mainly caused by the differences in economic development, industry structure and population. The magnitudes and components of primary particulate emissions are expected to change dramatically with rapid economic development.

Regional distribution of emissions
In recent years, China has implemented several new emission standards for power plants, cement plants and industrial boilers.
The strengthened emission standards will result in a major reduction of PM emissions in key emitting sectors. Industrialized regions, especially megacities such as Beijing and Shanghai, are taking more active measures to prevent the release of health-damaging pollutants. Energy consumption and industrial production are still growing; however, the increased penetration of effective PM control equipment will remove most of the coarse fraction of PM emissions, leading to an increasing trend in the PM 2.5 /TSP ratio. The (Ca + Mg)/TSP ratio is expected to decrease in industrialized regions, due to the abatement of emissions from industrial processes. Developing regions will tend to move toward the characteristics of industrialized regions.