Pollution Characteristics of Atmospheric Carbonyls during One Haze Event in Nanning , south China

This study was the first to investigate pollution characteristics of atmospheric carbonyls in one haze event (July 17–22, 2012) in Nanning, south China. It was found that (1) acetaldehyde (29.74 ± 6.06 μg m^(-3)), formaldehyde (11.20 ± 1.72 μg m^(-3)), and acetone (9.18 ± 11.22 μg m^(-3)) were the most abundant in haze days, (2) concentrations and O_3 formation potentials of ambient carbonyls in haze days were significantly higher than those on normal days, and (3) visibility and wind speed in haze days were lower than those on normal days, indicating that haze days represented favorable pollution conditions for carbonyls. Diurnal variations of ambient carbonyls in haze days showed a pattern of two peaks occurring in two traffic rush-hour periods due to positive traffic emissions. Average concentration ratio of formaldehyde/acetaldehyde (C_1/C_2) in haze days (0.39 ± 0.10) was slightly lower than that (0.87 ± 0.23) on normal days, and the ratios in two traffic rush-hour periods were close to those in non rush-hour periods, likely implying that traffic emissions might not be a major source for ambient carbonyls. Correlation among formaldehyde, acetaldehyde, acetone, and total carbonyls was good (R^2 = 0.49-0.85) in haze days and excellent (R2 = 0.80-0.98) on normal days, indicating that the sources of ambient carbonyls in haze days were more complex compared to normal days.


Sampling
The sampling site is located on the roof of a building (round 20m above ground level) in the campus of Guangxi University, which is surrounded by urban roads, residential buildings and business offices in urban center of Nanning (see Fig. 1 in the main manuscript).In this study, haze episodes were characterized by visibility of ≤10km (CMA, 2003;Wang et al., 2006;Tan et al., 2009;Guo et al., 2012a).were also collected for data comparison of haze episode versus normal days.The detail sampling procedure was similar to the descriptions in our previous studies (Duan et al., 2012;Guo et al., 2014a).In this study, meteorological data of visibility, wind speed, relativity humidity, temperature and pressure were collected from http://www.wunderground.com, and sunlight intensity was collected from the local meteorological department in Nanning named Nanning weather station, which were provided for us in time during the sampling periods.
Meanwhile, data of sunlight intensity and weather conditions were also recorded by the staff who engaged in the air sampling on the field.
Briefly, samples were collected by drawing air with a sampling pump (Gast, USA) through DNPH-coated cartridges.A potassium iodide denuder was connected to the upstream of the cartridges to avoid the interference of ambient O 3 .In addition, a breakthrough experiment was carried out by connecting two cartridges in series before sampling, indicating that no carbonyls were detected in the back cartridge when the sampling interval was 2h at a flow rate of 2L/min.Carbonyls were collected at four intervals (including 0700-0900 (in the morning), 1130-1330 (at noon), 1700-1900 (in the late afternoon) and 2200-2400 (in the midnight)) in haze episodes and normal days in summer (July 16-23, 2012), which were collected for 2h via a flow rate of 2L/min.The information of sampling, weather and meteorological conditions (including visibility, wind speed, relativity humidity, temperature, sunlight intensity and pressure) were listed in Table 1S.After sampling, each cartridge was firstly wrapped in a Teflon bag with aluminum foil and then stored in a 4°C refrigerator until analysis.The sampling program of each day included one laboratory blank and one field blank, and total 24 carbonyl samples and 12 blank samples were obtained.

Analysis
The analysis procedure of carbonyl samples was based on the EPA TO-11A method of United States (US EPA, 1996) and our previous studies (Lü et al., 2009;Guo et al., 2014b).
Briefly, all cartridges were eluted with 2mL of ACN into a 2mL volumetric flask and then a 10μL was injected into the HPLC system (HP1100, Agilent, USA) through an auto-sampler.
Relative standard deviation for duplicate analysis was lower than 5%.Method detection limits were 0.05-0.15μg/m 3 .One laboratory blank and one field blank were analyzed and the result in the blank cartridges did not show negative effects on field sampling according to the EPA blank criteria of United States (USEPA, 1996).In this study, carbonyls' statistical analyses was conducted by personal correlation matrix using the SPSS software and described in the reference Lü et al. 2009.

REFERENCES:
CMA (China Meteorological Administration), Beijing (2003) Guo, S. , He, X., Chen, M., Tan, J. and Wang, Y. (2014a).Photochemical Production of  Note: the references in the Table 2S were presented in this Supplementary material and the main manuscript.Blank refers to the data were not available in the relative references.
Note: the references in the Table 3S were presented in this Supplementary material and the main manuscript.

Fig. 1S .
Fig. 1S.Annual trend for number of haze days during 2001-2014 for ambient air in Nanning, south China.
Figure 1S plots annual trend of haze days during 2001-2014 in Nanning, and Figure 2S shows the plot of visibility variation during this sampling in summer of Nanning (July 16-23, 2012).It can be seen that average daily visibility was round 11-16km on July 16-18, and then reached the lower values of 5-8km on July 21-23.In addition, there was a raining event in the two days on July 19-20.Therefore, the days of July 21-23 were regarded as one haze episode, and air samples from July 17-19 wind speed, relativity humidity, temperature and pressure were collected from http: //www.wunderground.com.b Data of sunlight intensity and weather conditions were recorded by the staff who engaged in the air sampling on the field.
Fig. 1S.Annual trend for number of haze days during 2001-2014 for ambient air in Nanning, south China.

Table 3S .
A comparison of C 1 /C 2 concentration ratios during sampling in Nanning with other urban areas.

Table 1S .
The information of sampling time and meteorological conditions during sampling in Nanning, south China.

Table 3S .
A comparison of C 1 /C 2 concentration ratios during sampling in Nanning with other urban areas.