Observation Analysis on Microphysics Characteristics of Long-lasting Severe Fog and Haze Episode at Urban Canopy Top

A field experiment based in a 255 m meteorological tower in Tianjin was conducted from Dec. 29, 2016 to Jan. 8, 2017, to study microphysical characteristics of the urban canopy top and their effect on a long-lasting severe Haze and Fog (HF) episode. The results show that the gradients of the PM2.5 concentration in the vertical direction varied greatly during the clean days but less so during HF days, which was consistent with the variation in the PBL height and turbulent activity. During HF1-1, PM2.5 concentrations at 120 m were obviously higher than those at the surface and at 200 m. Wind shear was one of the important factors due to the accumulation of pollutants at 120 m. During HF1-2, PM2.5 concentrations at the three levels declined with a mode of “cliff.” The vertical gradient of concentrations between 120 and 200 m was small, but it was larger between the surface and 120 m. The wet scavenging effect of the fog, the damaged inversion layer, and the strengthened turbulence were closely related to the “cliff” decline in concentration. When the fog transformed into haze during HF1-3, the turbulent energy rapidly decreased, and the atmospheric layer again became stable, with the height of the PBL being nearly 120 m.


INTRODUCTION
Haze and Fog (HF) episodes are disastrous weather phenomenon occurring in the atmospheric boundary layer (Yao et al., 2003;Parrish and Zhu, 2009;Zhao et al., 2011;Sun et al., 2014;Guo et al., 2016), which plays important role in reducing the visibility by scattering and absorbing sunlight (Xue et al., 2011;Han et al., 2012), exerting detrimental effects on human health (Pope and Dockery, 2006;Tie et al., 2009), and also changing the radiation balance directly and indirectly (Bellouin et al., 2005;Ramachandran and Kedia, 2010).Most of the previous field experiment mainly conducted at ground level, which have improved our knowledge of complicated mechanism of HF episodes, from the synoptic, meteorological conditions, chemical compositions and source apportionment of aerosol (Gao et al., 2016;Liu et al., 2016;Chen et al., 2017).Vertical structure of temperature, humidity, wind and planetary boundary layer height have an important impact on urban air pollutants (Vakeva et al., 1999;Han et al., 2009;Sun et al., 2016).Currently, the most widely applied instruments in the field measurement of vertical observation for air pollutants include aircraft, remote sensing, the tethered balloon and tower.However, the traditional tethered balloon platform was limited with the payload and could only carry the lightweight instrument (Li et al., 2015;Han et al., 2018).Aircraft measurement was expensive and mainly focused on the mid-and high altitudes with low vertical resolution (Zhang et al., 2009;Lee et al., 2016).Helicopters are much cheaper and easier to maneuver in low-level flights, vertical and spatial variations of ozone and PM 2.5 were observed and studied during high pollution episodes with helicopters (Crosman et al., 2017).Remote sensing devices have blind observe spots near the ground, which was very important for the study of mechanism of pollution events (Tang et al., 2016).Tower is set up at a fixed site and have high observe resolution within the height of the tower, for example, there is a 325-m tower in Beijing and 255-m tower in Tianjin, which are used for the measurement of meteorological and air pollutants (Tian et al., 2013;Han et al., 2015a, b;Sun et al., 2015Sun et al., , 2016;;Wang et al., 2016).
Urbanization has a substantial effect on urban meteorology, which contributes to increased urban HF weather by decreasing urban wind speed and pollutant diffusion capability (Li et al., 2012;Liu et al., 2017).As the building height increases the dispersion of particles in the canyon becomes weaker (Scungio et al., 2013), the difference in wind direction with height is significant under the conditions of low wind speeds.In winter, the particulate matter was mostly confined to a surface layer less than 50 m in height from the surface and the temperature inversion events was the main factor contributing to the high pollutant concentration in the evening (Trompetter et al., 2013).It's also believed that the planetary boundary layer (PBL) has strong influences on vertical diffusivity and surface concentrations of air pollutants (Lena et al., 1999;Wang et al., 2016).The PBL height is a key variable to describe the urban boundary layer structure.Methods for retrieving PBL height from Lidar backscatter profiles have been presented and developed (Cohn et al., 2000;Munkel et al., 2007;Du et al., 2013;Lotteraner et al., 2016;Huang et al., 2017).Atmospheric turbulence within the boundary layer is crucial for the exchange of the energy and important for developing and dissipating fog and the height of fog depends on the intensity of turbulence (Li et al., 2015).The turbulence kinetic energy was large and showed obvious diurnal variations during clear days, while was small during heavy pollution periods, the horizontal wind speed first started to increase at high altitude and the strong upward motion was an important dynamic factor to end the pollution episode (Ren et al., 2018).In addition, the intensity of turbulence also related to the vertical profile of wind and temperature, which affect the diffusion of the pollutants (Liu et al., 2011;Han et al., 2015b).
Tianjin has experienced rapid urbanization, and now is one of the core cities and severely polluted region in Beijing-Tianjin-Hebei (BTH) region (Han et al., 2014;Wu et al., 2015).Some studies on the vertical distribution of HF episodes were conducted in Tianjin city based on observed data (Zhang et al., 2011;Shi et al., 2012;Tian, et al., 2013;Han et al., 2014;Lyu et al., 2018).Little has been known about the effect of the microphysics of meteorological conditions on the HF episodes in the lower troposphere (Ye et al., 2015).In this paper, a field experiment was conducted based on a 255-m meteorological tower in Tianjin from Dec. 29, 2016 to Jan. 8, 2017.Microphysics characteristics of urban boundary structure at urban canopy top and its effect on a long-lasting severe HF episode were analyzed.

Observation Site
The data used in this study were obtained from a 255-m meteorological tower, the atmospheric boundary layer observation station of China Meteorological Administration, which is located in the south of the Tianjin (39°04′N, 117°12′E) shown in Fig. 1.The site changed to urban centers gradually as a result of the rapid urbanization during the past several decades (Han et al., 2009).Now it is a district with a mixture of residential, commercial and traffic areas located between the Youyi Road and Liujiang Road.There is a residential district on the southwest side, a busy road to the east, an expressway about 150 m to the north, and no industrial pollution sources nearby.The vertical variations of meteorological elements and pollutant mass concentrations in the atmospheric boundary layer during the HF episodes can be detected effectively with the tower.

Instrument and Data Processing
Conventional gradient observations include 15 levels of wind speed, wind direction, air temperature and relative humidity, and 3 levels of PM 2.5 concentration measured with TEOM RP1400a (Thermo Fisher Scientific Inc., USA).The atmospheric visibility was measured using Belfort 6000 (Belfort Instrument Company, USA).The above hourly average data were calculated and used in the following analysis.
Fast-response instruments were installed to acquire turbulence fluxes.The fluctuations of three wind components and sonic temperature were measured by sonic anemometers (model CSAT3, Campbell Scientific, Inc., USA) mounted at the end of the southern support arms of the tower (40 m, 80 m, 120 m and 220 m).Turbulence measurements were conducted at a frequency of 10 Hz and processed with 30-min time interval with the TK3 software.Turbulent shear stress is the expressive force of turbulent momentum transport, caused by increases and decreases in the intensity of turbulence.The turbulence kinetic energy (TKE) was calculated with the following equation: where u, v, w are wind components at horizontal and vertical direction, the over bar means temporal average and the prime denotes the fluctuation from the mean value.The turbulent momentum flux τ is expressed by Eq. ( 2) with ρ being the air density: The friction velocity is given by: An aerosol Lidar (WUXI CAS Photonics Co., Ltd., China) was employed to obtain the profiles of backscatter coefficient, with which PBL height can be expected to reveal.Gradient method was used to derive PBL height in this paper.We selected the maximum of the negative gradient of the backscatter coefficient to be the top of the mixed layer.Meanwhile, meteorological parameter data observed at the 255-m tower were used to complement the Lidar because of its dead zone.The detailed method to calculate PBL height with vertical temperature profile can be found in our previous paper (Han et al., 2009).

Overview of the Haze and Fog Episodes in BTH Regions
From Dec. 29, 2016, to Jan. 9, 2017, BTH regions experienced a serious HF episodes, which led to "orange" air pollution alert and "red" fog alert at the same time.The PM 2.5 hourly maximum concentration were 508 µg m -3 , 475 µg m -3 , 325 µg m -3 and 320 µg m -3 for Beijing, Baoding, Tianjin and Cangzhou, separately.Conspicuous differences regarding PM 2.5 concentration changes were discovered in various locations (Fig. 2).The substantial transportation of regional pollution and stagnant weather conditions favorable for the accumulation of local pollution were the two main causes of rapid increase in pollutant concentration.
Meteorological conditions at BTH regions were stagnant during the entire pollution period and were controlled by similar atmospheric circulation.A straight westward flow is dominated at 500 hpa (Fig. 3), and weak pressure system maintained at the surface (Fig. 4).Time series of pollutants and relevant meteorological conditions were shown in Figs. 5 and 6. High humidity, low wind provided the favorable weather conditions for the formation and maintenance of HF episode (Han et al., 2014;Wang et al., 2016).To better analyze microphysics characteristics of boundary layer and its effect on this HF episode, the whole HF event was divided into the following period: Dec. 29 (Clear1); from Dec. 30 to Jan. 4 (HF1), Jan. 5 to Jan. 6 (Clear2), Jan. 7 to Jan. 8 (HF2), Jan. 9 (Clear3).On Dec. 29, Tianjin was mainly affected by the north wind, with the relative humidity at 40-50%, visibility greater than 30 km.PM 2.5 mass concentration was only 20 µg m -3 .From the night of Dec. 29, controlled by weak high pressure, the wind gradually turned to southwest.Until Dec. 31, relative humidity was approximately 100%, the minimum visibility was less than 50 m, and the maximum hourly concentrations of PM 2.5 was 332 µg m -3 .Due to the temperature drops at night of Dec. 31, the atmosphere supersaturated vapor condensed and rime and snow appeared in the morning of Jan. 1 (Fig. 7), PM 2.5 dropped down to 120-170 µg m -3 affected by the wet removal mechanism.Then from night of Jan. 1 to Jan. 4, the weak north wind and southwest wind alternated, and PM 2.5 concentration fluctuated and haze maintained.After a short cleaning process, relative humidity rose to 90% on Jan. 7, the heavy fog formed again with the minimum visibility was less than 100 m.Until the morning of Jan. 9, the cold high pressure ended the HF events.
Other than the wind and relative humidity, the planetary boundary layer (PBL) plays important roles in affecting atmospheric environment.The thinner the mixing layer, the weaker is the vertical atmospheric diffusion process (Han et al., 2009;Quan et al., 2013).The difference between clear period and heavy period was obvious in Fig. 8.For clear    period, the PBLH during the day was over 1000 m, the PBLH at night was over 400 m.For HF period, the PBLH during the day was below 500 m, the PBLH at night was below 200 m.

Microphysics Characteristics of this Episode at Urban Canopy Top Vertical Distribution of PM 2.5 Concentration
The vertical distribution of PM 2.5 concentration at surface, 120 and 200 m were depicted in Fig. 9.The three kinked line exhibited a similar variation trend in general.The gradients of PM 2.5 concentration in the vertical direction varied greatly during the clean days but smaller during HF days, which was consistent with the change of PBLH.HF1 was further subdivided into 3 stages: HF1-1 was from night of Dec. 29 to the morning of Dec. 31.PM 2.5 concentration at 120 m was obviously higher than those at surface and 200 m.The distribution of PM 2.5 in the vertical direction showed obvious changes during HF1-2 period (from noon of Dec. 31 to noon of Jan. 2).PM 2.5 concentration at three levels declined with a mode of "cliff."The concentration of PM 2.5 declined from 332 to 120 µg m -3 at surface, from 410 to 25 µg m -3 at 120 m, and from 321 to 20 µg m -3 at 200 m.The vertical gradient of PM 2.5 concentration between 120 and 200 m was small, but larger between surface and 120 m.At noon on Jan. 1, the surface concentration was 145 µg m -3 and 150 µg m -3 higher than those at 120 m and 200 m.During HF1-3 period (from noon of Jan. 2 to noon of Jan. 5), the vertical distribution of PM 2.5 showed different variation mode.The highest value observed at 120 m, followed by the surface, and the lowest was observed at 200 m.

Vertical Structure of the Wind and Relative Humidity
Temporal and vertical variation of the wind and relative humidity at urban canopy top were shown at Fig. 10.Vertical diffusion condition below 250 m was good during clear period.The vertical distribution of relative humidity was uniform with a value less than 70%.Under the influence of urban building, the wind speed changed less between surface and 40 m, while increased obviously at the height of 200 m., The vertical distribution of PM 2.5 concentration was closely related to the vertical structure of wind.The vertical distribution of wind and humidity changed a lot during HF period.During HF1-1 period, southwest wind maintained at the lower and upper levels, then turned to easterly wind at middle level at the night of Dec. 29.Wind shear was one of the important factor due to accumulation of pollutants in this layer, which brought the highest concentration of PM 2.5 at 120 m.Southwest wind increased obviously at higher altitude in the early morning of Dec. 30.Wind speed at 200 m was up to 5.2 m s -1 , and PM 2.5 concentration at 200 m decreased to 90 µg m -3 .In the morning of Dec. 30, wind speed at higher layer decreased and PM 2.5 concentration increased again.At HF1-2 period, the relative humidity in the middle and upper levels increased first and earlier than that on the ground.The lasting periods with relative humidity greater than 95% at higher levels were much longer than that at lower levels, suggesting a downward transition from higher humidity layer.Fog can not only provide a medium for the formation of secondary pollutants and increased the PM 2.5 concentration (Han et al., 2014;Quan et al., 2014), but also had capacity to scavenge fine particles during severe haze episodes (Wang et al., 2014).Under the wet scavenging effect, PM 2.5 concentration at three levels decreased obviously.At the same time, PBLH decreased to 100 m.The 120-m and 200-m platform are generally outside the PBLH.Due to the dynamical stability of the NPBL, air pollutants at surface layer are normally trapped inside the NPBL and rarely mix with the pollutants outside the PBLH.Very different distribution characterizations of PM 2.5 were measured inside and outside the PBLH (Han et al., 2009(Han et al., , 2015b)).PM 2.5 concentrations at 220 m and 120 m were lower obviously than surface.Urban buildings can affect the meteorological condition and the diffusion capabilities of the pollutants (Liu et al., 2015;Ye et al., 2015).During HF1-3, wind directions observed below 40 m were scattered than those at higher levels.Wind speeds at higher layers increased with a value of 4.2 m s -1 at 180 m and relative humidity decreased to 60% accordingly.PBLH continued to maintained low value during this period, average PBLH at night was 160 m.The 120-m platform is located at the top of the PBLH, while the 200-m platform is generally outside the PBLH, that was the reason why the maximum PM 2.5 concentration was observed at 120 m, and the minimum concentration observed at 200 m.It turned to westerly wind in the morning of Jan. 8.From then on, the northwest wind maintained and wind speed was 2 m s -1 , relative humidity dropped significantly below 70%.

Vertical Structure of Turbulence across Tower Levels
Turbulence is a major factor affecting atmospheric diffusion and is thus a good indicator of atmospheric stability (Wang et al., 2014).Fig. 11 depicted vertical structure of turbulent energy, turbulent momentum flux and frictional velocity.The variation tendency at three levels was similar.The amplitude of variation was small at 40 m, while greater at 220 m.At Clear1 period, the turbulent energy increased with height, with 0.96, 1.54 and 1.97 m 2 s -2 for the height of 40, 120 and 200 m.From the afternoon of Dec. 29, the turbulent energy decreased with a high speed at higher levels than lower level.During Clear2 period, compared to the HF1 period, turbulence energy at three levels all increased, the average turbulence energy at 40 m was about 1.6 m 2 s -2 and 1.0 m 2 s -2 at 120 and 200 m.During Clear3 period, the average turbulent energy at 40 and 200 m increased to 3.2 m 2 s -2 , and the time of average turbulent energy began to increase at 200 m was slightly earlier than at 40 m.
During HF1-1 period, the average turbulent energy at three levels decreased with the average value of 0.34, 0.34 and 0.30 m 2 s -2 for 40, 120 and 200 m, and PM 2.5 concentration increased quickly on Dec. 30.The vertical gradient of the average turbulent energy was small.The mean turbulent energy had obvious diurnal variation character, and the mean turbulent kinetic energy were 0.28 and 0.53 m 2 s -2 for night and day respectively.During HF1-2 period, dense fog was observed in the morning of Dec. 31.Turbulence energy at three levels all increased and maintained high value within the whole fog period.Due to the cooling effect of long-wave radiation of water droplets, the inversion layer has been damaged gradually, and turbulence had been strengthened.On the contrary, when fog transformed to haze during HF1-3 period, turbulence energy rapidly decreased to 0.07, 0.11 and 0.08 m 2 s -2 for 40, 120 and 200 m separately.During HF2 period, the average turbulent energy had diurnal variation characteristic, turbulence energy at night was very weak with a value of 0.08 m 2 s -2 , and the mean turbulent energy during the daytime was different, 0.37 and 0.84 m 2 s -2 for Jan. 6 and Jan. 7 respectively.The variation of friction velocity and momentum flux was consistent with the turbulent energy; no more details were described in this paper.

CONCLUSION
This paper investigated the microphysical structure of the boundary layer and its impact on the formation and evolution of the HF episodes from Dec. 29, 2016 to Jan. 8, 2017.The whole episode was divided into three clear periods and two HF periods, and HF1 was also subdivided into 3 stages.The gradients of the PM 2.5 concentration in the vertical direction varied greatly during the clean days but less so during HF days, which was consistent with the variation in the PBL height and turbulent activity.During HF1-1, the PM 2.5 concentration at 120 m was obviously higher than at the surface and at 200 m.Obvious changes were found during HF1-2, when PM 2.5 concentrations at the three levels declined with a mode of "cliff."The vertical gradient of concentrations between 120 and 200 m was small, but it was larger between the surface and 120 m.During HF1-3, the highest PM 2.5 concentration was observed at 120 m, followed by those at the surface and at 200 m.Wind shear was one of the important factors due to the accumulation of pollutants at 120 m.The wet scavenging effect of the fog, the damaged inversion layer, and the strengthened turbulence were closely related to the "cliff" decline in concentration.In contrast, when the fog transformed into haze during HF1-3, the turbulent energy rapidly decreased, and the atmospheric layer again became stable, with the height of the PBL being nearly 120 m.

Fig. 6 .
Fig. 6.Time variations of surface visibility, relative humidity and wind in Tianjin.

Fig. 7 .
Fig. 7. Fog in the evening of Dec. 31 and rime in the morning of Jan. 1.

Fig. 8 .
Fig. 8. Time serious of day and night average of the PBLH.

Fig. 9 .
Fig. 9. Vertical variation of the hourly average PM 2.5 at surface, 120 and 200 m.

Fig. 10 .
Fig. 10.Temporal and spatial variation of the wind and relative humidity.