The mixing layer is an important meteorological factor that affects
air pollution. In this study, the atmospheric mixing layer
height (MLH) was observed in Beijing from July 2009 to December 2012
using a ceilometer. By comparison with radiosonde data,
we found that the ceilometer underestimates the MLH under conditions of
neutral stratification caused by strong winds, whereas it
overestimates the MLH when sand-dust is crossing. Using
meteorological,
The mixing layer is formed when discontinuous turbulence exists due to discontinuities in temperature stratification between the upper and lower layers of the atmosphere. The atmospheric mixing layer height (MLH) is an important meteorological factor that affects the vertical diffusion of atmospheric pollutants and water vapour concentrations; therefore, it impacts the formation and dissipation of air pollutants (Aron, 1983; Stull, 1988). Continuous observations of the MLH are helpful for improving the parameterizations of boundary layer models, and they play an important role of improving the simulation accuracy of meteorological models and optimizing the simulation results for pollutants.
Three primary observation methods are used to determine the MLH:
meteorological radiosondes, aeroplane surveys, and ground-based
remote sensing. As the most conventional observation approach,
meteorological radiosonde profiles utilize a large number of
observation stations distributed globally and provide high-quality data.
However, because of the high cost of the observation,
only two observations at 00:00 and
12:00
Acoustic radar (sodar), laser radar (lidar) and electromagnetic radar (Doppler radar) are the three methods used to perform ground-based remote sensing. Sodar can obtain the vertical profiles of wind and temperature, and these can be used to calculate the MLH. Doppler wind radar can obtain variations of the wind vectors at different altitudes and identify the mixing layer through wind shear. Lidar can obtain the vertical profile of the aerosol concentration and discern the atmospheric MLH by calculating the height at which sudden changes in the profile occur.
Beyrich (1997), Seibert et al. (2000), and Emeis et al. (2008)
conducted reviews of these three methods, comparing their
advantages and disadvantages. The
sodar detection height is usually less than 1000
Beijing, located on the North China Plain, is the centre of politics, culture, and economics in China. With the rapid development of the economy and the concomitant increase in energy usage, serious air pollution and heavy haze occurs frequently (Tang et al., 2009, 2012, 2015; Xin et al., 2010; Wang et al., 2014; Zhang et al., 2014; Yang et al., 2015). Previous studies of Beijing have indicated that visibility declines dramatically when the concentration of particles increases; the weather conditions typically include high relative humidity (RH), stable atmospheric stratification, and low wind speed (WS) with southerly flow during the polluted period (Ding et al., 2005; Liu et al., 2014; Zhang et al., 2015).
Although many studies have provided
detailed descriptions of other weather conditions during
heavy pollution periods, variations in atmospheric MLH are
not well understood. As a key meteorological factor, MLH
has a strong influence on the occurrence, maintenance, and
dissipation of heavy pollution. For most areas in northern China, the
meteorological radiosondes can only acquire the MLH in the
morning (08:00
To compensate for the deficiencies in the aforementioned studies, a ceilometer was used to conduct continuous high-resolution observations for 3 years and 6 months (from July 2009 to December 2012) in Beijing. By comparing the obtained data with multiple meteorological and pollutant data sets, we verified the applicability of the ceilometer and obtained the temporal variations of the MLH over 3 years. By combining the meteorological data, we were able to determine the variations of the mixing layer and the atmospheric diffusion capability in different seasons. Finally, we used visibility as an index to classify the degree of air pollution, and analysed the thermal/dynamic parameters inside the mixing layer under different degrees of pollution; then, we delineated the influence of MLH on air pollution and revealed the critical meteorological factors that affect the formation and dissipation of heavy air pollution in Beijing.
Site description and instrument list. BJT refers to the Beijing tower; ZBAA is the international standard weather station.
To understand the characteristics of the mixing layer in the Beijing area, we conducted observations for 3 years and 6 months (from 15 July 2009 to 16 December 2012) in Beijing. The observation sites, parameters, and time periods are shown in Fig. 1 and Table 1.
Topography and the observation sites.
The site used to measure the MLH was built in the courtyard of
the Institute of Atmospheric Physics, Chinese Academy of Sciences,
to the west of the Jiande bridge in the Haidian district, Beijing (ID: BJT).
This site is located between the north third and the north
fourth ring road, and the Beijing–Tibet motorway is on the
eastern side. The geographic location of the station is
39.974
The instrument used to observe the MLH
was a single-lens ceilometer (CL31, Vaisala, Finland). This instrument
utilizes pulsed diode laser lidar technology (910
The conventional
meteorological data during the same period included temperature,
RH, WS, and wind direction observations at 8, 15, 32, 47, 65, 80,
100, 120, 140, 160, 180, 200, 240, 280, and 320
To identify the sand-dust crossing, the ratio of
To illustrate the variations of the chemical compositions in particles,
the ground-based observations of organic matters (
The meteorological radiosondes were measured by the
international standard weather station (ID: ZBAA) that is located
outside the south second ring road in the Fengtai district, Beijing,
10
The meteorological radiosondes observed at station ZBAA
included two categories: conventional observations, which were
conducted at 08:00
Statistics of the thermal/dynamic parameters under different degrees of air pollution.
WS: wind speed; RH: relative humidity;
PM concentration proved to be a good indicator for characterization of the degree of
air pollution. However, haze was defined by the visibility in China as shown in
Table 2 (CMA, 2010), and it was remarkably negatively correlated with PM
concentration (Yang et al., 2015). Because PM data were
occasionally missing, visibility was used as an index to
classify the degree of air pollution. Visibility at station ZBAA, which was obtained
from the Department of Atmospheric Science, College of Engineering, University
of Wyoming (
Because the lifetime of particles can be several days or even weeks,
the distribution of particle concentrations in the MLH is more
uniform than that of gaseous pollutants. However, the particle
concentration in the mixing layer and that in the free atmosphere are
significantly different. In the attenuated backscatter coefficient
profile, the position at which a sudden change occurs in the profile indicates
the top of the atmospheric mixing layer. In this study, we used the
Vaisala software product BL-VIEW to determine the MLH by selecting the location with
the maximum negative gradient (
A number of methods have been developed for analysis of the mixing layer
through the meteorological radiosonde (Beyrich, 1997; Seibert
et al., 2000; Wang and Wang, 2014). In this study, we calculated
the MLH for the convective and stable states,
respectively. For the convective state, we used the Holzworth
method (Holzworth, 1964, 1967), which is the method most widely applied to
obtain the MLH by analysing profiles in the
Previous studies with ceilometers did not resolve issues concerning the applicability of ceilometers in Chinese areas with high aerosol concentrations. According to the methods described in Sect. 2.1.2, 260 and 540 effective observations were obtained for the convective and stable states, respectively. The MLH data acquired by meteorological radiosondes and by ceilometer were compared for the two types of weather conditions (Fig. 2). Using the MLH calculated by the radiosondes as a reference, the comparison showed that the MLH observed from the ceilometer was overestimated or underestimated in a portion of the samples.
Comparison of MLH between radiosondes and the ceilometer according
to visibility for convective
Because the ceilometer determines the MLH by measuring the attenuated
backscatter profile, if the concentration of atmospheric particles is
relatively low, it will be difficult to determine the MLH based on
a sudden change in the backscatter profile, and use of this method will lead to a higher
absolute error (MLH
Comparison of MLH between radiosondes and the ceilometer according
to RH for convective
To investigate why the ceilometer results produced underestimations,
we analysed those samples with good visibility and small absolute error.
The results showed that although the visibility was good, the absolute error
of the MLH was still small when the aerosol concentration showed
large differences in the vertical direction. After taking the RH
and wind vectors into account, we found that underestimations were
always accompanied by low RH and strong northerly wind (Figs. 3 and 4).
The local meteorological conditions in Beijing indicated that this
kind of meteorological condition is usually caused by the bypass of
a cold air mass. When strong northerly winds with dry and clear
air masses prevail in Beijing, atmospheric aerosols spread rapidly
to the downstream region, resulting in a dramatic decrease in local
aerosol concentration and good visibility. In addition, the dry air
mass suppresses the liquid-phase and heterogeneous reactions of the
gaseous precursors and the hygroscopic growth of aerosols can also
be neglected. Therefore, the formation of aerosols cannot compensate
for the transportation loss, leading to low and uniform aerosol
concentrations in the vertical direction. Once the aerosol concentration
becomes uniform in the vertical direction, the ceilometer cannot
calculate the MLH correctly through sudden changes in the attenuated backscatter
profiles, resulting in serious underestimations. An analysis of the
relationship between the
Comparison of MLH between radiosondes and the ceilometer according
to wind vectors for convective
Virtual potential temperature and wind speed
With respect to overestimations of the ceilometer results,
we may take the meteorological radiosonde at 14:00
Because the detected aerosol layers are not only the result
of ongoing vertical mixing but also always originate from
advective transport or past accumulation processes, interpreting
data from aerosol lidars is often not straightforward
(Russell et al., 1974; Coulter, 1979; Baxter, 1991; Batchvarova et al., 1999).
Therefore, improving the algorithm cannot resolve the underestimations
and overestimations of the ceilometer observations; the only
option that can be used to rectify the MLH is to eliminate
the data with large absolute error. After determining the reasons for the
underestimations and overestimations, the elimination is much
easier to implement. For underestimations, the meteorological data were used to eliminate
the periods when cold air passed with a sudden change in temperature
and WS. For overestimations, we referred to the sand-dust weather almanac
to identify the sand-dust days firstly (CMA, 2012, 2013, 2014, 2015).
Using the principal described in Sect. 2.1.1, the exact
times of sand-dust starting and ending were determined as the times which
the ratio of
Comparison of MLH between the ceilometer and radiosondes for
convective
After the screening process, the post-elimination ceilometer data and meteorological radiosondes are strongly correlated, with a correlation coefficient greater than 0.9, demonstrating the effectiveness of the elimination method (Fig. 6). Consequently, the elimination results are good. This method replaces the time-consuming method of filtering the data manually and is of great practical value for future measurements of MLH with ceilometers.
Monthly variations in the effective rate, wind speed, and
RH
To provide a detailed description of variations in the MLH, we selected continuous measured MLH and meteorological data over a 3-year period (from December 2009 to November 2012). First, the availability was verified after the MLH elimination by the aforementioned method. The results of the evaluation indicate that the availability in different seasons is significantly negatively correlated with WS and positively correlated with RH (Fig. 7a). For spring and winter seasons with large WS and low RH, the availability is low, whereas for summer and autumn seasons with small WS and high RH, the availability is high. In particular, the availability is lowest in January at 63.5 % and highest in June at 95.0 %. The successful retrieval of MLH over the 3-year period is approximately 80 %, much higher than in a previous study (Muňoz and Undurraga, 2010).
Monthly variations of the MLH from December 2009 to November 2012 in Beijing.
Using the validated data, we analysed seasonal variations over 3 years.
The results showed that the changes of the monthly mean were similar
in different years, and no inter-annual trend can be found (Fig. 8).
Therefore, we examined the averaged seasonal variation, and the monthly
mean of the daily minimum, average, and maximum were calculated, respectively.
The daily minimum of the mixing layer was
high in winter and spring, and low in summer and autumn. The maximum
monthly mean of the daily minimum MLH was 351
Compared with the daily
minimum MLH, both the monthly mean of the daily average and the maximum MLH
exhibited different seasonal variations. As shown in Fig. 7b, two platform
periods (from March to August and from October to January) and two
transitional periods (February and September) occur for the monthly
average MLH. The MLH is similar from October to January
at approximately 500
Previous studies have suggested that the seasonal variation in the MLH may be related to radiation flux (Kamp and McKendry, 2010; Muňoz and Undurraga, 2010), but our study was not entirely consistent as shown in Fig. 7b. Although spring had a significantly higher total radiation flux than summer, the MLH in spring is equal to that in summer. This is because more data were eliminated for winter and spring, especially for weather with dry wind and relatively high MLHs. Thus, using the monthly mean of MLH is not a good method by which to analyse the reasons for MLH variations.
To gain a better understanding of the MLH variations, we use the
daily mean instead of the monthly mean to do the analysis.
As the most simple framework in which we can analyse the MLH
variations in Beijing, we consider the thermodynamic model of
the mixing layer growth (Stull, 1988), as follows:
Correlation between the sensible heat (
We analysed the diurnal variations of MLH on a monthly basis and
found that the MLH develops in four stages: from 09:00 to
14:00
Daily variations in MLH in spring and summer
As shown in Sect. 3.2, the monthly average MLH is similar between spring
and summer. However, when the daily growth rates in spring and summer
were compared using the
The development of MLH is mainly related to the turbulent energy
and the production of the turbulent energy is closely related to
two components: the heat flux caused by radiation
(
To avoid the impact of near-surface buildings on the wind measurements, we
selected the wind vector at 100
When this regional circulation occurs along with
surface cooling that occurs at night in summer, the cold air near the surface
forms a shallow down-sliding flow from the northeast to the southwest.
The cold air flows into the North China Plain and accumulates in a cold
pool, increasing the thickness of the inversion layer,
and the thickness of the mixing layer gradually decreases. After sunrise,
the radiation increases; the MLH increases rapidly under the impact
of thermal buoyancy lift, and this type of cold drainage flow is
maintained until 12:00
In summary, the mountainous wind in summer causes the mixing layer to decline gradually at night; this also suppresses the development of the mixing layer before noon, and the prevalence of plain winds after noon causes the mixing layer to increase rapidly. Therefore, compared to the spring, the regional circulation in summer produces a concave-down variation in the rapid development stage of the MLH in summer. Although some interpretations for the influences of mountain plain winds are given, more intensive observations over northern China are suggested in order to analyse this phenomenon by meteorological radiosondes and additional observations.
Variations in net radiation, RH, and wind speed
To analyse variations in the thermal dynamic parameters inside
atmospheric mixing layers under different degrees of pollution,
visibility was used as a reference. WS, RH,
In summary, when clear days change to slight haze, the WS,
Monthly variations in visibility,
To verify these results, we examined the TKE budget equation. If
we presume a horizontal average and neglect the advection of wind,
the forecast equation of the TKE can be written as follows
(Stull, 1988; Garratt, 1992):
To differentiate the contribution of horizontal and vertical
turbulence to the TKE, the shear and buoyancy
terms in the TKE forecast equation were analysed as in the previous study
(Ye et al., 2015). When visibility decreases from 10 to 5
At least one previous study indicates when the MLH decreases, the concentration of atmospheric particles increases and visibility decreases (Tang et al., 2015). However, analyses of the MLH and particle concentration or visibility indicates that the correlation of these is not strong (Li et al., 2015). We analysed the correlation between daily averages of MLH and visibility in this study and found that the correlation between them is poor, with a correlation coefficient of only 0.08; this finding is consistent with the results of a previous study (Li et al., 2015).
According to the discussion in Sect. 3.4.1, in addition to MLH,
WS is another factor in controlling air pollution. The MLH
and WS represent the vertical and horizontal diffusion
capabilities of pollutants, respectively. Synthesizing these
two factors, the product of the MLH and the WS (ventilation coefficient: VC)
is usually used as an index to measure the capability of
atmospheric diffusion; a higher VC indicates stronger
capability (Tang et al., 2015). As shown in Fig. 12a, the VC is
highest in spring with approximately 2000
Correlation coefficients (
To obtain a clear understanding of the relationship between the atmospheric MLH and air pollution, we analysed the correlation between daily averages of the MLH and visibility according to the RH and found that the relationship between them showed significant differences under different RH. When the RH was lower than 80 %, the correlation between the MLH and visibility was poor, but when the RH exceeded 80 %, the correlation coefficient of these two measurements significantly increased to as much as 0.72 (Table 3 and Fig. 13). If we assume that no transport from other regions occurs, local contributions (local emissions and secondary formation) will dominate the particle concentration; if we further suppose that the local emission is constant every day, due to the dominant role of the aqueous, heterogeneous, and hydroscopic processes for the formation of particles in Beijing (Guo et al., 2014; Sun et al., 2013), there will be little difference in the formation of particles under a fixed RH, and the column concentration in the MLH will be almost constant. Under such circumstances, the relationship between the MLH and visibility should be strong. Thus, poor correlation between the MLH and visibility indicates a significant influence from regional transportation, and their good correlation indicates the dominant role of local contributions. Tang et al. (2015) found that in light pollution, regional transport contributes heavily, whereas in heavy pollution, local contributions dominate. Because low and high RH correspond to light and heavy pollution, respectively, the two conclusions are strongly consistent.
Correlation between the MLH and visibility according to RH in Beijing.
To clarify the contributions of local emissions and secondary
formation during high RH conditions, we analysed the chemical
compositions in
Mass percentage of
From the aforementioned analyses, the reasons for the relationships observed under conditions of low and high RH can be clearly understood. Under low RH condition, since the significant impact of the regional transportation, the processes of local emissions, regional transportation, and physicochemical formation jointly dominate the concentration of atmospheric particles; thus, the poor correlation between the MLH and visibility occurs due to the multi-source of particles. For high RH, the RH plays an important role in transforming the trace gases to aerosols. Thus, an increase in the RH is favourable 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. The strong correlation of these factors under high RH indicates the dominant role of local secondary processes in heavy pollution.
Overall, the high correlation between the MLH and visibility under high RH indicates that humidity-related physicochemical processes is the primary source of atmospheric particles in heavy pollution and that the dissipation of atmospheric particles mainly depends on the vertical diffusion capability, which is dominated by the atmospheric MLH. From the aforementioned conclusion, the MLH and RH are extracted as the key meteorological factors for the evolution of heavy air pollution, a finding that is relevant to the dissipation and formation of the atmospheric particles.
After determining the critical meteorological factors for
air pollution, the cause of the poor visibility in summer can
be analysed according to the aforementioned conclusions.
As shown in Fig. 7a, summer exhibits higher RH and lower WS.
The meteorological conditions in summer shows
that low WS is a prerequisite for hazy days
and that high RH is conducive to the formation of particles.
In addition, due to the intrusion of the west Pacific subtropical high,
summer is occupied by more than 70 % by the highest frequency of cloud fraction
Continuous high-resolution observations of MLH are required to understand the characteristics of the atmospheric mixing layer in the Beijing and North China Plain areas. To acquire the high-resolution observations of MLH, a study using a ceilometer was performed from July 2009 to December 2012 in the Beijing urban area.
Based on a comparison with radiosondes, we determined
that the ceilometer underestimates the MLH during near-neutral
stratification caused by strong winds and that it
overestimates the MLH during dust crossing. By combining
meteorological,
The characteristics of the MLH indicate that the MLH in Beijing
is low in autumn and winter, and high in spring and summer.
There is a significant correlation between the
By applying visibility as an index for the classification of degree of
air pollution, it is found that in comparison with
clear days, changes in the
Although the correlation between the daily MLH and visibility is very poor, the correlation between them is significantly enhanced when the RH increases. The high correlation between the MLH and visibility under high RH indicates that humidity-related physicochemical processes is the primary source of atmospheric particles under heavy pollution, whereas the dissipation of atmospheric particles depends primarily on the vertical diffusion capability, which is dominated by the atmospheric MLH.
The aforementioned results provide reliable basic data for better portraying the structure of the boundary layer and improving the parameterizations of the boundary layer in meteorological models. Studies of the atmospheric mixing layer and its thermal dynamic parameters at different stages of air pollution reveal the critical meteorological factors for the formation, evolution, and dissipation of heavy pollution, thus providing useful empirical information for improving atmospheric chemistry models and the forecasting and warning of air pollution.
This work was supported by CAS Strategic Priority Research Program Grants (nos. XDB05020000 and XDA05100100), the National Natural Science Foundation of China (nos. 41230642 and 41222033) and the National Data Sharing Infrastructure of Earth System Science. We also gratefully acknowledge the Department of Atmospheric Science, College of Engineering, University of Wyoming for the provision of the meteorological data used in this publication. Edited by: C. Hoose