Aerosol Vertical Distribution and Typical Air Pollution Episodes over Northeastern China during 2016 Analyzed by Ground-based Lidar

The industrial city of Shenyang in northeastern China has undergone a period of rapid development; long-term aerosol vertical properties could be relevant to more clearly understanding local emissions and their regional transportation. Aerosol optical depth (AOD), planetary boundary layer (PBL) height, and the vertical profiles of extinction coefficient, were measured and analyzed with ground-based Lidar during 2016 in Shenyang. Ground-level particulate matter mass concentrations, meteorological parameters, backward trajectories, and Moderate Resolution Imaging Spectroradiometer products were used to study the pollutant sources in four cases using the potential source contribution function and concentration-weighted trajectory methods. The results indicate that the AOD was 0.10 ± 0.10 to 0.23 ± 0.34 from January to May, and approximately 0.49 ± 0.39 in July. The PBL height was highest in March (1318.7 ± 696.5 m) and lowest in winter (877.1 ± 508.1 m to 950.7 ± 762.3 m). The mass concentrations of PM10, PM2.5, and PM1.0 were highest in January at 148.2 ± 77.8 μg m, 106.0 ± 58.8 μg m, and 33.8 ± 20.5 μg m; and lowest in June at 56.2 ± 27.8 μg m, 33.7 ± 18.3 μg m, and 9.3 ± 6.0 μg m, respectively. The concentrations of SO2 and CO were higher in winter and lower in summer, whereas O3 concentrations were higher in summer and lower in winter. The monthly extinction coefficient was affected by dust events in spring and new particle generation in summer, as well as by biomass-burning and coal-burning emissions in autumn and winter. Four pollution sources—from northwestern, eastern, northern, and northeastern China—were selected to analyze the different paths and sources of pollutants affecting Shenyang. The results of this paper will be helpful in the study of continuous year-round aerosol vertical properties and the regional pollution features of northeast China.

The vertical distribution of aerosols has attracted considerable attention because of the potential of air pollution from the ground to reach the upper atmosphere (Krautstrunk et al., 2000;Kompalli et al., 2014;Sugimoto et al., 2015;Oleniacz et al., 2016).A ground-based lidar device is a direct remote sensing instrument with high spatial and temporal resolution that can provide vertical aerosol profiles (Tesche et al., 2007;Heese and Wiegner, 2008;Wu et al., 2012;Hänel et al., 2012;Cottle et al., 2014;Noh et al., 2017).The continental-scale lidar network, EARLINET (the European Aerosol Research Lidar Network), has been established to provide aerosol optical profiles over large temporal and spatial scales (Pappalardo et al., 2004(Pappalardo et al., , 2013)).
In China, aerosol vertical distributions have been studied in the majority of regions (Huang, et al., 2012;Quan et al., 2013;Sun et al., 2013;Tang et al., 2015;Xia et al., 2015;Zheng et al., 2015;Tang et al., 2016).Wu et al. (2013) and Ansmann et al. (2005) have presented the vertical structure of atmospheric layers in the Pearl River Delta in southern China to examine regional air.Qin et al. (2016) observed external aerosol transportation in three regions of eastern China using ground-based lidar.Uno et al. (2014) observed a shallow aerosol layer over Beijing during a pollution event in January 2013 from ground-based lidar data.However, few studies have focused on the aerosol vertical properties detected by ground-based lidar in northeastern China (Zhao et al., 2013a, b;Hu et al., 2014;Che et al., 2015;Zhao et al., 2015).One of the few such studies was conducted by Zhao et al. (2017) who calculated the atmospheric mixing layer height (MLH) under hazy and foggy conditions in a number of cities in northeastern China.
The present study investigated the aerosol vertical properties detected by ground-based lidar in northeastern China from January to December 2016.This is the first long-term study of aerosol vertical properties characterized by ground-based lidar in northeastern China.The aim of this paper is to provide continuous vertical aerosol data over a whole year in northeastern China, to enhance the understanding of regional transportation and local emissions.

SITES, METHOD, AND DATA
Shenyang is the main political, economic, and cultural center of northeastern China (41.77°N,123.50°E,60.0m a.s.l.), and is representative of the human activities and local industrial emissions of urban industrial areas in this region.The observation site was located on the roof of the Northeast Regional Meteorological Observation Center of China.
The ground-based Lidar-D-2000 (Wuxi CAS Photonics Co., Ltd, China) has been installed at the observation site since December 2015 and was standardized and calibrated during this period to ensure that the data produced were quality controlled.This lidar provides backscattering at 532 nm and 355 nm based on Mie scattering theory, with a temporal resolution of approximately 1 min and a vertical resolution of approximately 7.5 m.
In this study, visibility, AOD, PBL heights and aerosol extinction coefficients were derived from the backscatter signal at 532 nm during 2016.The planetary boundary layer is about 1-1.5 km from the ground and closely related to the temperature distribution or atmospheric stability in the atmosphere.We used the mixing layer height to describe the effect of mechanical or thermal turbulence between the upper and lower layers in the boundary layer.The MLH was calculated based on different grades of atmospheric stability (A, B, C, D, E, and F) according to dynamic and thermodynamic factors (Zhao et al., 2017).The MLH (h) can be calculated as h = a s U 10 /f when the atmospheric stability grade is A, B, C, or D; when the atmospheric stability grade is E or F, the MLH can be calculated as h = b s √U 10 /f.U 10 denotes the wind speed at 10 m (m s -1 ); a s (0.037, 0.06, 0.041, and 0.019 for grades A, B, C, and D) and b s (1.66 and 0.70 for grades E and F) represent the mixing layer coefficients in China; f is the Earth rotation parameter; Ω denotes the angular velocity (7.29 × 10 -n rad/s), and φ indicates the geographic latitude (°) with the function f = 2Ωsinφ.
The PM mass concentration (including PM 10 , PM 2.5 , and PM 1.0 ) and the concentrations of gaseous pollutants (including SO 2 , CO, and O 3 ) were obtained from the China air quality online monitoring and analysis platform (https://www.aqistudy.cn/);vertical profiles of wind speed and temperature were obtained from the University of Wyoming (http://weather.uwyo.edu/).
To determine the air source, 72-h back trajectories were calculated using the TrajStat software (http://www.meteothinker.com/products/trajstat.html).Two models with a weight function applied-potential source contribution function (WPSCF) and concentration-weighted trajectory (WCWT)were used in this study to illustrate the spatial distributions of potential PM 2.5 sources on the prevailing transport trajectories (Wang et al., 2006;Xin, 2016).The daily averaged PM 2.5 mass concentrations and the PBL height data, as well as the wind field at 10 m, from the European Centre for Medium-Range Weather Forecasts' ERA-Interim dataset (http://apps.ecmwf.int/datasets/)were obtained at a grid resolution of 0.5° × 0.5° to analyze the variations in the regional weather field during polluted days.The spatial distribution of AOD was illustrated by the daily data from the Moderate Resolution Imaging Spectroradiometer (MODIS) Level 3 Atmosphere Products using methods combining the dark target and deep blue algorithms at 550 nm (Levy et al., 2007).

Variation of Monthly Visibility, AOD, PBL Height, MLH, PM, and Gaseous Pollutants in Shenyang
The monthly variations of visibility, AOD, PBL height and MLH are shown in Figs.1(a)-1(d).Visibility was higher in March, April, and May, with values from 22.3 ± 12.8 km to 24.3 ± 11.0 km, and lower in July, with a value of approximately 14.1 ± 10.4 km.The higher visibility in spring and autumn may have been due to pollutants being dispersed by the stronger winds.The decreased visibility in summer was partly due to the increasing water content in the atmosphere, while the increase in fine-mode particles with higher scattering ability was also influential in lowering visibility (Xia et al., 2007;Zhang et al., 2015).Visibility began to decrease from September, with a value of 16.9 ± 12.1 km, reaching the minimum documented visibility in December, with an average value of approximately 11.2 ± 9.1 km.Visibility degradation in winter was partially caused by an increase in the use of coal combustion to provide heating as the temperature dropped, and the corresponding rise in pollutants.AOD did not increase as much as visibility decreased, which indicates that the vertical column of aerosol in this region was relatively constant.
Visibility was more strongly influenced by the concentration of pollutants at the surface.As Fig. 1 The relatively low MLH from September to December (from 434.8 ± 500.0 m to 403.5 ± 372.7 m) also contributed to the accumulation of pollution, resulting in worse air quality conditions.Fig. 1(e) shows the monthly variations of coarse-and fine-mode particles.The coarse-and fine-mode particles followed the same trend, being lower in summer and higher in winter.This result shows that the level of the PM concentration in this region is constant.The highest mass concentrations of PM 10 (148.2 ± 77.8 µg m -3 ), PM 2.5 (106.0 ± 58.8 µg m -3 ) and PM 1.0 (33.8 ± 20.5 µg m -3 ) occurred in January.The higher particle loadings in winter were caused by local emissions from heating sources, coupled with a temperature inversion that acted to hinder the diffusion of pollutants.Moreover, the low PBL height is another important contributing factor in the deterioration of air quality in winter (Sun et al., 2004;Guinot et al., 2007).The strong winds in spring enhancing wind erosion and resuspended dust may have been the cause of higher aerosol loadings in the atmosphere at that time of year.The concentrations of PM 10 , PM 2.5 , and PM 1.0 at ground level were lower in June, at 56.2 ± 27.8 µg m -3 , 33.7 ± 18.3 µg m -3 , and 9.3 ± 6.0 µg m -3 , respectively.The higher precipitation in the summer likely reduced the concentration of aerosols in this month through wet deposition.This result also implies that fine particles were not the main factor affecting visibility in the summer, with higher levels of water vapor in the air being the likely cause of the poorer visibility.
Figs. 1(f)-(h) illustrate the monthly variations of gaseous pollutants.The concentration of SO 2 was notably higher in January (1.12 ± 0.37 µg m -3 ) and December (1.44 ± 0.62 µg m -3 ), and lower during the rest of the year (0.72 ± 0.25 µg m -3 to 1.00 ± 0.38 µg m -3 ).The higher SO 2 concentrations in winter may have been due to coal combustion from central heating activities in northeastern China.At the end of the heating period, the SO 2 concentrations began to decline.As Fig. 1(g) shows, higher CO concentrations were also apparent in winter, with values of 50.52 ± 11.26 mg m -3 and 57.42 ± 17.26 mg m -3 in January and December, respectively, whereas lower CO concentrations were recorded in July, with a value of 27.61 ± 9.09 mg m -3 .The increase of CO from September to December is likely to have been due to biomass burning and heating sources in winter, which can lead to high levels of CO regionally.Unlike the other gases, the variation in O 3 showed no clear pattern, as Fig. 1(h) shows.The lowest concentrations of O 3 occurred in February and September, with values of 206.10 ± 82.51 µg m -3 and 164.8 ± 81.48 µg m -3 , respectively.The highest concentration of O 3 occurred in July, reaching 259.48 ± 93.87 µg m -3 .In other months, O 3 concentrations were fairly constant at approximately 220-240 µg m -3 .Strong solar radiation is one of the main factors that causes an increase in O 3 concentrations in summer.

Variations of Diurnal Visibility, AOD, MLH and PM in Shenyang
Fig. 2 depicts the diurnal patterns of visibility, AOD, MLH and PM.Visibility showed a minimum (14.4 ± 10.7 km) at approximately 10:00 (LT) in the morning (Fig. 2(a)).Visibility gradually increased along with the reduced anthropogenic activity following the morning rush hour.The diurnal variation of AOD and visibility exhibit opposite trends (Fig. 2(b)).A higher AOD occurred at approximately 12:00 (LT), with a value of 0.37 ± 0.36.The pollutants from the morning rush hour accumulated to a maximum at noon, which led to the maximum AOD values.The MLH exhibits a strong unimodal distribution pattern (Fig. 2(c)),  , respectively, at 21:00 (LT), which corresponded to the morning and evening rush hours.The particle mass concentration decreased in the afternoon, with the concentrations of PM 10 , PM 2.5 and PM 1.0 at 14:00 (LT) being 63.7 ± 66.6 µg m -3 , 40.4 ± 41.7 µg m -3 , and 11.0 ± 9.7 µg m -3 , respectively.The increasing concentration of PM at night corresponds to a significant reduction in the MLH, with the value being approximately 0.5 km at 21:00 (LT).Therefore, ground-level PM concentration may be strongly related to the aerosol vertical transport at this time.

Variations in the Monthly Series of Extinction Coefficient in Shenyang
The extinction coefficient at 532 nm was derived from the time-height cross-section of the ground-based lidar.There was a clear monthly variation in the extinction coefficient, as Fig. 3 shows.
The extinction coefficient was 0.31 km -1 to 0.38 km -1 at 0.2 km in January and February, respectively.However, a lower aerosol loading is suggested by the small extinction coefficient of 0.16 km -1 at 0.4 km in March.This phenomenon may be due to the diffusion effect of the larger surface wind speeds in the spring.In April, the extinction coefficient had two maxima: 0.23 km -1 at 0.5 km and 0.16 km -1 at 1.5 km.The extinction coefficient then became high overall, at approximately 0.18 km -1 , below 1.0 km in May.The extinction coefficient began to increase both at the surface and in the upper atmosphere from April to May due to frequent dust events from local and regional transportation.Dickerson et al. (2007) reported the meteorological Fig. 3. Monthly vertical profiles of the extinction coefficient during 2016 in Shenyang.Zhao et al., Aerosol and Air Quality Research, 18: 918-937, 2018 924 mechanisms of dust transportation during aircraft observations over northeastern China.Wu et al. (2012) noted that dust aerosols from the northwest contributing to episodic aerosol events were mainly from a northwesterly direction.Wang et al. (2015) and Pu et al. (2015) have determined that dust storms traveling far from their original sources could have a significant influence on the local air quality.Thus, the dust aerosol layer could comprise many sources during transportation to northeastern China in the spring.Particularly, the aerosol layer shows a vertical distribution of two sublayers in June, with the corresponding extinction coefficients at 0.2 km and 1.0 km being 0.20 km -1 and 0.34 km -1 , respectively.This contrasts with the unimodal distribution in the other months, indicating that the upper aerosol layer may have been formed due to secondary particles generated by photochemical reactions in summer and then transported vertically by turbulence (Wu et al., 2015).According to Zhao et al. (2017), the longer and stronger solar irradiation in summer could lead to higher fine-particle levels in the northeastern China.With the onset of the biomass-burning season in northeastern China and the surrounding area, the altitude of the near-surface aerosol pollution increased from July to October.Cheng et al. (2010) found that a high BC concentration in October at Tongyu in northeastern China was attributable to biomass burning during harvest time.The mixing state of the extinction coefficient distribution indicates the contribution of both human activities and natural primary particle emissions to the aerosol layers.Li et al. (2013) indicated that biomass-burning emissions and activities for keeping warm were both sources of aerosol during October in northeastern China.The ground-level extinction coefficient exhibited a clear increasing trend in November and December.In November, the peak value of the extinction coefficient was 0.56 km -1 at 0.2 km.In December, the extinction coefficient reached the annual maximum value of 0.69 km -1 at 0.2 km.Coal-burning emissions, with corresponding meteorological conditions, contributed greatly to the aerosol pollution in winter months.

Case Studies
Surface observations and vertical profiles of aerosols were obtained to analyze the variational characteristics and potential sources in four pollution cases during 2016 in Shenyang.Based on the different pollution sources from northwestern, eastern, northern and northeastern China, the four cases selected were 4 to 7 March 2016, 30 March to 2 April 2016, 23 to 26 October 2016, and 5 to 8 November 2016.

Major Dust Pollutants from Northwestern China
The pollution process of 4-7 March 2016 represents a dust transportation event.During this period, the maximum daily averaged PM 2.5 of 168.2 ± 108.3 µg m -3 occurred on 6 March 2016 (Fig. 4(a)).The values of PM 10 , PM 2.5 , and PM 1.0 reached 547.9 µg m -3 , 436.0 µg m -3 , and 138.2 µg m -3 , respectively, at 08:00 (LT) in the early daytime (Fig. 4(e)).In the vertical profiles of temperature and wind speed at 08:00 (LT) and 20:00 (LT), a temperature inversion (0.5 km) and a low surface wind speed (approximately 1.0 m s -1 ) were observed (Figs. 5(a) and 5(e)).These conditions were not conducive to the dispersion of pollutants, and the aerosol extinction was relatively high near the surface.The wind and pressure fields at the surface and at 850 hPa revealed that Liaoning province, of which Shenyang is the capital city, was in the northwest flow field due to the combined effect of a low vortex and a cold anticyclone on the west side of Lake Baikal, with a wind speed of 10-16 m s -1 on 6 March 2016.These weather conditions were favorable to dust transportation from Inner Mongolia to eastern Liaoning province.
Fig. 6(a) shows the vertical structure evolution from the lidar in this case, and the aerosol extinction coefficient at 0.2-5.0km is shown in Fig. 7(a).The distribution of the extinction coefficient shows two aerosol layers, at 0.6 km and 1.0 km, with extinction coefficients of 2.3 km -1 and 1.6 km -1 , respectively, on 4 March.The ground-level pollution worsened on 5 March, with an extinction coefficient of 1.8 km -1 at 0.1 km, as well as two other aerosol layers at 0.5 km and 0.8 km, with extinction coefficients of 1.0 km -1 and 1.1 km -1 , respectively.This multipeak distribution of aerosol shows the pollution sources at different heights.The extinction coefficient decreased considerably, to 0.4 km -1 , at 0.1 km on 6 March, while the vertical profile maintained its layered structure up to 3 km, with the next highest value of 0.35 km -1 at 1.8 km.On 7 March, there was an apparent extinction layer in the upper atmosphere, with an extinction coefficient of 0.9 km -1 at 1.6 km, indicating transportation at higher altitude.Two lower aerosol layers were observed, at 0.3 km and 0.5 km, with extinction coefficients of 1.0 km -1 and 1.4 km -1 , respectively.
The trajectories of air pollutants on 6 March 2016 were grouped into two main clusters, as depicted in Fig. 8(a).Cluster-1 and cluster-2 constitute 70.83% and 29.17%, respectively.Cluster-1 indicates that the major sources of pollution came from northwestern China, via central Inner Mongolia, and were directly transmitted eastward to Shenyang.Cluster-2 shows another route of travel, with a lower contribution, to Shenyang.This indirect route began in central Inner Mongolia, moved southward to the Beijing-Tianjin-Hebei region and the central part of the Shandong Peninsula, and then headed southwest to Shenyang.The most likely source areas covered two regions, one being central Inner Mongolia and the other most of the Beijing-Tianjin-Hebei region and the central part of the Shandong Peninsula.In both regions, the WPSCF values for PM 2.5 were over 0.9 (Fig. 8(e)).In addition, the spatial distribution of the WCWT indicates that the main contributions to PM 2.5 were from Inner Mongolia, with high aerosol loadings of more than 280 µg m -3 (Fig. 8(i)), whereas the northern part of the Beijing-Tianjin-Hebei region and the central part of the Shandong Peninsula had lower values of approximately 100-130 µg m -3 , thus contributing less PM 2.5 .
The spatial distribution of the daily averaged PM 2.5 mass concentration and the daily wind field at 10 m, as well as the PBL height, are shown together in Fig. 9(a).The figure shows that the northwest wind prevailing over northeastern China during this case brought the pollutants to Shenyang    from northwestern China.In another route, the southwest wind brought pollutants from northern and eastern China.The pollution was more severe when the PBL height was 200 m, but was dispersed when the PBL height increased to 1200 m.Although the aerosol loading was not as high as that in other eastern areas, the corresponding daily spatial distribution of AOD was approximately 0.45-0.65 in Shenyang, which was higher than the surrounding areas in Liaoning province, as recorded by the MODIS (Fig. 10(a)).

Major Anthropogenic Pollutants from Eastern China
This next case discusses the mixed pollution transported from northwestern and eastern China to Shenyang.The daily averaged PM 2.5 concentration was higher on 31 March and 1 April, with values of 115.8 ± 158.4 µg m -3 and 110.5 ± 85.4 µg m -3 , respectively (Fig. 4(b)).The hourly concentration of PM 10 , PM 2.5 , and PM 1.0 showed a significant increase to maximum values of 898.9 µg m -3 , 532.9 µg m -3 , and 100.7 µg m -3 at 13:00 (LT) on 31 March (Fig. 4(f)).Temperature inversion occurred at approximately 1.5 km, hindering the dispersion of pollutants in the upper atmosphere.In addition, the wind speed was higher at the ground (approximately 5 m s -1 ), favoring the vertical diffusion of pollutants from the surface, resulting in a decrease in the aerosol extinction at the near surface, and an increase in the extinction in the upper atmosphere (Figs. 5(b) and 5(f)).From the wind and pressure fields at the surface and at 850 hPa on the night of 31 March, Liaoning province was at the front of the southwest flow field at the end of a low vortex, with southwest winds reaching from 8 to 12 m s -1 at 850 hPa.The ground flow field was still affected by the southerly wind at a low pressure front with sustained wind speeds of approximately 4 m s -1 .
Fig. 6(b) shows the aerosol extinction coefficient evolution process from the vertical structure of the lidar measurements.The extinction coefficient at 0.2-5.0km was weaker, with a value of approximately 0.1 km -1 at 1.0 km on 30 March (Fig. 7(b)).On 31 March, the distribution of the aerosol extinction grew from the ground to 1.0 km, with a maximum extinction coefficient of 0.3 km -1 occurring at 0.6 km.A mixed-status layered structure was observed on 1 April, showing pollution mixing between the ground and upper atmosphere.The maximum extinction coefficient of 1.5 km -1 occurred at 0.7 km.The extinction coefficient in the upper atmosphere was approximately 0.4 km -1 at 1.0-2.0km.From 2 April, higher extinction coefficients were observed above 2 km, with a maximum value of 0.8 km -1 at 2.1 km, clearly indicating high-altitude transportation.The extinction coefficient gradually decreased with height between 2.0 km and 5.0 km.
The trajectories of air pollutants on 31 March in 2016 were also grouped into two clusters, as depicted in Fig. 8(b).The air pollutants originating from cluster-1 were from eastern China, including the north of Shandong Peninsula, contributing the highest proportion at 75.00%.Cluster-2 contributed 25.00%, and came from northwestern China.The spatial distribution of potential PM 2.5 sources is shown with WPSCF values in Fig. 8(f).The most likely source area, with WPSCF values greater than 0.8, corresponded to a substantial portion of eastern China (Henan and Shandong Peninsula).However, cluster-2 made almost no contribution to the pollutants from northwestern China.
Temperature inversion occurred at 0.5 km at 08:00 (LT) and 20:00 (LT), along with a low near-surface wind speed of 2.0 m s -1 (Figs.5(c) and 5(g)).These weather conditions were not conducive to the dispersion of pollutants near the ground.A low vortex located in eastern Inner Mongolia strengthened gradually and formed a closed system at 850 hPa, to the east of Liaoning Province, on 24 October.The wind at ground level gradually changed from northerly to southerly, with wind speeds of approximately 2-4 m s -1 in Liaoning Province, transporting pollutants from northern China. Fig. 6(c) shows the vertical profile of the aerosol extinction coefficient derived from the lidar measurements during this period.The extinction coefficient at 0.2-5.0km indicates no marked aerosol extinction layer on 23 October (Fig. 7(c)).Multilayer aerosol pollution was observed at 1.0 km, 2.3 km and 3.0 km, with extinction coefficients of approximately 0.4 km -1 , 0.2 km -1 , and 0.2 km -1 , respectively, on 24 October.This multipeak distribution of extinction coefficients represents aerosol contamination layers from different sources in the upper layer.The maximum aerosol pollution layer occurred near ground level on 25 October.The pollution aerosol layer shows a separation into two sublayers that stretched between 0.2 km and 0.6 km, with extinction coefficients of approximately 1.7 km -1 and 0.6 km -1 , respectively.An aerosol extinction layer was also present in the upper atmosphere at 2.4 km, with a coefficient of 0.1 km -1 .On 26 October, a clear unimodal distribution of extinction coefficients in the upper atmosphere was present at 3.3 km, with a value of approximately 0.4 km -1 .This structure indicates high volume transportation of aerosol in the upper layer.
The transportation of air pollutants was detected in two trajectories on 24 October 2016, as shown in Fig. 8(c).Cluster-1 air pollutants came from northwestern China, contributing a large proportion (66.67%), and traveling via eastern Inner Mongolia southward to Shenyang.Cluster-2 stemmed from northern China and the south-central Beijing-Tianjin-Hebei region, and accounted for 33.33%.The WPSCF values in Fig. 8(g) reveal that the most likely PM 2.5 source areas, with WPSCF values of more than 0.9, were concentrated in the Beijing-Tianjin-Hebei region, and the contribution from northwestern China was relatively small.The daily averaged PM 2.5 mass concentrations and the wind field at 10 m shows that the southwest wind prevailing over northeastern China brought anthropogenic pollutants from northern China to Shenyang during this case (Fig. 9(c)).The corresponding PBL height in the wind fields at 10 m varied from 400 m to 1200 m during the pollution process.In this period, the daily spatial distribution of AOD from MODIS in Shenyang was lower than the surface particle concentrations, which may have been due to the complex and changeable weather conditions (Fig. 10(c)).

Accumulation of Local Pollutants in Northeastern China
In the final case, the accumulation of locally emitted pollutants in northeastern China was examined.The maximum daily averaged PM 2.5 mass concentration was 112.3 ± 58.9 µg m -3 on 5 November 2016 (Fig. 4(d)).The hourly PM 10 , PM 2.5 , and PM 1.0 mass concentrations increased to peak values of 355.1 µg m -3 , 256.6 µg m -3 , and 91.1 µg m -3 , respectively, at 08:00 (LT), when temperature inversion occurred at approximately 2.0 km and the lower vertical wind speed was approximately 2.2 m s -1 (Figs.5(d) and 5(h)).Temperature inversion in the upper layer may also have contributed to preventing the diffusion of pollutants transported from the ground below 1.5 km.In the daytime of 5 November, the system moved gradually into the north flow field at the back of the transverse groove.The ground field was also affected by the northerly winds behind the low pressure, in which pollutants were easily trapped.   . 6(d) shows the vertical structure evolution of the lidar in this case, and the aerosol vertical distributions from 0.2 km to 5.0 km are shown in Fig. 7(d).Unlike the other events, the aerosols accumulated significantly near the surface in this case, with the peak extinction coefficients below 2.0 km.The extinction coefficient was higher near the ground on 5 November.The peak value of the extinction coefficient was approximately 2.0 km -1 at 0.2 km, with a thicker aerosol layer from the ground to 1.0 km.On 6 November, the extinction coefficient near the surface decreased rapidly and showed a marked distribution at high altitude.The peak value of the extinction coefficient was approximately 1.1 km -1 at 0.5 km on 7 November.The aerosol had a multilayer structure, with an upper and a lower layer.The highest value of the extinction coefficient was approximately 2.9 km -1 at 0.6 km, with the extinction coefficient in the upper layer being approximately 0.8 km -1 at 1.1 km.On 8 November, the upper aerosol layer gradually decreased and the lower layer was thicker from the ground to 1.0 km, with a smaller extinction coefficient of approximately 0.2 km -1 .
The trajectories of the air pollutants on 5 November 2016 were grouped into two clusters, as depicted in Fig. 8(d).
Cluster-1 accounted for the largest proportion at 62.50%, and came from biomass-burning events in northeastern China, proceeding via Heilongjiang and Jilin provinces, southward to Shenyang.Cluster-2 brought pollutants from another source from northwestern China, and accounted for 37.50%.These patterns of air mass trajectories were mainly associated with local pollution caused by human activities in northeastern China.The WPSCF pattern in Fig. 8(h) shows that the most likely PM 2.5 source areas cover most of Liaoning, Jilin, and Heilongjiang, with WPSCF values over 0.9.The other PM 2.5 source came from Inner Mongolia, with smaller WPSCF values of less than 0.5.Fig. 8(l) clearly demonstrates that the source regions with the largest contribution of PM 2.5 were Liaoning and Heilongjiang, with a WCWT value of approximately 160 µg m -3 .The northeastern part of Inner Mongolia had lower WCWT values (less than 100 µg m -3 ), and is thus considered the area contributing the least PM 2.5 .
The spatial distribution of the daily averaged PM 2.5 mass concentration and the wind field at 10 m shows that a northeastern wind was prevailing over northeastern China during this case, bringing biomass-burning pollutants from Harbin southwest to Shenyang (Fig. 9(d)).The daily wind fields at 10 m and the PBL height also show an increase in the PBL height from 400 m to 1200 m with the decrease in pollution.The corresponding spatial distribution of daily AOD from MODIS shows a considerably higher pollution level in Shenyang and its surrounding area, with AOD values of more than 1.45 (Fig. 10(d)).

SUMMARY
In this study, long-term observations of vertical profiles were obtained from ground-based lidar measurements in Shenyang during 2016.The corresponding data of PM mass concentrations, meteorological data, and MODIS products were used to analyze the pollutant sources and regional transportation in Shenyang, based on the backward trajectory, PSCF, and CWT methods.
The results indicate that high aerosol concentrations and lower PBL height were the main factors contributing to visibility deterioration.Diurnal visibility gradually increased with reduced anthropogenic activity following the morning rush hour, with an opposite trend in AOD.The features of the MLH displayed a unimodal distribution pattern, with a peak at 13:00 (LT) of approximately 1208.6 ± 637.0 m.The coarse-and fine-mode particles were lower in summer and higher in winter, and exhibited a bimodal diurnal pattern with peaks at 08:00 (LT) and 21:00 (LT).The concentrations of SO 2 and CO were higher in winter and lower in summer, whereas O 3 concentrations were higher in summer and lower in winter.The distribution of the extinction coefficient indicated contributions from both human activities and natural primary particle emissions to the aerosol layers.The maximum monthly extinction coefficient of 0.69 km -1 occurred in December at 0.2 km, and was due to coalburning emissions in winter.
Four cases were selected according to the different pollution sources from northwestern, eastern, northern and northeastern China.During 4-7 March 2016, the wind and pressure fields at the surface and at 850 hPa indicated favorable conditions for dust transportation from Inner Mongolia to eastern Liaoning province, with a multipeak distribution in the extinction coefficient showing the pollution sources of different levels in the atmosphere.During 30 March to 2 April 2016, Liaoning Province was in front of the southwest flow field, with southwest winds of 8 to 12 m s -1 at 850 hPa bringing pollutants from eastern China, and a mixed status layered structure showing pollution mixing between the ground and the upper atmosphere.During 23-26 October 2016, the wind at ground level gradually changed from northerly to southerly, with wind speeds of approximately 2-4 m s -1 in Liaoning Province, transporting pollutants from northern China and with a peak layer structure in the upper layer.During 5-8 November 2016, the ground field was affected by the northerly winds behind an area of low pressure, in which pollutants were easily trapped at ground level, with the highest extinction coefficients below 2.0 km.

Fig. 1 .
Fig. 1.Monthly variations of (a) visibility, (b) AOD, (c) PBL height, (d) MLH, (e) PM, (f) SO 2 , (g) CO, and (h) O 3 during 2016 in Shenyang.The boxes represent the 5 th to 95 th percentile distribution with the dots and solid lines within each box representing the mean and median, respectively.

Fig. 2 .
Fig. 2. Diurnal variations of (a) visibility, (b) AOD, (c) MLH, and (d) PM during 2016 in Shenyang.The boxes represent the 5 th to 95 th percentile distribution with the dots and solid lines within each box representing the mean and median, respectively.

Fig. 8
Fig. 8(k) shows the two main source regions potentially affecting PM 2.5 in Shenyang.The spatial distribution of the WCWT values greater than 130-160 µg m -3 indicates that the pollutants came from the northern part of China.The lower WCWT values of approximately 70-100 µg m -3 from Inner Mongolia suggest that this region made a lower contribution to PM 2.5 .The daily averaged PM 2.5 mass concentrations and the wind field at 10 m shows that the southwest wind prevailing over northeastern China brought anthropogenic pollutants from northern China to Shenyang during this case (Fig.9(c)).The corresponding PBL height in the wind fields at 10 m varied from 400 m to 1200 m during the pollution process.In this period, the daily spatial distribution of AOD from MODIS in Shenyang was lower than the surface particle concentrations, which may have been due to the complex and changeable weather conditions (Fig.10(c)).

Fig
Fig. 6(d)  shows the vertical structure evolution of the lidar in this case, and the aerosol vertical distributions from 0.2 km to 5.0 km are shown inFig.7(d).Unlike the other events, the aerosols accumulated significantly near the surface in this case, with the peak extinction coefficients below 2.0 km.The extinction coefficient was higher near the ground on 5 November.The peak value of the extinction coefficient was approximately 2.0 km -1 at 0.2 km, with a thicker aerosol layer from the ground to 1.0 km.On 6 November, the extinction coefficient near the surface decreased rapidly and showed a marked distribution at high altitude.The peak value of the extinction coefficient was approximately 1.1 km -1 at 0.5 km on 7 November.The aerosol had a multilayer structure, with an upper and a lower layer.The highest value of the extinction coefficient was approximately 2.9 km -1 at 0.6 km, with the extinction coefficient in the upper layer being approximately 0.8 km -1 at 1.1 km.On 8 November, the upper aerosol layer gradually decreased and the lower layer was thicker from the ground to 1.0 km, with a smaller extinction coefficient of approximately 0.2 km -1 .The trajectories of the air pollutants on 5 November 2016 were grouped into two clusters, as depicted in Fig.8(d).Cluster-1 accounted for the largest proportion at 62.50%, and came from biomass-burning events in northeastern China, proceeding via Heilongjiang and Jilin provinces, southward to Shenyang.Cluster-2 brought pollutants from another source from northwestern China, and accounted for 37.50%.These patterns of air mass trajectories were mainly associated with local pollution caused by human activities in northeastern China.The WPSCF pattern in Fig.8(h)shows that the most likely PM 2.5 source areas cover most of Liaoning, Jilin, and Heilongjiang, with WPSCF values over 0.9.The other PM 2.5 source came from Inner Mongolia, with smaller WPSCF values of less than 0.5.Fig.8(l)clearly demonstrates that the source regions with the largest contribution of PM 2.5 were Liaoning and Heilongjiang, with a WCWT value of approximately 160 µg m -3 .The northeastern part of Inner Mongolia had lower WCWT values (less than 100 µg m -3 ), and is thus considered the area contributing the least PM 2.5 .The spatial distribution of the daily averaged PM 2.5 mass concentration and the wind field at 10 m shows that a northeastern wind was prevailing over northeastern China during this case, bringing biomass-burning pollutants from Harbin southwest to Shenyang (Fig.9(d)).The daily wind fields at 10 m and the PBL height also show an increase in the PBL height from 400 m to 1200 m with the decrease in pollution.The corresponding spatial distribution of daily AOD from MODIS shows a considerably higher pollution level in Shenyang and its surrounding area, with AOD values of more than 1.45 (Fig.10(d)).