SEARCH

SEARCH BY CITATION

Keywords:

  • elemental carbon;
  • Beijing;
  • emission sources

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Measurements
  5. 3. Characteristics of Sources in Beijing
  6. 4. Temporal Variations of EC, CO, CO2, and Meteorological Parameters
  7. 5. Statistical Analysis
  8. 6. Correlation Analysis
  9. 7. Evaluation of Emission Inventory in Beijing
  10. 8. Summary and Conclusions
  11. Acknowledgments
  12. References
  13. Supporting Information

[1] Concentrations of elemental carbon (EC), carbon monoxide (CO), and carbon dioxide (CO2) were measured in Beijing between 2005 and 2006. EC was measured every hour with a semicontinuous thermal optical analyzer. The observed concentrations were rather uniform over a distance of about 50 km from the observation site. The annual average concentrations of EC and CO were 6.9 μgC m−3 and 1120 parts per billion by volume, respectively. The concentrations of these species increased with decreasing near-surface wind speed (WS). The slopes of the CO-CO2, EC-CO2, and EC-CO correlations are used to estimate major EC and CO sources. In the weak wind regime (WS ≤ 2.0 m s−1), the median EC, ΔEC/ΔCO2, and ΔEC/ΔCO (except for winter) increased in the late evening and remained high until early morning. The traffic of heavy duty diesel trucks during nighttime was about 20 times higher than that during daytime. These results indicate a dominant contribution of exhaust from diesel vehicles to the nighttime EC. In winter, the nighttime CO and ΔCO/ΔCO2 ratio were largely higher than those in the other seasons. The most likely cause is the increase in the CO emissions from the exhaust of gasoline vehicles at low temperature. The ΔEC/ΔCO2 ratio in winter was lower than that in fall, indicating no significant additional EC emissions. The diurnal variations of EC, CO, CO2, and ΔEC/ΔCO were similar between weekdays and weekends. The slopes of the CO-CO2-EC correlations are compared with the CO-CO2-EC ratios derived from a published emission inventory in the Beijing area.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Measurements
  5. 3. Characteristics of Sources in Beijing
  6. 4. Temporal Variations of EC, CO, CO2, and Meteorological Parameters
  7. 5. Statistical Analysis
  8. 6. Correlation Analysis
  9. 7. Evaluation of Emission Inventory in Beijing
  10. 8. Summary and Conclusions
  11. Acknowledgments
  12. References
  13. Supporting Information

[2] Elemental carbon (EC) or black carbon (BC) is produced by incomplete combustion of fossil fuels, biofuels, and biomass. EC strongly absorbs solar visible radiation and efficiently heats the atmosphere, resulting in a significant contribution to the radiative forcing of the atmosphere [e.g., Intergovernmental Panel on Climate Change, 2007]. The heating of the atmosphere by EC also changes the vertical and horizontal thermal structure of the atmosphere [e.g., Crutzen and Andreae, 1990; Jacobson, 2001; Ramanathan et al., 2001; 2007; Menon et al., 2002; Conant et al., 2002]. In addition, EC has a deleterious effect on human health [Watson et al., 1994; Rudell et al., 1994; Gong et al., 2003; McCreanor et al., 2007].

[3] Asia is estimated to have been the largest EC source region in the year 2000 [Streets et al., 2003; Ohara et al., 2007]. Among the Asian countries, China is the largest EC source, although uncertainties in the estimates are large. The Beijing area contributes dominantly to the EC emissions in northeastern China [Guinot et al., 2007]. EC transported from this area strongly influences EC concentrations in Asian outflow over the western Pacific [Shiraiwa et al., 2008; Sahu et al., 2009]. Reliable estimates of EC emissions in northeastern China are required for improved evaluations of the impacts of EC on air quality and climate in East Asia and regions downwind. A quantitative understanding of the temporal variations in Beijing is critically important for this purpose because Beijing is located in a large source area in northeastern China.

[4] Our previous studies of EC in Tokyo have shown that highly time-resolved EC measurements are very useful in interpreting temporal variations of EC emissions [Kondo et al., 2006; Takegawa et al., 2006]. In addition, carbon monoxide (CO) and carbon dioxide (CO2) have been shown to be key tracers for characterizing sources of EC because they are also produced by the combustion of fossil and biofuels. In previous studies of EC in Beijing, the time resolutions of EC measurements were between 1 day and 1 week, as summarized in Table 1. In addition, no simultaneous CO or CO2 measurements were made in these studies. Because of the limitations of these measurements, the temporal variations of EC emissions in the Beijing area were not fully characterized.

Table 1. Previous Studies of EC in Beijing, China
ReferenceMeasurement PeriodSample Integration TimeSitesEC μgC m−3Method of EC Determination
  • a

    TOR, thermal-optical reflectance.

  • b

    TOT, thermal-optical transmittance.

He et al. [2001] and Yang et al. [2005]Jul 1999 to Sep 20001 weeka residential and a downtown siteSpring: 6.6 Summer: 6.9 Fall: 9.4 Winter: 7.8TORa
Zheng et al. [2005]Jan, Apr, Jul, and Oct 200024 hfive urban and rural sitesJan.: 2.03 Apr.: 2.91 Jul.: 3.06 Oct.: 4.01TOTb
Sun et al. [2004]summer and winter seasons from 2002 to 20033 dayindustrial and residential sitesSummer: 6.6 (industrial) Winter: 21.9 (residential)C/H/N elemental analyzer
This studywinter, spring, summer, and fall seasons from Nov 2005 to Oct 20061-hurban site on a university campusWinter: 6.7 Spring: 6.2 Summer: 6.4 Fall: 8.8TOTb

[5] In this study, we measured EC in Beijing using a semicontinuous analyzer with 1-h time resolution together with CO and CO2 for the first time. The major objective of this study is to establish statistically significant EC, CO, and CO2 levels in Beijing during the four seasons. We also discuss the diurnal variations of these species and their mutual correlations in different seasons, especially in relation with their sources and meteorological conditions. Further, the slopes of the CO-CO2-EC correlations are compared with the CO-CO2-EC ratios derived from a published emission inventory for the Beijing area [i.e., Streets et al., 2003]. This provides a measure of the uncertainties in the emission inventory.

2. Measurements

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Measurements
  5. 3. Characteristics of Sources in Beijing
  6. 4. Temporal Variations of EC, CO, CO2, and Meteorological Parameters
  7. 5. Statistical Analysis
  8. 6. Correlation Analysis
  9. 7. Evaluation of Emission Inventory in Beijing
  10. 8. Summary and Conclusions
  11. Acknowledgments
  12. References
  13. Supporting Information

2.1. Measurement Site

[6] The in situ measurements of EC, CO, and CO2 were made at a surface site in Beijing. The city of Beijing (39.9°N, 116.4°E) is partially surrounded by Hebei province and located about 100 km northwest of Tianjin city, which is the largest city near Beijing, as shown in Figure 1a. The populations of municipal (metro) Beijing and Tianjin are about 17 (13) and 11 million, respectively. Figure 1a also shows the EC emission rates with a spatial resolution of 0.5° × 0.5° in latitude and longitude by Streets et al. [2003]. Beijing and Tianjin are two highly concentrated sources of EC in northeast China. Beijing is located close to the northern edge in the highest EC region of northeastern China, as seen from Figure 1a.

image

Figure 1. (a) Locations of Beijing, the measurement site at Peking University (PKU), and the Yufa site. The grayscale areas represent the annual anthropogenic emissions of EC for the year 2000 at 0.5° × 0.5° by Streets et al. [2003]. The region within the gray square identifies the Beijing study area (39.5°N, 40.5°N–116.0°E, 117.0°E) used to compare emission rates by inventory data with the observations in this study. (b) Map of Beijing region and the PKU site. The fourth and fifth ring roads are numbered.

Download figure to PowerPoint

[7] The observation site is on the top of one of the buildings of Peking University (PKU) (39.99°N, 116.30°E), 15 m above ground level. The PKU campus is located north and northwest of the urban center of Beijing. The fourth ring road is located about 650 m south of the PKU site (Figure 1b). The traffic density of this road is high. A one-way street is located about 800 m east of the observation site.

[8] The observations were conducted from November 2005 to October 2006, namely in winter (25 November 2005 to 16 January 2006), spring (27 March to 16 May 2006), summer (12 August to 16 September 2006), and fall (30 September to 31 October 2006). The EC and CO data were obtained over four seasons in 2005–2006. The CO2 data were not available in the spring of 2006.

2.2. Instruments

[9] EC mass concentrations were measured by a semicontinuous EC and organic carbon (OC) analyzer (model RT3052, Sunset Laboratory Inc., Beaverton, Oregon, USA), with a 1-h time resolution. The air samples taken from the roof of the PKU building were introduced into a cyclone with a 50% effective cutoff diameter of 2.5 μm (PM2.5) at 8 L min−1 to eliminate coarse particles and then delivered to a denuder to remove gas phase OC. Aerosol samples were collected on a 1.13 cm2 quartz filter for 40 min, and then the concentrations of EC and OC were analyzed by a thermal optical transmittance (TOT) method based on the National Institute for Occupational Safety and Health (NIOSH) protocol [Birch and Cary, 1996], with a slight modification to the heating temperatures according to previous studies [Lim and Turpin, 2002; Subramanian et al., 2006]. Using this protocol, carbon from OC sufficiently evolves with a final temperature of 700°C, whereas the protocol with a higher temperature of 870°C possibly underestimates the EC concentration. The samples were heated sequentially in four steps, 300°C, 450°C, 600°C, and 740°C, in a completely oxygen-free atmosphere for OC analysis. After the sample was cooled, the oven temperature was then stepped up to 870°C under a 10% oxygen environment for EC and pyrolized carbon (PC) analysis. All the carbon components of carbonaceous aerosols (OC and EC) were converted to CO2 and detected with a nondispersive infrared absorption (NDIR) CO2 sensor. OC and EC were automatically quantified by dividing their peak areas by the internal calibration peak made by methane gas (5% CH4 in He). The correction for the PC conversion of OC to EC was performed by monitoring the transmittance of a pulsed He-Ne diode laser beam at 670 nm through the quartz fiber filter during the sample analysis. The zero level of EC was measured by collecting particle-free ambient air through a particle filter placed upstream of the denuder. The average zero level (±3σ) was 0.07 (±0.03) μgC m−3 during the measurement periods. The detection limit was defined to be 0.03 μgC m−3. Calibration of total carbon (TC) = OC + EC was performed by loading a known amount of sucrose (about 86 μg) onto the filter before starting every seasonal measurement period. The uncertainty of the TC measurement was estimated to be about 7%, determined by the standard error of the sucrose calibration and the routine methane calibration. Uncertainties of the EC-to-OC split, which may depend on the temperature protocols, can lead to additional errors [e.g., Boparai et al., 2008]. We estimated this error to be about 22% from the difference of the EC values obtained in Tokyo by using NIOSH and Interagency Monitoring of Protected Visual Environments (IMPROVE) temperature protocols [Kondo et al., 2006]. We assume the same errors for the present measurements considering the very similar operating conditions to those in Tokyo.

[10] An NDIR-based gas analyzer (Model 48C, Thermo Environmental Instruments, USA) was used to measure CO at an integration time of 1 min. The sample air was dried using a nafion dryer to reduce the interference from water vapor. The background signal (zero level) was routinely measured every 2 h by supplying purified air into the sample line. Overall calibrations were performed at the beginning of each measurement period by supplying a CO standard (6.5 ppm by volume (ppmv) CO in air). CO2 was measured using an NDIR-based instrument with a 10-s time resolution (Model LI 7000, LI-Cor, Inc., USA). The sample line for the CO2 instrument was shared with the CO instrument. The CO2 calibration was made using a standard gas bottle with 628 ppmv CO2 in air.

[11] Meteorological parameters (ambient temperature, relative humidity, wind speed (WS), and wind direction (WD)) were measured at the sampling point with an integration time of 10 min. In addition to the in situ data, we used the vertical profiles of aerosol measured by a lidar system operated at the Sino-Japan Friendship Center for Environmental Protection (39.90°N, 117.16°E) to estimate the atmospheric mixed layer height (MLH). The aerosol backscattering coefficients up to 24 km in altitude were measured with a height resolution of 6 m at wavelengths of 1064 nm and 532 nm every 15 min [Sugimoto and Lee, 2006]. The concentrations of aerosols are generally high below the MLH and low above it. The MLH can be determined from the falling edge of the lidar backscattering intensity. The MLH was defined here as the height where the normalized gradient of the lidar backscattering intensity at 1064 nm had a minimum [Takegawa et al., 2009]. All the gas data and the meteorological parameters were merged onto the 1-h interval of the EC data.

3. Characteristics of Sources in Beijing

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Measurements
  5. 3. Characteristics of Sources in Beijing
  6. 4. Temporal Variations of EC, CO, CO2, and Meteorological Parameters
  7. 5. Statistical Analysis
  8. 6. Correlation Analysis
  9. 7. Evaluation of Emission Inventory in Beijing
  10. 8. Summary and Conclusions
  11. Acknowledgments
  12. References
  13. Supporting Information

[12] Streets et al. [2003] developed an emission inventory of gases and primary aerosols, including EC, CO, and CO2, for Asia corresponding to the year 2000 with a spatial resolution of 0.5° × 0.5° in latitude and longitude. Despite continued improvements [Suntharalingam et al., 2004; Tan et al., 2004; Streets et al., 2006], uncertainties of the estimates are still large, i.e., 484% for EC, 156% for CO, and 16% for CO2 for China [Streets et al., 2003].

[13] Large EC sources are located in the urban center of Beijing and in the area south and southeast of Beijing, while little EC is emitted in the northern area, as shown in Figure 1a. The spatial patterns of the CO and CO2 emissions are similar to those of EC in this region [Streets et al., 2003]. In this study, we defined the Beijing study area (39.5°–40.5°N, 116.0°–117.0°E) (the area enclosed within the gray solid line in Figure 1a), which covers the urban region of the Beijing Province. The validity of this definition is discussed in section 7.

[14] Table 2 shows contributions of different sources of EC, CO, and CO2 in the Beijing study area based on the emission inventory [Streets et al., 2003, 2006]: (l) fossil fuel and biofuel combustion from the domestic sector and (2) the nondomestic sector, including industry, small power generation, and transportation. The nondomestic sector comprises a large component of EC (61%), CO (81%), and CO2 (83%) emission sources. Industry and transportation contribute predominantly to the emissions from the nondomestic sector (Streets et al. [2003] and unpublished results, 2007). A larger portion from the domestic sector (39%) is estimated to contribute to the EC emissions than for CO (19%) and CO2 (17%). We note here that large uncertainties are given in estimating contributions of the individual sources [Streets et al., 2003]. In section 7, we evaluate the contributions of important sources in the Beijing study area based on our observations.

Table 2. Relative Contributions of Different sources to EC, CO, and CO2 Emissions in the Beijing Study Area for the Year 2000a
 Domestic SectorNondomestic SectorbTotal
Fossil FuelBiofuelIndustryPowerTransportation
  • a

    Values in parentheses are emissions of each species from the domain in units of Gg Species yr−1. Relative contributions are in %. Beijing study area is (39.5°N, 40.5°N–116.0°E, 117.0°E). From Streets et al. [2003].

  • b

    Contributions of each source in the non-domestic sector were estimated by provided inventory data set (D. G. Streets et al., unpublished data, 2007).

EC23 (2.0)16 (1.4)37 (3.2)2 (0.2)22 (1.9)100 (8.7)
CO14 (278)5 (110)32 (667)1 (17)48 (984)100 (2056)
CO214 (9487)3 (2084)41 (28178)25 (16907)17 (11271)100 (67928)

[15] Coal is a dominant energy source in China. Beijing consumed about 31 million tons of coal in 2005, accounting for 56% of the total energy consumption (Beijing Statistical Year Book 2006, available at http://www.bjstats.gov.cn/tjnj/2006-tjnj/index-english.htm). In Beijing, coal is used for industrial coal-burning boilers and residential coal combustion including cooking and heating (heating season from 15 November to 15 March) [Yang et al., 2005]. The use of coal may constitute substantial portions of the emissions of CO and carbonaceous aerosols.

[16] Traffic is an important anthropogenic emission source due to the rapid growth in the number of vehicles in Beijing [Guinot et al., 2007]. Figure 2a shows the diurnal variation of the traffic of all vehicles, heavy duty vehicles (HDVs), and heavy duty diesel trucks (HDDTs) measured on the northern part of the fourth ring road (40.0°N, 116.3°E) on 26 June (Thursday) 2008. HDVs include buses, coaches, and HDDTs. The number of HDVs can be considered as the upper limit of heavy duty diesel vehicle (HDDV) traffic. The total number of vehicles was high during daytime and low during nighttime. The number of HDDTs and their ratio to the total vehicles increased at 2200 LT and remained high during nighttime (Figures 2a and 2b). HDDTs are allowed to enter the urban area inside the fifth ring road (Figure 1b) from 2200 to 0600 LT, under Beijing traffic regulations (Beijing Traffic Management Bureau, available at http://www.bjjtgl.gov.cn/Inquiries/roadControl.asp (in Chinese)). The traffic of HDVs and HDDTs constituted about 10–20% of the traffic during nighttime, while they fell close to 0% during daytime (Figure 2b). Figure 2c shows the diurnal variation of the traffic of all vehicles and HDDTs measured at the same place on 29 June (Sunday) 2008. The ratio of HDDTs to the total number of vehicles is shown in Figure 2d. The daytime total traffic of vehicles on Sunday was similar to that on Thursday, within 20%.

image

Figure 2. Diurnal variations of (a) the traffic density of total vehicles, heavy duty vehicles (HDVs), and heavy duty diesel trucks (HDDTs) on Thursday, (b) the fraction of HDVs and HDDTs on Thursday, (c) the traffic density of total vehicles and HDDTs on Sunday, and (d) the fraction of HDDTs on Sunday. The left axis shows the total of number vehicles per hour, and the right axis shows the number of HDVs or HDDTs per hour (Figures 2a and 2c). The observations were conducted on the north-fourth ring road (40.0°N, 116.3°E) on 26 June 2008 for Figures 2a and 2b and on 29 June 2008 for Figures 2c and 2d.

Download figure to PowerPoint

4. Temporal Variations of EC, CO, CO2, and Meteorological Parameters

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Measurements
  5. 3. Characteristics of Sources in Beijing
  6. 4. Temporal Variations of EC, CO, CO2, and Meteorological Parameters
  7. 5. Statistical Analysis
  8. 6. Correlation Analysis
  9. 7. Evaluation of Emission Inventory in Beijing
  10. 8. Summary and Conclusions
  11. Acknowledgments
  12. References
  13. Supporting Information

[17] Figure 3 shows time series of the hourly average values of EC, CO, and WS in four seasons. EC and CO showed correlated temporal variations, often anticorrelated with WS. The annual and seasonal statistics of each species are summarized in Table 3 and Figure 4. The EC concentrations ranged from 0.1 to 34.5 μgC m−3 with a median value of 5.9 μgC m−3. CO concentrations ranged from about 100 to 8330 parts per billion by volume (ppbv) with a median value of about 840 ppbv.

image

Figure 3. Time series plots of EC and CO concentrations with wind speed (WS) at the PKU site, averaged in 1-h intervals, for winter, spring, summer, and fall between 2005 and 2006.

Download figure to PowerPoint

image

Figure 4. Seasonal variations of (a) EC, (b) CO, and (c) CO2 concentrations. The central lines and dots are the seasonal medians and means, respectively. The bottoms and tops of the boxes are the 25th and 75th percentiles.

Download figure to PowerPoint

Table 3. Statistical Summary of the Concentrations of EC in PM2.5, CO, CO2, and Wind Speed in Beijinga
 EC, μgC m−3CO, ppbvCO2, ppmvWind Speed, m s−1
  • a

    Values in parentheses are standard deviations. NA, nonavailable data because of instrumental error. WS, wind speed.

Annual average6.9 (±5.3)1118 (±1015)438 (±37)2.6 (±1.7)
Median5.98414312.0
Minimum0.11013500.4
Maximum34.5832865013.7
Number of data points3403331624833665
Average
Winter6.7 (±6.3)1557 (±1529)443 (±44)3.0 (±1.9)
Spring6.2 (±4.2)790 (±548)NA2.9 (±1.9)
Summer6.4 (±3.7)838 (±446)423 (±26)2.3 (±1.4)
Fall8.8 (±5.6)1197 (±713)446 (±28)1.9 (±1.9)
Average (Weak Wind Conditions: WS2.0 m s1)
Winter11.1 (±6.9)2784 (±1489)474 (±37)1.4 (±0.4)
Spring8.2 (±4.3)1157 (±553)NA1.3 (±0.4)
Summer7.8 (±3.5)993 (±438)432 (±24)1.4 (±0.4)
Fall10.0 (±5.5)1382 (±682)453 (±27)1.3 (±0.4)

[18] EC, CO, and CO2 concentrations were measured at the Yufa site (39.5°N, 116.3°E) in the same period during the summer season in 2006 [Takegawa et al., 2009]. The Yufa site is located in a rural area about 53 km south of the PKU site (Figure 1a). Figure 5 shows time series of EC, CO, and CO2 observed at the PKU and Yufa sites. The average concentration of EC at the Yufa site was about 30% lower than that at the PKU site. No significant average differences in the concentrations of CO and CO2 were shown between the two sites. Irregular variations at timescales shorter than 1 day were observed more frequently at the Yufa site, especially for CO and CO2. The diurnal and temporal variations at timescales longer than 1 day of these species were similar in amplitude and pattern at the two sites. These similarities indicate that the concentrations of these species were rather uniform over distances of about 50 km from the PKU site.

image

Figure 5. Time series plots of EC, CO, and CO2 concentrations at the PKU site and the Yufa in summer 2006.

Download figure to PowerPoint

[19] The mean EC concentration for each season ranged between 6.2 and 8.8 μgC m−3, with a slightly higher value in fall (Table 3). The mean CO concentrations were higher in winter and fall (1560 and 1200 ppbv, respectively) than in spring and summer (790 and 840 ppbv, respectively). In particular, the average CO in winter was markedly high (Figure 4), with a very large 1σ value of 1500 ppbv (Table 3). The low and very high CO was generally associated with strong northerly winds and weak winds, respectively. The causes of the very high CO in winter are discussed in section 6.2. The CO2 concentration in summer was lower than that during other seasons (Figure 4), likely due to stronger absorption by vegetation.

[20] Figure 6 shows statistics of the meteorological parameters (temperature and WS). The very low temperature and relatively high WS in winter should be noted. Figure 7 shows the distributions of WS and the frequency of WD in all four seasons. The mean annual WS was 2.6 m s−1. Fractions of strong (WS > 5.0 m s−1), moderate (2.0 m s−1 < WS ≤ 5.0 m s−1), and weak (WS ≤ 2.0 m s−1) wind conditions in winter were about 19, 38, and 43%, respectively. Northerly winds were dominant in winter and northerly and southerly winds dominated in the other seasons. In particular, in spring northerly and southerly winds were dominant, with large variability in the WS. Weak winds prevailed in summer and fall (54% and 64%, respectively), and the seasonal average WS was lowest in fall.

image

Figure 6. Meteorological conditions: (a) atmospheric temperature and (b) wind speed in each season during the observation period. The central lines and dots are the seasonal medians and means, respectively. The bottoms and tops of the boxes are the 25th and 75th percentiles.

Download figure to PowerPoint

image

Figure 7. Distributions of wind speed and the frequencies of various wind directions in each season. The gray, dark gray, and black areas indicate weak (WS ≤ 2.0 m s−1), moderate (2.0 m s−1 < WS ≤ 5.0 m s−1), and strong (WS > 5.0 m s−1) winds, respectively.

Download figure to PowerPoint

5. Statistical Analysis

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Measurements
  5. 3. Characteristics of Sources in Beijing
  6. 4. Temporal Variations of EC, CO, CO2, and Meteorological Parameters
  7. 5. Statistical Analysis
  8. 6. Correlation Analysis
  9. 7. Evaluation of Emission Inventory in Beijing
  10. 8. Summary and Conclusions
  11. Acknowledgments
  12. References
  13. Supporting Information

5.1. Data Classification

[21] Generally, aerosol concentrations at the surface are lower during faster outflow of air from the source region [Kondo et al., 2006], and this is seen in the EC concentration in Beijing as shown in Figure 8. The median WS in fall is lowest, as discussed in section 4, resulting in the highest median value of EC. The WS increased from fall to winter and the median EC concentration decreased accordingly.

image

Figure 8. Scatterplot of the EC concentration versus wind speed in each season. The plot symbols represent median values for each season and the bars are the 25th and 75th percentiles.

Download figure to PowerPoint

[22] The MLH also controls the near-surface concentrations of EC, CO, and CO2 over the source regions because it controls the volume of air into which the emitted species are mixed. The MLH over land is generally lower during nighttime (e.g., less than a few hundred meters) and reaches higher levels during daytime (e.g., greater than 1 km). Enhanced instability associated with greater vertical turbulence in the afternoon increases the downward transport of horizontal momentum, resulting in strong near-surface WS, reducing the concentrations of these species at the surface. Because the WS directly affects the concentrations at the surface, we use WS as a primary parameter to extract air masses impacted by primary emissions near Beijing.

[23] In addition, the WD is also an important factor determining the EC concentration. Figure 9 shows the concentrations of EC, CO, and CO2 as functions of WS for northerly and southerly winds. The concentrations of these species under weak wind conditions were much higher and dramatically decreased with increasing WS irrespective of WD. These concentrations reached stable levels at a WS of about 4.0 m s−1. We interpret the concentrations at stronger winds as representative of the background levels of air masses, which were transported over 350–400 km within a day. The background concentrations in southerly air were much higher than in northerly air. The EC and CO emissions north of Beijing are much lower than south of Beijing, as discussed in section 3. This estimate is qualitatively consistent with the observed north-south difference in the background levels. Under low-wind speed conditions, the concentrations of each species are interpreted to be more strongly controlled by emissions in Beijing. We focus on the weak wind (WS ≤ 2.0 m s−1) data hereafter.

image

Figure 9. Relationships between the median values of concentrations of (a) EC, (b) CO, and (c) CO2 and wind speed according to wind direction, for (top) northerly (271°–90°) and (bottom) southerly (91°–270°) winds in each season. The bars represent 25th and 75th percentiles.

Download figure to PowerPoint

5.2. Diurnal and Seasonal Variations

[24] Figures 10a10c show the diurnal variations of the median values of EC, CO, and CO2 during weekdays (from Monday to Friday) for each season. As shown in Figure 10a, the EC concentrations started to increase at around 1800 LT and remained high during the nighttime. The higher nighttime EC concentrations could have been due to higher EC emissions during the nighttime than those during the daytime and/or to more efficient accumulation of EC in the atmosphere due to lower WS and MLH. Even with a WS of less than 2.0 m s−1, EC increased with the decrease in WS, as seen from Figure 9a.

image

Figure 10. Diurnal variations of the median values of (a) EC, (b) CO, and (c) CO2 concentrations for each season under weak wind conditions during working days. The bars represent 25th and 75th percentiles.

Download figure to PowerPoint

[25] The diurnal variation of EC was more or less similar in all seasons. The pattern of the diurnal variations of CO is similar to that of EC to some extent in spring, summer, and fall (Figure 10b), although the night/day ratios of CO are much lower than EC. In winter, the CO concentrations were much higher than those in the other seasons. The highest CO in winter is discussed in section 6.2. The amplitude of the diurnal variations of CO2 concentrations in summer and fall was larger than that in winter (Figure 10c).

[26] The background concentrations of EC, CO, and CO2 ([EC]0, [CO]0, and [CO2]0, respectively) were defined as the 1.25th percentile of the EC, CO, and CO2 values (1.96σ) for each season. They varied slightly according to season, as summarized in Table 4.

Table 4. Background Concentrations of EC, CO, and CO2 During the Measurement Periodsa
 [EC]0, μgC m−3[CO]0, ppbv[CO2]0, ppmv
  • a

    NA, nonavailable data because of instrumental error.

Winter0.58175393
Spring0.64147NA
Summer0.68139383
Fall1.10161401

6. Correlation Analysis

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Measurements
  5. 3. Characteristics of Sources in Beijing
  6. 4. Temporal Variations of EC, CO, CO2, and Meteorological Parameters
  7. 5. Statistical Analysis
  8. 6. Correlation Analysis
  9. 7. Evaluation of Emission Inventory in Beijing
  10. 8. Summary and Conclusions
  11. Acknowledgments
  12. References
  13. Supporting Information

6.1. Slope of the Correlations and Emission Ratios

[27] Absolute concentrations of EC, CO, and CO2 are affected by their emissions and meteorological conditions, although the contribution of each effect is not quantitatively separated in this study. In order to characterize their important sources, we calculated the correlations of CO-CO2, EC-CO2, and EC-CO. The slopes of the correlations (ΔCO/ΔCO2, ΔEC/ΔCO2, and ΔEC/ΔCO) were determined using a bivariate regression analysis incorporating the errors in both variables. Figure 11 shows the CO-CO2, EC-CO2, and EC-CO correlations for winter and fall under weak wind conditions. The derived slopes in different seasons are summarized in Table 5. The diurnal variations of the slopes for each season are shown in Figure 12. The slopes of the correlations were evaluated every 2 hours. We excluded the slopes with a correlation coefficient (r2) less than 0.4 from the present analyses.

image

Figure 11. Scatterplots of (a) CO versus CO2, (b) EC versus CO2, and (c) EC versus CO for winter and fall under weak wind conditions. The solid line is the linear regression line determined by bivariate regression analysis.

Download figure to PowerPoint

image

Figure 12. Diurnal variations of (a) ΔCO/ΔCO2, (b) ΔEC/ΔCO2, and (c) ΔEC/ΔCO for each season under weak wind conditions during working days. The slopes with a correlation coefficient (r2) less than 0.4 were excluded.

Download figure to PowerPoint

Table 5. ΔEC/ΔCO, ΔCO/ΔCO2, and ΔEC/ΔCO2 Ratios Under Weak Wind Conditions During the Measurement Periodsa
 ΔCO/ΔCO2ΔEC/ΔCO2ΔEC/ΔCO
ppbv/ppmvr2μgC m−3/ppmvr2μgC m−3/ppbvr2
  • a

    NA, nonavailable data because of instrumental error. WS ≤ 2.0 m s−1.

Winter43.90.820.140.710.00350.75
SpringNANANANA0.00340.34
Summer32.90.320.120.590.00480.48
Fall30.20.710.190.800.00580.51

[28] Table 6 summarizes the emission ratios of CO/CO2, EC/CO2, and EC/CO for various types of sources in China. These values are derived from emission factors (mass of the species emitted per mass of fuel burned) given in previous studies [Streets et al., 2003; Cao et al., 2007, 2008; Westerdahl et al., 2009]. The ranges of each emission ratio, shown in Table 6, cover various combustion processes and different provinces. Table 7 summarizes the emission ratios of these species for the vehicle sector in Beijing [Westerdahl et al., 2009]. For this estimate, we used the emission factors of CO2 measured in Germany and the United States [Sullivan et al., 2004] because of the lack of data in Beijing. These ratios were compared with the slopes of observed CO-CO2-EC correlations to estimate the dominant source effects in Beijing.

Table 6. Emission Ratios of EC, CO, and CO2 Calculated by Emission Factors for Various Types of Sources in China
 DomesticVehicleIndustryBiomass Burning
CoalaBiofuelaDiesela,b,c,dGasolinea,b,c,dCoalaCrop ResidueaRice Strawe
CO/CO2f57–6253–1194.8–2524–6123–409564
EC/CO2g0.11–3.71.5–1.80.17–1.200.02–0.180.04–0.80.850.55
EC/COh0.002–0.0590.015–0.0280.007–0.2370.0003–0.0070.002–0.0190.0090.009
Table 7. Emission Ratios of EC, CO, and CO2 Calculated by Emission Factors for Vehicles in Beijing
 Heavy Duty Diesel Vehiclea,bLight Duty Gasoline Vehiclea,b
CO/CO2c25.247.3
EC/CO2d0.810.17
EC/COe0.0320.0036

6.2. CO-CO2 Correlation

[29] The diurnal variations of ΔCO/ΔCO2 are shown in Figure 12a. About half of the ΔCO/ΔCO2 ratios in summer were excluded especially during the daytime (r2 < 0.4). The ΔCO/ΔCO2 ratios in fall decreased around 2200 LT and remained low during nighttime. The ΔCO/ΔCO2 ratios were about 20–30 ppbv/ppmv during the nighttime and were about 40 ppbv/ppmv near noon, indicating stronger influences of sources with lower emission ratios during the nighttime. These ratios are much lower than the emission ratios for domestic sources (53–119 ppbv/ppmv) (Table 6) and similar to the emission ratios for vehicles (25–47 ppbv/ppmv) (Table 7). This indicates the large contribution of vehicles to CO emissions in Beijing.

[30] In winter, the CO levels and ΔCO/ΔCO2 ratios were much higher during nighttime (Figures 10b and 12a). The diurnal pattern of the ΔCO/ΔCO2 ratios in winter was different from that in fall, reaching a broad maximum after midnight. The high nighttime CO and ΔCO/ΔCO2 ratios in winter were due to (1) CO sources specific to winter with higher CO/CO2 emission ratios, including emissions by domestic heating and/or (2) increased emissions from the same combustion processes as occurring during the other seasons.

[31] To assess the first effect, we use the emissions ratios of CO and EC calculated by emission factors for various types of sources as summarized in Table 6. In winter, CO may be emitted additionally by coal or biofuel combustion for domestic heating (see section 3). If the increase in nighttime CO in winter of about 2980 ppbv from the spring-fall value was totally due to combustion for domestic heating, it should have led to an additional increase in EC of about 6.0–176 μgC m−3 for coal combustion and 45–83 μgC m−3 for biofuel combustion above the levels in spring-fall, according to the individual EC/CO emission ratios (Table 6). The predicted increase in EC is much larger than the observed increase in EC of 2.9 μgC m−3.

[32] For further analysis of the second effect, the ΔCO/ΔCO2 ratios during the nighttime (2000–0600 LT) are plotted versus atmospheric temperature for winter, summer, and fall in Figure 13a. The ΔCO/ΔCO2 ratio was highest at temperatures below 0°C. It decreased with increasing temperature. These results suggest that the CO emissions are temperature dependent. We showed the large contributions of vehicles to CO emissions in Beijing in this section. The emissions of CO from the engines of gasoline vehicles are removed by catalysts. About 90% of passenger vehicles in Beijing are equipped with three-way catalysts (http://www. issrc.org/ive). Most CO is emitted during the warm-up phase of vehicle, when the temperature of the catalyst is not sufficiently high, e.g., below 300°C, the so-called “cold start” period [Feest and Blaikley, 2001; Andrews et al., 2004; Weilenmann et al., 2005]. The duration of this period and therefore the total CO emissions during the cold start period increase at lower ambient temperature. CO emissions from diesel vehicles during the cold start period are about ten times lower than those from gasoline vehicles. CO2 emissions should not increase significantly even at low temperature because CO2 is not removed by the catalysts at any temperature. Thus it is likely that the increases in CO emissions from gasoline vehicles at low ambient temperatures are the main cause of the increase in CO and ΔCO/ΔCO2 ratios in winter.

image

Figure 13. (a) ΔCO/ΔCO2 ratios, (b) ΔEC/ΔCO ratios, and (c) ΔEC/ΔCO2 ratios versus ambient temperature during nighttime (2000–0600 LT).

Download figure to PowerPoint

6.3. EC Correlations With CO2 and CO

[33] The ΔEC/ΔCO2 ratios reached maximum values near midnight except for summer, when uptake of CO2 by vegetation might have significantly decreased the CO2 concentrations (Figure 12b). The ΔEC/ΔCO ratios also showed maximum values during the nighttime, except for winter (Figure 12c). The amplitudes of the diurnal variations of the ΔEC/ΔCO ratios were somewhat larger than those of the ΔEC/ΔCO2 ratios. The ΔEC/ΔCO ratios in fall were very similar to those in summer, somewhat different from the ΔEC/ΔCO2 ratios. The lower slope of the EC-CO correlation in winter is also seen from Figure 11c.

[34] These diurnal variations, together with those of EC, indicate large and regular sources of EC during the nighttime. The ΔEC/ΔCO ratios are considered to be a measure of the fraction of diesel vehicles to all the types of vehicles in the investigation of the ΔEC/ΔCO ratios measured in Tokyo [Kondo et al., 2006] because CO is emitted primarily from gasoline vehicles while EC emissions are dominated by diesel vehicles. Considering the diurnal variation of the fraction of HDVs (Figure 2b), the diurnal variation of the ΔEC/ΔCO ratios in Beijing indicates that diesel vehicles are dominant additional sources of EC during the nighttime.

[35] For a more quantitative discussion, the variations of ΔEC/ΔCO for vehicle exhaust ([ΔEC/ΔCO]vehicle) were calculated using the measured fractions of HDVs and light duty gasoline vehicles (LDGVs = Total − HDVs) and their EC/CO emission ratios:

  • equation image

where FHDV and FLDGV are fractions of HDVs and LDGVs, respectively, and ERHDDV and ERLDGV are emission ratios of HDDVs and LDGVs in Beijing (Table 7).

[36] Figure 14 shows the diurnal variations of the [ΔEC/ΔCO]vehicle ratios together with the observed ΔEC/ΔCO in spring, summer, and fall. The [ΔEC/ΔCO]vehicle ratios were low during daytime (0.004–0.005 μgC m−3/ppbv) and high during nighttime (0.006–0.009 μgC m−3/ppbv), in general agreement with the observed ratios. Considering the small anticipated seasonal variations in the traffic, the large increases in ΔEC/ΔCO during nighttime can be explained by the nighttime increases of the HDV fractions (Figure 2b). These were due to the increase in the traffic of HDDTs in the urban area in Beijing (Figure 2a).

image

Figure 14. Diurnal variations of the [ΔEC/ΔCO]vehicle (filled squares) and observed ΔEC/ΔCO ratios in spring (open triangles), summer (open circles), and fall (open squares). The [ΔEC/ΔCO]vehicle ratio was calculated by equation (1) using the on-road traffic fractions of LDGVs and HDVs (Figure 2). The observed ΔEC/ΔCO ratios with a correlation coefficient (r2) less than 0.4 were excluded.

Download figure to PowerPoint

[37] The low ΔEC/ΔCO ratios in winter (Figure 12c) are due to the enhanced CO emissions, as discussed in section 6.2. The ΔEC/ΔCO ratios during the nighttime are plotted versus ambient temperature in Figure 13b. The ratios increased by a factor of about 2.5 with temperature between −7°C and 23°C. It is likely that the increase in the ΔEC/ΔCO ratios is mainly due to the relative decrease in the CO emissions with increasing temperature.

[38] The ΔEC/ΔCO2 ratios are plotted versus temperature in Figure 13c. They did not show a monotonic increase with increasing temperature. The small temperature dependence indicates no large additional EC sources at cold temperatures with largely different EC/CO2 emission ratios, such as domestic heating by coal and/or biofuel combustion.

6.4. Weekend Effect

[39] Many urbanized regions have shown “weekend effects” that are characterized by lower levels of anthropogenic species on weekends than those on weekdays due to lower anthropogenic activity, including industrial activity and traffic, during the weekend [e.g., Cerveny and Balling, 1998]. Satellite observations have shown that column NO2 over the Beijing area did not show a significant weekend effect [Beirle et al., 2003]. However, there have been no reliable observations of this effect on EC, CO, and CO2 in this area.

[40] Figure 15 shows diurnal variations of EC, CO, and CO2 concentrations on weekdays and weekends (Saturdays, Sundays, and holidays) in Beijing for all seasons except for winter. Figure 16 shows the diurnal variations of the ΔEC/ΔCO ratios on weekdays and weekends for all seasons except for winter. The variations of these species and ratios were similar on weekdays and weekends. These results indicate that the diurnal variations of anthropogenic activity were not significantly different between weekdays and weekends. In fact, the traffic, which is an important source of EC and CO, was similar for weekdays and weekends at least during daytime (see section 3).

image

Figure 15. Diurnal variations of (a) EC, (b) CO, and (c) CO2 for weekdays and weekends under weak wind conditions except during the winter period. The bars show 25th and 75th percentiles.

Download figure to PowerPoint

image

Figure 16. Diurnal variations of the ΔEC/ΔCO ratio for weekdays and weekends under weak wind conditions except for the winter period.

Download figure to PowerPoint

6.5. Comparison With Other Urban Areas

[41] Table 8 compares EC concentrations and the ΔEC/ΔCO ratios in Beijing with those measured in other urban areas in Asia and North America [Baumgardner et al., 2002; Lim and Turpin, 2002; Park et al., 2005, 2007; Kim et al., 2006; Kondo et al., 2006]. The EC concentration in Beijing was highest among the cities listed.

Table 8. Comparison of EC and ΔEC/ΔCO Ratios in Beijing With Other Urban Areas
LocationSampling PeriodEC μgC m−3ΔEC/ΔCO μgC m−3/ppbvMethod of EC DeterminationReferences
  • a

    Median value.

  • b

    TOT, thermal optical transmittance.

  • c

    Mean value.

  • d

    TMO, thermal manganese dioxide oxidation.

Tokyo, Japan2003 to 20051.8a0.0057TOTbKondo et al. [2006]
Seoul, South KoreaDec 19994.5c0.0022TMOdPark et al. [2007]
Incheon, South KoreaJun to Oct 20044.3c TOTbKim et al. [2006]
Baltimore, USAFeb to Nov 20021.1c0.0027TOTbPark et al. [2005]
Atlanta, USAAug to Sep 19992.3c0.0032TOTbLim and Turpin [2002]
Mexico City, MexicoJan to Feb 2000 0.0013light absorptionBaumgardner et al. [2002]
Beijing, ChinaNov 2005 to Oct 20066.9c0.0035–0.0058TOTbthis study

[42] Emissions from vehicles were estimated to be important sources of EC in Beijing and Tokyo. High levels of EC and the ΔEC/ΔCO ratio were observed in the morning and sometimes during the late evening rush hours in Tokyo [Kondo et al., 2006]. The peaks of EC in the morning and/or evening were also observed in Incheon (Korea) [Kim et al., 2006], Baltimore [Park et al., 2005], Atlanta [Lim and Turpin, 2002], and Mexico City [Baumgardner et al., 2002]. By contrast, the timing of the EC and the ΔEC/ΔCO peaks in Beijing was shifted to midnight. The differences of the timing of the high EC and ΔEC/ΔCO in Tokyo and Beijing were due to the difference of the timing of the maximum diesel traffic in these cities.

[43] We compare the ΔEC/ΔCO ratios in winter between in Beijing and Tokyo. In Tokyo, the ΔEC/ΔCO ratio in the early morning in the winters of 2004 and 2005 were lower by about 20–30% than those in the other seasons [Kondo et al., 2006]. The low ratios were associated with the relatively high CO values. The monthly average temperatures in the early morning of these winters were about 2°–5°C. This is qualitatively consistent with the seasonal variation of the ΔEC/ΔCO ratio in Beijing. In Beijing, the median temperature in winter was lower by about 5°C than that in Tokyo (Figure 6), leading to a more pronounced decrease in the ΔEC/ΔCO ratio than in Tokyo.

[44] The weekend effects of EC are compared in Beijing and the other cities. In Tokyo, the ΔEC/ΔCO ratios for Sundays were lower by about 50% than weekday values, in reasonable agreement with changes in the traffic density of diesel vehicles [Kondo et al., 2006]. During the Atlanta Supersite Experiment in 1999 and the St. Louis Midwest Supersite Experiment in 2002, the mean EC concentrations at these sites were elevated on weekdays, reaching a maximum on Thursday and minimum on Sunday [Lim and Turpin, 2002; Bae et al., 2004]. These features were not observed in Beijing, as discussed in section 6.4.

7. Evaluation of Emission Inventory in Beijing

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Measurements
  5. 3. Characteristics of Sources in Beijing
  6. 4. Temporal Variations of EC, CO, CO2, and Meteorological Parameters
  7. 5. Statistical Analysis
  8. 6. Correlation Analysis
  9. 7. Evaluation of Emission Inventory in Beijing
  10. 8. Summary and Conclusions
  11. Acknowledgments
  12. References
  13. Supporting Information

[45] The slopes of the observed CO-CO2-EC correlations can be used to evaluate the emission ratios derived from emission inventory estimates, although the slopes can change due to mixing with background air after emission [McKeen and Liu, 1993; McKeen et al., 1996; Takegawa et al., 2003]. The validity of the approach is supported by a three-dimensional regional-scale model simulation for EC and CO in the Beijing area from 10 August to 11 September 2006 [Matsui et al., 2009]. They used the Community Multiscale Air Quality (CMAQ) model [Byun and Ching, 1999; Binkowski and Roselle, 2003], driven by the Weather Research and Forecasting (WRF) model [Skamarock et al., 2005], with a horizontal resolution of 9 km for the finest domain, which consisted of 54 × 54 grid cells with the center of 40°N and 116°E (Beijing). The Asian anthropogenic emission inventories with a grid resolution of 0.5° × 0.5° in latitude and longitude developed for the year of 2004 (D. G. Streets et al., unpublished data, 2007) were used in their simulation. According to their results, EC concentrations at the PKU observation site were most strongly influenced by the emissions in the areas within about 50–100 km from the urban center of Beijing, supporting the definition of the Beijing study area (39.5°–40.5°N, 116.0°–117.0°E), discussed in section 3. The EC-CO correlations calculated by the model agreed with the [EC/CO]total ratios derived from the emission inventory within about 20%. Here the [EC/CO]total ratios are defined as the ratios of the sums of their emissions from all sources and are different from the emission ratios for specific sources discussed in section 6.

[46] In this study, we evaluated EC, CO, and CO2 emitted from all sources given in the emission inventory for the year 2000 in the Beijing study area [Streets et al., 2003]. Table 9 summarizes [CO/CO2]total, [EC/CO2]total, and [EC/CO]total ratios derived from the emission inventory for the Beijing study area. We compared the observed ΔCO/ΔCO2, ΔEC/ΔCO2, and ΔEC/ΔCO ratios (Table 5) with the ratios shown in Table 9. For comparison with the emission inventories of Streets et al., we used the ratios of CO and EC to CO2 because the estimate of the CO2 emission is considered to be the most reliable among these species (section 3).

Table 9. Ratios of CO, CO2, and EC Emitted From All Sources Based on the Emission Inventory for the Year 2000 in the Beijing Study Areaa
[CO/CO2]total, ppbv/ppmv[EC/CO2]total, μgC m−3/ppmv[EC/CO]total, μgC m−3/ppbv
47.60.240.0050

[47] The [CO/CO2]total ratio of about 48 ppbv/ppmv (Table 9) is higher by a factor of about 1.5 than the observed ΔCO/ΔCO2 ratios of about 30–33 ppbv/ppmv in summer and fall (Table 5). This result suggests that Streets et al. [2003] overestimated the contributions of the emissions from domestic sectors with high CO/CO2 emission ratios (Table 6) and/or underestimated those from nondomestic sectors with low CO/CO2 emission ratios (Table 6), such as vehicle emissions, in the Beijing study area. In fact, the temperature dependence of ΔCO/ΔCO2 (Figure 13a) indicates dominant contributions of the emissions from gasoline vehicles in the Beijing area.

[48] The [EC/CO2]total ratio of 0.24 μgC m−3/ppmv (Table 9) is higher by factors of 1.3–2.0 than the observed ΔEC/ΔCO2 ratio of 0.12–0.19 μgC m−3/ppmv (Table 5). This result also suggests the overestimation of the contributions of domestic sources with high EC/CO2 emission ratios (Table 6) and underestimation of those of nondomestic sources, especially vehicle emissions with low EC/CO2 ratios (Table 6).

[49] The [EC/CO]total ratio of 0.0050 μgC m−3/ppbv (Table 9) is within the range of observed ratios of 0.0034–0.0058 μgC m−3/ppbv (Table 5), although the uncertainties in the CO and EC emissions are very large (section 3). This comparison does not necessarily support the estimates of the EC and CO emissions, considering their large uncertainties. These results indicate that the observed slopes of the CO-CO2-EC correlations are useful parameters in assessing the reliability of emission inventories.

8. Summary and Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Measurements
  5. 3. Characteristics of Sources in Beijing
  6. 4. Temporal Variations of EC, CO, CO2, and Meteorological Parameters
  7. 5. Statistical Analysis
  8. 6. Correlation Analysis
  9. 7. Evaluation of Emission Inventory in Beijing
  10. 8. Summary and Conclusions
  11. Acknowledgments
  12. References
  13. Supporting Information

[50] Concentrations of EC in PM2.5, CO, and CO2 were measured every hour at the PKU site near the fourth ring road in Beijing, China, between winter 2005 and fall 2006. The traffic of heavy duty diesel trucks (HDDTs) measured on the fourth ring road during nighttime was about 20 times higher than that during daytime.

[51] The similarity in the temporal variations of the measured species at the PKU and Yufa sites indicates they were distributed uniformly over horizontal distances of about 50 km from the PKU site. The annual median concentration of EC was 5.9 μgC m−3 at the PKU site. The EC concentration decreased with near-surface wind speed (WS). The median EC concentration was highest in fall and lowest in winter due to the WS dependence.

[52] The concentrations of CO2, CO, and EC measured under strong wind conditions (WS > 4.0 m s−1) represented their background levels. The background levels were much higher for southerly wind than those for northerly wind, reflecting much higher emissions of these species in the regions south of Beijing.

[53] For weak wind speed (WS ≤ 2.0 m s−1) conditions, EC, CO, and CO2 were well correlated throughout the measurement period due to the colocations of sources. In winter, CO and ΔCO/ΔCO2 ratios in the night were largely higher than those in the other seasons. CO emissions associated with coal or biofuel combustion for domestic heating are not the main cause of the high CO in winter, considering the lack of predicted increases in EC in winter above the levels in the other seasons. Instead, the increase in the CO emissions from vehicles is likely the dominant cause. The time required for catalysts to be heated to temperatures high enough for efficient CO removal is elongated at low temperatures, leading to higher CO emissions. The increase in the ΔEC/ΔCO ratios with ambient temperature supports this mechanism. The similarity in the ΔEC/ΔCO2 between winter and the other seasons indicates no significant additional EC sources in winter.

[54] The EC concentrations, ΔEC/ΔCO2 (except for summer), and ΔEC/ΔCO (except for winter) showed similar diurnal patterns, increasing in the late evening and remaining high until early morning. These diurnal variations indicate regular sources of EC throughout the year. The variations of ΔEC/ΔCO for vehicle exhaust ([ΔEC/ΔCO]vehicle) were calculated using the measured fractions of heavy duty diesel vehicles (HDVs) and light duty gasoline vehicles (LDGVs) and their EC/CO emission ratios (equation (1)). [ΔEC/ΔCO]vehicle during nighttime was higher by about 0.004 μgC m−3/ppbv than that during daytime, similar to the observed increase. The large increase in ΔEC/ΔCO during the nighttime can be explained by the increase of the HDV fraction due to HDV inflow into the urban area during the nighttime.

[55] The diurnal variations of EC, CO, and CO2 concentrations and the ΔEC/ΔCO ratios were similar for weekdays and weekends. This result indicates no significant weekend effects in Beijing.

[56] The observed slopes of the CO-CO2-EC correlations were used to evaluate the ratios of the emissions of these species from the all sources derived from the emission inventory of Streets et al. [2003] in the Beijing area. The comparison suggests that Streets et al. largely overestimated the contributions of the emissions from domestic sectors with high CO/CO2 and EC/CO2 emission ratios and/or underestimated those from nondomestic sectors, especially vehicle emissions, in the Beijing study area. These results indicate that the observed slopes are useful parameters in assessing the reliability of emission inventories.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Measurements
  5. 3. Characteristics of Sources in Beijing
  6. 4. Temporal Variations of EC, CO, CO2, and Meteorological Parameters
  7. 5. Statistical Analysis
  8. 6. Correlation Analysis
  9. 7. Evaluation of Emission Inventory in Beijing
  10. 8. Summary and Conclusions
  11. Acknowledgments
  12. References
  13. Supporting Information

[57] This work was supported by the Ministry of Education, Culture, Sports, Science, and Technology (MEXT), the global environment research fund of the Japanese Ministry of the Environment (B-083), and the Japanese Science and Technology Agency (JST). This study was conducted as a part of the Mega-Cities: Asia Task under the framework of the International Global Atmospheric Chemistry (IGAC) project.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Measurements
  5. 3. Characteristics of Sources in Beijing
  6. 4. Temporal Variations of EC, CO, CO2, and Meteorological Parameters
  7. 5. Statistical Analysis
  8. 6. Correlation Analysis
  9. 7. Evaluation of Emission Inventory in Beijing
  10. 8. Summary and Conclusions
  11. Acknowledgments
  12. References
  13. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Measurements
  5. 3. Characteristics of Sources in Beijing
  6. 4. Temporal Variations of EC, CO, CO2, and Meteorological Parameters
  7. 5. Statistical Analysis
  8. 6. Correlation Analysis
  9. 7. Evaluation of Emission Inventory in Beijing
  10. 8. Summary and Conclusions
  11. Acknowledgments
  12. References
  13. Supporting Information
FilenameFormatSizeDescription
jgrd15603-sup-0001-t01.txtplain text document1KTab-delimited Table 1.
jgrd15603-sup-0002-t02.txtplain text document1KTab-delimited Table 2.
jgrd15603-sup-0003-t03.txtplain text document1KTab-delimited Table 3.
jgrd15603-sup-0004-t04.txtplain text document0KTab-delimited Table 4.
jgrd15603-sup-0005-t05.txtplain text document0KTab-delimited Table 5.
jgrd15603-sup-0006-t06.txtplain text document1KTab-delimited Table 6.
jgrd15603-sup-0007-t07.txtplain text document0KTab-delimited Table 7.
jgrd15603-sup-0008-t08.txtplain text document1KTab-delimited Table 8.
jgrd15603-sup-0009-t09.txtplain text document0KTab-delimited Table 9.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.