In order to study the diurnal variations of VOC species, VOCs were intensively measured (eight samples per day) in different areas during 2009. VOC samples were also collected at the Xujiahui measurement site for three hours (from 6:00 to 9:00) from July 2006 to February 2010 using a 6 L Silonite canister equipped with a valve (model 29-10622, Entech Instruments Inc., USA). The internal Silonite coating improves long-term VOC storage, and the canisters have a large volume to provide detection of volatile chemicals down to low pptv range. These canisters meet or exceed the technical specifications of the US EPA method. The automatic VOC measurement system used in this study is shown in Figure 1.

Gas samples were pre-processed using a Model 7100 VOC pre-concentrator (Entech Instruments Inc., USA) and analyzed by a gas chromatography system (Agilent GC6890) equipped with a mass-selective detector (Agilent MSD5975N) with the capillary of 60 m length, 0.32 mm diameter, and a film thickness of 1.0 μm. The programmed temperature profile was used with helium as carrier gas and a flow rate of 1.5 mL·min⁻¹. The column temperature was controlled at an initial temperature of −50 °C for 3 minutes, increasing it to 170 °C at the rate of 4 °C·min⁻¹, and then switching to 220 °C at the rate of 14 °C·min⁻¹. VOC calibration standard samples were prepared by diluting 1.0 ppm standard gas mixtures with pure nitrogen gas with an Entech 4600 Dynamic Diluter. The relative standard deviation (RSD) of the relative response factor (RRF) for most VOCs ranges from 1.5 to 11.6%. The relative error for nine measurements (accuracy) ranged from 3.7 to 19%, and the precision for seven parallel samples ranged from 1.8 to 13.6%.

The three sampling sites (yellow stars in Figure 3) are located in the Xujiahui, Pudong, and Baoshan districts in the center, east and north areas of Shanghai, respectively. The Xujiahui sampling site is located in a busy commercial area. The Pudong sampling site is located in an urban administrative and office area, and the Baoshan sampling site is located in a large steel industry area. In order to gain insight into the influence of industrial factories on the three sampling sites, the distribution of industrial sites (smelters, steel, chemical factories, and coal burning power plants) in the surrounding area of Shanghai is shown in Figure 3, where it can be seen that the large smelters and steel factories are mainly located in the Baoshan and Jiading districts, which are in the north and northwest of Shanghai, the main chemical industry complex is mainly located in the west, southwest, and south regions of Shanghai, and the coal burning power plants are mainly located in the north, middle and south regions of Shanghai. Thus, wind directions become a crucial factor in controlling the VOC concentrations measured at the sampling sites, as will be discussed in Section 3.2.

In this study, the different VOC compounds related to ozone formation were also studied according to the propylene-equivalent concentration and the MIR method. The propylene-equivalent concentration method was proposed by Chameides et al. [11], and the following equation is used to calculate the propylene-equivalent concentration for each individual VOC: (1) Propy − equiv ( i ) = conc ( i ) × k OH ( i ) / k OH ( C 3 H 6 )where Propy-equiv(i) is defined as a VOC compound i on an OH reactivity-based scale, which is normalized to the reactivity of propylene; conc(i) is the concentration of a VOC compound i; and k_OH(i) is the rate constant for the reactivity of VOC compound i with OH radical; and k_OH(C₃H₆) is the rate constant for the reaction of C₃H₆ with OH radical. The rate constants were given by Atkinson and Arey [15].

The MIR, proposed by Carter, is a good indicator for deteriming the ozone formation potential of each individual VOC species. The scenarios of this method represent conditions when ozone formation is under VOC limited conditions as suggested by several studies in the Shanghai region [5,6,8,10]. The MIR method is defined by the following equation: (2) OFP ( i ) = conc ( i ) × MIR coefficient ( i )where OFP(i) is defined as the ozone formation potential of individual hydrocarbon i and MIR_coefficient(i) the maximum incremental reactivity coefficient of compound i, which is defined by Carter [14].

Table 1 summarized the measured VOC concentrations from August 2006 to February 2010 at the Xujiahui sampling site. The result indicates that the average total VOC concentration is about 41 ppbv, in which alkanes, alkenes, aromatics, and halohydrocarbons account for 36.5, 5.5, 24.9, and 22.5% of the total VOCs, respectively (see Figure 4). Other VOC species (including alcohol, ester, and ether) accounted for about 10.6% of the total VOC concentration. For individual species, toluene and propane have the highest concentrations, with average values of 4.62 and 4.56 ppbv, respectively.

Figure 5a shows the comparison of ambient VOC concentrations measured at different sampling sites (Xujiahui and Pudong) in Shanghai during 2009. The measured VOC concentrations at the two sampling sites are listed in Table 2. The result shows that the total VOC concentrations measured at the Xujiahui site (28.9 ppbv) were slightly higher than those measured at the Pudong site (24.4 ppbv). For different VOC species, the concentrations of alkanes, aromatics, and halohydrocarbons measured at the Xujiahui site were higher than those measured at the Pudong site. However, the concentrations of alkenes measured at Xujiahui site were lower than at the Pudong site. The detailed compositions of VOC concentrations at two sampling sites are illustrated in Figure 5b. The resulta indicate that alkanes, aromatics and halohydrocarbons were the dominant VOC groups at the two sampling sites.

The Baoshan sampling site is located in a large steel industry area in Shanghai (Figure 3). At the Baoshan site as well as the Xujiahui site, diurnal variation of VOCs was sampled (eight samples per day with a three hour interval) during May 2009. The resulting comparison of ambient VOC concentrations is shown in Figure 6a.

The measured VOC concentrations at the two sampling sites are listed in Table 3. The concentrations of alkanes, alkenes and toluene measured at the Baoshan site are considerably higher than those at the Pudong site, indicating that alkane, alkene and toluene emissions might be associated with steel production. This result is consistent with the measurements by Liu et al. [16], who reported that toluene concentrations are high in steel production areas. In addition, a high 1,2-dichloroethane concentration was detected which could be emitted from coal burning due to the large consumption of coal in Shanghai [9]. Otherwise, the compositions of VOCs measured in the two sampling sites are very different (see Figure 6b). The percentage of alkanes and alkenes are much higher in the Baoshan site (35.3% for alkanes and 6% for alkenes) than in the Pudong site (26.3% for alkanes and 3.1% for alkenes).

The general characteristics of the measured VOC concentrations in Shanghai were compared to similar measurements in different large cities (Table 4). The table shows the mean concentrations of VOCs measured in other important Asian cities (Seoul and Nagoya) and other metropolis in China (Beijing, Guangzhou, and Hong Kong) [17–21]. The comparison shows that the concentrations of VOCs in Shanghai are similar to those measured in Nagoya, but are much lower than those measured in Beijing, Guangzhou, Hong Kong and Seoul.

The seasonal variations of VOC concentrations and monthly averaged precipitation at the Xujiahui site are shown in Figure 7a. The result suggests that the VOC concentrations were lower when the precipitation was higher, except in summer. During summer, the averaged VOC concentration in July (60 ppbv) was higher than in June (43 ppbv) and August (36 ppbv). The wind directions in different months (Figure 8) show that the prevailing wind direction might have an important impact on the measured VOC concentrations. Figure 8 shows that in June and August, the prevailing wind directions were east and southeast. In this case, there were no major pollutant sources which have significant effect on the measured VOC concentration in the sampling sites. However, in July, the prevailing wind directions were south and southwest, and many major large industrial areas could have important effects on the measured VOC concentrations at the sampling sites.

Among the four VOC groups, alkanes were the dominant species throughout the year, except for February, and the seasonal variation was similar to that of total VOCs [i.e., higher in July (21 ppbv) and lower in March (10 ppbv)]. The seasonal variation of aromatics and alkenes was also similar to that of total VOCs. However, the seasonal variation of halohydrocarbons was different from other VOC groups, indicating that the sources of halohydrocarbons must be different from those of other VOCs.

Figure 7b also indicates that the propylene-equivalent concentrations of alkanes were lower than for aromatics and alkenes throughout the year. The propylene-equivalent concentration of alkenes reached its peak value in July (5.10 ppbv). The seasonal variation of alkenes was similar to that of aromatics. In this study, the MIR method was used to evaluate the contributions of VOC groups to ozone formation (see Figure 9). The result shows that aromatics play important roles in ozone formation, contributing to ozone formation by about 57%. This result suggests that even though the measured concentrations of alkanes were the highest, their contributions to ozone formation (20%) were less than those of aromatics, due to the lower OH reactivity.

The diurnal variations of ambient VOC concentrations from 25 August to 16 September are shown in Figure 10. The result suggests that the VOC concentrations are higher on weekdays (Monday to Friday) than in weekends (Saturday and Sunday), indicating that human activities have significant impact on the VOC emissions in Shanghai. The diurnal cycles of alkanes, alkenes and aromatics had a double-peak pattern on weekdays, especially during high traffic time (07:00–19:00) (shown in Figure 10). The two peaks occurred around 09:00 and 15:00–18:00, and were correlated to the morning and afternoon rush hours. In addition, in weekends, no clear double-peak patterns were indicated, which might due to the change in traffic pattern during the weekends. However, the diurnal variation pattern of halohydrocarbones was different with other VOCs, indicating that the different sources of halohydrocarbones.

There are two possible sources for isoprene. One is production by biogenic sources, which is normally considered as the major source of isoprene. Another possible source is from anthropogenic activity, especially in large cities [22]. Because biogenic isoprene emissions are very weak in winter, the concentration of isoprene from biogenic sources should be much lower in winter than in summer [23,24]. However, the measured isoprene concentrations in the Xujiahui site during 06:00–09:00 had a small variation in summer (0.15 ± 0.11 ppbv) and in winter (0.13 ± 0.17 ppbv). In addition, there is very little vegetation cover near the sampling site, consequently producing small amounts of the local isoprene emissions. Furthermore, the correlation coefficients among isoprene, propene and isopentane were calculated (see Figure 11). Because isopentane is a typical tracer for gasoline evaporation, the high correlation (R² = 0.61) between isoprene and isopentane suggests that isoprene was mainly related to industrial activities near the sampling site.