Characteristics of Pollutants and Boundary Layer Structure during Two Haze Events in Summer and Autumn 2014 in Shenyang , Northeast China

The characteristics of pollutants and the boundary-layer structure during two haze events in the summer and autumn of 2014 in Shenyang, Northeast China, were comparatively analyzed by using measurements of the mass concentrations of PM10, PM2.5, O3, NO2, SO2, and CO; vertical profiles of meteorological parameters from a 100-m high tower; and radiosonde data. The results showed that PM concentrations increased rapidly during the two haze events, resulting in visibility that decreased to 1400 and 405 m, respectively. The weak haze event on 16 June was characterized by high O3 but low NO2 mainly due to the photochemical reaction, while all the pollutants increased during the severe haze event on 31 October, which was affected by pollutant emissions and meteorological conditions. The PM2.5 concentration had a good correlation with friction velocity (u*) but did not have an obvious relationship with ' ' w  , which means that the haze events were largely affected by the dynamic effect of turbulence and less so by its thermal effect. According to the radiosonde data, a single inversion layer with an inversion intensity of 1.6°C/100 m existed during the weak haze event, whereas double inversions and even more occurred during the severe haze event, with the inversion intensity larger than 2–4°C/100 m. Such stable atmospheric conditions favored the accumulation of pollutants. Backward trajectory analyses showed that the weak haze event was probably caused by pollutant transport from North China, whereas the severe haze event was generated mostly by local pollutants.


INTRODUCTION
Haze is a phenomenon that usually occurs in the nearsurface layer and consists of fine airborne aerosols, including sulfate, nitrate, ammonium, particulate organic matter, and black carbon, as well as other chemical species (Liu et al., 2013).Haze pollution can reduce horizontal visibility, degrade air quality, cause harm to human health, and influence the atmospheric radiation budget by scattering and absorbing incident sunlight (Tie et al., 2009;Zhang et al., 2014;Tang et al., 2017).The atmospheric boundary layer (ABL) is an important meteorological factor that affects air pollution (Tang et al., 2016;Yang et al., 2017), and the vertical distributions of meteorological parameters within ABL strongly determine pollutant concentrations in the near-surface layer as well as their diffusion, reaction, settlement, and other processes (Sun et al., 2013;Ye et al., 2015;Platis et al., 2016).Moreover, the thermal and dynamic processes of ABL play different roles on the formation and Hebei region.Their results indicated that high concentrations of particles in the residual layer can lead to severe pollution the next morning through convective transport of particles to the surface.
However, such studies have rarely been conducted in Northeast China, mainly due to a lack of simultaneous measurements of pollutant concentrations and vertical distributions of meteorological parameters in the nearsurface layer (Wang et al., 2010;Fang et al., 2017).In fact, atmospheric haze problems have increased in Northeast China under the national five-year development plan in the region; thus, this area may become the fifth severe haze region (Chen et al., 2016).An extreme haze event on 8 November 2015 affected the public over an expansive area and showed an ultra-high PM 2.5 concentration in some locations such as Shenyang (http://news.sina.com.cn/c/nd/2015-11-08/doc-ifxknutf1607479.shtml).However, available studies on the haze pollution in Northeast China mostly focused on the distribution characteristics, chemical compositions, and optical properties of aerosols (Hong et al., 2011;Ma et al., 2011;Zhang et al., 2012;Zhao et al., 2015;Chen et al., 2016).
This work comparatively study the characteristics and evolution of air pollutants and the near-surface layer structure during a summer weak haze event and an autumn severe haze event in 2014 in Shenyang, mainly using the air quality monitoring data, the vertical distributions of meteorological parameters obtained from a 100-m tower, and radiosonde data.Formation mechanism of the two haze events were also discussed by investigating the relationship between air pollutants and meteorological parameters, the dynamic and thermal effect by turbulence, and the backward trajectory analyses.

Observations of Pollutant Mass Concentrations
The experiment was conducted in Shenyang, a large city in Northeast China and the provincial capital city of Liaoning province (Fig. 1(a)).The total population of Shenyang was over 8.29 million at the end of 2015 (Shenyang Statistics Bureau, 2016).A total of eleven air quality monitoring stations were located in different regions of Shenyang and mostly concentrated in the central urban region (Fig. 1(b)).Mass concentrations of atmospheric particles (PM 10 and PM 2.5 ) and gaseous pollutants (SO 2 , NO 2 , O 3 , and CO) were measured automatically and continuously at these stations, and AQI values were calculated from six pollutant concentrations according to the latest ambient air quality standards in China published in 2012(GB3095-2012).All observational data were reported hourly on the Liaoning real-time air quality publishing system (http://211.137.19.74:8089/).The averaged data of the eleven stations were used for further analysis.

Vertical Measurements of Meteorological Parameters
A 100-m high meteorological tower (black square in Fig. 1(b)) was set up in the southern urban region of Shenyang and belonged to Liaoning Meteorological Office.The landscape around the tower was virtually flat, with only one small building (< 10-m height) located to the east of the tower by the end of 2015.
Some meteorological instruments were installed on the tower to measure the vertical profiles of wind speed (U), air temperature (T a ), and relative humidity (RH) at five levels (10, 30, 50, 70, and 100 m), wind direction at three levels (30, 50, and 70 m), and air pressure at 8 m.All meteorological data were recorded with a temporal resolution of 1 min and were then averaged hourly.Additionally, some meteorological parameters at the surface, including U, T a , and RH, and their radiosonde data were obtained from a national meteorological station in Shenyang.

Measurements of Aerosol Optical Properties
Atmospheric visibility (VIS) measured by an FD12 visibility automatic observation instrument and aerosol optical depth (AOD) retrieved from a Cimel Electronique CE-318 sun-photometer were obtained from an atmospheric composition observational station that was located on the roof of the Northeast Regional Meteorological Center (123.50°E,41.77°N, 60 m) in Shenyang (Zhao et al., 2013a).Detail instrumental information is listed in Table 1.

Calculation of Atmospheric Turbulent Fluxes
Friction velocity (u * ) and scaling potential temperature (θ * ) are two important quantities in atmospheric turbulence studies.u * can be regarded as a simple expression of the momentum flux at the surface, and θ * is related to the vertical kinematic eddy heat flux ( ' ' w  ).Wind speed and air temperature profiles can be used to calculate u * and θ * based on Monin-Obukhov similarity theory (Garratt, 1992), according to Eqs. ( 1)-( 3): where κ = 0.4 is von Karman's constant, U ̅ (z) is the mean wind speed at 10 m height, therefore, z u = 10 m,  is the mean air temperature at specific heights with z θ1 = 10 m and z θ2 = 30 m, z 0 is the roughness length and taken as 0.25 m according to the terrain classification (terrain with scattered obstacles) in terms of z 0 value (WMO, 2008), L is the Obukhov length, Ψ M and Ψ H are the surface layer stability correction functions for momentum and heat, respectively, ( / )( ' ') s g w   is the buoyancy heat flux at the surface.

Temporal Variations in Pollutant Mass Concentrations during Two Haze Events
The Ministry of Environmental Protection of China has drafted definitions for the conditions that qualify a day as having haze pollution; that is a haze pollution day occurs when the average concentration of PM 2.5 is above 75 µg m -3 and visibility is less than 5 km for more than six consecutive hours due to an increasing concentration of fine particulate matter in the air.However, the criteria of VIS < 10 km and RH < 90% are commonly used to distinguish a haze day from the view of meteorology (Wu et al., 2005).According to the latter definition, a summer weak haze event and an autumn severe haze event in 2014 over Shenyang can be identified in Fig. 2. A weak haze event occurred on 16 June when the largest daily averaged AQI (= 166) in this month was observed.A severe haze event lasted from 12:00 Local Time (LT) on 30 October to 08:00 LT on 1 November 2014 and reached the upper limit of AQI (= 500) for ambient air quality standards in China on 31 October; it was the most severe haze pollution event in Shenyang in 2014.
During the weak haze event, the hourly mean AQI values gradually increased after the afternoon of 15 June and the maximum hourly AQI in 11 stations reached a peak of 315 at 17:00 LT on 16 June.Correspondingly, VIS decreased rapidly, ranging from 1.4 to 9.9 km.RH values ranged from 50%-90% on June 16, showing a negative correlation with VIS (Fig. 2(a)).During the severe haze event, the hourly mean AQI increased abruptly up to 500 after 12:00 LT on 30 October, and such high AQI level lasted for more than 36 hours.Meanwhile, VIS decreased significantly to a minimum value of 405 m, and RH ranged from 39% to 89% (Fig. 2(b)).
Temporal variation in the hourly mean mass concentrations of six pollutants (PM 10 , PM 2.5 , O 3 , NO 2 , SO 2 , and CO) as well as the ratio of PM 2.5 /PM 10 during the two haze events are shown in Fig. 3. Overall, PM 10 and PM 2.5 concentrations showed similar change over time.PM 2.5 and PM 10 concentrations began to increase in the nighttime on 15 June and reached a peak of 173.4 and 253.4 µg m -3 , respectively, around 18:00 LT on 16 June during the weak haze event.Before the weak haze event, PM 2.5 concentration exhibited two smaller peaks less than 100 µg m 3 on 14 and 15 June, which should be influenced by low ABL height at night.With the development of ABL height after sunrise, PM concentration at surface decreased gradually (Fig. 3(a)).The maximum concentrations of 912.9 µg m -3 for PM 2.5 and 1230.4 µg m -3 for PM 10 were observed around 02:00 LT on 31 October during the severe haze event (Fig. 3(b)).With the increase of PM concentrations, the ratio value of PM 2.5 /PM 10 increased to 0.80 approximately during the two haze events, that was larger than its annual mean ratio (= 0.69-0.72)observed during 2010-2012 in Shenyang (Zhao et al., 2013b), which means that fine PM pollution dominated during the two haze events.It should be noted that PM concentrations before the severe haze event remained a level that corresponded to the maximum PM in the summer haze event.Such high PM values are common in Northeast China during autumn and winter seasons.The reasons behind such difference are complex due to different photochemical  ) midnight (Fig. 3(a)).This is mainly because O 3 is formed in photochemical reactions involving volatile organic compounds (VOC s ) and nitrogen oxides (NO x = NO + NO 2 ) in the presence of heat and sunlight, resulting in the different regular patterns of diurnal change between O 3 and other pollutants (Seinfeld and Pandis, 2006;Tang et al., 2009Tang et al., , 2012)).SO 2 and CO concentrations exhibited similar diurnal variation with NO 2 .Lin et al. (2008) indicated that a higher SO 2 layer may exist above the nighttime and early morning inversion due to high chimneys in factories and power plants, which can probably explain that why SO 2 concentration peaks at 12:00 when the mixing layer is expected to develop well.However, such normal variation in NO 2 , SO 2 , and CO disappeared during the severe haze event, and all the gaseous pollutants increased at the daytime on 31 October, mainly due to the burning of crop residues in this region (Fig. 3(b)).
The diurnal pattern and variability of surface O 3 depend very much on the intensity of UV radiation, level and variations of O 3 precursors, and vertical and horizontal transport of O 3 (Lin et al., 2008).Comparing the two haze events, it can be found that the summer haze event was characterized by high O 3 pollution.The peak of O 3 during the weak haze event (= 250 µg m -3 ) was about three times as large as that during the autumn haze event mainly because the photochemical reaction to produce O 3 weakens in late autumn with decreasing temperature and UV radiation flux.

Formation and Evolution Mechanism of the Two Haze Events Stable Synoptic Conditions
According to the surface weather maps, a slowly migrating cyclone that originated in North China controlled conditions in Shenyang on 16 June during the weak haze event (Figs.4(a  surface usually had the largest values at afternoon, as found from June 13 to 17, 2014.U-values at all heights were smaller than 2 m s -1 at the beginning of the weak haze event, but they increased suddenly with the development of haze pollution.During the severe haze event, U at all heights almost remained smaller than 3 m s -1 , and the momentum transport in the near-surface layer were very weak due to the small vertical gradient in U (Fig. 5(a)).The different evolution in U indicates different formation mechanism of the two haze event.
Temperatures also showed regular diurnal variation during the day from 13 to 16 June, and the largest T a value on 16 June was 29.8°C.However, such diurnal variation in T a was very weak during the severe haze event after 31 October (Fig. 5(b)).Besides small wind speed, the formation of haze is also usually accompanied by humid air masses in the near-surface layer (Sun et al., 2013).RH at all heights gradually increased from 50% to 90% during the weak haze event, and during the severe haze event, RH remained high values above 70% (Fig. 5(c)).
The formation and evolution of haze pollution is also closely related to thermal inversion layer.Air temperature radiosonde curves at 08:00 and 20:00 LT during 13-16 June and from 29 October to 1 November 2014 in Shenyang are shown in Fig. 6.For the weak haze event, a weak inversion layer existed at 975 hPa at 20:00 LT on 16 June, with an inversion depth of 178 m and inversion intensity of 1.6°C/100 m, which was similar to those observed at 20:00 LT on 14 and 15 June (Fig. 6(a)).However, double or even multiple inversions occurred during the severe haze event from 29 October to 1 November (Fig. 6(b)), and the thickness and intensity of the low-altitude temperature inversion were higher than in the weak haze event (Table 2).During the severe haze event, the depth of the near-ground inversion layer gradually developed from 190 to 330 m at 08:00 LT on 29 and 30 October, and the inversion intensity was larger than 2-4°C/100 m.The second inversion layer existed between 850 and 925 hPa from 29 to 31 October, and its depth varied widely (76-884 m).The existence of a near-ground inversion and secondary inversion made the atmospheric conditions very stable and suppressed the vertical movement and turbulent exchange in the boundary layer (Wang et al., 2014;Ye et al., 2015;Wang et al., 2016).Such stable atmospheric conditions are favorable for the formation and maintenance of heavy haze pollution (Rao et al., 2008;Yang et al., 2015;Xie et al., 2010).
Using the profiles of wind speed and temperatures, turbulent fluxes including u * and ' ' w  , as well as their relationship with PM 2.5 concentration were analyzed here, because atmospheric turbulence plays an important role to control the formation and development of haze pollution.Fig. 7 shows the scatter plots of u * and ' ' w  against PM 2.5 during the haze development period that started from the beginning of the haze event and ends at the moment with the maximum AQI value.The u * values were smaller than 0.5 m s -1 during the development period of both haze events.A positive correlation exhibited between PM 2.5 concentration and u * during the weak haze event, with correlation coefficient R = 0.49 (Fig. 7(a)), whereas a negative correlation was shown between them during the severe haze event, with R = 0.56 (Fig. 7(b)), which probably means the weak dynamic turbulence favors the development of local heavy pollution, but the increasing dynamic turbulence helps the pollutants transport during the weak haze event.2.0 0.9 Moreover, PM 2.5 concentration increased under larger RH conditions during the severe haze event (Fig. 7(a)), but such relationship was not obvious for the weak haze event (Fig. 7(b)).Tang et al. (2016) indicated that an increase in the RH is favorable for the formation of particles from the liquid-phase, heterogeneous reactions and the hygroscopic growth processes, and the primary source of particles will change to local humidity-related physicochemical processes during heavy pollution periods, and local secondary processed is dominated in heavy pollution.PM 2.5 concentration had a weak positive correlation with ' ' w  during the severe haze event, and the values of ' ' w  concentrated between 0.2 and 0.4 K m s -1 , meaning that the thermal effect of turbulence is very weak for the development of heavy haze pollution, comparing with the dynamic effect of turbulence.However, no obvious relationship was shown between PM 2.5 and ' ' w  during the weak haze event, and the values of ' ' w  mostly remained smaller than 0.2 K m s -1 , but sometimes reached larger than 1 K m s -1 .Li et al. (2015) compared the turbulent characteristics between the fog day and haze day in Nanjing, and they found out that the haze day was associated largely with the mechanical turbulence that can be represented by u * but less with thermal turbulence that can be represented by sensible heat flux Q or ' ' w  , which is consistent with our results.However, ABL height is closely related to the thermal effect of turbulence in the atmosphere.Wang et al. (2015) observed that diurnal changes in averaged ABL height and PM 2.5 displayed a generally contrary correlation during July 2008 over the Jing-Jin-Ji region, indicating the important impact of the ABL height on the pollution strength of the surface air.

Characteristics of AOD and Effects of Meteorological Parameters on PM Concentrations
Table 3 lists the daily mean values and standard deviations of AOD (500 nm) from 14 to June 16, 2014 in Shenyang.Before the weak haze event occurred, the mean AOD value  Unfortunately, the AOD data during the severe haze event was not available due to a problem with the sun-photometer.
The correlation coefficients (R) between the hourly AQI values and U, T a , RH, and VIS at surface during the two haze events with an AQI > 100 are listed in Table 4.The AQI values were positively correlated with U on June 16, 2014 during the weak haze event, meaning that this event was much likely related to the long-range transport of air pollutants.The AQI values exhibited a strong positive correlation with T a (R = 0.81), but showed a weak negative correlation with RH (R = 0.24) and P a (R = -0.79).This likely indicates that warm air in the near-surface layer and low air pressure favored the accumulation of air pollutants and hence the development and maintenance of the haze event.However, during the heavy haze event from 28 October to 1 November, the R-values between the AQI and these meteorological parameters were relatively smaller, mainly due to the very stable atmospheric conditions and small variability in these parameters.Wang et al. (2016) also reported that the combination of high pollutant emissions, moist atmospheric conditions, and a near-surface temperature inversion in the boundary layer caused a heavy haze event in Xi'an, Northwest China.
Two-day back trajectory analyses at 100, 300, and 500 m above ground level (AGL) were conducted to explore the influence of the transport of pollutants within the local area under the hazy weather conditions using the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model from the National Oceanic and Atmospheric Administration (NOAA) (Draxler and Polph, 2003;Wang et al., 2010), as shown in Fig. 8.During the weak haze event, the air masses over Shenyang on this hazy day originated from the north China region where anthropogenic activities are high.Pollutants could be transported to the ground by downmixing movements from 08:00 LT on 15 June to 08:00 LT on 16 June, corresponding with the southerly and southwesterly winds prevailing in the near-surface layer as the AQI values increased above 100.The polluted air masses passed over the height under 500 m AGL, corresponding to the period of pollutant accumulation (Fig. 8(a)).However, for the severe haze event, the air masses at different heights showed different paths.The air masses under 100 m mainly originated at the northern part of Liaoning and the southern part of Jilin province, as well as the down-mixing movements of air masses from the southern part of Liaoning.These air masses circled over Shenyang, which means the local pollutant sources were very important for this haze event (Fig. 8(b)).Based on lidar ceilometer observations, Tang et al. (2015Tang et al. ( , 2016) ) and Zhu et al. (2016) presented the regional air pollution mechanism and concluded that regional transport played an important role during slight haze periods and local secondary formation dominated the air pollution   during heavy haze periods, which can provide reference to our studies.

CONCLUSIONS
The characteristics of air pollutants (PM 10 , PM 2.5 , SO 2 , NO 2 , O 3 , and CO) and the near-surface layer structure during a weak summer haze event and a severe autumn haze event in 2014 were comparatively analyzed based on the measurements of mass concentrations of pollutants from eleven air-quality monitoring stations and the profiles of wind speed, air temperature, and relative humidity observed from a 100 m-high meteorological tower in Shenyang, an urban city in Northeast China.
The weak haze event on June 16, 2014, was characterized by high O 3 pollution, with a maximum hourly mean concentration of 248.3 µg m -3 , while the severe haze event from October 30 till November 1, 2014, was characterized by high PM concentrations, with maximum PM 2.5 and PM 10 concentrations of 912.9 and 1,230.4µg m -3 , respectively.Mass concentrations of O 3 and NO 2 showed distinct diurnal variations, with O 3 peaking in the middle of the day but NO 2 peaking at midnight due to the photochemical O 3 production during daytime and titration during nighttime under high NO x concentrations.However, such diurnal variations in the gaseous pollutants were weak or even disappeared during the severe haze event, mainly due to irregular crop-residue burning in this region.
Atmospheric turbulence plays an important role in controlling the formation and development of haze pollution.The correlation between friction velocity u * and PM 2.5 concentration was positive during the weak haze event, with the correlation coefficient R = 0.49, but negative during the severe haze event, with R = 0.56, which probably indicates that the transport of pollutants contributed to the weak haze event but that the severe haze event was formed mainly by the accumulation of local pollutants.Compared with the dynamic effect of turbulence (relating to u * ), the thermal effect of turbulence (relating to ' ' w  ) had little effect on the formation and evolution of the two haze events.In addition, thermal inversion layers were observed from the radiosonde data during the two haze events.A single inversion layer with an inversion depth of 178 m and inversion intensity of 1.6°C/100 m existed during the weak haze event, whereas two or even more inversions occurred during the severe haze event.Such stable atmospheric conditions favor the formation and continuation of heavy haze pollution.
According to backward trajectory analyses, pollutant transport from North China probably contributed to the weak haze event, which is confirmed by an obvious positive correlation between air quality index (AQI) and wind speed (U), whereas mostly local pollutants generated the severe haze event.

Fig. 1 .
Fig. 1.(a) Geographical position of Shenyang, and (b) locations of eleven air quality monitoring sites (yellow circle) and a meteorological tower (black square) in Shenyang.

Fig. 5 .
Fig. 5. Vertical distributions of (a) wind speed, (b) air temperature, and (c) relative humidity within the near-surface layer as well as their variation at the surface (black line) during the two haze events in Shenyang.

Fig. 6 .
Fig. 6.Temperature (T a ) sounding curve of the Shenyang station on 08:00 LT and 20:00 LT during (a) a weak haze event and (b) a severe haze event in 2014.

Fig. 7 .
Fig. 7. Relationships between friction velocity (u * ) and vertical heat flux ( ' ' w  ) against PM 2.5 concentrations during the development periods of (a) a weak haze event and (b) a severe haze event in Shenyang 2014.Color bar means relative humidity (RH).

Table 1 .
Performance parameters of the FD12 and CE-318 instruments.Temporal variations in the hourly and daily mean air quality index and hourly mean atmospheric visibility and relative humidity during (a) a weak haze event from June 13 to 16 and (b) a severe haze event from October 30 to November 1, 2014 in Shenyang.The red double arrow represents the duration of haze events.Temporal variations in the hourly mean concentrations of PM 10 , PM 2.5 , SO 2 , NO 2 , O 3 , and CO and the ratio of PM 2.5 /PM 10 during (a) a weak haze event during June 13-17 and a severe haze event from October 28 to November 1, 2014 in Shenyang.Red double arrow represents the duration of haze event.
longer than that in most other regions in China, and it usually starts in October and lasts five months in Shenyang.During this period, more combustion of coal contributes to higher PM concentration.In addition, the open burning of crop residues, which is a traditional agricultural activity and usually occurs in October or November in the widespread

Table 2 .
Height, thickness, and strength of the temperature inversion at low altitudes at 08:00 and 20:00 LT each day during the two haze events in Shenyang 2014.

Table 3 .
Daily mean values and standard deviations of AOD (500 nm) during a weak haze event in Shenyang.

Table 4 .
Correlation coefficients between the AQI (> 100) and some meteorological parameters during the two haze events in Shenyang.