Hefei occupies a strategic location, bridging the Beijing–Tianjin–Hebei metropolitan cluster and the Yangtze River Delta economic hub. The study area lies in a transitional climatic zone, where temperature inversions occur frequently in winter. Such conditions often result in stagnant air that limits pollutant dispersion. In addition, the geography of Hefei plays a critical role in exacerbating air quality challenges. Hefei is surrounded by the Dabie Mountains in the west and the Huangshan ranges in the south. The open plains dominate its northern and eastern frontiers (Fig. 1b). These natural barriers hinder the southward dispersion of airborne pollutants while trapping those transported from industrialized northern regions within the basin. The unique combination of geographic constraints and atmospheric conditions has made Hefei a focal point for studying complex pollution mechanisms. Researchers here focus particularly on the synergistic effects of regional transportation, local emissions, and meteorological drivers on particulate matter formation (Fang et al., 2024; Huang et al., 2016; Liu et al., 2024).

The LGJ-05 aerosol LiDAR combines traditional radar technology and modern laser technology, with 355, 532, and 1064 nm as detection light sources. Measurements at 532 nm were utilized exclusively for this study. The LiDAR emits a laser pulse into the atmosphere to capture backscattered signals, thereby obtaining key aerosol optical characteristics. As a polarization-sensitive system, it can retrieve extinction coefficients and depolarization ratios to characterize aerosols in the atmosphere. During the transmission process, the laser pulse is scattered and extinguished by atmospheric aerosol particles. The intensity of backscattered light at different altitudes correlates with the scattering and extinction properties of aerosol and cloud particles at those altitudes. The backscattered light from spherical particles can maintain its polarization characteristics of the emitted pulse since the emitted laser pulse is linearly polarized. In contrast, non-spherical particles (such as dust particles and ice crystals in cirrus clouds) depolarize the backscattered light due to their irregular shapes and asymmetric interaction with linearly polarized laser pulses. The LiDAR could detect the echo signals of the parallel and vertical components in the backscattered light, enabling acquisition of the vertical profile of the depolarization ratio of atmospheric aerosol particles. It could enable three-dimensional monitoring of the atmosphere in real time through active remote sensing, with the capacity to invert spatiotemporal distribution information such as aerosol extinction, depolarization, and water vapor mixing ratio in the atmosphere. The aerosol extinction coefficient reflects the attenuation capacity of aerosol particles to incident light, while the depolarization ratio indicates the non-spherical nature of aerosol and cloud particles. A higher extinction coefficient typically indicates a larger concentration of aerosol particles or stronger light-scattering/absorbing properties. However, greater depolarization values denote a higher abundance of non-spherical aerosol particles (e.g., dust, rugged industrial aerosols) or mixed-phase cloud constituents. By contrast, lower depolarization ratios indicate a dominance of spherical particles or simpler particle structures.

Since the instrument is capable of receiving data below 30 km, the effective detection range of the aerosol LiDAR can cover 0.2–6 and 0.2–15 km for daytime and nighttime, respectively. It offers a vertical spatial resolution of 7.5 m and a temporal resolution of 10 min. The observations were conducted at the National Meteorological Observation Station in the northwest of Hefei (31.96° N, 117.06° E) from March 2021 to May 2023.

Hourly concentrations of PM_2.5 and PM₁₀ were obtained from ground-based monitoring stations in Hefei, which are operated and maintained by the China National Environmental Monitoring Centre (CNEMC) (Liu et al., 2017). These pollutant data were obtained from professional monitoring instruments, with the LGH-01E aerosol mass concentration monitor used for PM₁₀ and the LGH-01B for PM_2.5, both of which apply the beta attenuation method. All instruments are calibrated regularly according to national standards, and the data undergo strict quality control procedures, including hourly, daily, and annual audits, as described in the China Environmental Monitoring Quality Assurance and Quality Control Manual. It should be noted that we focused on PM_2.5 and PM₁₀ concentrations during winter and spring, respectively. All comparisons between PM_2.5 and PM₁₀ pollution episodes were conducted across both seasons. The hourly resolution of the dataset allows for capturing diurnal variations in pollutant levels, while the multi-pollutant data facilitates analysis of the differences in meteorological responses of pollutants in Hefei.

The observed meteorological variables were obtained from the China Meteorological Administration (CMA) (http://data.cma.cn/en, last access: 16 March 2025), measured by the Vaisala PTB210 Digital Barometer for atmospheric pressure, the Vaisala HMP155A Temperature and Relative Humidity Probe for air temperature (T) and relative humidity (RH), the EL15-2C Wind Direction Sensor for wind direction (WD), and the EL15-1C Wind Speed Sensor for wind speed (WS). All meteorological instruments are routinely maintained and calibrated to ensure the accuracy and reliability of observational data.

This study aims to investigate the impact of the synoptic system on PM_2.5 and PM₁₀ pollution. The ERA5 reanalysis dataset, freely accessible via the Copernicus Climate Change Services platform (https://cds.climate.copernicus.eu/datasets, last access: 23 March 2025), serves as the data source. Vertical atmospheric data within the ERA5 dataset are interpolated to 37 pressure levels. This interpolation yields comprehensive data ranging from the Earth's surface to the upper atmosphere. Given its high spatial and temporal resolution, the ERA5 dataset has been widely employed in extreme weather events, climate prediction, and air pollution studies (Fan et al., 2021; Zhang et al., 2020b).

This study collected hourly geopotential height, vertical wind velocity, temperature, and humidity data. The data correspond to multi-pressure levels (500, 700, 725, 750, 775, 800, 825, 850, 875, 900, 925, 950, 975, and 1000 hPa). The spatial resolution of our dataset was 0.25° × 0.25° (zonal and meridional) and supports a detailed and accurate exploration of the synoptic-particulate matter relationship. The data were extracted for winter and spring during 2021–2023, consistent with the periods of the surface pollutant and aerosol LiDAR observations.

To explore the potential of aerosol LiDAR observations and improve data reliability, this study extracted the extinction coefficient and depolarization ratio data at 532 nm from the LiDAR system for the period from March 2021 to May 2023. In the quality control process, specific criteria were established to ensure the reliability of the retrieved parameters. Specifically, it covers three steps for outlier detection, removal of spurious points, and temporal consistency analysis. Outlier detection was conducted to eliminate all records outside the normal value range according to the parameter intervals provided by the instrument, with specific thresholds applied based on the signal-to-noise ratios (SNR) of the parallel (P) and perpendicular (S) channels. The extinction coefficient at 532 nm was calculated using data from the P channel, which was considered reliable only when the corresponding SNR exceeded three. The depolarization ratio was derived from the P and S channels, and data were accepted only when both channels had SNRs greater than three. The P channel detects linearly polarized light in the parallel direction, while the S channel detects linearly polarized light in the perpendicular direction. Based on the time series of SNR records, the individual extinction and depolarization data were cross-referenced. Temporal consistency analysis was conducted by identifying records that deviated by more than three standard deviations from the mean as outliers. The original LiDAR data have a 10 min temporal resolution. After quality control, these data were averaged to hourly resolution to be consistent with the hourly ground-based pollutant observations. To further improve data reliability, this study calculated the credibility ratio of the extinction coefficient and depolarization ratio at different altitudes. Based on the criterion of a 60 % valid data availability rate, the effective detection ranges for the 532 nm of the extinction coefficient and depolarization ratio were 0.2–1.8 km.

Figure 2 shows the variations in the concentrations of PM_2.5 and PM₁₀ pollutants from 2021 to 2023. Both PM_2.5 and PM₁₀ pollutants exhibited a clear seasonal pattern, with concentrations higher in winter and lower in summer (Liu et al., 2021; Wang et al., 2024). In addition, PM₁₀ concentrations were relatively high during the spring months of 2022 and 2023, likely driven by increased dust events such as sand-dust storms that commonly occurred during this season. High PM_2.5 / PM₁₀ ratios (≥ 0.5) during winter suggest a dominant contribution from fine particulate matter, whereas lower ratios (≤ 0.5) in spring indicate an increased presence of coarse particles (Liu et al., 2015). The episodic PM₁₀ peaks observed in spring 2022–2023 coincided with a pronounced decrease in the PM_2.5 / PM₁₀ ratio in Hefei. This phenomenon reflected intensified pollution of coarse-mode aerosols. Although winter months also suffered elevated PM₁₀ pollution, the concurrent high PM_2.5 concentrations complicated the separation of dust-related contributions. In contrast, the lower PM_2.5 / PM₁₀ ratios in spring facilitated the distinct observation of dust events due to the reduced interference from fine particulate matter.

The hourly PM_2.5 and PM₁₀ concentrations exhibited clear diurnal variation on both polluted and clean days (Fig. 3). Polluted days showed significantly higher PM_2.5 levels, with hourly averages ranging from 100 to 125 µgm-3, compared to the stable 35–55 µgm-3 range observed on clean days. The PM_2.5 concentrations on polluted days displayed a bimodal pattern. The concentrations remained consistently high throughout the morning period from 04:00 to 11:00 (Beijing time, BJT), with the peak at 10:00–11:00 (BJT) (Dai et al., 2020). Valleys in PM_2.5 concentrations occurred at 16:00 (BJT), possibly associated with enhanced atmospheric dispersion from midday heating that facilitates the dilution of particulate matter. Concentrations initiated an upward trend from 17:00 (BJT), with a gradual increase culminating in a peak between 21:00 and 22:00 (BJT). In contrast, clean days exhibited a subtle morning and overnight peaks. For PM₁₀, polluted days presented pronounced peaks at 10:00–12:00 and 21:00–22:00 (BJT), coinciding with increased dust resuspension or construction activities (Yu et al., 2020). However, valleys occurred at 07:00 and 19:00 (BJT). These may be due to reduced surface disturbances during dawn and enhanced vertical mixing in the afternoon. Clean days maintained relatively low hourly mean PM₁₀ concentrations (50–70 µgm-3) with minor fluctuations. The time series curves of pollutants had different peak times and amplitudes, indicating distinct origins. PM_2.5 was more influenced by continuous anthropogenic emissions, while PM₁₀ was sensitive to episodic coarse-particle events (Deng et al., 2023; Wang et al., 2024). These hourly dynamics highlighted the role of diurnal meteorological cycles in modulating particulate matter distribution, with polluted days amplifying both primary emissions and secondary aerosol formation (Liu et al., 2021).

In addition to the impacts exerted by surface-level pollution, the investigation of vertical distribution is also of paramount significance for a comprehensive understanding of air quality dynamics. The extinction coefficient at 532 nm shows seasonal and vertical variation. Vertically, extinction coefficients were highest near the surface and gradually decreased with altitude in all seasons. This vertical decline arises from rapid reductions in aerosol loading due to vertical mixing, gravitational sedimentation, and dilution with increasing height, as well as the dominance of weaker molecular scattering above the planetary boundary layer (Wang et al., 2021). Winter presented the highest extinction values among the four seasons (Fig. 4a) (Chen et al., 2023; Wang et al., 2024). This was likely related to enhanced emissions from domestic heating and stagnant meteorological conditions that restrict vertical mixing (Zhong et al., 2018). In contrast, summer exhibited the lowest extinction coefficients at 532 nm across the profile, suggesting the effective dispersion of aerosols due to strong convection and a high planetary boundary layer (Li et al., 2015). Spring and autumn showed intermediate levels. The most significant differences between seasons appeared below 1 km, especially in the 0.2–0.6 km range (Zhong et al., 2018). Above 1.0 km, the extinction values in all seasons converged toward high altitude, indicating a reduced aerosol presence in the upper layer (Liu et al., 2024).

The depolarization ratio also exhibited seasonal differences (Fig. 4b). With increasing altitude, the contribution of non-spherical particles increases, leading to a steady rise in depolarization ratio. This is attributed to the fact that non-spherical particles exhibit stronger anisotropic light scattering than nearly spherical fine-mode aerosols that dominate near the surface (Wang et al., 2020b). In the lower layer below 0.6 km, spring showed the highest depolarization ratios, which were attributed to a greater proportion of non-spherical particles such as mineral dust or internally mixed soot (Biuki et al., 2022). This may result from springtime dust resuspension and aged industrial emissions. In contrast, summer had the lowest depolarization ratios, implying that aerosols were dominated by more spherical particles, such as those formed through secondary processes (e.g., sulfate or organic aerosols) (Sun et al., 2013). In autumn and winter, values were moderate and stayed between the two extremes. However, the seasonal pattern changed above 0.6 km. Winter showed the highest depolarization ratio, while spring became the lowest. The vertical profile revealed that the change in particle shape with height strongly depends on seasons (Wang et al., 2004). In winter, the low ambient temperatures in the upper atmosphere promote the formation of non-spherical particles (e.g., ice crystals and irregular aerosols), which substantially enhance the depolarization ratio due to their anisotropic light-scattering properties (Haarig et al., 2017; Wang et al., 2021; Yu et al., 2021). In addition, elevated aerosol layers were dominated by long-range transported dust with pronounced nonspherical properties. However, aerosols in the upper layer were more spherical and less influenced by dust in spring (Wang et al., 2020b).

The extinction coefficient is significantly higher on PM_2.5-polluted days than on PM₁₀ polluted days. This is attributed to the optimal size match between fine-mode particles (≤ 2.5 µm) and visible light (0.4–0.7 µm), as well as their chemical components, which are more efficient at scattering and absorbing light (Wang et al., 2021; Chen et al., 2023). The extinction coefficients at 532 nm exhibited clear differences between polluted and clean days for both PM_2.5 and PM₁₀, with strong vertical and diurnal variation (Fig. 5). For PM_2.5, polluted days were characterized by significantly higher extinction values below 0.9 km, especially during the nighttime and the period from early morning to near noon (Wang et al., 2024). This enhancement was mainly attributed to the stable boundary layer and limited vertical mixing during the night, which favored the accumulation of aerosols near the surface. Strong solar radiation enhances photochemical reactions, leading to substantial secondary aerosol formation at noon and a lift in altitude (Fig. 5a). However, extinction values near the surface (below 0.4 km) were relatively low during 12:00–15:00 (BJT), with high values concentrated mainly around 0.6 km. This midday vertical uplift of the aerosol layer corresponds to the decrease in near-surface PM_2.5 shown in Fig. 3, as enhanced convective mixing lifts fine particles upward and weakens near-surface extinction. Hence, strong daytime turbulence promotes the vertical transport of aerosols, resulting in an elevated and vertically expanded aerosol layer (Fig. 5a) (Wang et al., 2024). During this period (12:00–15:00 BJT), the enhanced extinction was concentrated around 0.6 km, extending vertically from the surface to this altitude, rather than peaking near the ground. The upward expansion of aerosols was driven by boundary layer development and persistent emission sources (Dai et al., 2020). In contrast, clean days showed low extinction values throughout the day, with little vertical variation and no significant peak. For PM₁₀, extinction coefficients at 532 nm were also increased on polluted days (Fig. 5b). However, the enhancement of extinction values extended to a higher altitude compared to that of PM_2.5, reaching up to 1.2 km. The primary increase occurred during the night and early morning, with a minor secondary rise observed around midday (Li et al., 2020). Clean-day extinction remained uniformly low with minimal temporal variation (Fig. 5d).

For the depolarization ratio, distinct patterns also emerged between polluted and clean days for both PM_2.5 and PM₁₀ (Fig. 6). For PM_2.5, depolarization ratios near the surface (below ∼ 0.6 km) were relatively low on polluted days, particularly during nighttime and early morning (Fig. 6a). This can be attributed to the accumulation of fine, spherical particles in the stable boundary layer, which tend to have lower depolarization ratios. As altitude increased, depolarization ratios rose. This indicates the presence of more irregularly shaped or coarser particles aloft, possibly from transported dust or aged aerosols. In contrast, clean days showed uniformly low depolarization ratios across most heights, consistent with the dominance of fine, spherical aerosols under favorable dispersion conditions (Fig. 6c). For PM₁₀, depolarization ratios at 532 nm on polluted days exhibited a different behavior (Fig. 6b). Starting around 08:00 (BJT), elevated depolarization values were observed across a broad vertical layer, extending up to ∼ 1.2 km throughout the day. This suggests that coarser particles, which typically have higher depolarization ratios, were more effectively mixed vertically and persisted at higher altitudes compared to PM_2.5. The broader vertical distribution of PM₁₀ depolarization signals reflects the larger size and potentially different sources (e.g., resuspended dust, industrial emissions) of these particles, which are less confined by boundary layer dynamics than finer PM_2.5. The increased depolarization ratio at noon during PM₁₀-polluted days is attributed to midday boundary layer growth and frontal dust transport, elevating non-spherical coarse particle concentrations.

The difference between polluted and clean days highlighted a strong increase in near-surface extinction during early morning on polluted days, indicating suppressed vertical mixing and increased emission accumulation (Fig. 5e and f). For PM_2.5, the differences were concentrated near the ground, especially below 0.9 km. The strongest enhancement appeared between 07:00 and 14:00 (BJT). A subtle lifting with height was observed around midday. Noon solar radiation maximizes surface heating, which causes severe thermal convection. The lower atmosphere becomes unstable as a result of the hot Earth transferring energy upward. It could break the stable stratification and drive upward air motion. Pollutants trapped near the surface are lifted, reducing extinction in the lower layer (< 0.4 km) while increasing it aloft as vertical mixing intensifies (Wang et al., 2024). For PM₁₀, the enhancement tended to be localized and less vertically extended. The coarse particles may be lifted in upstream source regions, but their vertical transport weakens significantly along the trajectory. As a result, the transport altitude of PM₁₀ reaching Hefei was largely confined below 1.2 km (Shim et al., 2022). Compared to PM_2.5, the high-value region for PM₁₀ on polluted days has a higher vertical extension, and the period of strong extinction is shorter. The particles of PM₁₀ are relatively large and heavy, which leads to rapid sedimentation and thus a short duration of heavy pollution. Pollutants tend to accumulate in the middle and low altitudes (0.4–1.2 km) on polluted days.

To further analyze the vertical aerosol properties under different pollution levels, the mean extinction coefficient and depolarization ratio at 532 nm were calculated across three representative conditions (Fig. 7). Unlike the temporal variation that emphasized vertical profile distributions (Figs. 5 and 6), these curves highlight altitude-related trends (e.g., gradual changes, peak positions, and inflection points) for the comparison between polluted, clean, and difference conditions. For PM_2.5, the extinction coefficient decreased exponentially with height and intensified sharply below 0.4 km on polluted days (Fig. 7a) (Wang et al., 2020a). The near-surface value reached up to 1.4 km-1 at 0.4 km, which was nearly 3 times higher than that on clean days. The difference was most significant within the boundary layer, especially between 0.3–0.6 km. For PM₁₀ (Fig. 7b), although the overall extinction values were lower than those of PM_2.5, the difference between polluted and clean days remained obvious. The polluted-clean difference for PM₁₀ was more confined below 1.2 km. This discrepancy may be attributed to that the coarse-mode particles tend to concentrate closer to the surface due to their limited vertical transport.

The vertical distributions of the depolarization ratio under different pollution levels are presented in Fig. 7c and d for PM_2.5 and PM₁₀, respectively. For PM_2.5, the depolarization ratio on clean days was higher than that on polluted days across the entire vertical column up to 1.8 km (Fig. 7c). The depolarization ratio of PM_2.5 increases with height. The most significant difference in depolarization ratios between polluted and clean days was observed around 1 km. Above this height, the difference narrowed but clean days still showed slightly higher values. Unlike near-surface PM_2.5, which is dominated by local pollution sources, PM_2.5 at higher altitudes is mainly associated with long-range transport in the free troposphere, leading to higher depolarization ratios. Higher altitudes have more PM_2.5 particles from long-range transport in the free troposphere, which possess higher depolarization ratios (Vakkari et al., 2021; Wang et al., 2021). Even at higher altitudes, the residual influence of cleaner, non-spherical particle sources in the background air kept the clean-day depolarization ratio higher (Li et al., 2020). For PM₁₀, the depolarization ratio of polluted days was higher than that of clean days across the entire vertical range (Fig. 7d), reflecting a greater abundance of coarse-mode and irregularly shaped particles on polluted days. Below 1.2 km, the depolarization ratio of PM₁₀ was distinctly higher than that of PM_2.5 on polluted days. This difference emphasized the dominant influence of coarse, non-spherical particles (e.g., dust and mechanically suspended material) in the PM₁₀ fraction, particularly near the surface (Shim et al., 2022). On PM₁₀-polluted days in autumn, the depolarization ratio increased with height, which is mainly also affected by the long-distance transport of dust in the free troposphere.

Figure 8 presents the 3-hourly variation of extinction coefficient and depolarization ratio at 532 nm on polluted and clean days across three altitude layers (0–0.5, 0.5–1, and 1–1.8 km). Below 0.5 km, the extinction coefficients were markedly higher on polluted days for PM_2.5 than that on clean days, with peak values occurring at 03:00–12:00 (BJT) (Fig. 8a), which was consistent with the variation in surface pollutant concentrations (Fig. 3). The depolarization ratio for PM_2.5 was lower on polluted days in contrast with clean days. This is due to the dominance of spherical fine particles (e.g., sulfates, organic aerosols) from anthropogenic and secondary sources (Fig. 8a and c). These particles scatter light less directionally compared to the irregular coarse particles (e.g., dust) prevalent on clean days. In contrast, both the extinction coefficient and depolarization ratio for PM₁₀ were high on polluted days (Zhang et al., 2020a). At the layer between 0.5 and 1 km, the extinction coefficient for PM_2.5 showed a significant peak around 12:00 to 15:00 (BJT) on polluted days, which was consistent with the upward transport of pollutants to around 0.6 km during the same period as shown in the diurnal vertical structure (Fig. 5). The extinction values decreased significantly and remained at low levels above 1 km (Fig. 8e). However, the depolarization ratio remained relatively high during polluted periods. The depolarization ratio in winter was higher than that of spring, mainly because PM_2.5 pollution contains a higher proportion of irregular particles from coal-fired fly ash and industrial emissions (Fig. 8f). The low temperatures inhibit the diffusion and settlement of coarse particles. In contrast, days in spring for PM₁₀ pollution were more affected by coarse particles such as windblown dust. Although these particles are irregular in shape, their large particle size makes them prone to settling. Moreover, the strong atmospheric fluidity in spring led to a slightly lower depolarization ratio compared to winter. Overall, the greatest contrast between polluted and clean days was observed in the lowest layer, where both particle concentration and shape varied most distinctly with pollution level and time of day (Zhong et al., 2018).

Figure 9 shows the vertical profiles of the relationships between surface pollutant concentrations and aerosol optical properties (extinction coefficient and depolarization ratio at 532 nm) on polluted and clean days. For PM_2.5, the extinction coefficients showed a clear positive correlation with surface PM_2.5 concentration throughout the profile below 0.9 km on polluted days, marking a key boundary for vertical variation in particle composition (Fig. 9a). The values on polluted days increase sharply near the surface, suggesting a local accumulation. A negative correlation was observed between the depolarization ratio and PM_2.5 concentration below 0.9 km on polluted days, as elevated PM_2.5 levels corresponded to a higher fraction of fine, spherical particles with low depolarization (Zhang et al., 2020a). In contrast, clean days showed a consistently positive correlation between depolarization ratio and surface pollutants throughout the vertical profile.

The increased depolarization levels are linked to greater proportions of non-spherical particles, such as dust or mechanically suspended matter (Vakkari et al., 2021). The depolarization ratio showed consistently positive correlations with surface PM₁₀ concentration on polluted days, especially below 1.2 km (Fig. 9b). On clean days, the correlation between the depolarization ratio and PM₁₀ concentration remained near zero, reinforcing that the depolarization ratio is a distinguishing feature of coarse-mode aerosols. These results revealed the contrasting aerosol properties between PM_2.5 and PM₁₀ pollution. PM_2.5 events were dominated by fine, spherical particles that reduce depolarization, while PM₁₀ events were closely associated with the presence of coarse, non-spherical particles that enhanced depolarization signals (Biuki et al., 2022).

Temporal variations of the extinction coefficient and depolarization ratio at 532 nm exhibited a clear association with PM_2.5 and PM₁₀ concentrations, effectively highlighting severe pollution periods (Fig. 10). Severe PM_2.5 pollution was evident at 10:00 on 2 January 2023 (BJT), with the maximum hourly concentration reaching 159.6 µgm-3. The peak of the extinction coefficient matched the actual maximum concentration of PM_2.5, with fine particulate matter exhibiting strong light-scattering properties at this time (Fig. 10a). However, the depolarization ratio of PM_2.5-polluted days was markedly decreased during this peak period (Fig. 10c). Physically, the extinction coefficient scales with aerosol concentration and size, while the depolarization ratio is governed by particle morphology. Spherical fine particles such as hygroscopic sulfates and nitrates dominating this PM_2.5 event, leading to increased extinction and decreased depolarization. A dramatic peak in PM₁₀ concentration was observed, with the highest hourly value of 1440 µgm-3 recorded at 22:00 p.m. (BJT) on 11 April 2023. The severe PM₁₀ event was reflected by a pronounced increase in the extinction coefficient (Fig. 10b) (Sun et al., 2013). In addition, the depolarization ratio exhibited high values during this period, which aligned with the elevated extinction coefficient and was consistent with the irregular morphology of coarse particles that contributed to both stronger light scattering and greater depolarization (Fig. 10d). Overall, the profiles of the extinction coefficient and depolarization ratio correspond directly to the temporal patterns of PM_2.5 (1–8 January 2023) and PM₁₀ (11–13 April 2023) concentrations (Fig. 10). Peaks in pollutant concentrations were reflected in enhanced extinction coefficients and characteristic depolarization ratio patterns (Zhong et al., 2018). This confirms that these optical metrics effectively identify severe pollution episodes.

Figure S1 in the Supplement shows boxplots of the concentrations of PM_2.5 and PM₁₀ in temperature (T), relative humidity (RH), surface pressure (PRS), and wind speed (WS) bins, respectively. Overall, the relationships of temperature with PM_2.5, PM₁₀ were not significant (Fig. S1a and e). The RH exerted distinct influences on PM₁₀ and PM_2.5 concentrations (Fig. S1b and f). Higher PM_2.5 concentrations were associated with increased RH, primarily due to elevated humidity promoting the condensation of gaseous precursors (e.g., sulfur dioxide, nitrogen oxides) onto pre-existing PM_2.5 particles (Fig. S1b) (Wang et al., 2004). Additionally, it accelerated secondary aerosol formation via aqueous-phase chemical reactions in the atmosphere, thereby elevating PM_2.5 levels (Yang et al., 2022). In contrast to PM_2.5, the relationship between PM₁₀ concentrations and RH was inverse (Fig. S1f). Higher RH promotes moisture uptake and growth of coarse PM₁₀ particles , accelerating their sedimentation and wet deposition, thereby reducing ambient concentrations (Gao et al., 2020; Ma et al., 2023). PRS tended to increase with rising concentrations of PM_2.5 and PM₁₀ (Fig. S1c and g). The average pressure values were lower on clean days (PM_2.5 ≤ 75 µgm-3, PM₁₀ ≤ 150 µgm-3) and became higher on polluted days. The high-pressure systems promote atmospheric stability and suppress vertical mixing, thereby facilitating the accumulation of particulate matter near the surface (Li et al., 2015). Wind speed exhibited a decreasing trend with increasing PM_2.5 concentrations (Fig. S1d and h). Lower wind speeds during heavy pollution periods indicate weak horizontal dispersion, which contributes to the persistence and accumulation of fine particles in the boundary layer. In addition, the polar plot of PM_2.5 concentrations shows a directional pattern, with the highest values predominantly associated with winds from the northwest (Fig. S2a). The cities of northern Anhui Province, Henan Province, and even further north were characterized by dense industrial activities and frequent wintertime heating emissions, which contribute to elevated PM_2.5 levels (Qian et al., 2024a; Shi et al., 2018). For PM₁₀, the concentration distribution was more spatially confined but still shows a dominant contribution from the northwest area (Huang et al., 2016). The most intense PM₁₀ concentrations were clustered in the northwest, appearing localized and patchy (Fig. S2b).

The vertical wind shear (VWS) is a key dynamic factor affecting pollutant dispersion, as it influences mechanical turbulence and vertical mixing within the boundary layer (Deng et al., 2023). Figures 11 and 12 illustrate the diurnal variation of VWS on polluted and clean days, respectively. For PM_2.5, weak shear persisted throughout the boundary layer on polluted days when temperature inversions and stable stratification typically dominate the lower atmosphere (Fig. 11a–h). Clean days showed stronger shear during most periods, enhancing upward transport and dilution of fine particles (Fig. 12). Notably, the difference in VWS between polluted and clean days was more pronounced in the lower boundary layer (1000–900 hPa) than in the upper boundary layer (875–700 hPa) (Zhang et al., 2020b). Hence, suppressed vertical mixing near the surface contributed significantly to the accumulation of PM_2.5. Surface air quality tends to deteriorate significantly due to pollutant accumulation near the ground, especially in cases where shear is insufficient in the lower layer (Wang et al., 2024), which was consistent with the trend that the extinction coefficient at 532 nm near the surface is higher than that in the upper layer (Figs. 7 and 8).

Contrary to the expectation that polluted days generally exhibited weak VWS, the VWS associated with PM₁₀ on polluted days was not consistently lower than that on clean days (Figs. 11i–p and 12i–p). In particular, dust storms with high PM₁₀ levels are typically accompanied by intensified VWS, which is conducive to dust uplift and transport under dynamic atmospheric conditions (Yang et al., 2019). The large VWS associated with PM₁₀ would clear the air in the upper air at the beginning but later contributed to raising surface PM₁₀ levels through regional transport at the near-surface and through downward transport of aerosol particles from the upper air. During the onset phase of dust intrusion, intensified shear facilitates both horizontal and vertical transport of dust particles (Biuki et al., 2022). From 00:00 to 11:00 (BJT), the VWS in the upper layer (850–700 hPa) gradually increases on PM_2.5-polluted days. At night, the near-surface air cools, forming a stable temperature inversion layer that inhibits vertical mixing. Hence, the stable structure would lead to weak VWS in the upper layer. As the sun rises in the mornings, the inversion layer gradually dissipates, and vertical mixing shifts from being suppressed to active (Fig. 11i–l). The VWS in the upper layer gradually increases due to the mixing effect. In addition, PM₁₀ pollution often occurs during cold front passage. The vertical gradient of wind direction and speed changes suddenly near the frontal surface when cold and warm air masses converge. Once the front is located in a certain pressure range, the wind shear may be significantly higher than that in the surrounding layers. The weaker VWS between adjacent pressure layers (e.g., near 825–850 hPa) may correspond to the fault distribution (Fig. 11i–l). The sinking of cold air also leads to an increase in VWS at the lower level.

To clarify the synoptic influences on PM_2.5 and PM₁₀ pollution, we analyzed the distributions of air temperature across different pressure levels from March 2021 to May 2023. During PM_2.5 pollution days, temperatures at all levels were consistently higher than those on clean days. The warming showed a notable pattern, with the most pronounced trend initially observed at 875 hPa, (Fig. 13b). The enhanced upper-level warming relative to the lower-level led to a thermally stable stratification, which inhibited vertical mixing and promoted the accumulation of PM_2.5 near the surface. In contrast, PM₁₀-polluted days exhibited negative temperature differences below 875 hPa (Fig. 13d), indicating the intrusion of cold air masses. These cold air masses, often accompanied by long-range dust transport, not only bring in PM₁₀ particles from upwind regions but also suppress vertical mixing in the lower atmosphere due to their higher density. Notably, the altitude of the cooling (875 hPa) coincided with the effective aerosol layer height identified in our LiDAR-based analyses (Figs. 8, 11, and 12), which supported the conclusion that dust particles were primarily transported within this level.

To further investigate the vertical thermal and humidity differences between clean and polluted days, we focus on three key pressure levels (1000, 850, and 500 hPa). During PM_2.5-polluted periods, the warming at 850 hPa was greater than that near the surface at 1000 hPa in Hefei (Fig. S3f and i). This vertical temperature structure created a pronounced inversion, which inhibited the vertical mixing and enhanced the atmospheric stability. The lower troposphere also exhibited elevated relative humidity, particularly in Hefei and nearby regions. Elevated relative humidity suppresses turbulent exchange and reduces dilution capacity, and hence forms a shallow, moist, and stable boundary layer that favors pollutant persistence and contributes to poor air quality. In contrast, the clean days showed a lower humidity in the boundary layer and a weaker vertical temperature gradient, allowing enhanced upward motion and effective removal of surface aerosols (Deng et al., 2023). PM₁₀ pollution episodes occurred under a markedly different meteorological background (Fig. S4). Specifically, the relative humidity at both 1000 and 850 hPa was significantly lower on PM₁₀-polluted days than on clean days (Fig. S4f and i). Aloft cold air intrusions occurred on PM₁₀ pollution days, creating favorable conditions for the large-scale transport of dust into Hefei. Under such dry and calm conditions, the accumulation of PM₁₀ is likely driven by mechanical resuspension or regional dust transport rather than secondary aerosol formation processes enhanced by moisture (Li et al., 2020). Overall, these findings reveal that PM_2.5 episodes over Hefei are typically governed by moist and thermodynamically stable boundary layers under subsiding air masses, while PM₁₀ events are more influenced by dust-laden airflow and dry boundary-layer dynamics. The distinct thermal and humidity structures observed across the vertical profile emphasize the importance of differentiating pollution types when diagnosing meteorological drivers.

The vertical wind velocity was investigated at three pressure levels to reveal the dynamic mechanisms on PM_2.5 and PM₁₀ pollution days (Figs. 14 and 15). To better elucidate the meteorological characteristics between polluted and clean conditions, we calculated the differences in vertical velocity between these two scenarios, as their broadly similar large-scale background circulation patterns obscure the subtle but critical dynamic anomalies that drive pollution formation. On PM_2.5-polluted days, the atmosphere around Hefei exhibits a clear subsidence trend at 850 and 1000 hPa (Fig. 14). The subsidence reflects the existence of high-pressure systems and stagnant synoptic conditions, which contribute to the accumulation of fine particles. This inhibition of the upward motion of vertical wind corresponds with a stable thermal structure, which can effectively restrict vertical exchange and compress the boundary layer. During PM₁₀ pollution events, subsidence at 500 hPa is significantly stronger than that on clean days (Fig. 15). However, Hefei is located in a unique region defined by upward motion at 850 hPa, with a comparable weak upward movement also observed at 1000 hPa. The subsidence at 500 hPa stabilizes the atmosphere and suppresses the upward dispersion of particles. Meanwhile, upward motion at 850 hPa and near the surface (1000 hPa) may be caused by local convergence or orographic lifting, allowing the uplift and recirculation of dust within the lower troposphere. This dynamic difference could reflect more complex interactions between local surface sources and large-scale synoptic. Upward motion at 850 hPa supports the interpretation that PM₁₀ events are influenced by weak convective movements or vertical recycling of particles within a confined layer rather than stagnation alone. These findings reveal that PM_2.5 and PM₁₀ pollution in Hefei are governed by contrasting vertical dynamic regimes. PM_2.5 events are closely tied to a uniformly subsiding atmosphere, which favors the trapping and accumulation of fine particles. In contrast, PM₁₀ events are characterized by a layered structure, where upper-level subsidence and lower-level ascent work in tandem to recirculate coarse particles, rather than simply trapping them. This distinction underscores the need for targeted, pollutant-specific strategies in air quality forecasting and management.