Temporal and spatial variation in recent vehicular emission inventories in China based on dynamic emission factors

Abstract

The vehicular emission trend in China was tracked for the recent period 2006–2009 based on a database of dynamic emission factors of CO, nonmethane volatile organic compounds (NMVOC), NO_x, PM₁₀, CO₂, CH₄, and N₂O for all categories of on-road motor vehicles in China, which was developed at the provincial level using the COPERT 4 model, to account for the effects of rapid advances in engine technologies, implementation of improved emission standards, emission deterioration due to mileage, and fuel quality improvement. Results show that growth rates of CO and NMVOC emissions slowed down, but NO_x and PM₁₀ emissions continued rising rapidly for the period 2006–2009. Moreover, CO₂, CH₄, and N₂O emissions in 2009 almost doubled compared to those in 2005. Characteristics of recent spatial distribution of emissions and emission contributions by vehicle category revealed that priority of vehicular emission control should be put on the eastern and southeastern coastal provinces and northern regions, and passenger cars and motorcycles require stricter control for the reduction of CO and NMVOC emissions, while effective reduction of NO_x and PM₁₀ emissions can be achieved by better control of heavy-duty vehicles, buses and coaches, and passenger cars. Explicit provincial-level Monte Carlo uncertainty analysis, which quantified for the first time the Chinese vehicular emission uncertainties associated with both COPERT-derived and domestically measured emission factors by vehicle technology, showed that CO, NMVOC, and NO_x emissions for the period 2006–2009 were calculated with the least uncertainty, followed by PM₁₀ and CO₂, despite relatively larger uncertainties in N₂O and CH₄ emissions. The quantified low uncertainties of emissions revealed a necessity of applying vehicle technology- and vehicle age-specific dynamic emission factors for vehicular emission estimation, and these improved methodologies are applicable for routine update and forecast of China's on-road motor vehicle emissions.

Implications:

This paper tracks the temporal and spatial variation characteristics in recent vehicular emission inventories in China based on dynamic emission factors. The fact that CO and NMVOC emissions kept growing at reduced rates and the NO_x, PM₁₀, and GHG emissions continued rising rapidly reveals that it was insufficient to bring down the rapid growth of NO_x, PM₁₀, and CO₂ emissions by merely tightening emission standards and improving fuel quality of motor vehicles. The results will assist decision makers to formulate effective control policies for China's vehicular emissions. The improved methodologies are applicable for routine update of China's vehicular emission inventories.

Introduction

In the recent years of 2005–2009, the population of on-road motor vehicles in China has been soaring ceaselessly, increasing by about 46.5% at an average growth rate of about 10.0%, and reached about 170 million in 2009. Meanwhile, ambient air quality in about 37.6%, 39.5%, 42.5%, and 33.3% of the Chinese cities failed to meet the Grade-II limits of the Chinese National Ambient Air Quality Standard in 2006, 2007, 2008, and 2009, respectively, and in many cities, particularly the big and medium ones, urban air pollution has been characterized of coexistence of coal combustion-induced and motor vehicle exhaust-induced pollution (MEP, 2007–2010). Recent source apportionment studies have revealed that motor vehicles were the major contributor to the ambient nonmethane volatile organic compounds (NMVOC) concentrations in Beijing (Song et al., 2007a), were one of the major sources of atmospheric polycyclic aromatic hydrocarbons in the coastal city of Dalian, China, contributing about 11% and 79%, respectively, during the heating and non-heating periods (Zhang et al., 2007), and made up about 8.5% and 7.8% of ambient PM_2.5 in Beijing in winter and summer, respectively, in 2004 (Song et al., 2007b). Moreover, motor vehicles have been recognized as the major source of urban air pollution in many Chinese cities in 2010 by the Ministry of Environmental Protection of China, and the frequently occurring regional haze and ozone pollution in China (Chan and Yao, 2008; Ran et al., 2009; Wang et al., 2009; Ma et al., 2010) is partly or mainly caused by the fine particle emissions and the NO_x and volatile organic compounds (VOC) emissions from motor vehicles (Duan et al., 2008; Geng et al., 2008; Liu et al., 2008; Wang et al., 2008a; Shao et al., 2009a; Shao et al., 2009b; Cheng et al., 2010).

The motor vehicle source has experienced fast and significant advances and changes in the recent period 2005–2009, particularly in the engine technologies, exhaust treatment technologies, emission standards, the population growth and fleet compositions, and transportation fuel quality. All these are factors that affect the actual emission characteristics of motor vehicles in recent years and require scrutinized updates of the effects of these factors on emission factors. With these rapidly changing influencing factors, emission inventories of on-road motor vehicles in China require continuous and long-term update and refinement. To tackle the challenges for rapid responses and updates of emission factors and improvement of the accuracy of domestic estimation of vehicular emission inventories in China, efforts have been made and resulted in some recent reports of local emission factors of Euro II, Euro III, Euro IV, and CNG buses in Beijing by mobile platform measurements (Wang et al., 2011a), and of Euro III and Euro IV diesel buses using a portable emission measurement system (PEMS) and electric low-pressure impactor (Liu et al., 2011). In addition, emission factors of PM₁₀, CO, hydrocarbons (HC), and NO_x for Euro 0 through Euro III diesel trucks in Xi'an and Beijing (Liu et al., 2009) and the CO, HC and NO_x emission factors for Euro 0 through Euro 4 light duty gasoline vehicles in three Chinese megacities were measured using PEMS (Huo et al., 2012a). These recent studies obtained fresh emission factors in certain cities of China for gasoline light-duty vehicles and diesel buses or trucks, with an emphasis on PM₁₀ emission factor measurement. However, a complete emission factor database including both gaseous and particulate pollutants based on dynamic update of the recent major influencing factors for all major vehicle categories at the provincial level has yet to be established. Despite the completion of emission estimation of the gaseous, particulate, and speciated NMVOC emissions from on-road motor vehicles in China from 1980 to 2005 (Cai and Xie, 2007, 2009), of the gaseous and particulate emissions from rural vehicles in China in 2006 (Yao et al., 2011), and of the vehicular emission inventories in a few cities like Beijing and Hangzhou for 1995 (Hao et al., 2000), 2004 (Huo et al., 2009), and 2004–2005 (Guo et al., 2007), and in 22 selected cities in 2007 (Huo et al., 2011), emission inventories of major gaseous and particulate pollutants from on-road motor vehicles in China have not been reported for the recent period 2006–2009, mostly due to the lack of a complete database of dynamic emission factors that can reflect the rapid influencing factors just described. Therefore, COPERT 4, the latest vehicular emission factor model, which replaces COPERT III and updates in the methodology, particularly for Euro 4 and later vehicles (Gkatzoflias, 2007), was adopted to develop a database of dynamic emission factors of major air pollutants from all categories of on-road motor vehicles in China at the provincial level. Based on the developed dynamic emission factor database, the recent emission trend of on-road motor vehicles in China was tracked, providing decision-making proof for vehicular emission control in China and the latest vehicular emission inventory for air quality management and traffic-related air pollution mitigation practices in China.

Methodology

Emission estimation

Emissions of 31 provinces/autonomous regions/municipalities on the Chinese mainland, with Hong Kong Special Administrative Region, Macau Special Administrative Region, and Taiwan province excluded, were estimated using eq 1, and were further aggregated to represent the national inventories for the period 2006–2009.

(1)

where is the emission of pollutant plt in province pro in year y in tons; is the population of vehicle category i at age j in province pro in year y in thousands; is the vehicle miles traveled (VMT) under driving cycle (DC) k by vehicle category i at age j in province pro in year y in thousands of kilometers; is the emission factor of pollutant plt for vehicle category i under DC k in province pro in year y in grams per kilometer; and is the deterioration rate of for vehicles at age j in grams per kilometer.

Vehicle population, fleet compositions, and age distribution

Vehicle population and fleet compositions

The Chinese provincial population of nine vehicle categories, which are small-size passenger cars (SPC, with a maximum passenger number of nine), mini passenger cars (MiniPC, with engine displacement smaller than 1.0 L), light-duty trucks (LDT, with gross vehicle weight between 1.8 tons and 6.0 tons), mini trucks (MiniDT, with gross vehicle weight below 1.8 tons), heavy-duty trucks (HDT, with gross vehicle weight above 14.0 tons), medium-duty trucks (MDT, with gross vehicle weight between 6.0 tons and 14.0 tons), large-size passenger cars (LPC, with a maximum passenger number above twenty), medium-size passenger cars (MPC, with a maximum passenger number between nine and twenty), and motorcycles (MC) classified according to Chinese statistical protocol, was obtained from official statistical yearbooks (CATARC, 2007–2010; NBS, 2007–2010a) for the period 2006–2009. These nine categories were grouped into five types to facilitate discussion of results: passenger cars (PC) including SPC and MiniPC, light-duty vehicles (LDV) including LDT and MiniDT, heavy-duty vehicles (HDV) including HDT and MDT, buses and coaches (BC) including LPC and MPC, and MC including both two-stroke and four-stroke MCs. Figure 1, which shows the annual vehicle population by type and per thousand inhabitants for the period 2000–2009, reveals that the total vehicle population in China has been growing very rapidly, with the vehicle fleet in 2009 increased by 3.24 and 1.62 times that in 2000 and 2005, respectively. Besides, Figure 1 shows that MC and PC were the major vehicle categories, constituting about 64.0% and 26.6%, respectively, of the total population in 2009. Meanwhile, with its population in 2009 increasing by 8.22 and 2.62 times compared to those in 2000 and 2005, respectively, the proportion of SPC, which dominated PC and were mainly privately owned, has been rising dramatically from 9.7% in 2000 to 24.6% in 2009. Consequently, vehicle population per thousand inhabitants in China has tripled during the past 10 years and reached 129.2 in 2009. Still, China has a great potential to increase the motor vehicle fleet due to its flourishing economy and still low number of vehicles per thousand inhabitants, compared to those in the United States, Western Europe, and Japan.

Figure 1. Vehicle population by type and per thousand inhabitants in China for the period 2000–2009.

By applying the established conversion relationships between the Chinese and European vehicle categories (Cai and Xie, 2007), vehicle fleet compositions for the period 2006–2009 were shifted to the European categories defined in COPERT III. Since COPERT 4 has an increased number of vehicle categories especially for trucks and buses, the COPERT III HDV classifications were further converted to the COPERT 4 ones, based on the conversion sheet developed by the model developers (EMISIA, 2011), and on the fact that the Chinese HDV were dominated by the class with a gross vehicle weight of 13–16 tons (HJGY, 2011). Therefore, it was assumed all Chinese HDV belonged to the 13–16 tons class, and were not converted to any higher tonnage HDV classes in COPERT 4.

Vehicle age distribution

Vehicle age distributions (VADs) have significant effects on the accuracy of estimated vehicular emissions (Zachariadis et al., 2001). VADs for LPC, MPC, SPC, MiniPC, HDT, MDT, LDT, and MiniDT at the provincial level were identified according to the available historical records of new vehicle registration by category since 1999 (NBS, 2000–2010), with an assumption that there had been a zero scrappage rate for vehicles less than 20 yr old for the period of 1980–1999, since vehicles in the 1980s and 1990s remained very expensive commodities in China and were seldom scrapped until reaching a very long service.

Dynamic emission factors

COPERT 4, the latest European motor vehicle emission factor model, was adopted to determine the emission factors by vehicle category based on the developed parameterization of COPERT (Cai and Xie, 2007). Such emission factors were further updated dynamically for the period 2006–2009, to quantify the effects of rapid variation and advances in engine technologies, tightening exhaust emission standards, improving fuel quality, emission factor deterioration due to accumulative mileage (AM) traveled, and inspection/maintenance (I/M) level based on additional parameterization in COPERT. Consequently, a dynamic emission factor database including emission factors of CO, NMVOC, NO_x, PM₁₀, SO₂, CO₂, CH₄, and N₂O for major vehicle categories at the provincial level was developed to model the temporal and spatial variation characteristics of recent vehicular emission inventories in China.

Advances in engine technologies

With the technology maturity of the electronic injection (EI) and direct injection (DI) engines, the indirect injection engines and the carburetor engines have been phased out due to their failure to meet more stringent exhaust emission standards. Considering the fact that the exhaust emission standards for motor vehicles in China have been improving, it is assumed that all Euro I or later gasoline vehicles in China were equipped with DI engines, while none of the Euro 0 vehicles were DI engine equipped.

Improvement and implementation of exhaust emission standards

Emission standards of different stages were introduced and implemented for gasoline and diesel vehicles and MC with different timetable among Beijing, Shanghai, Guangzhou, and other provinces (MEP, 2005, 2007, 2008), as shown by Table 1.

Table 1. Implementation timetable for emission standards at different stages for each province of China

Vehicle category	Province	Euro 0	Euro I	Euro II	Euro III	Euro IV
Gasoline vehicles	Beijing	Prior to 1999	1999–2001	2002–2004	2005–2007	2008–2009
	Shanghai and Guangzhou	Prior to 1999	1999–2003	2004–2006	2007–2009	2010–
	Other provinces	Prior to 2001	2001–2004	2005–2008	2009–
Diesel vehicles	Beijing	Prior to 1999	1999–2001	2002–2005	2006–2008	2009–
	Shanghai and Guangzhou	Prior to 1999	1999–2003	2004–2007	2008–2010
	Other provinces	Prior to 2001	2001–2004	2005–2009	2010–
MC	Nationwide	Prior to 2002	2002–2004	2005–2008	2009–

Although the Euro III emission standard should come into effect in provinces other than Beijing, Shanghai, and Guangzhou in 2008 (MEP, 2005), these provinces could not be supplied with gasoline and diesel that meet the fuel quality requirement for Euro III vehicles until 2009. Therefore, the Euro III emission standard in these provinces is assumed to be implemented in 2009.

In China, “the new standard for new vehicles, and the old for the old ones” was the regulation and practice when new emission standards were implemented. Therefore, vehicle population by category, model year, and emission standard at the provincial level for the period 2006–2009 can be identified based on the recognized VAD, Table 1, and eq 2:

(2)

where , , , , , and are the total population of vehicle category i in province pro and the number of vehicle category i conforming to Euro 0, Euro I, Euro II, Euro III, and Euro IV emission standard, respectively, in province pro.

Accumulative mileage based deterioration coefficient of emission factors

Vehicles of the same category but of different ages generally have diverse AM and thus differ in the pollutant emission levels due to differences in the I/M level, the abrasion of auto parts, and catalyst aging, showing a deterioration tendency of higher emisison factors with higher AM. Therefore, a more accurate estimation of vehicular emissions requires consideration of the deterioration rates of emission factors of vehicles due to AM (Corvalan and Vargas, 2003). In this study, the deterioration rates of CO, VOC, and NO_x emission factors were calculated by linking to the AM, the I/M level, and DC, as shown by eq 3:

(3)

where is the deterioration corrected emission factor for vehicle category i at age j under DC k with I/M level m in province pro in year y in grams per kilometer; is the deterioration rate per unit mileage traveled for vehicle category i under DC k with I/M level m in grams per kilometer; and is the accumulative mileage for vehicle category i at age j under DC k with I/M level m in province pro in year y in kilometers.

varies with vehicle categories, DCs and the I/M levels. Table 2 shows the deterioration rates of CO, VOC, and NO_x for Euro I, Euro II, Euro III, and Euro IV gasoline PC and LDV based on the COPERT methodology (Ntziachristos et al., 2009).

Table 2. Deterioration rates (g/km) of CO, VOC and NO_x for Euro I, Euro II, Euro III, and Euro IV gasoline PC and LDV based on the COPERT methodology

		for gasoline PC and LDV
		Euro I and Euro II vehicles		Euro III and Euro IV vehicles, and Euro I and Euro II vehicles with enhanced I/M scheme
Pollutant	Enginedisplacement	Urban DC (<19 km/hr)	Rural DC (>63 km/hr)	Urban DC (<19 km/hr)	Rural DC (>63 km/hr)
CO	≤1.4L	1.523	1.689	7.129	1.502
	1.4L-2.0L	1.148	9.607	2.670	2.234 ^b
	>2.0L	9.243	2.704	2.670	0.781 ^b
NOx	≤1.4L	1.598	1.220	3.986 ^a	3.043 ^b
	1.4L-2.0L	1.598	1.220	3.986	3.043 ^b
	>2.0L	1.598	1.220	3.986	3.043 ^b
VOC	≤1.4L	1.215	6.570	3.419	1.849 ^b
	1.4L-2.0L	1.232	9.815	3.467 ^c	2.762 ^b
	>2.0L	1.208	6.224	3.399 ^c	1.751 ^b
Notes: ^aAssumed the same as that for PC and LDV with engine displacement of 1.4 L–2.0 L.
^bAssumed proportional to the deterioration rate ratio of rural DC over urban DC of Euro I and Euro II vehicles.
^cAssumed proportional to the deterioration rate ratio of urban DC of Euro I and Euro II vehicles over urban DC of Euro III and Euro IV vehicles.

When the driving speed falls between 19 km/hr to 63 km/hr, the is calculated by linear interpolation following eq 4, and for driving speeds below 19 km/hr or above 63 km/hr, the is independent of speed and equals and , respectively:

(4)

where and are the at the driving speeds of 19 km/hr and 63 km/hr, respectively, and is the driving speed.

The deterioration rates of CO, VOC, and NO_x emission factors under the urban (20 km/hr), rural (40 km/hr), and freeway (80 km/hr) DCs were calculated separately based on Table 2 and eq 4.

Moreover, considering the recently widely installed on-board diagnostic (OBD) systems on new vehicles, the for vehicles with enhanced I/M scheme were adopted for the study period 2006–2009. However, deterioration of diesel vehicle emission factors due to aging or AM was neglected since diesel engines usually have less deterioration level than gasoline engines (Hao et al., 2000b; Chiang et al., 2008; Hu et al., 2011).

Improvement of fuel quality

The rapid introduction and implementation of more stringent emission standards during the period 2006–2009 meant more stringent requirements for fuel quality, particularly the sulfur contents of gasoline and diesel. Accordingly, the gasoline and diesel standards for motor vehicles in China have been updated and new regulations have been announced (MEP, 2005, 2007, 2008), focusing on the decrease of the sulfur contents. Table 3 shows the sulfur contents of gasoline and diesel for Euro 0, Euro I, Euro II, Euro III, and Euro IV motor vehicles in Beijing, Shanghai, and other provinces of China.

Table 3. Sulfur contents (ppm) of gasoline and diesel for Euro 0, Euro I, Euro II, Euro III, and Euro IV motor vehicles in Beijing, Shanghai and other provinces of China

		Euro 0	Euro I	Euro II	Euro III	Euro IV
Gasoline	Beijing and Shanghai	800	500	500	150	50
	Other provinces	800	500	500	150	150
Diesel	Nationwide	2000	2000	2000	350	50

Based on Table 1 and Table 3, the sulfur content and SO₂ emission factors for each vehicle category at the provincial level for 2006–2009 were dynamically updated.

Moreover, fuel properties like the sulfur content, density, distillate temperature, aromatics content, polycyclic aromatics content, oxygenate content, cetane number, and so on affect CO, NMVOC, NO_x, and PM₁₀ emission factors. The default fuel effect correction factor algorithms linking emission factors of gasoline and diesel PC, LDV, HDV and BC to fuel properties in COPERT 4 (EEA, 2011) were adopted to quantify the effects of fuel quality improvement, particularly the decrease of sulfur contents on CO, NMVOC, NO_x, and PM₁₀ emission factors.

Mileages

Provincial VMT of LPC, MPC, HDT, MDT, LDT, and MiniDT

Due to lack of official reporting on annual VMT by vehicle category, the provincial VMT values of LPC, MPC, HDT, MDT, LDT, and MiniDT for the period 2006–2009 were indirectly calculated based on the provincial passenger person-kilometers (PPK) and the freight ton-kilometers (FTK) using eq 5 (Cai, 2011), since provincial PPK and FTK in China are fulfilled by the commercial LPC and MPC, and by commercial HDT, MDT, LDT, and MiniDT, respectively:

(5)

where is the VMT of vehicle category i in province pro in year y; is the PPK or FTK in province pro in year y; is the proportion of PPK or FTK fulfilled by vehicle category i in province pro in year y; is the actual boarding or loading rate for vehicle category i in province pro in year y; is the commercial population of vehicle category i in province pro in year y; and is the seat number or load capacity of vehicle category i in province pro in year y.

It is noticed that the total population of LPC and MPC and that of HDT, MDT, LDT, and MiniDT did not equal the commercial passenger vehicles or freight vehicles (NBS, 2000–2010). Therefore, is quantified by assuming that the gap between the total population of LPC and MPC and that of the commercial passenger cars was filled by MPC, and the gap between the total population of HDT, MDT, LDT, and MiniDT and that of the commercial trucks was filled by MiniDT or MiniDT and LDT, which have the lightest gross vehicle weights of all categories of trucks. In addition, was obtained from national and provincial statistical yearbooks, and was determined according to the regulation Standard for Classification of Chinese Motor Vehicles (GB9417-89), as summarized in Table 4.

Table 4. The number of seats for LPC and MPC and load capacity (tons) of HDT, MDT, LDT, and MiniDT in China

Number of seats		Load capacity
LPC	MPC	HDT	MDT	LDT	MiniDT
41	31	15	10	3.9	1

Due to lack of statistical records of , this parameter was determined based on the assumption that was positively proportional to the product of the commercial vehicle population and the seat number of LPC and MPC, or the load capacity of HDT, MDT, LDT, and MiniDT, as expressed by eq 6 (Cai, 2011):

(6)

The provincial VMT values of commercial LPC, MPC, HDT, MDT, LDT, and MiniDT for the period 2006–2009 were calculated by the PPK and FTK statistics using eq 5, while the VMT values of the noncommercial MPC were quantified by using eq 7, and the VMT of the noncommercial MiniDT or MiniDT and LDT is based on the most recent domestic surveys, which was 30,000 kilometers per year (Huo et al., 2012b).

Provincial VMT of passenger cars and MC

A decrease in VMT of passenger cars in China has been observed over the recent years based on multiyear and multicity surveys (Huo et al., 2012b). Based on the recent time-series VMT data for passenger cars from 2002 to 2009 (Huo et al., 2012b) and the passenger car ownership per 1000 people, it is found that the VMT of passenger cars in China can be well predicted by the level of passenger car ownership per 1000 people, as proven by the very high Pearson correlation coefficient of 0.963 shown in Figure 2. This developed VMT prediction function, as shown in eq 7, is used to calculate the province specific VMT for passenger cars, by taking into account the provincial differences in passenger car ownership levels:

Figure 2. Correlation between VMT of passenger cars and passenger car ownership per 1000 people in China based on historical data for 2002–2009.

(7)

where is the VMT of passenger cars, and is the passenger car ownership per 1000 people.

For the VMT of MC for the period 2006–2009, we took into account the fact that MC in China are still mainly and widely used in the rural areas, and the MC population in cities has been decreasing and transferring to the rural areas due to more and more stringent restriction for MC in cities. Therefore, it is assumed that the provincial VMT of MC decreased with the increase of urbanization rate in each province. Based on the 2009 VMT of 5600 km revealed by the recent investigation conducted by Huo et al. (2012b), the provincial VMT values of MC for the period 2006–2009 were calculated using eq 8:

(8\textrm{a})

(8\textrm{b})

where , , and are the VMT of MC in province pro in year y, y+1 and 2009, respectively; , , and are the urbanization rates of province pro in year y, y+1 and 2009, respectively, obtained from EIU (2011); 5600 was the national average VMT for MC in 2009 in kilometers; and 46.6% was the national average urbanization rate in 2009.

VMT of each vehicle category under urban, rural, and freeway DCs

The determined provincial VMT of each vehicle category were further allocated to the urban, rural, and freeway DCs based on the developed methodology (Cai and Xie, 2007). The required data of the urban vehicle population proportion of the total vehicle population and the freeway length proportion of the total transportation length in each province were obtained from China City Statistical Yearbooks (NBS, 2007–2010b) and China Statistical Yearbooks (NBS, 2007–2010a), respectively.

Monte Carlo uncertainty analysis

The Monte Carlo method, which was recommended by IPCC as the methodology to quantify the uncertainties of emission inventories of various sources, was adopted to quantify the uncertainties of the recent emission trend of CO, NMVOC, NO_x, PM₁₀, CH₄, N₂O, and CO₂, following IPCC's detailed procedures (IPCC, 2001): First, the probability density functions (PDFs) of CO, NMVOC, NO_x, PM₁₀, CH₄, N₂O and CO₂ emission factors, as well as the VMT of each vehicle category, were determined; second, stochastic simulations were performed for the emission factors and VMT based on the determined PDFs, and the Monte Carlo estimation of annual emissions of CO, NMVOC, NO_x, PM₁₀, CH₄, N₂O, and CO₂ was obtained using eq 9, which takes into account explicitly the potential uncertainty sources; finally, stochastic simulations were stopped when the variation of the mean value of the simulations was convergent within 1% at a 95% confidential interval (CI):

(9)

where is the Monte Carlo estimated emission for pollutant plt in year y; is the vehicle population of vehicle category i conforming to emission standard es under DC k in province pro in year y; is the dynamic emission factor for ; and is the VMT of vehicle category i under driving cycle k in province pro in year y.

The uncertainties associated with emission factors for each vehicle category by emission standard and by driving cycle were quantified by taking into account both the intrinsic uncertainty of COPERT model-derived emission factors and the uncertainty of the representativeness of model emission factors for Chinese real-world emission factors. The former uncertainty was quantified by referring to the extensively developed lognormal PDF database of CO, NMVOC, NO_x, PM₁₀, CH₄, N₂O, and CO₂ emission factors for each vehicle category by emission standard and by driving cycle, which aimed at uncertainty estimation of COPERT emission factors (Kouridis et al., 2010a), while the latter uncertainty was quantified by developing lognormal PDF of pollutants based on recently available domestic real-world measurements of motor vehicle emission factors, as shown in Table 5. The lognormal model was selected because the uncertainty associated with emission factors is asymmetric, that is, there are no negative emission factors, but ultra-emitters may emit several times above the average.

Table 5. Emission factors (g/km) of CO, NMVOC, and NO_x by vehicle type in China based on domestic measurements and literature reports

	Emission standard	CO	NMVOC	NOx
PC	Euro 0	37.448^a, 29.67^b, 43.9^c, 17.07^d, 33.4^e, 23.9^f, RDC, 32.71^i	5.026^a, 3.68^b, 4.7^c, 2.39^d, 3.6^e, 2.0^f, RDC, 1.71^i	2.18^a, 5.05^b, 1.56^c, 2.25^d, 1.87^e, 1.63^f, RDC, 1.68^i, 1.82^i
	Euro 1	7.28^g, 3.01^g, 14.52^RDC, 11.3^e, 6.6^f, RDC	0.21^g, 0.12^g, 1.02^RDC, 0.70^e, 0.54^f, RDC	0.1^g, 0.38^g, 0.15^RDC, 1.02, 0.90^f, RDC
	Euro 2	2.39^g, 4.1^e, 3.04^f, RDC, 5.4^i, 4.41^i, 2.64^i,	0.58^g, 0.31^e, 0.27^f, RDC, 0.04^i, 0.16^i, 0.13^i	0.24^g, 0.47^e, 0.59^f, RDC, 1.54^i, 0.43^i, 0.16^i
	Euro 3	0.95^g, 2.2^e, 0.63^i	0.21^g, 0.09^e, 0.0044^i	0.26^g, 0.23^e, 0.058^i
	Euro 4, gasoline	0.40^e, 0.927^h, RDC	0.02^e	0.05^e, 0.148^h, RDC
	Euro 4, diesel	0.497^h, RDC		0.572^h, RDC
LDV	Euro 0	21.592^a, 39.31^b, 75.5^c, 31.94^d	3.612^a, 5.22^b, 6.2^c, 3.404^d	2.438^a, 5.09^b, 2.08^c, 2.31^d
HDV and BC	Euro 0	69.581^a, 11.44^b, 29.3^c, 70^d, 36.3^j	3.893^a, 3.92^b, 9.0^c, 3.10^d, 2.56^j	12.180^a, 25.5^b, 11.87^c, 10.11^d, 6.57^j
MC	Euro 0	9.126^a, 8.7^b, 28.8^c, 21.14^d	4.248^a, 1.84^b, 5.0^c, 3.49^d	0.261^a, 0.14^b, 0.13^c, 0.094^d
Notes: Sources: a, Wang et al., 2002; b, Liu, 2000; c, He et al., 1998; d, this study; e, Huo et al., 2011a; f, Yao et al., 2007; g, Wang et al., 2008b; h, Hu et al., 2011; i, Yao et al., 2006; j, Huang et al., 2006; RDC represents rural DC, and the numbers in other cases represent urban DC.

Due to lack of quantitative uncertainty estimation of LDV emission factors in COPERT, the qualitative indicators (Kouridis et al., 2010b) were used and a coefficient of variation (CV) of 0.2, 0.41, and 0.43 (Wisner et al., 2010) was assigned to LDV emission factors graded A, B, and C, respectively, to develop the lognormal PDFs.

Assuming that the uncertainty of COPERT derived emission factors and that of the domestic measurements were comparable, eqs 10 – 13 are used to combine the PDFs of both model-derived emission factors and domestic measurements, for the purpose of quantifying both uncertainties of the dynamic emission factors used for emission estimation in this study. Since domestic measurements mainly focused on CO, NMVOC, and NO_x, the uncertainty quantifications for PM₁₀, CH₄, N₂O, and CO₂ were based on the uncertainty associated with COPERT emission factors for these pollutants:

(10)

(11)

(12)

(13)

where ,and are the PDFs for , COPERT-derived and domestically measured , respectively; and , , and are the emission factors representative of best estimated real-world emission factors in China and the COPERT-derived and domestically measured ones, respectively.

The PDFs of VMT for various vehicle categories for the period 2006–2009 were determined using the lognormal probability distribution based on the calculated VMT and the literature reports for LDV, HDV and BC, and for PC and MC, for years 2005 and 2007, respectively, as shown in Table 6.

Table 6. Literature reported VMT of PC and MC in China for 2005–2009

	PC	LDV	HDV	BC	MC
2005	4.00^a, 4.83^b	3.00^a, 1.75^b	6.00^a, 3.26^b	6.00^a, 3.62^b	1.00^a
2006	2.09^c
2007	2.07^c; 3.49^a	4.01^c	4.01^c	4.99^c	1.5^a
2008	2.03^c
2009	1.94^c				0.56^c
Note: Sources: a, Huang and Tang, 2008; b, Zhang, 2005; c, Yu, 2007.

Given that the provincial vehicle population by category for the period 2006–2009 was from the best available data and was further broken down by VAD and emission standard, the precision of was believed credible, and thus the possible uncertainties of were neglected in this study. Consequently, a total of 14,144 lognormal PDFs were developed for VMT by vehicle category and for emission factors of the said seven pollutants by vehicle category, emission standard, and driving cycle at the provincial level for each year of the period 2006–2009.

Results and Discussion

Dynamic emission factor database

By quantifying the major influencing factors for emission factors, dynamic emission factor database of CO, NMVOC, NO_x, PM₁₀, SO₂, CO₂, CH₄, and N₂O for the period 2006–2009 was developed at the provincial level for each vehicle type under urban, rural, and freeway DCs. The dynamic emission factors of CO, NMVOC, and NO_x for gasoline PC in Beijing were illustrated as an example, as shown by Figure 3.

Figure 3. Dynamic emission factors of CO, NMVOC, and NO_x for gasoline PC in Beijing, indicating effects of emission standards, AM, engine technologies, and I/M levels.

Figure 3 shows that emission standards had a significant effect on the reduction of emission factors of CO, NMVOC and NO_x under urban, rural and freeway DCs. For CO, the Euro I, Euro II, Euro III and Euro IV vehicles had a reduction of 80.7%, 90.0%, 92.8%, and 97.6%, respectively, in comparison to the Euro 0 vehicles, under the urban DC. Meanwhile, the deterioration of emission factors due to age-based accumulated mileages was significant for NO_x, with a maximum (when AM reaches 120 or 160 km for Euro I and Euro II, and for Euro III and Euro IV vehicles, respectively) NO_x deterioration increased by a factor of about 5.3, 7.9, 6.7, and 7.9 for Euro I, Euro II, Euro III, and Euro IV vehicles, respectively, compared to the brand new vehicles. Moreover, the maximum VOC deterioration was also remarkable, increasing by a factor of about 2.7 and 4.2 for Euro I and Euro II vehicles, while the maximum CO deterioration was relatively lower, increasing by only a factor of about 1.21, 1.40, 1.21, and 1.71 for Euro I, Euro II, Euro III, and Euro IV vehicles, respectively, compared to the brand new vehicles. Dynamic emission factors of both gasoline PC and other vehicle types in other provinces resembled the characteristics, despite slight differences in absolute values. Therefore, it is clear that ignorance of the effects of emission standards and AM on emission factors would induce inaccurate estimation of the real-world emission factors for the recent period, and the dynamic emission factor database developed at the provincial level in this work would guarantee credible estimation of the recent vehicular emission trend in China.

Recent trend

Vehicular emissions of CO, NMVOC, NO_x, PM₁₀, CO₂, CH₄, and N₂O for the period 2005–2009 are summarized in Table 7, which shows that emissions continued increasing for the period 2006–2009, and emissions of these pollutants in 2009 had increased by about 31.1%, 35.8%, 57.6%, 64.6%, 105.7%, 88.2%, and 157.1%, respectively, in comparison with those in 2005.

Table 7. Emissions (thousand tons) of CO, NMVOC, NO_x, PM₁₀, CO₂, CH₄, N₂O, and carbon dioxide equivalent (CO₂e, calculated based on a 100-yr global warming potential of 25 and 298 for CH₄ and N₂O, respectively) from on-road motor vehicles in China for the period 2005–2009

Year	CO	NMVOC	NOx	PM10	CO2	CH4	N2O	CO2e
2005	16,243	3572	3563	261	226,521	228	21	226,522
2006	18,856	4117	3892	309	276,245	284	29	276,246
2007	20,150	4605	4422	348	327,834	332	37	327,836
2008	20,850	4648	4842	380	371,122	366	42	371,125
2009	21,293	4850	5616	430	465,914	429	54	465,916

The growth rate of CO emissions began to slow down in recent years, decreasing to 2.1% in 2009 compared to 13.2% in 2005, as shown by Figure 4a. The slowing down of CO emission growth was mainly due to two reasons: First, the rapid improvement of more stringent emission standards resulted in much lower emission factors for CO; taking gasoline SPC, for example, the emission factors for Euro III and Euro IV vehicles dropped by 76.4% and 91.8%, respectively, compared to that for the Euro I vehicles. Second, the proportion of vehicle population conforming to Euro II, Euro III and Euro IV emission standards increased from 21.8% in 2006 to 52.6% in 2009, while those only conforming to Euro 0 and Euro I emission standards decreased accordingly, which helped bring down the overall CO emission factor for the whole fleet. However, the rapid growth of vehicle population, particularly the PC and MC, was the major cause for the rebounded emissions in 2009, despite the positive reduction effect of improved emission standard adopted by more vehicles, which implied the necessity of slowing down the rapid growth rate of vehicle population and even capping the total vehicle population for vehicular CO emission control.

Figure 4. Recent trend (2006–2009) of (a) CO, (b) NMVOC, (c) NO_x, (d) PM₁₀, (e) CO₂, (f) CH₄, and (g) N₂O from on-road motor vehicles in China, in comparison to the historical period 1980–2005. (Continued)

The growth rates of NMVOC emissions also began to slow down during the period 2005–2009, with a yearly average growth rate of 8.0%, compared to 15.4% for the period 1980–2005, as shown by Figure 4b. The primary reasons were the same with CO, and the fact that NMVOC emissions continued increasing also benefited from the lower reduction percentages in NMVOC emission factors than CO emission factors with the emission standard improvement.

Similarly, the NO_x and PM₁₀ emissions continued rising at a much higher yearly average growth rate of 12.4% and 11.8%, respectively, than CO and NMVOC for the period 2005–2009, as shown by Figure 4c and Figure 4d. One major reason was that the PPK and FTK in China have been growing, resulting in a rise of the VMT of BC, HDV, and LDV, which were mainly diesel vehicles with relatively higher NO_x and PM₁₀ emission factors. Besides, the population of BC, HDV, and LDV increased as well by 18.7%, 42.6%, and 43.7%, respectively. In addition, emission factors of NO_x and PM₁₀ drop much less significantly compared to those of CO and NMVOC when emission standards improved from Euro 0 to Euro III, which was the case for new vehicles in most parts of China in 2009. Thus, the increased vehicle population and VMT of BC, HDV, and LDV had offset the reduction effect of improved emission standards on NO_x and PM₁₀ emission factors, causing a ceaseless growth of NO_x and PM₁₀ emissions during the period. Therefore, merely improving the emission standards of motor vehicles was not sufficient to turn around the rapid growth trend of NO_x and PM₁₀ emissions, and capping the size of PC population could be beneficial to curb NO_x and PM₁₀ emissions without affecting the public and freight transportation sector in China.

Meanwhile, the greenhouse gas (GHG) emissions of CO₂, CH₄, and N₂O also kept increasing at a yearly average growth rate of 19.7%, 17.5%, and 21.7%, respectively, mostly due to the huge and increasing vehicle population, and the little effect of emission standard improvement on the improvement of fuel economy, as shown by Figures 4e, 4f, and 4g. Therefore, improvement of fuel economy standards and gradually capping the total vehicle population in China are imperative measures to curb recent vehicular GHG emissions, when potential options, like biofuels and electric vehicles, to address the dramatically increasing vehicular GHG emissions and energy demand are premature for China.

Spatial distribution

Due to provincial differences in vehicle population, fleet compositions, timetable for emission standard improvement, dynamic emission factors, and VMT for various vehicle categories, emissions of CO, NMVOC, NO_x, PM₁₀, CO₂, CH₄, and N₂O at the provincial level showed remarkable differences. Table 8, which shows the provincial emissions of CO, NMVOC, NO_x, PM₁₀, CO₂, CH₄, and N₂O in 2009, revealed significant discrepancy in the spatial distribution of China's on-road motor vehicle emissions between different regions: The five provinces of Guangdong, Shandong, Jiangsu, Zhejiang, and Hebei, which are mainly located in the eastern and southeastern coastal areas and cover merely 7.4% of China's territory, accounted for about 41.8%, 43.6%, 34.9%, 35.4%, 38.8%, 43.8%, and 38.9%, respectively, of the national emissions of CO, NMVOC, NO_x, PM₁₀, CO₂, CH₄, and N₂O in 2009, and the western provinces of Xizang, Xinjiang, Neimenggu, Qinghai, Ningxia, Gansu, and Guizhou, which cover as much as 57.2% of China's territory, accounted for only 6.7%, 6.2%, 10.2%, 10.3%, 8.1%, 6.1%, and 7.9%, respectively, of CO, NMVOC, NO_x, PM₁₀, CO₂, CH₄, and N₂O emissions in 2009.

Table 8. Provincial and regional emissions (thousand tons) of CO, NMVOC, NO_x, PM₁₀, CO₂, CH₄, and N₂O from on-road motor vehicles in 2009

Region	Province	CO	NMVOC	NOx	PM10	CO2	CH4	N2O
NE	Heilongjiang	396	81	146	11	10,769	7	1.5
	Jilin	418	84	110	8	8922	8	0.7
	Liaoning	664	140	256	19	17,818	12	2.2
NW	Gansu	178	37	85	7	5358	3	0.7
	Ningxia	95	19	33	3	2296	2	0.2
	Qinghai	54	11	23	2	1512	1	0.1
	Shaanxi	476	101	134	10	10,466	9	1.5
	Xinjiang	323	67	110	8	7239	6	0.7
SW	Guizhou	307	62	107	9	8361	5	0.7
	Sichuan	1122	263	260	20	23,780	23	3.0
	Tibet	40	9	37	3	1655	1	0.1
	Yunnan	640	138	154	12	13,864	12	1.5
	Chongqing	250	58	117	9	7806	5	0.7
N	Anhui	598	132	183	14	13,022	11	1.5
	Beijing	620	130	170	13	20,158	11	2.2
	Hebei	1275	303	356	27	27,170	27	3.0
	Neimenggu	439	96	178	13	11,396	8	1.5
	Shandong	2151	451	368	29	37,556	40	4.5
	Shanxi	645	141	229	17	16,147	12	1.5
	Tianjin	303	69	102	9	8712	6	0.7
C	Henan	1111	241	283	21	20,917	21	2.2
	Hubei	692	159	161	12	12,910	14	1.5
	Hunan	548	127	150	11	11,729	11	1.5
	Jiangxi	451	103	180	13	11,043	9	1.5
SE	Fujian	699	168	138	11	13,262	15	1.5
	Guangdong	2223	581	573	45	52,010	51	6.0
	Guangxi	709	163	132	10	10,652	15	1.5
	Hainan	106	26	30	2	2275	2	0.2
	Jiangsu	1617	377	321	24	31,239	34	3.7
	Shanghai	503	110	148	11	13,220	10	1.5
	Zhejiang	1639	405	342	27	32,651	36	3.7
Total		21,293	4850	5616	430	465,914	429	54
Note: SE, N, C, SW, NW, and NE represent the southeastern, northern, central, southwestern, northwestern and northeastern regions of China, respectively.

Figure 5 shows the vehicular emission inventories in 2009 gridded at a high spatial resolution of using the established methodology (Cai and Xie, 2007), as well as the provincial emission growth from 2005 to 2009. One distinct characteristic of the spatial distribution of emissions in 2009 was that emission densities in the eastern areas were much higher than those in the western areas, which remained unchanged since 2005. Meanwhile, the areas with highest CO, NMVOC, NO_x, PM₁₀, and CO₂e emission densities per grid in 2009 accounted for 1.3%, 2.1%, 2.1%, 1.9%, 1.1%, and 1.9%, respectively, of China's territory, which almost doubled for NO_x, PM₁₀, and CO₂e compared to the situation in 2005, showing a rapidly expansive trend for high emission density areas in recent years. Although the national CO and NMVOC emissions increased at a reduced rate for the recent period 2005–2009, all provinces had increased emissions by 16.7–29.7% and 29.4–39.7%, respectively, for CO and NMVOC, and consequently the most polluted areas were still the Beijing–Tianjin-Hebei region, the Yangtze River Delta, the Pearl River Delta, and the Chengdu–Chongqing region, which experienced the fastest economic growth and motorization during the recent period.

Figure 5. Spatial distribution of gridded (a) CO, (b) NMVOC, (c) NO_x, (d) PM₁₀, and (e) CO₂e emissions at a high spatial resolution of in 2009 and provincial emission growth rate (the height of the bars indicating the growth rates (%)) from 2005 to 2009.

Therefore, priority should be put on the control of emissions in the eastern and southeastern coastal provinces like Jiangsu, Zhejiang, Shanghai, and Guangdong, as well as the northern areas of Beijing, Tianjin, and Hebei, and it is foreseeable that effective control of vehicular emissions in these heavily polluted areas will have a significantly positive impact on the alleviation of traffic-related air pollution in China.

Emission contributions by vehicle categories

Vehicle contributions to emissions of CO, NMVOC, NO_x, PM₁₀, and CO₂e for the period 2005–2009 are depicted in Figure 6. PC and MC continued to dominate in CO emissions after 2005, contributing over 80% of the emissions, as shown by Figure 6a. Besides, the PC contribution has been increasing recently, mainly due to the ceaseless growth of PC population, and the MC contribution has been decreasing, which was mainly because of the improvement of emission standards for MC and its VMT decrease.

Figure 6. Contributions of vehicle categories to national emissions of (a) CO, (b) NMVOC, (c) NO_x, (d) PM₁₀, and (e) CO₂e for the recent period 2005–2009.

Figure 6b shows that the PC and MC remained the dominant contributors to NMVOC emissions since 2005, due to the population growth of these vehicles, most of which were gasoline vehicles and had much higher NMVOC emission factors than the diesel ones. Meanwhile, the PC contribution to NMVOC emissions has been increasing and MC contribution has been decreasing because of the same causes for vehicle contributions to CO emissions. Meanwhile, the NMVOC emission contributions by HDV and BC, which were mostly diesel vehicles, showed a slight increasing trend for the period 2005–2009, due to the growth of their population and VMT stimulated by the rise of freight transportation demand.

Vehicle contributions to NO_x emissions were quite different from those to CO and NMVOC emissions. Figure 6c shows that the primary contributors were HDV and BC, which were mostly diesel vehicles and had higher NO_x emission factors than the gasoline ones, as well as PC, of which the population was large and the VMT was high, despite their relatively lower NO_x emission factors. The total NO_x emission contribution of HDV, BC, and PC was 88.3–88.9% for the period 2005–2009, with increasing contribution from HDV and BC and decreasing PC contribution.

The major contributors to PM₁₀ emissions were similar to those of NO_x, which were the HDV, PC and BC, as shown by Figure 6d. Particularly, the HDV contributed about 37.6-45.9%, due to their high PM₁₀ emission factors, and showed a clear increasing trend and therefore remained the primary contributor for the period 2005–2009. Meanwhile, the PC and BC accounted for about 21.2–17.74% and about 22.7–24.1%, respectively, for PM₁₀ emissions, with the remaining emissions produced by LDV and MC, of which the total contribution showed a decreasing trend from 18.5% in 2005 to 12.4% in 2009.

Figure 6e shows that PC was the primary contributor to CO₂e emissions, which accounted for about 39.1–56.6% and showed a notable increasing trend for the period 2005–2009, mostly due to the ceaseless growing population of PC and the currently relatively loose fuel economy standards for PC in China. It is foreseeable that the GHG emissions from PC in China will continue increasing with the persistent growth of PC population that could continue for another one or two decades unless more stringent control measures, like implementing tighter fuel economy standards and allowing a slower growth rate for a stabilized volume of PC population, can be implemented.

Therefore, PC and MC are the key vehicle categories that require stricter control for the reduction of CO and NMVOC emissions, by measures like slowing down the growth rate and capping the size of PC and MC population, while effective reduction of NO_x and PM₁₀ emissions can be achieved by better control of HDV, BC, and PC through fuel quality improvement, application of advanced tailpipe emission control technologies like lean NO_x catalyst systems and diesel particulate filter, and capping the population of such vehicle types, in addition to more stringent emission standards.

Comparison with other emission inventories and Monte Carlo uncertainty analysis

The recent emission inventories in this study are compared to the first official vehicular emission inventory in China (MEP, 2010): Both emission inventories indicated continuous growth trend for CO, NMVOC, NO_x, and PM₁₀ emissions in recent years, and the temporal variation characteristics in historical emissions were also consistent. In 2009, the MEP report estimated 40,188, 4822, 5833, and 590 thousand tons, respectively, for the emissions of CO, NMVOC, NO_x, and PM₁₀, in comparison to 21,293, 4850, 5616, and 430 thousand tons, respectively, for the same four pollutants in this study. Apparently, the major discrepancies are in CO and PM₁₀ emissions, which differed by about 89% and 37%, respectively, while the NMVOC and NO_x emissions agreed pretty well. Unfortunately, it is difficult to investigate the causes for the discrepancy or consistency in a transparent way, due to lack of the details in the data and methodology adopted by the MEP report. To evaluate the credibility of the CO emissions in this study, the results from other studies were compared: It is found that the CO emissions in Beijing, Shanghai, and Chongqing in 2007 in this study were only about 32%, 20%, and 6%, respectively, lower than the estimation for these cities by Huo et al. (2011), which indicates a much smaller difference than compared to the MEP report. Moreover, Wang et al. (2011b) estimated the CO emissions from passenger cars in 2005 to be 3160 thousand tons, which was about 20% lower than the estimation of 3950 thousand tons in this study, but both were significantly lower than the MEP's estimation of about 8000 thousand tons. Therefore, the CO emission estimation in this study is within the range of literature reports, and it is likely that the CO emissions by MEP (2010) were significantly overestimated. Therefore, the generally consistent temporal variation features and the emission quantities in comparison with recent studies justify the credibility of the recent emission characteristics revealed in this study.

The Monte Carlo uncertainties of CO, NMVOC, NO_x, PM₁₀, CH₄, N₂O, CO₂, and CO₂e emissions at the 95% CI for the period 2006–2009 are summarized in Table 9. The most uncertain emission calculations for the period 2006–2009 were for N₂O and CH₄, which were partly due to the relatively high uncertainties associated with the COPERT-based N₂O and CH₄ emission factors (Kouridis et al., 2010a), followed by PM₁₀, while CO, NMVOC, NO_x, and CO₂ emissions were calculated with the least uncertainty. Despite the relatively larger uncertainty in N₂O and CH₄ emissions, the uncertainty in CO₂e emissions was modest at [–14.1%, 17.3%] for 2006–2009 at the 95% CI, owing to the modest uncertainty of CO₂ emissions that dominated the CO₂e emissions. Moreover, these quantified relatively low uncertainties of emissions, which were based on dynamic emission factors and novel identification of provincial VMT of various vehicle categories, revealed a necessity of applying vehicle technology- and vehicle age-specific emission factors and VMT for vehicular emission estimation.

Table 9. Uncertainties (%) of CO, NMVOC, NO_x, PM₁₀, CH₄, N₂O, CO₂, and CO₂e emissions from on-road motor vehicles in China at the 95% CI for the period 2006–2009 by Monte Carlo stochastic simulations

Year	CO	NMVOC	NOx	PM10	CO2	CH4	N2O	CO2e
2006	[–15.5, 10.3]	[–13.2, 14.7]	[–14.2, 17.7]	[–23.3, 19.2]	[–14.5, 15.8]	[–37.3, 14.9]	[–77.3, 32.5]	[–15.0, 16.7]
2007	[–18.1, 12.1]	[–15.7, 12.6]	[–15.1, 15.6]	[–25.7, 18.2]	[–13.9, 17.7]	[–35.3, 10.4]	[–79.2, 41.4]	[–14.8, 17.3]
2008	[–17.9, 9.2]	[–14.8, 14.9]	[–13.9, 12.3]	[–22.7, 15.9]	[–13.8, 15.5]	[–33.7, 5.9]	[–80.3, 32.2]	[–14.5, 16.3]
2009	[–18.5, 13.0]	[–17.9, 15.7]	[–15.6, 13.8]	[–28.0, 18.3]	[–12.9, 14.3]	[–33.5, 11.2]	[–81.3, 17.6]	[–14.1, 14.8]

Conclusions

The recent trend of emissions from on-road motor vehicles in China was tracked by means of updated emission factor database for a complete vehicle fleet at the provincial level, to take into account the rapid changes of influential factors for emission factors of vehicular pollutants. The developed dynamic emission factors compensated for the deficit of real-world measurement and allowed convincing estimation of state-of-the-art vehicular emissions in the recent period 2006–2009. The results showed that despite the tremendous advances in emission standards of both gasoline and diesel vehicles, the emissions of major vehicular pollutants have yet to decrease from the consistent growth trend from 1980 to 2005. Improvement of more stringent emission standards and the increased proportion of vehicle population conforming to Euro III and Euro IV emission standards had slowed down the growth rates of CO and NMVOC emissions. The NO_x and PM₁₀ emissions, however, continued increasing rapidly, due to the growth of population and VMT of diesel vehicles, which offset the reduction effect of improved emission standards on NO_x and PM₁₀ emission factors. Furthermore, GHG emissions of CO₂, CH₄, and N₂O in 2009 doubled compared to those in 2005, due to the huge and growing vehicle population and little effect of emission standard improvement on fuel economy improvement. This recent emission trend revealed that merely improving emission standards of motor vehicles was insufficient to turn around the rapid growth trend of NO_x, PM_10, and GHG emissions or to prevent the continued growth of CO and NMVOC emissions, and future reduction of vehicular emissions can be expected by control of the growth rate of vehicle population, particularly PC, and capping the total vehicle population in China. The distinct emission intensities among provinces reveal the importance of emission control in the eastern and southeastern coastal provinces like Jiangsu, Zhejiang, Shanghai, and Guangdong, as well as the northern areas of Beijing, Tianjin, and Hebei, for mitigation of China's traffic-related air pollution.

Explicit Monte Carlo stochastic simulations showed that the uncertainties of the emission estimation for the recent period were quite acceptable, mostly owing to the improved methodologies to quantify the impacts of recent dynamic influencing factors for emissions factors and to calculate the provincial VMT of various vehicle categories. The methodologies developed in this study to dynamically update the emission factors and to quantify VMT of various vehicle categories are also applicable for routine update and forecast of China's on-road motor vehicle emissions.

Acknowledgment

We would like to thank the financial support from Research Program of Optimization of Stationing of Urban Ambient Air Quality Monitoring (200709001) funded by the Ministry of Environmental Protection of the People's Republic of China, and we also appreciate the financial support from Shanghai Tongji Gao Ting-yao Environmental Science and Technology Development Foundation, from the Doctoral Thesis Scholarship of China Development Research Foundation funded by General Motors, and from the Better Future Scholarship of Energy Foundation.