Atmospheric Environment 36 (2002) 3473–3484 Influence of meteorological conditions on PM 2.5 and PM 2.510 concentrations during the monsoon season in Hanoi, Vietnam P.D. Hien a, *, V.T. Bac b , H.C. Tham b , D.D. Nhan b , L.D. Vinh c a Vietnam Atomic Energy Agency, 59 Ly Thuong Kiet, Hanoi, Viet Nam b Institute for Nuclear Science and Technology, Cau giay, Hanoi 5T-160, Viet Nam c Upper Air Meteorology Station, Hanoi, Viet Nam Received 21 January 2002; accepted 10 April 2002 Abstract Twenty-four hour samples of air particulate matter with aerodynamic diameters from 2 to 10 mm (PM 10 ) and o2.5 mm (PM 2.5 ) were collected in Hanoi throughout 1 year since August 1998. The air sampler was located in a meteorological garden where routine surface observations and upper air radiosoundings were conducted. Very high PM 2.5 and PM 2.510 concentrations were observed in conjunction with the occurrence of nocturnal radiation inversions from October to December and subsidence temperature inversions (STI) from January to March. In the first case, the PM 2.510 fraction was much enhanced and particulate pollution was significantly higher at night than in daytime. During the occurence of STIs particulate mass was almost evenly distributed among the two fractions and no significant diurnal variations in concentrations were observed. In summer (May–September) particulate pollution was much lower than in winter. The multiple regression of 24-h particulate concentrations against meteorological parameters for both the winter and summer monsoon periods shows that the most important determinants of PM 2.5 are wind speed and air temperature, while rainfall and relative humidity largely control the daily variations of PM 2.510 , indicating the high abundance of soil dust in this fraction. As to turbulence parameters, among the determinants of 24-h particulate concentrations are the vertical gradients of potential temperature and wind speed recorded at 06.30 and 18.30, respectively. Meteorological parameters could explain from 60% to 74% of the day-to-day variations of particulate concentrations. r 2002 Elsevier Science Ltd. All rights reserved. Keywords: Coarse and fine particulate matter; Temperature inversions; Meteorology; Diurnal variations; Regression analysis 1. Introduction The rapid economic development since the introduc- tion of a market orientation reform in the late 1980s has dramatically changed the face of the 3.5 million Vietnam’s capital, Hanoi. In the meantime, uncontrolled growth of construction works, traffic and small manu- facturing activities has resulted in an increasing number of air pollution sources. Dusty atmosphere is visible, especially during the dry winter season. A systematic air particulate pollution study has begun in Hanoi since 1998. The 24-h PM 10 (particulate matter with aerodynamic diameters o10 mm) concentration varies substantially, from as low as 10 mgm 3 in the rainy monsoon months (June–August) to over 300 mgm 3 in the winter (October–March). The devel- opment of appropriate pollution abatement measures requires a thorough understanding of the nature of major emission sources and atmospheric conditions governing the variations of air particulate concentra- tions, particularly those relevant to pollution episodes. The variations of atmospheric conditions in Hanoi are governed by large-scale air circulations which are *Corresponding author. Fax: +84-4-9424133. E-mail address: [email protected] (P.D. Hien). 1352-2310/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII:S1352-2310(02)00295-9
12
Embed
Influence of meteorological conditions on PM 2.5 and PM 2.5−10 concentrations during the monsoon season in Hanoi, Vietnam
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Atmospheric Environment 36 (2002) 3473–3484
Influence of meteorological conditions on PM2.5 and PM2.5�10
concentrations during the monsoon season in Hanoi, Vietnam
determine the gravimetric masses of collected materials
using a Mettler balance placed in a dedicated room with
controlled temperature and humidity. The filters were
acclimatised in the room condition for 3–4 days prior to
weighing. The readability of the balance is 1mg. A 210Po
electrostatic charge eliminator was used to neutralise
charges accumulated on the filters before weighing.
4. Meteorological data treatment
4.1. Surface observation parameters
Surface observation meteorological parameters re-
corded at every 3-h interval include temperature,
pressure, RH, WS, wind direction and sunshine dura-
tion. Rainfall was recorded as a 24-h total value. The
seasonal averages of meteorological parameters are
given in Table 1. For simplicity, the 1 October and 1
May are assigned to the beginning of the winter and
summer, respectively. For the summer, only a period
from May to July 1999 was analysed. The two summer
months of 1998 (August and September) were not
included in the regression analysis. WS and RH show
little seasonal variations, while temperature and rainfall
were much higher in summer than in winter.
4.2. Radiosoundings
Vertical profiles of temperature and WS in the
boundary layer were analysed to derive atmospheric
Fig. 1. High-pressure systems in East Asia in winter (a) and summer (b).
P.D. Hien et al. / Atmospheric Environment 36 (2002) 3473–3484 3475
turbulence parameters. For this purpose, the actual
temperature (T) was converted into potential tempera-
ture (y), thus enabling us to easily distinguish three typesof buoyancy according to the gradient Dy=Dz; namelystable (Dy=Dz > 0), neutral (Dy=Dz ¼ 0) and unstable
(Dy=Dzo0). It was found that high particulate levels
were recorded in conjunction with the occurrence of
NRIs in the first winter period (October–December) and
STIs in the second winter period (January–March). The
criteria adopted for these temperature inversions were
taken as in Heffter (1983), Marsik et al. (1995) and
Calori and Carmichael (1999), namely
Dy=DzX0:005 Km�1;
yT � yBX2K; ð1Þ
where Dy=Dz is the potential temperature gradient in the
inversion layer and yT and yB refer to the potential
temperatures at the top and the base of the inversion
layer.
NRI appears at dawn, extending from the ground to
about 100–150m. Over the night the NRI layer moves
up to a higher and higher elevation, reaching a few
hundred meters in the early morning (Fig. 2a), and
presumably disappears thereafter as the sunlight warms
the ground. STI usually persists for days during which it
can be observed in both the morning and evening
soundings with the inversion layer height varying within
several hundred meters above the ground (Fig. 2c).
The WS profiles are illustrated in Figs. 2b and d. WS
increases with height reaching the first maximum at
some elevation below 1000m. No significant relation-
ship exists between the height of this maximum and the
top or bottom of temperature inversions. The profiles of
RH and dew point have good correlations with that of
potential temperature. For this reason upper air data on
these parameters were not included in the regression
analysis.
4.3. Atmospheric turbulence
To characterise atmospheric turbulence, the mixing
depth is usually derived from the vertical profile of
temperature. A literature survey, however, did not find
an overall acceptable definition and criteria for the
practical determination of the mixing depth that could
encompass a wide range of atmospheric stability and a
variety of its governing physical processes (Beyrich,
1997). For daytime convective conditions the mixing
depth was estimated by using a temperature profile
intersection scheme developed by Holzworth (1967). For
nighttime stable conditions, several profile-derived
heights have been proposed, e.g. the height of the NRI
or the first WS maximum. (Mahrt et al., 1982; Baxter,
1991; Berman et al., 1999; Lena and Desiato, 1998;
Seibert et al., 2000). However, several researchers e.g.
Aron (1983), Lena and Desiato (1998) and Seibert et al.
(2000) noted that the mixing depth estimated by the
above methods in general poorly correlate with air
pollutant concentrations.
In our work, the gradients of potential temperature
Dy=Dz and WS Du=Dz between two elevations z1 and z2in the surface layer were used for characterising atmo-
spheric turbulence. Such a simple empirical method is
Table 1
Summary statistics of 24-h average particulate mass concentrations and meteorological parameters
Notation October 98–March 99 May 99–July 99
Mean S.D. Mean S.D.
Surface observations
Coarse mass (mgm�3) CO 69.8 52.3 27.6 15.0
Fine mass (mgm�3) FI 51.5 28.5 18.9 8.0
Wind speed (m s�1) WS 1.6 0.7 1.8 0.6
Air temperature (1C) T 21.5 3.5 28.5 2.3
Air pressure (mb) P 1014.6 5.5 1004.2 3.9
Relative humidity (%) RH 74.5 10.3 78.9 5.7
Sunshiness (h) SUN 3.4 3.5 4.9 3.2
Rainfall (mm) RAIN 0.9 3.8 8.0 20.3
Upper air observationsa
(Dy=Dz), 06.30 ðDy=DzÞm 0.61 0.39 0.55 0.26
(Du=Dz), 06.30 ðDu=DzÞm 1.06 0.64 0.92 0.52
(Dy=Dz), 18.30 ðDy=DzÞe 0.30 0.20 0.32 0.32
(Du=Dz), 18.30 ðDu=DzÞe 0.86 0.54 1.10 0.58
aThe values in (Km�1) for Dy=Dz and (s�1) for Du=Dz are multiplied by 100.
P.D. Hien et al. / Atmospheric Environment 36 (2002) 3473–34843476
showing a significant enhancement of the coarse mode
during NRI episodes. Soil particles thrown into the
atmosphere by street sweeping and traffic shaking in
evening rush hours are found to be a main cause giving
rise to the coarse mode enhancement as well as the very
high levels of PM2.5�10 during NRI episodes. As the
abundance of this soil dust component is inversely
related to the humidity of surface soil, the PM2.5�10
concentrations are much suppressed in humid condi-
tions during non-episode and STI periods. The above
findings provide guidance not only for forecasting
pollution episodes but also for developing abatement
measures.
Multiple regression analysis was applied to reveal
atmospheric parameters controlling the day-to-day
variations of particulate concentrations. PM2.5 is gov-
erned mainly by WS and air temperature, while rainfall
and RH largely control the daily variations of PM2.5�10,
indicating the high abundance of soil dust in the
PM2.5�10 fraction. Dusty air resulting from uncontrolled
construction works and unpaved roads and sidewalks is
common in urban areas of Vietnam (Hien et al., 2001).
As to parameters characterising atmospheric turbulence,
among the predictors of 24-h particulate concentrations
are the vertical gradients of potential temperature and
WS recorded at 06.30 and 18.30, respectively. These
controlling meteorological parameters dominate the
regression models for both the winter and summer
periods. Regression models could explain from 60% to
74% of the variances of 24-h particulate concentrations.
The remaining unexplained parts are associated mainly
with the variabilities of emission strengths and LRT air
pollutants.
Acknowledgements
This research was funded by the Ministry for Science,
Technology and Environment and was supported by
UNDP/IAEA/RCA Co-ordinated Project for Asia and
the Pacific on Air Pollution and its Trends. The authors
are grateful to the Hanoi Meteorological Station for the
kind assistance in providing routine surface observation
data. We also gratefully acknowledge contributions of
N.H. Quang and N.Q. Long (Institute of Nuclear
Fig. 6. The gradients of potential temperature and wind speed at 06.30 (a,b) and 18.30 (c,d) during the first winter period October–
December 1998 (see footnote a in Table 1 for the units of the gradients). The graphs show the association of ðDy=DzÞm and ðDu=DzÞewith the occurrence of NRIs, which is marked by full-height columns.
P.D. Hien et al. / Atmospheric Environment 36 (2002) 3473–3484 3483
Science and Technology) during the implementation of
this research project.
References
Aron, R., 1983. Mixing height—an inconsistent indicator of
potential air pollution concentrations. Atmospheric Envir-
onment 17, 2193–2197.
Baxter, R., 1991. Determination of mixing heights from data
collected during the 1985 SCCCAMP field program.
Journal of Applied Meteorology 30, 598–606.
Berkowicz, R., Prahm, L.P., 1982. Evaluation of the profile
method for estimation of surface fluxes of momentum and
heat. Atmospheric Environment 16, 2809–2819.
Berman, S., Ku, J.Y., Rao, S.T., 1999. Spatial and temporal
variation in the mixing depth over the northeastern United
States during the summer of 1995. Journal of Applied
Meteorology 38, 1661–1673.
Beyrich, F., 1997. Mixing height estimation from sodar data—a