Seasonal and Diurnal Characteristics of the Vertical Profile of Aerosol Optical Properties in Urban Beijing, 2017–2021

Abstract

Seasonal and diurnal characteristics of the vertical profiles of aerosol properties are essential for detecting the regional transport and the climatic radiative effects of aerosol particles. We have studied the seasonal and diurnal characteristics of the vertical distribution of aerosols in urban Beijing from 2017 to 2021 based on long-term Raman–Mie LiDAR observations. The influence of the vertical distribution of aerosols, the meteorological conditions within the boundary layer, the optical–radiometric properties of aerosols, and their interconnections, were investigated during a heavy haze pollution event in Beijing from 8 to 15 February 2020 using both meteorological and sun photometer data. The aerosol extinction coefficient was highest in summer (0.4 km⁻¹), followed by winter (0.35 km⁻¹), and roughly equal in spring and autumn (0.3 km⁻¹). The aerosol extinction coefficient showed clear daily variations and was different in different seasons as a result of the variation in the height of the boundary layer. During the haze pollution event, the particulate matter mainly consisted of scattered spherical fine particles and the accumulation time of pollutants measured via the AOD_440nm and PM_2.5 mass concentration was different as a result of the hygroscopic growth of the aerosol particles. This growth increased scattering and led to an increase in the aerosol optical depth. The vertical transport of particulate matter also contributed to the increase in the aerosol optical depth.

1. Introduction

“Aerosol” is a general term used to describe suspended solid and liquid particles in the atmosphere. The multi-phase system consisting of aerosols and the atmosphere, with diameters ranging from 0.001 to 100 μm, is referred to as atmospheric aerosols [1,2]. Aerosols form haze pollution, which affects both visibility and human health, after accumulating to a certain concentration under certain meteorological conditions [3,4,5,6,7,8,9]. Aerosols can affect the radiation balance of the Earth–atmosphere system through direct effects and can also indirectly affect the regional and global climate by influencing the microphysical properties and life cycles of clouds [10,11,12,13,14,15,16]. Aerosols have a crucial role in climate change at both regional and global scales, but there is considerable uncertainty in the assessment of the strength and importance of their effects [17,18,19,20]. The study of the optical–radiative properties of aerosols is one of the key solutions to reducing the uncertainties in aerosol radiative forcing.

The radiative power and effects of aerosols depend on their physical and optical properties and therefore studying these properties provides important information for predicting regional and global climate change [21]. Global long-term, ground-based observation networks (e.g., AERONET) and China’s CARSNET (China Aerosol Remote Sensing Network) have provided an observational basis for understanding changes in the spatial distribution of aerosols and have improved our understanding of the impact of aerosols on regional and global climate change [11,14,22]. However, there have been few studies of the vertical distribution of aerosols [23]. The vertical distribution of aerosols can change the vertical variation of the aerosol radiative heating rate and affect atmospheric stability, which, in turn, affects the generation and development of turbulence and clouds [24,25,26,27]. The vertical distribution of aerosols can also indicate the level of pollution and the evolution of pollution processes in a defined area [28,29]. Studies of the vertical distribution of aerosols are therefore important in improving boundary layer schemes, analyses of the mechanism of pollution processes, determining the boundary layer height and in predicting air quality [30,31].

The rapid development of China’s economy in recent years has led to much research on the causes of aerosol pollution and the optical properties of aerosols, especially in urban areas where high aerosol loads lead to reduced visibility, reduced solar radiation and frequent pollution events [32,33,34,35,36,37,38]. The meteorological conditions have a crucial role in ambient air pollution by both directly and indirectly influencing the emission, transport, formation and deposition of air pollutants [39,40,41]. Previous studies have found that the response of air quality to single or multiple meteorological factors varies by both city and season, with significant north–south differences in the response to humidity and wind speed [41,42,43,44,45]. Detailed information about the meteorological conditions within the boundary layer is therefore needed to investigate the influence of meteorological conditions within this layer on pollution events. This will help our understanding of the causes of pollution and the development of pollution prevention policies [46].

The Beijing–Tianjin–Hebei region is an important economic zone in northern China. Extreme haze events have frequently broken out in this region in recent years, causing great public concern [47,48,49]. Heavy pollution events (HPEs) were studied in the Beijing area during autumn and winter in 2017–2020 and the HPEs were divided into an early transport stage and a cumulating stage [49,50,51,52,53,54,55]. Southerly winds prevailed in the Beijing area during the transport stage and pollutants from south of Beijing were transported to the city. The solar radiation reaching the near-ground layer was reduced in the cumulating phase as a result of the extinction effect of particulate matter. This generated or enhanced the inversion layer and increased the relative humidity, which, in turn, inhibited turbulence, decreased the height of the boundary layer and inhibited the diffusion of pollutants. The higher relative humidity also promoted the hygroscopic growth of aerosols, accelerated liquid-phase and non-homogeneous-phase reactions, and increased the concentration of pollutants. This, in turn, led to a further deterioration in the meteorological conditions within the boundary layer. This two-way feedback mechanism has an important role in the formation and maintenance of heavy pollution in Beijing.

Most of the rapid increase in pollutants in these studies started from around 12:00 to 17:00 [49,50,51,52,53,54,55]. However, most studies have been based on overall measurements of a vertical atmospheric column without considering the vertical distribution of aerosols. There have been relatively few studies on the optical properties of aerosols and their influence on the development of pollution on a vertical scale. We therefore need detailed information on the vertical distribution and optical properties of aerosols to further investigate their interconnection with, and influence on, the meteorological conditions in the boundary layer. This will help us to improve and validate numerical models and satellite data [23].

We obtained vertical profiles of the aerosol extinction coefficient from long-term observations at the Raman–Mie LiDAR system located at the Chinese Academy of Meteorological Sciences (CAMS) from September 2017 to February 2021. We used these profiles to represent the vertical distribution of aerosols and studied the temporal variation of the vertical distribution of aerosols in urban Beijing. We then explored the interconnections between the vertical distribution of aerosols, their optical properties and the meteorological conditions within the boundary layer during HPEs in Beijing from 8 to 15 February 2020. These results not only enrich our understanding of the regional climate and the influence of local pollution events, but also contribute to validating numerical simulation results and improving LiDAR algorithms.

The remainder of this paper is organized as follows. Section 2 describes the observational sites, data and methods. Section 3 presents the climatological characteristics of the vertical distribution of aerosols and the interconnections between the vertical distribution of aerosols, their optical properties and the meteorological conditions within the boundary layer during HPEs in Beijing. Our detailed conclusions are given in Section 4.

2. Data and Methods

2.1. Sun Photometer and LiDAR Measurements and Site Descriptions

We used the GQB Ⅰ-03 atmospheric water vapor aerosol LiDAR at the CAMS observatory to obtain the vertical profiles of the optical properties of aerosols from September 2017 to February 2021. The instrument consists of six parts: a laser emission unit, a receiving telescope, a spectroscopic unit, a detector, an acquisition card and a control unit. The instrument emits a 532 nm wavelength laser and accepts 532 nm parallel and perpendicular and 607 and 660 nm Raman scattering signals. These signals are used to determine the physical properties of atmospheric aerosols, including the identification of non-spherical particles and the aerosol extinction coefficient, by detecting the scattering of the 532 nm laser using a polarization meter [56]. Raman-Mie LiDAR has a detection blind area of 200 m and a vertical resolution of 7.5 m. The aerosol extinction coefficient is retrieved using the algorithm of Fernald as follows, which considers the atmosphere to consist of air molecules and aerosols.

$$P\left(z\right)=CE\left[{\beta }_a\left(z\right)+{\beta }_m\left(z\right)\right]e^{-2{{\displaystyle \int }}_{z_0}^z\left[{\alpha }_a\left(z\right)+{\alpha }_m\left(z\right)\right]dz}$$

where P is the atmospheric backscattering echo power (W) at height z (km) received by LiDAR, C is the LiDAR system constant $\left(\mathrm{W}·{\mathrm{km}}^3·\mathrm{sr}\right)$, E is the LiDAR launch energy ($\mathsf{\mu }$J), β(z) is the backscattering coefficient$\left({\mathrm{km}}^{-1}·{\mathrm{sr}}^{-1}\right)$ at height z (km), α(z) is the extinction coefficient $\left({\mathrm{km}}^{-1}\right)$, a represents aerosols, and m represents air molecules.

Aerosol particles are not easily diffused vertically because the free atmosphere above the boundary layer is in a relatively static state. The aerosol concentration in the free atmosphere is therefore much lower than that in the boundary layer and the extinction coefficient decreases abruptly at a certain height with a gradient extreme, which we regard as the aerosol boundary layer height. We can obtain the aerosol boundary layer height by inversion of the vertical abrupt rate of change of the extinction coefficient. We use the aerosol boundary layer height to approximate the boundary layer height.

The sun photometer(CE-318, Cimel Electronique, Paris, France) is located on the roof of the CAMS building (at about 106 m above ground level) and was operated in February 2020 to obtain the aerosol optical depth (AOD) and the optical properties of the aerosols. The instrument has an uncertainty in cloud-screened AODs calculated by ASTPwin software(Cimel Electronique) were <0.01. It measures wavelengths of 340, 380, 440, 500, 670, 870, 1020 and 1640 nm and obtains the optical properties of aerosols from sky brightness and direct solar radiation measurements in the 440, 500, 670, 870, 1020 and 1640 nm channels, combined with direct solar AOD detection at the same wavelengths [13,14,22,57,58,59]. These inversion data include the coarse and fine mode aerosol volume concentrations, the Ångström exponent at 440–870 nm, the aerosol size distribution ($dV/dlnr$ in the range 0.05–15 μm), the single scattering albedo (SSA), the absorbing aerosol optical depth (AAOD), the absorbing Ångström exponent (AAE) and the direct aerosol radiative forcing (DARF) [11,56,58,59,60]. The computational model of DARF is as follows:

$$DARF=\left(F^{\downarrow A}-F^{\uparrow A}\right)-\left(F^{\downarrow C}-F^{\uparrow C}\right)$$

$$DAR{F}_{TOA}=\left(F_{TOA}^{\uparrow C}-F_{TOA}^{\uparrow A}\right)$$

$$DAR{F}_{BOA}=\left(F_{BOA}^{\downarrow A}-F_{BOA}^{\downarrow C}\right)$$

$$DAR{F}_{AH}=\left(DAR{F}_{TOA}-DAR{F}_{BOA}\right)$$

Where F^A denotes the net radiative energy in the presence of aerosols, F^C denotes the net radiative energy in the absence of aerosols, and the arrows denote the direction of the global radiative flux, with downward fluxes denoted by $\downarrow$ and upward fluxes denoted by $\uparrow$, DARF_AH is defined as the atmospheric heating term.

2.2. Meteorological and Pollutant Data

We used surface meteorological data from Haidian Station of the China Meteorological Administration, including the temperature, relative humidity, wind direction and wind speed (for location, see Figure 1). We obtained data on pollutants from the National Environmental Monitoring Center of China and used the arithmetic mean of 35 monitoring stations in Beijing to represent the overall air quality in Beijing [61]. Because Wanliu Station is close to the CAMS observatory (about 5 km), we used PM_2.5 data from Wanliu Station to represent the air quality in urban Beijing. The pollutant data included the hourly average data for PM_2.5, SO₂, NO₂ and CO, and the 8-h sliding average data for O₃.

We used daily sounding data at 08:00 and 20:00 from a sounding balloon at Beijing South Suburban Observatory (54511), including the pressure, temperature, temperature dew point difference, wind direction and wind speed. We obtained the convective available potential energy (CAPE) between 1000 and 100 hPa. When the CAPE is negative, it is expressed by the convective inhibition (CIN), which indicates that the lifting of pollutants is inhibited; dispersion will be strongly inhibited when the absolute value of CIN is >100 J kg⁻¹ [62,63].

$$CAPE=R_d{{\displaystyle \int }}_{EL}^{LFC}{T^{\prime }}_v-T_{v}dlnp$$

where R_d is the dry gas constant (R_d = 287 J·kg⁻¹K⁻¹), p is the pressure and ${T^{\prime }}_v-T_v$ is the difference between the virtual temperature of the micro-cluster of lifted air and the virtual temperature of the surrounding atmosphere. LFC is the level of free convection and EL is the equilibrium level [64].

The radar wind profiler deployed at Haidian Park (54399) provides the horizontal wind direction and speed and the vertical velocity below about 3 km above the ground with a vertical resolution of 120 m and a temporal resolution of 6 min, measured from 150 m [65,66]. We explored the interactions between the wind field and changes in the concentration of pollutant during HPEs.

3. Results and Discussion

3.1. Climatological Characteristics of the Vertical Distribution of Aerosols

As an important parameter of the aerosol extinction capacity, the extinction coefficient reflects the aerosol load in the atmosphere, whereas the aerosol vertical profile reflects the vertical distribution of aerosols [67]. Figure 2 shows the characteristics of the inter-month variation of the atmospheric extinction coefficient in urban Beijing from autumn 2017 to winter 2020. The area with the maximum change in the vertical profile of the extinction coefficient in all months is concentrated within 2 km above ground level (monthly average value of the extinction coefficient >0.1 ${\mathrm{km}}^{-1}$). The aerosol extinction coefficient decreases with increasing height, which suggests that the concentration of atmospheric aerosols in urban Beijing is highest near ground level [67].

The extinction coefficient is largest in summer, followed by winter, and is roughly the same in both spring and autumn, with mean values of the extinction coefficient at 300 m being about 0.4, 0.35 and 0.3${ \mathrm{km}}^{-1}$. The aerosol extinction coefficient within 2 km of ground level is highest in summer, especially in July, and the extinction coefficient at 300 m is 0.5${\mathrm{km}}^{-1}$. This suggests that aerosol particles are more likely to absorb moisture and grow under the high temperature and high humidity conditions in summer, increasing the extinction coefficient [68,69,70]. It also suggests that there is a broader distribution of aerosol particles in summer. This may be because the maximum height of the boundary layer in summer can develop to about 2 km and aerosol particles are transported to this altitude via turbulence [71,72]. The extinction coefficient within 0.5 km of the ground is also relatively large in February, October, November and December, with an average value of about 0.4 ${\mathrm{km}}^{-1}$. This is because buildings in Beijing are heated during autumn and winter, leading to higher emissions of pollutants, more frequent pollution events and higher aerosol mass concentrations, resulting in an increase in the extinction coefficient within 0.5 km of the ground surface [25].

In the range 1–2 km above ground level, the extinction coefficient is highest in summer, followed by spring and autumn, and smallest in winter. The extinction coefficient in the range 1–2 km varies between 0.1 and 0.2 ${\mathrm{km}}^{-1}$ in winter because the mean value of the boundary layer height is <1 km in this season [72,73] and aerosol particles are not easily transported to >1 km, resulting in a smaller extinction coefficient for aerosols above 1 km. Aerosols are transported to high altitudes during the frequent dust events in Beijing in spring [74,75,76] and the extinction coefficient is therefore largest at 2–3 km in spring. The mean value of the extinction coefficient is about 0.1 ${\mathrm{km}}^{-1}$ and it is almost unchanged with increasing height.

Figure 3 shows the daily variation of the atmospheric extinction coefficient in urban Beijing from autumn 2017 to winter 2020. In spring, turbulent motion transports aerosol particles to 0.5–1 km during the development of the boundary layer from 07:00 and the aerosol extinction coefficient gradually increases in this height range until reaching a maximum value around noon. The aerosol extinction coefficient within about 0.75 km is >0.3${ \mathrm{km}}^{-1}$. The height of the boundary layer decreases from 12:00 to 23:00 and from 00:00 to 06:00 and the aerosol particles are gradually pressed downward to below 0.5 km. The extinction coefficient of aerosols at 2–3 km altitude in spring is higher from 02:00 to 14:00 than in other seasons as a result of the frequent occurrence of sand and dust pollution in Beijing [74,75,76].

In summer, the boundary layer develops at 07:00 and turbulent motion brings aerosols from below 0.5 km to between 0.5 and 1.5 km. The extinction coefficient between 0.5 and 1.5 km gradually increases until it reaches a maximum at 13:00. The extinction coefficients are all >0.3${ \mathrm{km}}^{-1}$within about 1.25 km of the ground surface. Between 13:00 and 18:00, as the sun sets, the aerosols in the range 0.5–1.5 km above the ground surface are gradually depressed to below 0.7 km, the extinction coefficient within the range 0.5–1.5 km decreases, and the lowest extinction coefficient gradually increases to 0.45${ \mathrm{km}}^{-1}$. However, in the range 1–1.5 km, the extinction coefficient always remains between 0.15 and 0.3 ${\mathrm{km}}^{-1}$ because some aerosols are retained in the residual layer, resulting in a relatively high extinction coefficient between 1 and 1.5 km [77,78].

The daily variation in the vertical distribution of aerosols in autumn is similar to that in spring. In winter, with the development of the boundary layer from 08:00 to 14:00, the aerosol extinction coefficient between 0.5 and 1 km gradually increases and the aerosol extinction coefficient below about 0.6 km is >0.3 ${\mathrm{km}}^{-1}$. The boundary layer height decreases after 14:00 and the aerosol particles are gradually pressed down to below 0.5 km. From 12:00 to 18:00, the aerosol extinction coefficient in the range 0.5–2 km is significantly lower in winter than in other seasons. By contrast, the aerosol extinction coefficient below 0.5 km is higher than in other seasons. This is because the boundary layer height is relatively low in winter and pollutants are not easily transported to altitudes >0.5 km [72]. In addition, winter is the main heating season in Beijing and the emission of pollutants increases the aerosol extinction coefficient below 0.5 km [25].

3.2. Case Study of Haze Pollution from 8 to 15 February 2020

We investigated the effects of the vertical motion of the boundary layer on particulate matter and its optical properties during an extreme haze event and the link between the vertical distribution of aerosols and the meteorological conditions within the boundary layer. Based on observations from the Raman–Mie scattering LiDAR and sun photometer combined with meteorological data, we conducted a comprehensive analysis of an extreme haze event in Beijing from 8 to 15 February 2020.

3.2.1. Meteorological Conditions within the Boundary Layer

Figure 4 shows the variation of the PM_2.5 mass concentration, the Ångström exponent at 440–870 nm (AE_440–870nm), the AOD, the boundary layer height, the temperature, the relative humidity, the wind direction and the wind speed in urban Beijing from 8 to 15 February 2020. Figure 4a shows that a heavy pollution event occurred in Beijing from February 8 to 15. Pollutants began to accumulate from about midnight on February 8, when the PM_2.5 mass concentration increased suddenly from 15 to 180 μg m⁻³ and then remained steady at about 120 μg m⁻³. The concentration of pollutants increased suddenly from about 09:00 to 12:00 on 11 February, when the PM_2.5 mass concentration increased from 120 to 250 μg m⁻³, after which it remained at around 240 μg m⁻³. Precipitation occurred in the Beijing area from the afternoon of 13 February to 14 February and the pollutants gradually cleared via wet deposition. The PM_2.5 mass concentration decreased and the pollution event ended. We focus here on two cumulative growth processes of pollutants: from 12:00 on 8 February to 01:00 on 9 February (PE1) and from 18:00 on 10 February to 18:00 on 11 February (PE2). We analyzed the influence of the vertical motion of the boundary layer on particulate matter and its optical properties during HPEs and the connection between the vertical distribution of aerosols and the meteorological conditions within the boundary layer.

Figure 4 shows that the near-surface wind speed was low in PE1 and mainly dominated by westerly and easterly winds. The PM_2.5 mass concentration increased rapidly from 15 to 180 μg m⁻³ and the AOD showed an increasing trend similar to that of PM_2.5. The boundary layer height showed a fluctuating decreasing trend, which was influenced by the combined effect of the daily variation in the boundary layer and the aerosol radiation effect [21]. The near-surface temperature had a clear daily trend, with the minimum temperature increasing from −9 to −3 °C at the beginning of the period of relative pollution, indicating that the presence of an aerosol particle layer at night enhances the downward counter-radiation in the atmosphere and heats the underlying atmosphere, increasing the daily minimum temperature [79].

The changes in the relative humidity and PM_2.5 both showed an increasing trend. The change in the relative humidity was influenced by the daily change in the water vapor content and temperature [80]. Before sunset, the amount of shortwave radiation reaching the near-surface decreased as a result of aerosol extinction [21]. The near-surface region therefore cooled down, the saturated water vapor pressure decreased and the relative humidity gradually increased [80]. After sunset, the temperature near the ground gradually decreased and therefore the saturated water vapor pressure also decreased, resulting in a gradual increase in the relative humidity [80].

In PE2, the near-surface region was in a static state with a wind speed <2 m s⁻¹. The temperature and relative humidity showed clear daily variations, but the minimum value of the relative humidity increased from 30% on 10 February to about 45% on 11 February and the maximum temperature decreased from 13 to about 11 °C. This suggests that the amount of solar radiation reaching the near-surface decreased as a result of the extinction effect of pollutants, which led to a decrease in the near-surface temperature and an increase in the relative humidity [50,51].

The PM_2.5 mass concentration increased rapidly from 120 μg m⁻³ to about 250 μg m⁻³ at 09:00–12:00 on 11 February, whereas the boundary layer height decreased from 1.1 to about 0.6 km. This was caused by the accumulation of pollutants leading to radiative cooling near the ground, lowering the surface temperature and forming or enhancing the inverse temperature lamination. This inhibited the diffusion of vertical turbulence and led to a decrease in the boundary layer height [54,55].

The AOD_440nm increased from 0.4 to 1.6 from about 18:00 on 10 February to 08:00 on 11 February, whereas the Ångström exponent increased from 1.2 to 1.6, indicating that the accumulation of pollutants during PE2 was dominated by fine particles [14]. The increasing trends presented by both PM_2.5 and AOD_440nm were not synchronized in time. The PM_2.5 mass concentration increased about 2.1 times from 09:00 (about 120 μg m⁻³) to 12:00 (about 250 μg m⁻³) on 11 February, whereas the AOD_440nm increased about four times from 18:00 (about 0.4) on 10 February to 09:00 (about 1.6) on 11 February (Figure 4). Figure 4 shows that the increase in the AOD was synchronized with the increase in the relative humidity. The increase in the AOD may be because, at high particle concentrations (about 120 μg m⁻³), the hygroscopic growth of a large number of fine particles with increasing relative humidity increases the scattering of particles [32].

The increases in the SSA at 675, 870 and 1020 nm all showed different degrees of growth. The vertical profile of the extinction coefficient also showed different degrees of growth at each height, especially at the lowest layer, where there was an increase of about 3.3 times from 0.6 to 2.0 (Figure S1). This led to a decrease in the amount of radiation reaching the near-surface and an increase in the amount of radiation trapped in the boundary layer. This strengthened the inversion layer in the boundary layer and suppressed the development of turbulence. The height of the boundary layer decreased from 1.1 to 0.6 km from 09:00 to 12:00 on 11 February (Figure 4). Particulate matter accumulated continuously from high altitudes to the near-surface, which promoted the rapid growth in the mass concentration of PM_2.5.

The vertical transport of particulate matter may also explain the lack of synchronization in the incremental trends presented by the AOD_440nm and the PM_2.5 mass concentration. The PM_2.5 mass concentration is measured with the micro-oscillation balance and β-absorption methods [81], which respond to the near-surface mass concentration of PM_2.5, whereas the AOD reflects the whole aerosol column [21]. The pollutants may be suspended in the atmosphere, resulting in low PM_2.5 mass concentrations from near-surface observations, while the AOD shows an increasing trend.

Figure S1 shows that the aerosol extinction coefficient at 1–1.5 km continued to grow from 18:00 on February 10 to about 10:00 on 11 February. The aerosol extinction coefficient increased from about 0.1 to 0.3 ${\mathrm{km}}^{-1}$ at 1 km height and the high-value region of the aerosol extinction coefficient (>0.3 ${\mathrm{km}}^{-1}$) gradually thickened from <1 km to about 1.25 km, indicating the continuous accumulation of pollutants between 1 and 1.5 km. At the height of 1000 hPa, the Beijing area was located on the eastern side of a region of high pressure system and the near-surface winds were mainly northerly. With an eastward shift in the region of high pressure, the wind speed gradually decreased at the near-surface and was relatively stable, and the strengthening of the sinking airflow in Beijing. From 00:03 to 12:00 on 11 February, there was a sinking airflow with a wind speed of about 0.5 m s⁻¹ between 0 and 1.5 km, which continuously transported pollutants from altitudes of 1–1.5 km to the near-surface. This promoted the rapid increase in pollutants.

Figure 5 shows the contribution of CAPE to the variation in the PM_2.5 concentration. When the CAPE is negative, the diffusion of pollutants is inhibited and is expressed by the CIN [62,63]. During PE1 and PE2, the atmosphere had a stable stratification at 08:00 and 20:00 and the diffusion of pollutants was inhibited. The temperature within the boundary layer (1000–900 hPa) gradually increased with height during PE2 at 08:00 and 20:00 on 10–11 February, showing an inverse temperature stratification. This further enhanced the accumulation of pollutants.

The wind field is an important meteorological factor affecting the accumulation or dispersion of atmospheric pollutants [66]. Figure 6 and Figure 7 show the horizontal wind field profiles and vertical wind field plots, respectively, during this period of haze pollution. Positive values in the vertical wind field plots indicate a sinking airflow and negative values indicate a rising airflow. During PE1, Beijing was in the southeast of the low-pressure system from 12:00 to 18:00 and the near-surface winds were mainly southerly. Pollutants were continuously transported from southern Beijing to downtown Beijing, which accelerated the accumulation of pollutants [50]. With the eastward movement of the low-pressure system, the near-surface wind speed weakened and the lifting airflow gradually strengthened.

During PE2, the Beijing area was located on the eastern side of the high-pressure region and the near-surface winds were mainly northerly. The near-surface wind speed decreased with the eastward shift in the high-pressure region and the sinking airflow was enhanced. At around 12:00 to 18:00 on 11 February, a sinking airflow with wind speeds of 1.5–2.5 m s⁻¹ appeared below 500 m, which transported pollutants to the near-surface and promoted their accumulation, leading to a stable PM_2.5 mass concentration of about 250 μg m⁻³ (Figure 4).

3.2.2. Aerosol Optical Properties

Figure 8 shows the spatiotemporal distribution of the extinction coefficient and depolarization ratio of aerosols below 3 km from 8 to 15 February 2020. In general, aerosols, sea salt particles and the water droplets in clouds are almost spherical, with a relatively low depolarization, so they are not easily distinguished from atmospheric pollutants. By contrast, dust aerosols and ice crystals in ice clouds are non-spherical with a relatively large depolarization (>0.2), so they are easily distinguished from atmospheric pollutants [74].

Figure 8 shows that the polluted aerosol layer during this haze was mainly concentrated within 1 km and the depolarization ratio was <0.2. In PE1, the pollutants started to accumulate from about 13:00 on 8 February and the aerosol extinction coefficient gradually increased significantly from <0.1 to 1.3${ \mathrm{km}}^{-1}$ within 1 km. The aerosol concentration increased significantly and the depolarization ratio was <0.2 during the accumulation of pollutants, indicating that this process consisted mainly of the accumulation of spherical particles. In PE2, with the accumulation of pollutants, the height of the area with a high aerosol extinction coefficient gradually increased to about 1.0 km and the receding polarization ratio of this layer was relatively large (>0.2) from 12:00 to 18:00 on 10 February. There was an updraft of 1–1.5 m s⁻¹ below 500 m at about 12:00 on 10 February (Figure 7). This indicates that a dust-raising process that produced non-spherical particles occurred from 12:00 to 18:00. The receding polarization ratios were relatively low (<0.1) at all other times, indicating that a large number of water-soluble spherical particles were produced during this haze episode [21].

Figure 9 shows the trends of the SSA, AAOD, AAE, DARF, the volume distribution of coarse and fine particles, and the volume spectrum distribution of aerosols throughout the pollution process. The SSA440nm was between 0.95 and 0.975 throughout the heavy pollution event, indicating that scattering aerosol particles were predominant throughout. This may be because, under high relative humidity (daily average > 65%), a large number of fine modal particles absorb moisture and grow, thus increasing the scattering of aerosols [32]. SSA1020nm, SSA870nm, SSA675nm and SSA440nm gradually converged with the accumulation of pollutants, indicating that the scattering of each band was greatly increased as the pollutants accumulated.

The AAE is often considered to be a key factor in determining the types of particles absorbing radiation (e.g., black carbon, organic materials or mineral dust) [82]. Specifically, the AAE of mixtures of black carbon with organic or inorganic components is <1, the AAE of the black carbon emitted from the combustion of fossil fuels is roughly equal to 1, and the AAE of black carbon and mineral dust emitted from the combustion of biomass is roughly equal to 1.1 [21]. The AAOD in all four bands (440, 675, 870 and 1020 nm) showed an increasing trend with the accumulation of pollutants, especially from 10 to 11 February, when the AAOD_440nm increased from 0.016 to 0.054 and the AAE increased from 0.62 to 1.21. This indicates that the dust present on 10 February (Figure 8) was an important factor in enhancing absorption by aerosols [23].

The particle spectrum of the aerosol volume concentration showed a bimodal logarithmic distribution throughout the pollution process. The volume of fine modal particles increased gradually with the accumulation of pollutants and the peak concentration increased from 0.008 to 0.31 ${\mathsf{\mu }\mathrm{m}}^3{ \mathsf{\mu }\mathrm{m}}^{-2}$. The total volume increased from 0.01 to 0.33 ${\mathsf{\mu }\mathrm{m}}^3$, whereas the total volume of coarse modal particles increased from 0.009 to 0.061 ${\mathsf{\mu }\mathrm{m}}^3$ on 8–9 February, but then did not change much. This suggests that the aerosol particles accumulated throughout the pollution process, whereas some coarse modal particles accumulated on 8–9 February. As the volume of fine modal particles increased, their peak concentration and the radius of fine particles corresponding to the peak concentration also increased. For example, the peak concentration of fine modal particles on 10 February was 0.03${ \mathsf{\mu }\mathrm{m}}^3{ \mathsf{\mu }\mathrm{m}}^{-2}$ and the radius was 0.25 μm. However, as the pollutants accumulated, by 13 February the peak concentration of fine modal particles was 0.31${ \mathsf{\mu }\mathrm{m}}^3{ \mathsf{\mu }\mathrm{m}}^{-2}$ and their radius was 0.33 $\mathsf{\mu }\mathrm{m}$. This may be due to the accumulation of a large number of water-soluble particles (e.g., sulfate and nitrate particles) during the accumulation of pollutants. The hygroscopic growth of these particles at high relative humidity (daily average > 65%) results in a gradual increase in the size of fine particles as pollutants accumulate [83].

Figure 9 shows that the volume of both coarse and fine mode particles increased significantly on 8–9 February. The volume of fine mode particles increased from 0.01 to 0.054${ \mathsf{\mu }\mathrm{m}}^3$ and the volume of coarse mode particles increased from 0.009 to 0.061 ${\mathsf{\mu }\mathrm{m}}^3.$ The peak volume concentration of fine mode particles on 9 February was 0.037 ${\mathsf{\mu }\mathrm{m}}^3{ \mathsf{\mu }\mathrm{m}}^{-2}$, which corresponds to a radius of 0.2 μm for fine mode particles. The peak volume concentration of coarse mode particles was 0.04${ \mathsf{\mu }\mathrm{m}}^3{ \mathsf{\mu }\mathrm{m}}^{-2}$ with a radius of 3.9 μm, which indicates that both coarse and fine particles accumulated considerably during the accumulation phase on 8 and 9 February.

The SSA440nm increased slightly from 0.96 to 0.97, whereas the SSA increased substantially at 670–1020 nm, which may be due to its dependence on wavelength. The AAOD_440nm increased from 0.004 to 0.016 and the AAE increased from 0.15 to 0.56. This indicates that a mixture of black carbon and organic or inorganic components may have been present in the pollutants during this accumulation stage, resulting in enhanced absorption [84]. As the aerosol mass concentration increased, direct radiative forcing also increased to varying degrees, with the DARF at the base of the atmosphere decreasing from −15.7 to −61.5 W m⁻², suggesting that aerosol extinction greatly reduced the solar energy received at the Earth’s surface [21].

Figure 9 shows that the volume of fine mode particles increased significantly on 10–13 February, with the volume of fine mode particles increasing from 0.044 to 0.332${ \mathsf{\mu }\mathrm{m}}^3$. By contrast, the volume of coarse mode particles did not increase significantly compared with that on 8–9 February, which indicates that the aerosol particles accumulated in PE2 were mainly fine mode particles. The peak concentration of the volume of fine mode particles on 13 February was 0.31 ${\mathsf{\mu }\mathrm{m}}^3{ \mathsf{\mu }\mathrm{m}}^{-2}$ and the corresponding radius was 0.33 μm. The SSA440nm increased from about 0.96 to about 0.98, whereas the SSA670–1020nm gradually converged with the SSA440nm, indicating that the scattering of each band was greatly increased with the accumulation of pollutants. The increase in the AAOD_440nm from 0.016 to 0.061 and the AAE from 0.62 to 1.2 suggests the accumulation of black carbon or mineral dust aerosols from the burning of biomass in the PE2 phase. This can be explained by dust lifting from the surface on 10 February. The heating capacity of the atmosphere increased about 3.2 times from 10 February (23.2 W m⁻²) to 13 February (75.5 W m⁻²) as a result of the strong absorption of mineral dust aerosols.

3.3. Characteristics of Changes in Aerosol Precursors

The concentrations of CO are mainly influenced by the primary sources and the meteorological conditions, whereas PM_2.5 is also influenced by secondary sources [85,86]. CO can therefore be used to normalize the PM_2.5 concentrations to exclude the effects of primary emissions and meteorological conditions. We qualitatively assessed the intensity of secondary transformations in PM_2.5 using the PM_2.5/CO ratio (Figure 10) [81,85,87]. On 15 February (clean day), the daily average relative humidity was about 29% and the PM_2.5/CO ratio was low, with a mean value of 30.8 × 10⁻³ and a maximum value of 42.1 × 10⁻³. On February 13, when the pollution was higher, the daily average relative humidity was about 84% and the PM_2.5/CO ratio had an average value of 95.0 × 10⁻³, about 3.1 times higher than that on February 15. Comparing the PM_2.5/CO values on 13 and 15 February showed that secondary particulate matter was easily generated under conditions of high relative humidity [88,89,90].

The daily mean PM_2.5/CO values showed a fluctuating increasing trend as the pollution developed from 8 to 13 February. The daily mean PM_2.5/CO values increased about 1.3 times from the early (68.3 × 10⁻³) to late (95.0 × 10⁻³) stages of the pollution event. This indicates that secondary transformation intensified when the pollution accumulated in the late stage [87]. During PE1, the PM_2.5/CO ratio increased from about 40.0 × 10⁻³ to about 120.0 × 10⁻³, indicating that secondary transformation was gradually increasing during PE1. CO, NO₂ and O₃ all increased with the PM_2.5/CO ratio, but the SO₂ concentration did not change significantly and the incremental increase in the SSA (Figure 9) on 8–9 February was probably due to secondary sulfate. As the accumulation of pollutants increased [87], a large amount of SO₂ was converted to SO₄²⁻. In PE2, the PM_2.5/CO ratio showed a decreasing trend from about 100 × 10⁻³ to 80.0 × 10⁻³. The CO and SO₂ concentrations showed a clear increase, whereas O₃ showed clear daily variations, indicating that the role of secondary conversion gradually decreased. The SSA also appeared to increase and NO₂ may have been converted to NO₃⁻.

4. Conclusions

We have studied the vertical distribution of aerosols over Beijing from 2017 to 2021 based on long-term LiDAR observations. We have analyzed the influence of the vertical distribution of aerosols and their interconnection with the meteorological conditions within the boundary layer during an episode of heavy haze pollution in Beijing from 8 to 15 February 2020.

The vertical profile of the aerosol extinction coefficient was different in different seasons and at different altitudes. Below 2 km, the aerosol extinction coefficient was highest in summer, lower in winter, and lowest in spring and autumn; the lowest extinction coefficients were about 0.4, 0.35 and 0.3 ${\mathrm{km}}^{-1}$, respectively. The maximum height of the boundary layer in Beijing can reach about 2 km in the summer months. Aerosol particles can be transported to an altitude of about 2 km via turbulence and the high relative humidity in summer makes it easier for aerosol particles to absorb moisture and grow, resulting in the high aerosol extinction coefficient below 2 km in summer. At the upper level of 2–3 km, the aerosol extinction coefficient was highest (about 0.1${ \mathrm{km}}^{-1}$) in spring and roughly comparable in summer, autumn and winter (<0.1 ${\mathrm{km}}^{-1}$). This is due to the frequent occurrence of dust pollution events in Beijing in spring and the transport of dust aerosols to an altitude of 2–3 km. This leads to a larger aerosol extinction coefficient that does not change with height.

The vertical profile of the aerosol extinction coefficient showed a clear daily variation in all four seasons. The development of the boundary layer and the vertical distribution of aerosols were different in different seasons. The height of the high-value zone of the aerosol extinction coefficient was highest in summer, with a maximum value at about 1.25 km at 13:00. By contrast, the height of the high-value zone of the aerosol extinction coefficient was lowest in winter, with a maximum of about 0.7 km at 14:00. The maximum height in both spring and autumn was about 0.8 km at 12:00.

We have discussed the influence and interconnections of the vertical distribution of aerosols, the meteorological conditions within the boundary layer and the optical properties of aerosols during a haze event in Beijing. During PE2, the updraft promoted the accumulation of pollutants between 0.5 and 1.5 km. Together with the hygroscopic growth of aerosol particles, this led to a gradual increase in AOD_440nm, but the PM_2.5 mass concentration did not change significantly. A sinking airflow of 1–1.5 m s⁻¹ appeared around 11:00, which transported pollutants from high altitudes to the near-surface, promoting a rapid increase in the PM_2.5 mass concentration. Scattered spherical fine particles dominated this heavy haze pollution event.