Aerosols from surface emission can be transported upwards through convective mixing in the planetary boundary layer (PBL), which subsequently interact with clouds, serving as important sources to nucleate droplets or ice particles. However, the evolution of aerosol composition during this vertical transport has yet to be explicitly understood. In this study, simultaneous measurements of detailed aerosol compositions were conducted at two sites, namely urban Beijing (50 m above sea level – a.s.l.) and Haituo mountain (1344 m a.s.l.) during wintertime, representing the anthropogenically polluted surface environment and the top of the PBL, respectively. The pollutants from surface emissions were observed to reach the mountain site on daily basis through daytime PBL convective mixing. From the surface to the top of PBL, we found efficient transport or formation of lower-volatility species (black carbon, sulfate, and low-volatile organic aerosol, OA); however, a notable reduction in semivolatile substances, such as the fractions of nitrate and semivolatile OA reduced by 74 % and 76 %, respectively, during the upward transport. This implies that the mass loss of these semivolatile species was driven by the evaporation process, which repartitioned the condensed semivolatile substances to the gas phase when aerosols were transported and exposed to a cleaner environment. In combination with the oxidation processes, these led to an enhanced oxidation state of OA at the top of the PBL compared to surface environment, with an increase of oxygen to carbon atomic ratio by 0.2. Such a reduction in aerosol volatility during vertical transport may be important in modifying its viscosity, nucleation activity, and atmospheric lifetime.
Substances in the atmosphere, present as aerosol and gas phases, are subject to phase transformation during their lifetime (Pankow, 1994, 1987). Gas-to-particle partition processes thermodynamically determine the production of secondary aerosol mass and the constituents of gases through the condensation or evaporation process. The condensation process leads to gas molecular partitioning in the condensed phase, while the evaporation process occurs when aerosols were diluted in an environment with lower concentration (Donahue et al., 2006). Hereby, their physiochemical properties could be modified, such as the condensation results in enlarging aerosol size (Riipinen et al., 2011, 2012) or the production of new particles (Zhang et al., 2004; Kulmala et al., 2013); the evaporation led to a loss of particulate mass (May et al., 2015; Cubison et al., 2011). These factors have important impacts on altering the radiative interactions of aerosols by changing the mass of aerosols and, thus, direct radiative impacts (Tsigaridis et al., 2014; Wang et al., 2014) or by changing the number concentration and the ability of cloud condensation nuclei for the indirect radiative impacts (D'Andrea et al., 2013; Kuang et al., 2009). The repartitioned gases from aerosols during dilution could experience chemical evolution and further contribute to the modification of aerosol properties (Zhang et al., 2007; Robinson et al., 2007).
In regions with intense anthropogenic activities, such as megacities,
nitrate and sulfate are dominant inorganic aerosol chemical components due
to intensive emissions of gaseous precursors (anthropogenic NO
The vertical transport of aerosol and gases in the planetary boundary layer (PBL) from the surface to the lower free troposphere (FT) importantly determines the influence of anthropogenic emissions to the upper level of the atmosphere; for example, aerosols in the upper level of the PBL may have more important feedback effects in influencing the boundary layer dynamics (Li et al., 2017). The aerosol could be upward transported from surface to upper level through daytime convective mixing in the PBL (Garratt, 1994). Previous studies have used aircraft (Liu et al., 2020b; Zhao et al., 2019) or balloon platforms (Ran et al., 2016; Li et al., 2015) to measure the vertical profiles of PBL in situ; however, these lack sufficient temporal coverage or full chemical components. The evolution of aerosol properties during this vertical transport in the PBL on a daily basis has not been fully understood. In this study, simultaneous and continuous measurements of detailed aerosol compositions were performed at both surface and surface-influenced mountain sites using advanced instrumentation, which provides an opportunity to realize the high time resolution variations at different altitudes. The relative location of the mountain site with respect to the top of the PBL varies with the diurnal variation of PBL height (PBLH), which leaves the mountain site in the free troposphere most of the day and allows it to be influenced by PBL air masses around midday. Through comparing the difference in aerosol chemical compositions between the two sites, we aim to investigate the modification of compositions during the upward transport in the PBL and explore the generic mechanisms in driving the evolution of chemical composition.
Simultaneous measurements of detailed aerosol compositions (organics,
nitrate, sulfate, ammonium, chloride, and black carbon) and gaseous
pollutants (NO
Experimental overview.
Figure 1a shows the locations of surface and mountain sites and the spatial
distribution of mean aerosol optical depth during the observation period.
The surface site locates in the Institute of Atmospheric Physics (IAP), Chinese Academy of Science (39.97
Non-refractory aerosol chemical components, including nitrate (NO3), sulfate
(
Standard AMS data analysis software packages (SQUIRREL 1.59D and PIKA 1.19D)
were used to deconvolve the mass spectrum and obtain mass concentrations of
chemical components. Then, high-resolution mass spectra of OA for
Black carbon (BC) mass was measured by a single particle soot photometer
(SP2; DMT Inc.), following the calibration and data analysis processes by
Liu et al. (2020). Gaseous pollutants (i.e., NO
The three-dimensional air mass histories were calculated by the Numerical Atmospheric-dispersion Modeling Environment (NAME; Jones et al., 2007), which is a Lagrangian dispersion model following 3D trajectories of plume parcels by Monte Carlo methods. The meteorological data source uses the global configuration of UK Met Office's Unified Model. In order to calculate the historical air mass contribution, the model releases tracer particles at a nominal rate of 1 g s
We aim to classify the source influences on the mountain site from the local surface emission or wider regional area. Air mass history analysis showed a pronounced diurnal pattern of local air mass influence (as determined by the NAME dispersion model), peaking around midday (11:00–14:00 local time). The local air mass faction, as calculated from the dispersion model, is only used to indicate the predominant local air mass influence in the midday. The aerosol concentration contained in the air mass depends on the transport efficiency, reaction, and deposition rate of each aerosol type. The fraction of transported aerosols, even for the inert BC may not be quantitatively comparable with the air mass fraction. The maximum local influence was consistent with the most developed PBLH (Fig. 1c). This suggested the strongest influence of surface emissions to the mountain through midday convective mixing (CM; termed, hereafter, as the CM period). For a certain period (9–12 January), westerly air mass continuously influenced the mountain site (gray bar in Fig. 1d–h), which advected regional pollutants from the polluted high plateau, adding on the persistent local emission (termed the regional advection, RA, period). Note that the RA period was also influenced by the convective mixing of surface sources around midday but was combined with additional sources from other regions besides the surface emission. In this study, the statistical results of the RA period include the whole period marked with the gray bar in Fig. 1d–h, and the rest period is used for the statistics of the CM period.
Diurnal variations in key aerosol compositions at both
sites during convective mixing (CM) and regional advection (RA) periods.
Figure 2 gives the statistical diurnal variations in aerosol species during
CM and RA periods, respectively, with species of BC, OA, nitrate, and sulfate
shown in Fig. 2a–d and PMF-derived organic components in Fig. 2e–h. During the CM period, the chemically inert species, such as BC, showed a clear diurnal pattern on the mountain, with concentrations elevated by 82 % (from 0.19 to 0.34
Averaged chemical composition at both sites during the
convective mixing (CM) daytime period (11:00–14:00).
Aerosol chemical compositions showed remarkable differences between both
sites, even when the PBL was fully developed (Fig. 3), i.e., nitrate (23 %) and organics (54 %) dominated at surface, while sulfate (23 %) and organics (45 %) dominated at the top of PBL. Meanwhile, the characteristics of OA varied from POA dominated (59 %) to SOA dominated (64 %). Statistical analysis in Fig. 4 highlights the difference between both sites during the periods of CM midday (11:00–14:00), CM night (23:00–02:00), and RA, where CM midday represents the period with the most efficient convective mixing. The matched concentrations of BC and CO between surface and mountain demonstrated the capability of the boundary layer in transporting pollutants upwards in terms of atmospheric dynamics. For other species, such as organics and nitrate, there was still an enhancement on the mountain peaking around midday; however, the loadings were 61 % and 74 % lower than the surface at the same hours (Figs. 2b, c, 4a), which contrasted with the efficient transport of BC. The low-volatile species of sulfate at both sites showed no apparent (or only a broad) diurnal pattern and matched the concentration at midday, consistent with its gas precursor SO
Statistical analysis of chemical components at both sites.
The diurnal patterns of all PMF-resolved organic components on the mountain presented a midday peak feature at different amplitude (Fig. 2e–h), suggesting there was no additional source around the site, and the upward transport processes for these components was affected by various factors. Previous studies found that POA (e.g., HOA and CCOA) has substantial semivolatile materials and presents relatively high volatility (Cao et al., 2018). The comparison between the two sites showed significantly decreased semivolatile species (i.e., HOA, CCOA, and SV-OOA); their concentrations on the mountain were significantly lower than that at surface by 48 %, 24 %, and 76 % in the CM midday, respectively (Fig. 4a–b). The LV-OOA is a typical SOA which predominantly exists in aged air mass (Zhang et al., 2011). In contrast with semivolatile species, the low-volatile LV-OOA on the mountain showed a poorly defined diurnal pattern, and its concentrations in the midday were higher than that at surface by 52 %, which may be partially caused by further oxidation of relative fresh species in vertical transport process.
The results above demonstrated that, during vertical transport, the loss of
particulate masses only occurred for semivolatile substances (nitrate, POA,
and SV-OOA) but not for low-volatile species (BC, LV-OOA, and sulfate) or inert gas (CO). Meanwhile, these losses had occurred in a relatively dry condition without notable wet scavenging (Fig. S7b). Due to few anthropogenic
emissions on the mountain, the concentrations of gaseous precursors, such as
ammonia, nitric acid vapor, etc., should be significantly lower than that
in urban environment. This suggested that the evaporation process may have
played an important role in repartitioning the condensed phase rich in semivolatile species to the gas phase, which occurred when the activity of
semivolatile species in the condensed phase (molar fraction multiplied by
activity coefficient) was higher than the partial vapor pressure (relative
to equilibrium vapor pressure of pure substance under certain temperature; Pankow, 1994, 1987). Given the winter time period of the experiment, the mean
temperature shifted from
This evaporation process tended to occur along the path of vertical transport, with a higher loss rate when there is a larger gradient of concentration between the condensed phase and ambient air (Donahue et al., 2006; Shrivastava et al., 2006; Robinson et al., 2007). Previous studies indicated the dilution could be particularly important for biomass burning emissions (Li et al., 2021). The high concentrations of the condensed phase could be an important contribution to the gaseous precursors and, under certain conditions, could form secondary aerosol. In addition, these low-volatile aerosols transported to the free troposphere may have a longer lifetime and be transported for a longer distance (Liu et al., 2020a). Here we only observed the resultant compositions after being transported to the top of PBL, but at which atmospheric layer this process had mostly occurred remained inconclusive. There may be some production process, e.g., photochemical oxidation around midday, which contributed to some increase in the SV-OOA concentration around midday (Fig. 2g). The overall reduced SV-OOA suggested its net loss, which was when evaporation prevailed in production processes, during upwards transport.
During CM night, pollutants (except O
Characterization of elemental ratios at both surface and
mountain sites.
The Van Krevelen triangle (VK) diagram (Fig. 5a) could indicate the likely
oxidation pathway of OA by investigating different degrees of changes on the
oxygen or hydrogen over the carbon element ratio, depending on the manner of adding functional groups (Van Krevelen, 1950). For example, the replacement of a hydrogen atom with an alcohol/peroxide group (
Differences in the
Difference in mass spectra between the mountain and surface
sites (mountain minus surface) for
For CCOA and LV-OOA,
The increase in the oxidation state could be caused by the evaporation process losing the less oxidized and more volatile species, and the evaporated gases could be further oxidized to partition to a more oxidized phase. The evaporative loss occurred in a relatively short timescale, i.e., a few hours of vertical transport induced by daytime convective mixing of boundary layer, as reflected by the 76 % decrease in semivolatile organic masses and 74 % of nitrate. These losses occurred in a relatively dry condition without notable wet scavenging; therefore, the evaporative loss tended to dominate. For the other period rather than midday on the mountain, there were additional primary sources contributing neither particles nor VOCs; thus, most of the vapors repartitioned from the condensed phase in the midday may stay and be subjected to oxidation at the mountain. These continuous inputs of gases to the top of boundary layers on a daily basis serve as a source of precursors to be oxidized and contribute to an important fraction of highly oxidized SOA.
All of these processes lead to a consistent manner of enhancing the oxidation state of OA at the top of PBL, which may modify the hygroscopicity and viscosity of OA (Koop et al., 2011). Combining with the more efficiently transported and less-volatile species, these processes consistently led to the overall decreased volatility of aerosols at the top of the PBL, where cloud formation is initialized, influencing the activities of both cloud condensation nuclei and ice nuclei. These lower-volatile aerosols could be transported to a longer distance in the free troposphere, thus having a longer lifetime. This also implies that the aerosol characteristics at the surface may not represent those at upper levels, and the evolution during transport should be considered when evaluating the contribution of surface emissions to cloud particle nucleation and their atmospheric lifetime.
All data in this paper are available from the authors upon request (liuquan620@126.com).
The supplement related to this article is available online at:
DD, MH, and HH led and designed the study. QL, DL, YW, KB, WG, PT, DZ, SL, CY, GT, YW, KH, SD, QG, FW, and SK were involved in collecting, processing, and analyzing of surface and mountain data. QL and DL carried out the data analysis and wrote the paper. QL and all authors contributed to the discussions.
The contact author has declared that neither they nor their co-authors have any competing interests.
Publisher’s note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The authors would like to kindly thank the dedicated efforts of many people from IAP and BJWMO at the observation sites. Moreover, we thank the two anonymous reviewers for the helpful comments and discussions.
This research has been supported by the National Key Research and Development Program of China (grant no. 2016YFA0602001), the National Natural Science Foundation of China (grant nos. 41975177, 41875167, 41875044, and 41775138), and the Beijing Municipal Natural Science Foundation (grant nos. 8192021 and 8194065).
This paper was edited by Andreas Petzold and reviewed by two anonymous referees.