The Eighth Asia-Pacific Conference on Wind Engineering,
December 10–14, 2013, Chennai, India
Effect of atmospheric stability on urban pollutant concentration
Ruichiro Yoshie, Tingting Hu
1
Professor of Architecture, Tokyo Polytechnic University, Iiyama, Atsugi, Japan, [email protected]
2
Former Ph.D student of Tokyo Polytechnic University, Iiyama, Atsugi, Japan, [email protected]
ABSTRACT
The purpose of this study is to investigate and generalize the influence of atmospheric stability on pollutant
concentration within urban areas. For this purpose, observation data of pollutant concentration were analyzed,
and wind tunnel experiments of gas dispersion within street canyons in unstable, neutral, and stable turbulent
boundary layers were conducted. It was found that normalized non-dimensional concentration C* became higher
with the increase of atmospheric stability, and the ratio between C* under neutral condition and non-neutral
condition was almost independent of measuring locations (flow field around measuring locations).
Keywords: urban pollutant concentration, atmospheric stability, environmental impact assessment
Introduction
For environmental impact assessment of air pollution, a Gaussian plume model is
usually used in Japan. This model is applicable for pollutant dispersion from a high chimney
(Fig. 1a), but it is obvious not appropriate for pollutant dispersion within an urban area (Fig.
1b). Even so, it is used for this situation in the environmental impact assessment in Japan.
One of the reasons why wind tunnel experiments (WT) or Computational Fluid Dynamics
(CFD) are not commonly used for environmental impact assessment for air pollution is that
too many cases of wind tunnel experiments or CFD simulations would be required because
various classes of atmospheric stability would have to be considered. (Eg. 16 wind directions
×10 stability classes = 160 cases). Another reason is that few institutes have a thermally
stratified wind tunnel. If a general function independent of wind direction, urban shape, and
location expressing the effect of atmospheric stability on pollutant concentration could be
proposed, it would become possible to conduct WT/CFD for only neutral conditions and
convert the results (pollutant concentrations under neutral condition) into those in other
atmospheric stability conditions by using the proposed general function. We attempted to find
the general function that expresses the effect of atmospheric stability on pollutant
concentration from observation data and wind tunnel experiments.
(a) from high chimney
(b) within urban area
Figure 1 Dispersion of pollutant
Proc. of the 8th Asia-Pacific Conference on Wind Engineering – Nagesh R. Iyer, Prem Krishna, S. Selvi Rajan and P. Harikrishna (eds)
c 2013 APCWE-VIII. All rights reserved. Published by Research Publishing, Singapore. ISBN: 978-981-07-8011-1
Copyright doi:10.3850/978-981-07-8012-8 191
1012
Proc. of the 8th Asia-Pacific Conference on Wind Engineering (APCWE-VIII)
Analysis of observation data
We firstly attempted to find the general function that expresses the effect of atmospheric
stability on pollutant concentration by analyzing observation data at air pollution monitoring
stations of the Tokyo Metropolitan Government (TMG). Observation data in this study
consisted of two parts. One was hourly averaged Nitrogen oxides (NOx) concentration data
observed at 25 general air pollution monitoring stations in Tokyo 23 ward area (Fig. 2) The
other part was meteorological data including wind speed and direction, solar radiation and
cloud cover of Japan Meteorological Agency (JMA). We downloaded these data for five years
from 2006 to 2010.from the TMG website and the JMA website shown below.
ίhttp://www.kankyo.metro.tokyo.jp/air/air_pollution/result_measurement.html
ίhttp://www.jma.go.jp/jma/index.html
Tokyo 23 ward area
Figure 2 Air pollution monitoring location in Tokyo[4]
Procedure of data analysis is as follows.
Step 1. Define hourly atmospheric stability classes from 3:00 to 21:00 based on the modified
Pasquill–Gifford stability [Table 1: Environmental research and control center (2000)]
using above meteorological data for five years. The data from 22:00 to 2:00 were
omitted because there are no cloud cover data to evaluate atmospheric stability classes
during that time zone.
Table 1 Modified Pasquill-Gifford stability classes
Nighttime cloud cover
Daytime Solar radiation (T) (kW/m2)
Surface wind speed at
10m above ground
T • 0.6
0.6 > T • 0.3
0.3> T • 0.15
T <0.15
8~10
5~7
0~4
(U) (m/s)
U<2
A
A-B
B
D
D
G
G
2” U <3
A-B
B
C
D
D
E
F
3” U <4
B
B-C
C
D
D
D
E
4” U <6
C
C-D
D
D
D
D
D
6”U
C
D
D
D
D
D
D
Note: The nighttime refers to the period when there is no solar radiation
Solar radiation data are used during daytime and cloud cover data are used during nighttime.
The first and last hour of nighttime should be defined as neutral condition “D” regardless of cloud cover.
Table 2 Atmospheric stability classes
stability class
definition
corresponding
number
A
very
unstable
A-B
1
1.5
B
B-C
unstable
2
C
slightly
unstable
2.5
3
1013
C-D
3.5
D
E
slightly
neutral stable
4
5
F
stable
G
very
stable
6
7
Proc. of the 8th Asia-Pacific Conference on Wind Engineering (APCWE-VIII)
Step 2. For each station normalize hourly NOx concentration value by the formulation below.
(1)
C* =CUH2/Q
where C* is non-dimensional normalized NOx concentration; C is observed NOx
concentration (ppb); U is wind speed (m/s); H is the representative length (m) and it is
always constant; Q is the emission rate (m3/s). But in the actual procedure, only CU
(ppb㨯m/s) was calculated for C* because H and Q can be eliminated after the following
assumption and procedure.
Step 3. For each station, classify the C* by 133 groups based on day and hour (7 days and 19
hours from 3:00 to 21:00). We assume that the NOx pollutant emission rate Q is the
same if the day and hour is the same.
Step 4. For each station, divide the above 133 groups into 2128 groups (133*16) based on 16
wind directions
Step 5. For each station, divide the above 2128 groups into 10640 groups (2128*5) based on 5
wind speed (U) ranges (0.5<U<2m/s; 2”U<3m/s; 3m/s”U<4m/s; 4m/s”U<6m/s; 6
m/s”U). The first range 0.5<U<2m/s is different from the first range in table 1
(U<2m/s). We omit wind speed data less than 0.5m/s because the wind direction under
very weak wind condition is not stable and always changing.
Step 6. For each station, select C* under neutral conditions “C*n” on each day, hour, wind
direction, and wind speed range.
Step 7. For each station, average the selected C*n to obtain the 5 years averaged C*n,ave values
for each day, hour, wind direction, and wind speed range.
Step 8. For each station, divide all C* data by C*n,ave to get the ratio that expresses the stability
effect for each day, hour, wind direction, and wind speed range. (Q and H are
eliminated in this step because the same Q and H exist in both denominator and
numerator. As described in step 3, we assumed that Q is the same if the day and hour
is the same. This is because industrial activities, traffic volume, and human activities
in residential houses and office buildings which emit pollutants are considered to be
almost the same if the day and hour is the same.)
Thus, Stability Effect Ratio = C*/ C*n,ave
(2)
Fig. 3 shows an example of C*n at station No.104 for NE wind direction at each hours. 5
years averaged value of C*n,ave and ± 1ı (standard deviation) are plotted in the figure. The
vertical dotted lines correspond to 00:00 (midnight). The daily variation of C*n,ave in hour is
regular from Mondays to Sundays. C*n,ave increases from the minimum value at 3am to the
maximum value at 9:00, because 9:00 is the rush hour.
80
C*n(ppb⋅m/s)
60
40
20
0
-20
0 (Sun) 24 (Mon) 48 (Tue) 72 (Wed) 96 (Thu) 120 (Fri) 144 (Sat) 168
Time (Hour) (Station No.104. H = 4m)
Figure 3 Cn in NE wind direction (station No.104)
We attempted to integrate all the monitoring stations data to find a general relationship
between atmospheric stability and NOx concentration. The index “Stability Effect Ratio”
(SER) was proposed for this purpose. It is the ratio between C* and C*n,ave at each hour, each
1014
Proc. of the 8th Asia-Pacific Conference on Wind Engineering (APCWE-VIII)
2
1
0
-1
1
8098
2
3
4
5
6
Atmospheric stability
7
1
61
297
2
1
0
122
-1
1
327
54
2089
185
2
3
4
5
6
Atmospheric stability
(d) No. 138 station ( N )
7
n,m
6
208
1213
1
0
1022 2112 483 14765
-1
1
329
2
3
4
5
6
Atmospheric stability
7
3
3
194
1671
1
0
-1
1
228
92
2
3
4
5
6
Atmospheric stability
73
200
1138
1
0
1094 2059 473 14453
-1
1
314
2
3
4
5
6
Atmospheric stability
7
(c) No.138 station
(all wind directions)
2
80
2301 105
2
(b) No. 104 station
(all wind directions)
Stability effect ratio C*/C*
n,m
Stability effect ratio C*/C*
359
73
2
(a) all data
3
2253 108
n,m
2686952796 11945366462
3
n,m
29318
Stability effect ratio C*/C*
5206
7
(e) No. 138 station ( NE )
Stability effect ratio C*/C*
58116 2665 1843
n,m
3
Stability effect ratio C*/C*
Stability effect ratio C*/C*
n,m
day and each wind direction for each station. All averaged values of the SER ± 1ı (all stations,
days, hours, wind directions, and wind speed ranges) are plotted against stability classes in
Fig. 4a. The numbers in horizontal axis express the atmospheric stability classes shown in
table 2. Data numbers of each class are written in the figure. The stability effect ratio (SER) is
less than 1 under unstable atmospheric conditions and larger than 1 under stable atmospheric
conditions. However the standard deviations are very large probably due to the variability
(uncertainty) of atmospheric condition and pollutant emission rate. In order to check whether
the SER in Fig. 4a is independent of urban shape and location, two additional figures are
added: SER (all days, hours, wind directions, and wind speed ranges) at Station No. 104 (in
Shinjuku-ku) and station No. 138 (in Edogawa-ku) were displayed in Fig. 4b and c. The
tendencies of Fig. 4b-c are similar to Fig. 4a. Not only these two stations but also all the
stations have the similar tendency (figures are omitted). Thus, there is a possibility that the
Stability Effect Ratio (SER) is independent of urban shape and location. In addition, in order
to check whether the SER in Fig. 4a is independent of wind direction, three additional figures
are added (Fig. 4d-f). SER at Station No. 138 (in Edogawa-ku) under N, NE and S wind
directions were shown in figures 4d-f, respectively. Some differences were found depending
on wind direction. However data number is small. So we cannot judge whether the
atmospheric stability ratio is independent of wind direction from these data.
3
166
47
12
19
60
2
1
0
48
-1
1
251
138 1705
20
2
3
4
5
6
Atmospheric stability
7
(f) No. 138 station ( S )
Figure 4 Stability effect ratio
Wind tunnel experiments
As mentioned above, it was found there is a possibility that the Stability Effect Ratio
(SER) was independent of wind speed, urban shape and location. But the deviation of the
ratio was large probably due to variability (uncertainty) of atmospheric condition and
pollutant emission rate. Thus in order to reduce the variability (uncertainty) of atmospheric
condition and pollutant emission rate, wind tunnel experiments were conducted in a thermally
stratified wind tunnel at Tokyo Polytechnic University (TPU). Fig. 5 shows the experimental
setup. The unstable, neutral, and stable turbulent boundary layers were generated by heating
1015
Proc. of the 8th Asia-Pacific Conference on Wind Engineering (APCWE-VIII)
or cooling wind tunnel floor and air with 23 thin roughness elements which were made of
aluminum angles in the upstream. They create a long, rough upwind fetch to generate a
turbulent boundary layer. Total 9×14=126 cubic blocks were put in the turbulent boundary
layers in a downstream (Figs. 5 and 6) to represent building blocks. Each building block has
the same configuration: L (60mm) × W (60mm) × D (60mm). The city blocks were spaced
60mm apart in both x and y direction. Tracer gas ethylene (C2H4) was released from a line on
the floor. Fig. 7 shows measuring points. The locations of the measuring points were selected
so that various flow patterns (reverse flow, upward flow, downward flow in the street canyons
and flow on the roads) were included.
1200
500 5
al u m
i nu m
00
angle
23× @
20 0
4 4 00
d coo
ling p
anels
1000
x
Heati
ng an
street
1 8 40
can y o
n mo
460
00
z
9370
Temp
e
profi rature
le car
t
y
del
12
Wind
1670
Figure 5 Experimental setup (units: mm)
37mm
Gas emission line
W
W
W
W
W
W=60mm
300mm
1620mm
980mm
Figure 6 Arrangement of building blocks
Gas emission line
Gas
emission line
(a) top view
(b) side view
Figure 7 Measuring points
1016
Proc. of the 8th Asia-Pacific Conference on Wind Engineering (APCWE-VIII)
Table 3 summarizes the atmospheric stability conditions. Totally 5 cases were
conducted. The reference height was 0.32m, and the velocity at this height of inflow boundary
was set as reference velocity (UR). The atmospheric stability was characterized by Bulk
Richardson’s number (Rib). Bulk Richardson number is expressed as the following.
Rib =
gH R × (TR − Ts )
(T0 + 273) × U R 2
Where g is the acceleration due to gravity [m/s2]; HR is the reference height [m]; TR is the
temperature at the reference height [oC]; TS is the ground surface temperature [oC]; T0 is the
average inflow temperature [oC]; UR is the velocity at the reference height (m/s). The Rib for
5 inflow profiles were summarized in the last row of Table 1. Values of Rib ranged from -0.23
(unstable) to 0.29 (stable).
Table 3 Atmospheric stability conditions (HR =0.32m)
Unstable
Weakly
unstable
Neutral
Weakly
stable
Stable
1.4
1.8
1.8
1.5
1.2
TR (°C)
9
10
11
48
55
TS (°C)
49
41
11
16
14
ǻT (°C)
40
31
0
32
41
T0 (°C)
13
14
11
48
55
-0.23
-0.1
0
0.13
0.29
UR (m/s)
Rib
Results of wind tunnel experiments
The correlations for normalized non-dimensional concentration C* between neutral
condition and non-neutral conditions were investigated and shown in Fig. 8. The nondimensional concentration C* is defined as follows.
C ⋅ U R ⋅ H R2
C* =
q
Where C is the measured concentration [-]; q is the gas emission rate [m3/s]
All the measured concentration data (measuring positions, see Fig.7) in the wind
tunnel tests were included in the figures. There were totally 92 measuring points for each case.
Those measuring positions located in different places. Some were in the street while some
were in the canyon between blocks. In addition, four heights in vertical direction were
included. As shown in the figures, data were plotted almost on a single straight line. Thus the
ratio between C* under non-neutral conditions and neutral condition were almost independent
of the measurement locations (the flow filed around the measuring positions). The slope
became steeper with the increase of the Bulk Richardson Number, which means nondimensional concentration became higher with the increase of the Bulk Richardson number.
Fig. 9 shows the ratio between C* under non-neutral condition and C*n under neutral
condition obtained from the experimental results. This is the Stability Effect Ratio (SER) on
pollutant concentration (SER= C*/ C*n). Averaged SER ± ı (the standard deviation) of all the
1017
Proc. of the 8th Asia-Pacific Conference on Wind Engineering (APCWE-VIII)
measuring points are plotted in the figures. As shown in the figures, with the increase of Rib,
the SER increases. Since the standard deviation is relatively small, the SER is almost
independent of the locations or flow fields.
40
30
Unstable (C ) (Rib=-0.23)
y=0.704
2
R =0.896
20
*
*
Unstable (C ) (Rib=-0.1)
40
10
0
0
10
20
30
30
20
10
0
40
y=0.611x
2
R =0.848
0
10
*
Neutral (C )
(a) Rib=㧙0.1
Stable (C ) (Rib=0.29)
40
30
*
20
*
Stable (C ) (Rib=0.13)
40
(b) Rib=㧙0.23
40
y=1.315x
2
R =0.955
10
0
20
30
*
Neutral (C )
0
10
20
30
30
20
10
0
40
y=1.378x
2
R =0.883
0
10
*
Neutral (C )
20
30
40
*
Neutral (C )
(c) Rib=0.13
(d) Rib=0.29
Figure 8 Correlations of C* (Neutral VS. non-neutral)
2.0
1.6
C*
C*n
1.2
0.8
0.4
0.0
㪄㪇㪅㪊
㪄㪇㪅㪉
㪄㪇㪅㪈
㪇㪅㪇
㪇㪅㪈
㪇㪅㪉
㪇㪅㪊
Rib
Figure 9 Stability Effect Ratio (SER= C*/ C*n)
㪇㪅㪋
Conclusions
Five year’s observation data of pollutant concentration were analyzed, and wind
tunnel experiments were conducted for investigating the effect of atmospheric stability on
urban pollutant concentration. It was found that the Stability Effect Ratio (SER) on pollutant
concentration (SER= C*/ C*n) became higher with the increase of atmospheric stability, and it
was almost independent of the locations or flow fields. If the function of the SER is universal
one, we can predict pollutant concentration under non-neutral condition from experimental or
CFD results under neutral condition by using this function. It will bring about a drastic change
1018
Proc. of the 8th Asia-Pacific Conference on Wind Engineering (APCWE-VIII)
in the environmental impact assessment of air pollution in Japan. We will continue further
investigation to confirm the generality of the SER function by changing urban configurations
in wind tunnel experiments.
References
Bureau of Environment, Tokyo Metropolitan Government (2006-2010),
http://www.kankyo.metro.tokyo.jp/air/air_pollution/result_measurement.html
Environmental research and control center (2000), “Regulation manual of total amount of nitrogen oxide”,
pp.80-82, (in Japanese)
Japan Meteorological Agency (2006-2010), http://www.jma.go.jp/jma/index.html
1019