1 Introduction

Black carbon (BC) is an important atmospheric aerosol component and is produced by the incomplete combustion of carbonaceous materials (Bond and Bergstrom 2006). Recent industrial development and rapid increases in world population have led to the consumption of large amounts of fossil fuels, which, together with anthropogenic biomass burning and automobile exhaust emissions, have resulted in a significant increase in the concentration of BC in the atmosphere (Ruppel et al. 2014). BC aerosol can affect the climate by modifying the radiative properties of the atmosphere (Reddy and Boucher 2007). BC may be the third most important component of global warming after CO2 and CH4 (Bond et al. 2013). The optical absorption properties of BC can greatly reduce atmospheric visibility (Moosmüller et al. 2009). BC can also carry toxic substances and heavy metals into the body and induces respiratory and cardiovascular problems (Highwood and Kinnersley 2006; Borm et al. 2004). Because BC can have substantial impacts on global climate change, regional ambient air quality, and human health, BC has become an increasingly important field of atmospheric research in recent years (Bond et al. 2013). In response, numerous counties have established national-wide BC monitoring networks (Kirchstetter et al. 2017; Kutzner et al. 2018; Zhang et al. 2008, 2012). However, most existing BC studies have been ground-based. Modeling studies suggested that substantial uncertainties of BC radiative forcing were arise from limited measurement of BC vertical profiles (Samset et al. 2013; Zarzycki and Bond 2010). As a result, field measurements of BC vertical profiles are useful for reducing the uncertainties of BC radiative forcing in climate models.

In China, ground-based BC observations had been performed in various locations (Zhang et al. 2008; He et al. 2009b; Wu et al. 2009, 2012; Zhuang et al. 2014, 2015; Chen et al. 2016; Ji et al. 2017). Globally, ground-based BC measurements have been made from polar regions (Sharma et al. 2004) to urban areas (Chen et al. 2016). However, measurements of BC vertical distributions remain limited. Wu et al. (2009) conducted parallel BC measurements at the foot and top of a hill using two aethalometers. Aircraft-based measurement is a useful tool for studying BC vertical profiles. For example, Schwarz et al. (2008) conducted aircraft-based BC measurements in Costa Rica. Vaishya et al. (2018) reported vertically resolved light absorption profiles at Indo-Gangetic Plain (IGP) using aircraft measurements and found significant atmospheric warming induced by BC. Wang et al. (2018a) found a decreasing trend of light absorption with height within 0–3 km, and the light absorption vertical profiles were influenced by the evolution of a boundary layer. The influence of boundary layer height on BC vertical profiles was also observed by aircraft measurements in India (Safai et al. 2012) and Nepal (Singh et al. 2019). In the arctic region, BC was found to be more abundant in the upper polar dome (Willis et al. 2019). Aircraft measurements in the Yangtze River Delta found increased organic carbon to elemental carbon ratio with height (Ren et al. 2018). However, the high cost of aircraft measurements limits its applications in field studies. Besides aircraft, balloon sounding was also useful for probing the BC vertical distribution, especially for height below 1000 m, whose range was usually not covered by aircraft measurements due to the takeoff and landing process. Ran et al. (2016) explored the variability of BC vertical profile measurements at the North China Plain using a tethered balloon. (Li et al. 2015) studied BC profile variations in Shanghai. Lu et al. (2019) studied BC vertical profiles in the boundary layer in the Yangtze River Delta. Tethered balloon–based vertical profiling of BC was also been reported in India (Babu et al. 2011; Bisht et al. 2016; Mishra et al. 2018), New Zealand (Trompetter et al. 2013), Italy (Ferrero et al. 2011; Ferrero et al. 2014), and Arctic region (Ferrero et al. 2016; Markowicz et al. 2017). In recent years, unmanned aerial vehicles (UAV) were also applied for BC vertical profiling. UAV-based BC vertical profiling studies had been conducted in the Yellow Sea (Ramana et al. 2010), Indian Ocean (Corrigan et al. 2008), Yangtze River Delta (Li et al. 2018), and Pearl River Delta (Liu et al. 2020). UAV provides more flexibility in field deployments, but UAV is limited in payload capacity and temporal coverage. Nevertheless, vertical BC profile measurements, and especially time-resolved measurements, remain limited. In comparison, the meteorological mast can provide continues measurements at different heights. Xie et al. (2019) studied the light absorption vertical profile in Beijing during winter using the Beijing meteorological mast. Wang et al. (2018b) found that constant BC profiles accounted for 37% of all profiles during winter haze episodes in Beijing. Recent studies have suggested that the interactions between BC and the boundary layer can cause a “dome effect” that leads to air pollutant accumulation (Wendisch et al. 2008; Ding et al. 2016). Considering the limited number of existing studies on BC vertical distributions and the importance of BC vertical profile on evaluating BC’s climate effect, there is a strong need for time-resolved BC vertical distribution measurements in order to better characterize the environmental effects of BC.

Shenzhen, which is located in the southern part of Guangdong Province (in southern China) on the eastern bank of the Pearl River Estuary, is an important international gateway and home to numerous high-tech companies. Shenzhen ranked 11th cleanest of 74 major Chinese cities in terms of PM2.5 in December 2017, with a monthly average PM2.5 concentration of 46 μg m−3 (CNEMC 2018). The air quality improvement of Shenzhen in recent years may be partially attributed to the active pollution control measures implemented by the local government. For example, the number of public electric buses in Shenzhen has surpassed the total number of electric buses in major metropolitan areas in the USA, and 90% taxis in Shenzhen have been electrified (Keegan 2018). As a coastal city, Shenzhen is an ideal location to examine the characteristics of BC from local vehicular emissions. Thus, this study aims to characterize the temporal dynamics of BC vertical distributions near the surface layer in the highly urbanized areas. Field measurements were conducted at the 356-m meteorological tower in Shiyan, Shenzhen, on December 6–15, 2017. Equivalent BC (eBC) concentrations were measured at different heights, and the vertical distributions were analyzed in combination with meteorological data.

2 Field measurements

2.1 Observation site

The observation site location is shown in Fig. 1. The Shenzhen meteorological tower (SZMT) is located in a protected area in the Tiegang Reservoir, Bao’an District, Shenzhen, and surrounded by orchards. Although the area within the reservoir was covered by vegetation, the areas beyond the reservoir had been fully urbanized as shown in Fig. 1. An expressway is located near the site, and expressway vehicle emissions are one of the sources affecting the observations. Open burning of straw and trash was also observed within the reservoir area during the observation period occasionally, which may have affected the measurement results. With a height of 356 m, the SZMT is the tallest meteorological mast tower in Asia and the second tallest in the world. The SZMT platform was designed to enable integrated vertical measurements of meteorological and environmental parameters (Li et al. 2020).

Fig. 1
figure 1

Map of the sampling site and photo of the 356-m Shenzhen meteorological tower (SZMT)

Full size image

2.2 The advantage of meteorological masts over other tower types

TV towers have difficulties for meteorological and pollutant measurements (Deng et al. 2015), primarily because the surrounding airflow may be disturbed by the solid bulk cross section of the tower structure. In contrast, meteorological masts are much smaller and feature a hollow structure, which is much more conducive to ventilation and minimizes airflow disturbances. In addition, the meteorological mast cross section is identical at all heights, enabling uniform flow through the mast at all altitudes. These features ensure consistent measurement accuracy at different heights. Meteorological masts also have an advantage over other approaches, including unmanned aerial vehicles (UAVs) (Liu et al. 2020; Li et al. 2018; Pikridas et al. 2019), tethered balloons (Ferrero et al. 2011; Li et al. 2015; Ran et al. 2016; Lu et al. 2019), and airships (Rosati et al. 2016) in that they can support continuous time-resolved measurements at different heights (He et al. 2009a). For example, Wu et al. (2015) conducted PM2.5 sampling at four heights (10, 40, 120, and 220 m) using a meteorological mast in Tianjin, finding that mass percentage of SO42−, NO3−, and OC increased with height. The 325-m meteorological tower in Beijing has been used to study the vertical distribution of PM2.5 (Sun et al. 2013; Wang et al. 2018b; Zhou et al. 2018), and the Amazon Tall Tower Observatory (ATTO) has proven useful for studying atmospheric composition and physical properties (Andreae et al. 2015). Finally, long-term studies have been undertaken using the Zotino Tall Tower Observatory (ZOTTO) in Central Siberia (Chi et al. 2013).

2.3 Instrumentation and deployment

Two types of micro aethalometers were used in this study: the MA200 (AethLabs, CA, USA), which provides eBC measurements at five wavelengths, and the AE51 (AethLabs, CA, USA), which reports eBC at 880 nm only. Micro-aethalometers employ the same principle as rack-mount aethalometers (Hansen et al. 1984). Both MA200 and AE51 were operated at a flow rate of 150 ml/min for sampling. The AE51 collects samples on a quartz filter strip, which requires manual replacement once the light attenuation (ATN) value exceeds 100 (unitless).

The MA200 uses a filter cartridge with 15 sampling spots, making it more suitable for multiple-day sampling. The filter advancing threshold for ATN was 100 for MA200 in this study. MA200 equipped with 5 LED light sources, covering the wavelength from 375 to 880 nm. That enables the determination of wavelength dependency of the samples. The meteorological tower has 13 vertically distributed observation platforms (at 10, 20, 40, 50, 80, 100, 150, 160, 200, 250, 300, 320, and 350 m); the observations for this study were performed at ground level (2 m) and 50, 100, 200, and 350 m (Fig. S1). Three AE51 instruments (placed at 2, 50, and 100 m), and two MA200 instruments (placed at 200 and 350 m) were used, and eBC data were obtained at a total of five altitudes. Due to a power supply interruption, eBC data were not available at 350 m on December 12–15. Meteorological parameters such as temperature, relative humidity, wind speed, and wind direction were also measured.

2.4 Data processing

The AE51 and MA200 are capable of acquiring eBC data at high time resolution (1 s), but are subject to optical and electronic noise at lower eBC concentrations. The Optimized Noise-reduction Averaging (ONA) algorithm (Hagler et al. 2011) was therefore applied for signal noise reduction. The ONA treatment results are shown as black lines in Fig. S2. The data logging interval used in this study is 1 min for AE51 and 1 s for MA200. The higher time resolution used in MA200 leads to a noisier signal of the raw data (shown as red dots in Fig. S2). It is worth noting that the negative values of the raw data do not reflect any physical meanings of the eBC concentrations. These are just raw data that need ONA treatment. The data fluctuation was minimized after ONA treatment as shown by the black lines in Fig. S2. After ONA treatment, the data were further corrected for the loading effect following the algorithm proposed by Virkkula et al. (2007). The loading correction equations are shown below:

$$ {\mathrm{BC}}_{\mathrm{corrected}}=\left(1+k\times \mathrm{ATN}\right){\mathrm{BC}}_0 $$
(1)
$$ {k}_i=\frac{1}{{\mathrm{ATN}}_{\left({t}_{i,\mathrm{last}}\right)}}\left(\frac{{\mathrm{BC}}_0\left({t}_{i+1,\mathrm{first}}\right)}{{\mathrm{BC}}_0\left({t}_{i,\mathrm{last}}\right)}\right) $$
(2)

where BC0 is the raw BC concentration reported by the aethalometer; ATN is the light attenuation measured by aethalometer; k is an empirical constant that can be derived from Eq. 2; ti, last is the time of the last measurement data for filter spot i; and ti + 1, first is the time of the first measurement data for the next filter spot. For each spot (sampling cycle), a factor ki is obtained to correct the data according to Eq. 1. As shown by the blue lines in Fig. S2, the divergence of BC before and after filter change was effectively minimized by the correction scheme.

An Igor Pro (WaveMetrics, Inc., Lake Oswego, OR, USA) based program (MA Toolkit) was developed to incorporate both ONA and loading correction for MA200 and AE51 (Fig. S3).

Inter-instrument comparisons were conducted to ensure consistency between the MA200, AE51, and stationary reference instruments. In this study, a rack-mounted AE33 aethalometer (Drinovec et al. 2015) was used as the reference instrument. As shown in Fig. S4, the MA200 and AE51 correlate well with the AE33. As a result, the given slopes were used to correct the AE51 and MA200 data to ensure data consistency in this study. The AE51 had been proved as a suitable tool for ambient measurements in previous evaluation studies (Cheng and Lin 2013; Viana et al. 2015). We conducted extensive field tests on the MA200, and the systematic characterization of MA200 will be discussed in a separate paper in detail (Wu et al. 2020). Another two Igor Pro–based programs, Histbox (Wu et al. 2018) and ScatterPlot (Wu and Yu 2018), were used for data visualization (box plot and scatter plot).

3 Results and discussion

3.1 Characteristics of eBC concentration changes at different heights

Figure 2 shows eBC concentration frequency distributions at five heights during the campaign. The average eBC concentrations at 2, 50, 100, 200, and 350 m are 6.6 ± 3.6, 5.4 ± 3.3, 5.9 ± 2.8, 5.2 ± 1.8, and 4.9 ± 1.4 μg m−3, respectively, showing a decreasing trend with height (Fig. 2f). The mode concentrations of eBC (from 2 to 350 m, respectively) are 4.1 ± 0.5, 3.3 ± 0.3, 4.1 ± 0.4, 4.3 ± 0.3, and 4.4 ± 0.2 μg m−3. eBC features a bimodal distribution at 2 and 50 m, and a unimodal distribution at 100, 200, and 350 m. This implies that episodic events affect the surface levels (2 and 50 m) more than they do the upper levels (100, 200, and 350 m).

Fig. 2
figure 2

eBC concentration frequency distributions at the five sampling heights as shown in a–e. The solid line red represents the lognormal distribution function fitted curve, and N represents the amount of data. Box plot of eBC at different heights is shown in f

Full size image

3.2 Linear relationships between eBC concentrations at different heights

Linear relationships between eBC concentrations at different heights are shown in Fig. 3. It should be noted that considering the measurement uncertainties and the relatively short sampling period, the results shown here should not be interpreted to a high degree of accuracy. The slopes represent the ratios of eBC at 50, 100, 200, and 350 m to eBC at 2 m, which are 0.82, 0.73, 0.35, and 0.20, respectively (Fig. 3). All p values of slopes at different heights were 0.000, suggesting that all slopes are significantly different from zero. The slopes at 50 and 100 m are similar, but those at 200 and 350 m decrease substantially. In contrast, the intercept increased with height, from 0.03 to 3.1. In addition, the relatively larger intercepts at the upper two levels (2.19 at 200 m and 3.10 at 350 m) suggest a large fraction of the concentration variations cannot be explained by the linear relationship. Besides intercept, R2 decreases with height, from 0.56 to 0.22 as shown in Fig. 3. The large intercept and low R2 found at 200 and 350 m imply that the variations of BC at these two heights were less associated with ground-level BC comparing with 50 and 100 m.

Fig. 3
figure 3

Linear relationships between eBC concentrations at different heights

Full size image

3.3 eBC diurnal variations at different heights

eBC diurnal variations are shown at five heights on the SZMT in Figs. 4 and 5. The eBC diurnal variations clearly show bimodal patterns at all five heights, with peaks appearing between 09:00 and 10:00 in the morning and between 19:00 and 21:00 in the evening. Minima occur from 13:00 to 14:00 in the afternoon and from 06:00 to 07:00 in the morning. Vehicle emissions likely drive the elevated eBC near the morning and evening rush hours. The evening peak is higher in magnitude and broader, which may be associated with traffic regulations in Shenzhen, where trucks are banned in urban areas from 7:00 to 22:00. Because trucks typically have higher emission factors, this truck ban may contribute to the large evening peak observed in this study. Higher emission combining the generally shadow mixing height during the evening leads to the enrichment of BC near the surface as shown in Fig. 5d. The minimum from 13:00 to 14:00 is associated with the gradual decrease in traffic volume after the morning rush hour and the increase in solar radiation in the afternoon, which causes strong atmospheric turbulence (i.e., mixing) and elevation of the boundary layer (i.e., dilution) (Tiwari et al. 2013). With the full development of the mixing layer, the vertical profiles are straight in the afternoon (Fig. 5c). The magnitude of the morning peak is comparable at all five heights, while the magnitude of the evening peak decreases significantly with height. This may imply that the emission source of the evening peak is closer to the sampling site than the emission source of the morning peak, as an emission source closer to the measurement site is expected to produce a more pronounced eBC vertical gradient.

Fig. 4
figure 4

Diurnal variations in eBC at five heights. Red circles represent hourly averages, the lines inside the boxes indicate the hourly medians, the upper and lower boundaries of the boxes represent the 75th and 25th percentiles, and the whiskers above and below each box represent the 95th and 5th percentiles

Full size image
Fig. 5
figure 5

Temporal variations of eBC vertical profiles

Full size image

3.4 Temporal variations in eBC and PM2.5 at different heights

Temporal variations in eBC and PM2.5 are shown at five heights in Fig. 6. The eBC and PM2.5 trends are similar at all five heights during most of the sampling period. The eBC-to-PM2.5 ratio from ground to 350 m is about 15%, 8%, 11%, 9%, and 10%, respectively. The eBC/PM2.5 mass ratio could potentially be an indicator for PM2.5 composition and sources. For the same height, occasionally the eBC/PM2.5 ratio could be lower (e.g., December 9, 2017, at ground level), which implies the increase of non-BC fraction in PM2.5 (e.g., secondary aerosol formation). The occasional increase or spikes of the eBC/PM2.5 ratio (e.g. December 10, 2017, 2 and 50 m) may suggest the passing by of a primary emission plume that is rich in BC-containing contents (e.g., individual high emission vehicles or combustion events). The differences in eBC/PM2.5 between different heights can also provide some qualitative source information. For example, eBC/PM2.5 is highest at 2 m, in agreement with the fact that most BC particles are emitted at the ground level.

Fig. 6
figure 6

Temporal variations in eBC and PM2.5 at five heights. PM2.5 (shaded areas) is plotted on the right y-axis, while eBC (solid lines) is plotted on the left y-axis

Full size image

3.5 Effect of meteorological factors on eBC temporal variations

On December 6, the region was experiencing the final stages of a cold episode. December 7–11 featured a cold high-pressure degeneration process, during which a stagnant zone formed near the ground, causing heavy pollution; December 12–15 featured southward movement of the cold air mass. Temporal variations in temperature (T), relative humidity (RH), solar radiation (R), and eBC are shown in Fig. 7. Solar radiation drives both T and RH; T increases as solar radiation increases, leading to decreased RH in the afternoon. Both T and the amplitude of the variations in RH decrease with increasing height. Under sufficient solar radiation (December 6–11), decreases in T lead to more variable RH diurnal patterns; however, under cloudy conditions, when solar radiation is blocked (December 12–15), the diurnal variations in RH are much less pronounced at all heights. It can be seen from Fig. 7 that the trend of eBC concentration is similar to the trend of RH, which is generally opposite to the trend of T.

Fig. 7
figure 7

Temporal variations in temperature (T), relative humidity (RH), solar radiation (R) and eBC at five heights. Temperature (solid lines) are plotted on the left axis; RH (shaded areas) and eBC (solid lines) are plotted on the right axis

Full size image

Figure 8 shows the effect of wind speed on eBC concentrations at five altitudes. Wind speed increases significantly with height, measuring two-fold higher at 350 m than at ground level. High wind speeds likely contribute to the decrease in eBC with height (Bibi et al. 2017). The eBC episode during December 9–10 was likely driven primarily by the low wind speed during this period, confirming the key role of wind speed, among other meteorological factors, in the observed eBC concentrations.

Fig. 8
figure 8

Temporal variations in eBC and wind speed at five heights. eBC (solid lines) is plotted on the left axis. Wind speed (WS; shaded areas) is plotted on the right axis

Full size image

3.6 Influence of wind direction on eBC concentrations

Figure 9 shows eBC wind-rose plots that were generated using ZeFir (Petit et al. 2017) for the five sampling heights. The results suggest that high eBC concentrations come from different directions at different heights. At 2 m, high eBC concentrations arise mainly from the west, but occur relatively infrequently since northwest winds are dominant. At 50 m, high eBC episodes arise mainly from the southwest. At 100 m, elevated eBC is associated with both southerly and westerly winds. At 100, 200, and 350 m, northeasterly winds are dominant; however, high eBC events arise from different directions at different heights, confirming that local primary emission plumes contribute to these events.

Fig. 9
figure 9

eBC wind-rose plots for the five sampling heights. The left column shows the distributions of eBC concentration as a function of wind speed and wind direction at the five heights, where the color represents the eBC concentration. The right column shows the probability of wind arising from the given wind direction, where the color represents probability

Full size image

3.7 Backward trajectory cluster analysis

Backward trajectory cluster analysis has proven useful for exploring relationships between air pollutants and their air mass origins (Stein et al. 2015). In this study, 72-h backward trajectories were calculated on an hourly basis in Metinfo (Wang 2014) using GDAS data from the National Oceanic and Atmospheric Administration (NOAA). The resulting backward trajectories were then subjected to cluster analysis, producing four clusters (C1–C4).

Figure 10 shows the 72-h backward trajectory clusters and corresponding eBC concentrations at 100 m. During the sampling period, the region was affected by a northeastern monsoon and thus dominated by air masses transported from the northeast. The fractional contributions of the four clusters are, in descending order: C4 (45%) > C3 (22%) > C1 (20%) > C2 (13%). The highest eBC concentrations are found in C3 (7.8 ± 3.4 μg m−3), followed by C1 (5.8 ± 1.5 μg m−3), C4 (5.6 ± 1.6 μg m−3), and C2 (5.6 ± 2.5 μg m−3). C3 originates from inland central China, and C1 and C4 possess shorter paths than do C2 and C3, implying lower wind speeds during air mass transport. The results identify central China as an important long-range contributor to eBC levels in Shenzhen, which consisted with previous research on eBC in Shenzhen (Cheng et al. 2018).

Fig. 10
figure 10

Seventy-two-hour backward trajectory clusters at a height of 100 m. The box plot shows eBC concentrations for the different clusters

Full size image

3.8 Absorption Ångström exponent variations

The absorption Ångström exponent (AAE) can be used to quantify the wavelength-dependent absorption characteristics of BC aerosols. AAE is calculated using light absorption at two wavelengths via (Moosmüller et al. 2011):

$$ \mathrm{AAE}\left({\lambda}_1,{\lambda}_2\right)=-\frac{\ln \left({\sigma}_{abs,\lambda 1}\right)-\ln \left({\sigma}_{abs,\lambda 2}\right)}{\ln \left({\lambda}_1\right)-\ln \left({\lambda}_2\right)} $$
(1)

In this study, 375 nm and 880 nm were used to determine AAE375–880. Due to limited multi-wavelength data availability at 200 m, the AAE375–880 at 350 m was used in the diurnal variation analysis.

Diurnal variations in AAE375–880 at 350 m are shown in Fig. 11, in which the hourly average values range from 1.2 to 1.5. The AAE375–880 diurnal maximum occurs at 16:00, while the diurnal minimum is found at 23:00. The local AAE375–880 minimum at 10:00 may arise from an increase in freshly emitted BC particles during the morning rush hour (with a 1–2-h delay due to the required vertical dispersion). The decrease in AAE375–880 in the evening may be associated with a combination of evening rush hour and truck regulation (trucks are banned from urban areas from 7:00 to 22:00). The increase in AAE375–880 in the afternoon may be associated with the aging of BC particles (Lack and Cappa 2010). The AAE375–880 at 350 m (1.31) is slightly higher than that at 200 m (1.26), as shown in Fig. S5. Our hypothesis for the increase in AAE375–880 with height was associated with particles aging during vertical transport. Since the difference is relatively small (1.31 vs. 1.26), more evidence is needed to verify this hypothesis in future study.

Fig. 11
figure 11

Diurnal variations in AAE375–880 measured at 350 m. The red circles represent the hourly averages, the lines inside the boxes indicate the hourly medians, the upper and lower boundaries of the boxes represent the 75th and the 25th percentiles, and the whiskers above and below each box represent the 95th and 5th percentiles

Full size image

4 Conclusions

In this study, eBC variations were measured at five altitudes (2, 50, 100, 200, and 350 m) using the 356-m Shenzhen meteorological tower (SZMT). During the observation period, the average eBC concentrations were, from 2 to 350 m, 6.6 ± 3.6, 5.4 ± 3.3, 5.9 ± 2.8, 5.2 ± 1.8, and 4.9 ± 1.4 μg m−3, respectively. The temporal variations in eBC at different heights were well correlated, and the eBC concentrations decreased with height; the ground-level eBC concentration was 1.2 times that at 50 m, 1.4 times that at 100 m, 2.8 times that at 200 m, and 4.9 times that at 350 m. At all five heights, the eBC diurnal variations featured a bimodal pattern, with peaks appearing between 09:00 and 10:00 in the morning and between 19:00 and 21:00 in the evening. The magnitude of the evening peak decreased with height.

The eBC-to-PM2.5 ratio was relatively stable throughout the campaign. eBC episodes were largely driven by low wind speed, and, considering the increase in wind speed with height, wind speed likely played a key role in the observed eBC concentrations, among other meteorological factors. eBC wind-rose analysis suggested that high eBC events arose from different directions at different heights, confirming the contribution of local primary emission plumes. In the backward trajectory cluster analysis, air masses from central China exhibited much higher eBC levels than the remaining three backward trajectory clusters. The AAE375–880 at 350 m showed clear diurnal variations, which may have been associated with both eBC aging and varying fractional contributions of freshly emitted eBC. A slightly increasing trend in AAE375–880 with height was observed.

Although meteorological towers enable time-resolved vertical measurements, the vertical resolution is relatively coarse (e.g., five heights in this study) and is limited by the availability of observation decks of a tower. In contrast, battery-powered UAV can provide higher vertical resolution (e.g., 1~10 m) and better vertical coverage (e.g., ~ 1 km). However, current battery technology limits the UAV flight time of a single flight (less than 1 h), leading to poor temporal coverage. Incorporating these two complementary approaches for co-located measurements in the future study can provide more insights into the characterization of BC vertical profiles.