Page 8 - 201905
P. 8
JOUR NAL O F E NVI R ONM E NT AL S C I E NC ES 79 (2019 ) 1 – 10 3
(peroxy acetyl nitrate), acetonitrile and methyl chloride (CH 3 Cl)
which are typical tracers of specific emissions were selected as
inputs for the model. PAN is a secondary product and could be
used as the tracer for secondary formation in the PMF model
(Guven and Olaguer, 2011), while ACN and CH 3 Cl are tracers of
biomass burning (Yuan et al., 2010). The detailed description of
the model configuration and data processes can be found in
previous studies (Ling et al., 2017; Yuan et al., 2012). Briefly, the
model was tested with different numbers of factors and the
most optimum source profiles and contributions were deter-
mined based on the correlations between the modeled and
observed data as well as the comparisons of the modeled
profiles with the results from emission-based measurements
and other PMF model simulations (HKEPD, 2015; Wang et al.,
2014; Yuan et al., 2013; Liu et al., 2009).
2. Results and discussion
2.1. Overall results of isoprene, MACR and MVK
The time series of isoprene, MACR and MVK concentrations as
Fig. 1 – Location of the sampling site and its surrounding well as the meteorological parameters during the study period
areas in the Pearl River Delta region (blue dot is the location if are shown in Fig. 2. The overall average mixing ratios of
Heshan district in Jiangmen city). isoprene, MACR and MVK during the entire sampling period
(30 days in total) were 151 ± 17, 91 ± 6 and 79 ± 6 pptv, with
the median values of 60, 59 and 38 pptv, respectively. The
levels of isoprene dropped to an average value of 75 ± 7 pptv
while their concentrations were quantified based on the due to the decrease in temperature from 02 November
calibration curves of standard and zero air samples. The onwards (p < 0.05). In this study, daily variations of MACR
configuration, quality control and assurance, and the analyt- and MVK at the HS presented three patterns: (1) comparable
ical performance of the system have been provided in the mixing ratios were observed for MACR and MVK for ~10%
previous study (Shao et al., 2009). (three sampling days in total) of the total sampling days (p >
0.05) with average mixing ratios of isoprene, MACR and MVK
1.3. Positive matrix factorization (PMF) model being 52 ± 11, 37 ± 4 and 36 ± 6 pptv, respectively; (2) the
mixing ratio of MACR was much higher than MVK on eighteen
The PMF (version 4.0) model was applied to the observed VOC sampling days ([MACR] > [MVK], p < 0.05) with the average
data to explore the sources of MACR and MVK at the HS. mixing ratios of isoprene, MACR and MVK being 147 ± 20,
Although, a total of 58 NMHCs were identified in the present 85 ± 8, and 39 ± 3 pptv, respectively; and (3) higher mixing
study, not all were utilized in PMF due to the fundamental ratios were observed for MVK than MACR on nine sampling
assumption of non-reactivity and/or mass conservation of the days ([MVK] > [MACR], p < 0.05) with average mixing ratios of
model (Zhang et al., 2012). Furthermore, the aim of this study isoprene, MACR and MVK being 193 ± 38, 106 ± 10, 153 ±
was to investigate the source attributions of MACR and MVK 13 pptv, respectively. The second pattern was consistent with
based on the observed NMHC data rather than identifying the those observed at an urban site in Beijing during summer
sources of the NMHCs (He et al., 2018). As such, species that non-haze days and winter haze days (Rao et al., 2016). This
are typical tracers of the sources were selected as the input for indicates that biogenic emissions and photooxidation of
the PMF model (Chen et al., 2013). It should be noted that a isoprene were not important sources of MACR and MVK.
single VOC may be a tracer for several emissions. For example, Furthermore, the higher MACR and MVK levels in Beijing were
isoprene is an important biogenic NMHC in the PRD region. related to variations in gasoline and diesel vehicle emissions
Aromatics, such as benzene, toluene, ethylbenzene, and as indicated by the principal component analysis (Rao et al.,
xylenes, could be emitted from both vehicular exhausts and 2016). The third pattern was consistent with the findings of
solvent usage while alkenes, such as ethene, propene, and previous studies which indicated that the mixing ratios of
butene, are tracers for incomplete combustion (i.e., vehicular MVK were usually higher than those of MACR if photooxida-
emissions) (Guo et al., 2011; Lau et al., 2010; Ho et al., 2009; Liu tion of isoprene was the dominant source of MACR and MVK
et al., 2008). Furthermore, previous studies have demon- (Guo et al., 2012 and references therein). This could be
strated that C 3 –C 4 alkanes are typical tracers of vehicular attributed to the longer lifetime of MVK as well as the higher
exhaust (although they could also be emitted from liquefied production rate of MVK from photooxidation of isoprene in
petroleum gas (LPG)) while C 5 –C 6 alkanes, n-octane, and their contrast to MACR. Nevertheless, the large daily variations of
isomer are mainly related to gasoline vehicle emissions (Liu et MACR and MVK indicated the influence of other sources other
al., 2008). In total, 18 NMHCs, together with MACR, MVK, PAN in addition to photooxidation of isoprene at the HS.