Correlation of Atmospheric Visibility and
Meteorological Factors in Southern Taiwan
Control # 576
Horng-Yu Yang
Department of Civil Engineering, Institute of Civil & Disaster Prevention Engineering,
China Institute of Technology
245, Yen-Chiu-Yuan Road, Sec. 3rd, Taipei, Taiwan, ROC
Chung-Shin Yuan
Institute of Environmental Engineering, National Sun Yat-Sen University
70, Lian-Hai Road, Kaohsiung, Taiwan, ROC
ABSTRACT
The objective of this study was to investigate the correlation of atmospheric
visibility and meteorological factor with weather patterns in Kaohsiung Area.
Daytime atmospheric visibility was observed around noon at two observation sites in
metropolitan Kaohsiung, Taiwan in the years of 1999 and 2000. Both seasonal and
diurnal variation patterns of atmospheric visibility were observed in the region. The
visibility its field of observation were consecutively conducted to investigate the
influence of pollution sources and meteorological factors on urban air quality. The
results from regular observation of visibility from November 1998 to April 2000
indicated that most frequently observed visibility range from 2 to 8 kilometers.
Approximately 79.3% of these days are with visibility less than 8 kilometers, which
observed mainly during in the wintertime. However, intensive observation of
visibility and measurement of scattering coefficient from January 8-16 to March
24-30 2000 indicated that atmospheric visibility had an opposite correlation with
scattering coefficient for relative humidity less than 70%. Moreover, daily observation
data indicated that the worst visibility occurred in early morning at 6:00am and the
best visibility occurred in the late afternoon at 1:00pm. In addition, seasonal variation
of visibility indicated that the best and the worst visibility seasons were summertime
and wintertime in Kaohsiung city. This study concluded that the relationship of the
atmospheric visibility and meteorological factor with weather patterns are very
closed .
INTRODUCTION
The atmospheric visibility in Kaohsiung city has been regularly observed to
characterize the variation of atmospheric visibility since November 1998. Three
levels of atmospheric visibility were set as good (＞8 km), moderate (2-8 km), and
poor (＜2 km). For investigating the development of an innovative technology to
determine atmospheric visibility, the feasibility of using atmospheric visibility as
ambient air quality index. Kaohsiung city was selected for this particular study
owning to its worst ambient air quality, especially impaired atmospheric visibility,
over the Taiwan Island. With rapid population growth and increasing industrialization
over the past three decades, poor ambient air quality has become one of the major
environmental concerns of the general public in Kaohsiung city. In order to
characterize the visual air quality, the atmospheric visibility has been regularly
observed in the Kaohsiung city since November 1998. General speaking, the
atmospheric visibilities observed at Kaohsiung area are very poor and are usually
lower than 10 kilometers (or 6.2 miles). The most frequently observed atmospheric
visibility ranged from 2 to 4 kilometers that is much poorer than most mega cities
over the world. Previous study characterize the visibility of Kaohsiung city into three
levels: good (＞8 km), moderate (2-8 km), and poor (＜2 km). (1) Moreover, previous
investigation also suggested that atmospheric visibility might be used as an alternative
ambient air quality of the Kaohsiung city. (2)
Atmospheric visibility is defined as the greatest distance in a given direction at which
an object can be visually identified with unaided eyes. The object could be a dark
object positioned prominently against the sky on the horizon in the daytime or a
known, preferably unfocused moderately intense light source at nighttime. Generally,
atmospheric visibility varies significantly with regions and seasons. Meteorological
parameters of various weather patterns (including relative humidity，wind direction，
wind speed and mixing height) played important roles on the reduction of visibility in
Metropolitan Kaohsiung. This study attempted to observe the influence with
meteorological parameters on visibility in the area. This study concluded that
visibility might be used as an indicator of ambient quality. In order to characterize and
further improve the ambient air quality of Kaohsiung city, both central government
and local agency have paid many attentions on visual air quality, mainly atmospheric
visibility, since 1998. As part of the visual air quality improvement project sponsored
by National Science Council of Republic of China and Environmental Protection
Department of Kaohsiung Municipal Government, this investigation studied the
statistical analysis of field observed atmospheric visibility, the feasibility of using
atmospheric visibility as ambient air quality index, and finally the determination and
forecasting of atmospheric visibility using an automated atmospheric visibility
observation system.
METHODOLOGY
A. Field observation of atmospheric Visibility
This study was based on both regular and intensive observations of atmospheric
visibility in the Kaohsiung city. Regular observation was conducted to elucidate the
temporal and spatial variation of atmospheric visibility, while intensive observation
was conducted to elucidate the diurnal variation of daytime atmospheric visibility.
Regular observation of atmospheric visibility was performed each day at 11:00 am
and 2:00 pm from November 1998 to December 2001. On the other hand, four
intensive observation campaigns were conducted in January-May 1999 and
January-March 2000, which were conducted each hour from 7:00 am to 5:00 pm.
Concurrently, Kaohsiung aerosols were sampled in situ and meteorological data were
obtained from Kaohsiung Meteorology Station for further statistical analyses.
Atmospheric visibility was observed without visual aids at two observation sites,
Kaohsiung Meteorology Station and Farshing Temple, as shown in Figure1.At
Kaohsiung Meteorology Station, the observations were made facing northward with
an observation sector of 60o, while those at Farshing Temple were made facing
eastward with an observation sector of 120o. Hence, the field observed atmospheric
visibility herein was also named as sector visibility. Objects were pre-identified prior
to observation at each site for determining the atmospheric visibility. The location of
each pre-identified object was mapped with global positioning system (GPS) and the
GPS data was then used to determine the distance from the observation site to each
object with an error lower than 50 m. In this study, eleven and seventeen objects were
positioned and observed from Kaohsiung Meteorology Station and Farshing Temple,
respectively.
Fig. 1. The atmospheric visibility its location of observation sites and aerosols its sampling
site in the metropolitan Kaohsiung.
B. Statistical Analysis
Both daily and seasonal variations of atmospheric visibility were plotted with date in order
to understand the influence of meteorological factors (mainly relative humidity) on the
trend of visibility. Daily visibility data observed from 1999 to 2000 was further compared
with PM10 concentration and Pollutants Standard Index (PSI) to evaluate the relationship
between atmospheric visibility and ambient air quality. A regression model described the
relationship between visibility and PM10 concentration was also established and validated.
Results from statistical analysis were used to evaluate the possibility of using atmospheric
visibility as an indicator for ambient air quality. Furthermore, the regression model was
applied to reconstruct the PM10 concentration in previous period while there was no field
measured PM10 concentration available.
RESULT AND DISCUSSION
Temporal and Spatial Variation of Atmospheric Visibility
Regular observation of daytime atmospheric visibility was performed at 11:00 am
and2:00 pm at two observation sites in metropolitan Kaohsiung. Figure 2 plots the
cumulative distribution of atmospheric visibility observed from November 1998 to
December 2001. It showed that the field observed atmospheric visibilities in
metropolitan Kaohsiung were always lower than 21 km, ranging from 0.7 to 20.6 km.
According to the levels of atmospheric visibility (i.e. 2 and 8 km) regulated by the
Environmental Protection Department of Kaohsiung, the cumulative percentage of
visibility less than 2 and 8 km were 7.8 and 76.2%, respectively. Approximately 50 %
of the field observed atmospheric visibilities were lower than 5 km, indicating that the
atmospheric visibility in metropolitan Kaohsiung has been seriously impaired.
Moreover, regular observation of daytime atmospheric visibility at Kaohsiung
Meteorological Station yielded an average atmospheric visibility of 6.1 and 7.2 km at
11:00 am and 2:00 pm, respectively.
Percentage of frequency with atmospheric
Visibility
visibility(km)
less than shown (%)
Figure 2. Cumulative distribution of daytime atmospheric visibility
observed in the Kaohsiung city in the years of 1999-2001.
100
9
90
80
70
8
60
50
7
40
30
6
20
10
0
5
0
2
4
6
8
10 12 14 16 18 20 22 24
Atmospheric visibility (km)
The results4were consistent with previously reported monthly average atmospheric
6
8
14 (3)
visibility of 7.2
km (1975-1994)
with10a range of 12
3.3-13.10 km.
Figure16
3 illustrates18
the monthly variation of daytime atmospheric visibility
during the investigation
hr
period from November 1998 to December 2001. A significant seasonal variation of
atmospheric visibility was observed in the Kaohsiung city. The atmospheric visibility
was
generally higher in summer and lower in late autumn and winter. Moreover, the
hsiung.
mean seasonal atmospheric visibilities in spring, summer, fall, and winter were 5.4,
9.1, 8.2, and 3.4 km, respectively. Low atmospheric visibility observed in late autumn
and winter was mainly due to the high frequency of fog formation, resulting from the
radiation inversion occurred in the early morning.During the intensive observation
periods, daytime atmospheric visibility was observed at two observation sites each
hour from 7:00 am to 5:00 pm.
Figure 3. Seasonal variation of daytime atmospheric visibility observed in the
Kaohsiung city from November 1998 to December 2001.
24
Kaohsiung Meteorology Station (11:00 am)
Kaohsiung Meteorology Station (2:00 pm)
Farshing Temple (11:00 am)
Farshing Temple (2:00 pm)
22
20
9
18
16
14
8
12
Visibility (km)
10
7
8
6
4
6
2
Nov./01
Sep./01
Jul./01
May./01
Mar./01
Jan./01
Nov./00
Sep./00
Jul./00
May./00
Mar./00
Jan./00
Nov./99
Sep./99
Jul./99
May./99
5
Mar./99
Jan./99
Nov./98
0
Month/Year
4
6
8
10
12
14
16
18
Figure 4 illustrates the diurnal variation of daytime atmospheric visibility and
hr
compares the atmospheric visibility observed at two sites. The atmospheric visibilities
observed at Kaohsiung Meteorology Station (northward) were slightly higher than
those
at Farshing Temple (eastward). Atmospheric visibility tended to increase in the
hsiung.
morning and then leveled off at 2:00-3:00 pm in the afternoon. During four intensive
observation periods, the atmospheric visibilities in three winter periods were markedly
higher than those in the early summer period. On average, the lowest mean
atmospheric visibilities at 7:00 am were 5.7 and 4.8 km at Kaohsiung Meteorology
Station and Farshing Temple, respectively. While, the highest mean atmospheric
visibilities at 3:00-5:00 pm were 7.9 km and 7.5 km at Kaohsiung Meteorology
Station and Farshing Temple, respectively. It is quite consistent with previously
reported diurnal variation of atmospheric visibility observed for the past two
decades.(4) In that study, the lowest atmospheric visibility observed at 7:00 am were
around 5 km at Kaohsiung Meteorology Station, while the highest atmospheric
visibility observed at 3:00-5:00 pm were around 8 km. The phenomenon was
attributed to the following two reasons. First of all, relative humidity was usually
higher in the early morning, causing fog formation, which scattered visible light and
hence impaired atmospheric visibility. Secondly, the primary fine particles emitted
from motor vehicles in rush hours would enhance the effects of light scattering and
absorption, and thus reduced the atmospheric visibility due to the condensation and
nucleation of moisture onto the fine particles in the early morning.
Figure 4. Diurnal variation of daytime atmospheric visibility
observed in the metropolitan Kaohsiung during the intensive
observation periods of 1999-2000.
12
Farshing Temple
Kaohsiung Meteorology Station
10
Jan. 28~Feb. 6, 1999
8
6
4
2
0
12
May 21~30, 1999
10
Farshing Temple
8
Kaohsiung Meteorology Station
Visibility (km)
6
4
2
0
12
Jan. 8~16, 2000
10
8
6
4
2
0
12
March 24~30, 2000
10
8
6
4
2
0
6
8
10
12
14
16
18
Local time
Establishment of Visual Standard Index (VSI)
In this study, a Visual Standard Index (VSI) was originally established to compare with
traditional Pollutants Standard Index (PSI) of ambient air quality. VSI used atmospheric
visibility as a solely indicator, while traditional PSI considers fine criteria air pollutants
(PM10, O3, SO2, NOx and CO). In order to establish the visual air quality standard,
correlation between field measured atmospheric visibility and PSI value was initially
investigated for four various observation situations (i.e. two periods at two observation
sites). Table 1 show that the measured atmospheric visibility was moderately correlated
with PSI value. Ambient air quality with a PSI value of 100 was equivalent to the
atmospheric visibility of approximately 4.0Km in metropolitan Kaohsiung. A five-level
VSI was established to describe the ambient air quality. The range of atmospheric visibility
levels were <0.5 Km, 0.5~2.0 Km, 2.0~4.0 Km, 4.0~10 Km, >10 Km, respectively, for
very poor, poor, moderate, good, and very good ambient air quality in urban area shown in
Figure 5.The distribution of field measured atmospheric visibility in metropolitan
Kaohsiung was shown in Figure 6. It shows that the cumulative percentage of atmospheric
visibility less than 1.0 Km, 2.0Km, 4.0 Km, 10.0 Km was approximately 2.09%, 16.59%,
43.17%, and 84.29%, respectively. It shows that the ambient air quality and atmospheric
visibility in metropolitan Kaohsiung were much worse than most mega cities in the world.
Table 1. In Comparison with the five level visual standard index and pollutants
standard index (PSI) in the metropolitan Kaohsiung.
Air Query Level
very poor
atmospheric visibility
<1
(Km)
PSI value
>300
atmospheric visibility
2.09
(%)
poor
1-2
moderate
2-4
good
4-10
very good
>10
150-300
14.50
100-150
26.58
50-100
41.12
<50
15.71
Fig 5. A five level visual standard index and the cumulative percentage frequency of
atmospheric visibility in metropolitan Kaohsiung (November 1998 – June 2001)
Fig 6. Distribution of field measured atmospheric visibility in metropolitan Kaohsiung
(November 1998 – April 2000)
Influence of Meteorological Factors on Atmospheric Visibility
During the intensive observation period, atmospheric visibility had a tendency to
continuously increase in the morning and then leveled off at approximately 7:00 AM to
3:00 PM. It was quite consistent with previous reported cyclic diurnal variation of visibility
observed for the past twenty years. In this study, the worst visibility occurred at 7:00 AM
were 5.7 Km and 4.8 Km at Kaohsiung Meteorological Station and Farsing Temple,
respectively. While the best visibility occurred at 3:00 to 5:00 PM were 7.9 Km and 7.5
Km at Kaohsiung Meteorological Station and Farsing Temple, respectively. These results
suggested that a correlation between atmospheric visibility and meteorological factors,
mainly relative humidity, mixing height. High relative humidity up to 100% in the early
morning resulted in fog formation, which would scatter visible light and thus reduce the
atmospheric visibility.
In addition to relative humidity, wind direction and wind speed also played important roles
on the dispersion of aerosol particles and thus had influence on atmospheric visibility. This
study revealed that low atmospheric visibility usually occurred in late autumn and winter
as strong northeastern monsoon blew. High atmospheric visibility occurred in summer
when southwestern monsoon blew. Another important meteorological factor is Ventilation
Index (VI) that is defined as the multiple effects of wind speed and mixing height. A
moderate correlation between atmospheric visibility and Ventilation Index (VI) is as
follows:
Equation 1.
Y = 0.5512X + 595.87; R2= 0.6124
Where:
Y=VI (Ventilation Index)
X=visibility
and regression model illustrated in Fig. 7.
Ventilation Index
Fig. 7. Dependency of atmospheric visibility upon ventilation Index
Validation for Atmospheric Visibility Forecasting
A multiple regression model was established to correlate atmospheric visibility with
air pollutants and meteorological factors is shown below. Among them, NO2 and SO2
are the precursors of secondary nitrate and sulfate aerosols. Relative humidity
represents the moisture content of aerosol particles. Although fine particles (0.26μm
< dp <0.9μm) were thought as the most efficient aerosols to scatter visible light,(6)
PM10 was used as the input variable since it was routinely measured in situ. The
correlation coefficients of the regression models for various weather patterns are listed
in Table 2.
Table 2. Correlation coefficients of multiple regression models for various
weather patterns.
Weather Pattern
Correlation Coefficient (R)
Cold Front System
0.90
Northwestern Monsoon System
0.78
Continental High-Pressure System
0.92
Subtropical High-Pressure System
0.93
Low-Pressure System
NA
Other Systems
0.91
NA: data is not enough to correlate visibility with air pollutants and
meteorological parameters
Equation 2.
log(Lv) = a1PM10 + a2NO2 + a3SO2 + a4O3 + a5RH + a6WS + a7WD1 + a8WD2 +
a9 WD3 + a10WD4 + a11WD5 + a12WD6 + a13WD7 + a14WD8 + a15
where:
Lv = atmospheric visibility (km)
PM10 = particulate matter with dp≦10μm (μg/m3)
SO2, NO2, O3 = concentration of gaseous pollutants (ppbv)
RH = relative humidity (%)
WS = wind speed (m/s)
WD1-8 = wind direction (clockwise, “1” for the northeast and “8” for the north).
Equation (2) was then used to determine the atmospheric visibility with known input
variables and could be further compared with the observed atmospheric visibility.
During this investigation period, the accuracy of atmospheric visibility forecasting
based on three visibility levels was approximately 86.7% with root mean square of
1.84 km. The correlation coefficient of forecasted and observed atmospheric visibility
was 0.75. Approximately 92.0% of the forecasted atmospheric visibility was within an
error of 3.0 km, while about 77.9% of the forecasted atmospheric visibility was within
an error of 2.0 km. Moreover, around 41.6% of the forecasted atmospheric visibility
lied within an error less than 1.0 km.
These results implied that the forecasting of atmospheric visibility was feasible that
can be achieved by approximately 70.0% for an error of 1.84 km. Further correlation
analysis of weather patterns with field measured atmospheric visibility indicated that the
worst atmospheric visibility as well as ambient air quality were usually occurred in winter
with weather patterns of high pressure outflow Style I (Fig. 8) and of high pressure outflow
circumfluence (Fig. 9). Regression models between atmospheric visibility and ambient air
pollutants (i.e. SO2, NOx and PM2.5) as well as meteorological factors (i.e. relative
humidity, wind speed, wind direction, and mixing height) were also developed for sixteen
weather patterns.
Fig. 8. Weather Pattern : broken high- pressure outflow style Ι
Fig. 9. Weather Pattern: high-pressure outflow circumfluence
CONCLUSIONS
1. Moreover, a multiple regression model considering both ambient air pollutants
(PM10, NO2, SO2, and O3) and meteorological parameters (relative humidity, wind
speed, and wind direction) was successfully developed to forecast the atmospheric
visibility for various weather patterns.
2. The atmospheric visibility was then forecasted with the multiple regression models
at 5:00 pm each day. Based on a hundred of effective atmospheric visibility data
forecasted from June 15th to October 15th, 2003, the accuracy was approximately
96.0% with a root mean square of 1.79 km on the basis of visibility levels (i.e. good,
moderate, and poor). The correlation coefficient of the forecasted and the observed
atmospheric visibility was 0.80.
3. Diurnal variation of atmospheric visibility indicated that atmospheric visibility was
highly influenced by meteorological factors, especially the relative humidity of the
ambient atmosphere.
4. Atmospheric visibility has a close correlation with seasonal variation and weather
patterns style. In this study, we find that the worst visibility occurred in winter. High
PM10 concentration and unhealthful PSI usually occurred at weather patterns of the high
pressure outflow style I and high pressure outflow circumfluence.
5. Results from visibility observation shows that the visual range in summer is wider than
that of in winter, it can be affected by both weather patterns and local circulation (ex.
sea-land breeze etc.) in metropolitan Kaohsiung. In winter, atmospheric visibility is
major affected by radiant inversion due to continental cold high pressure so air pollution
diffuses difficultly and atmospheric visibility is worse. In summer, atmospheric visibility
is affected by convective effect due to tropic low or subsidence effect that makes
temperature increase and moisture vapor decrease so that air pollution diffuses easily and
atmospheric visibility is better. The best atmospheric visibility occurred in summer,
while the worst atmospheric visibility occurred in late autumn and winter.
6. Generally, atmospheric visibility in the afternoon is better than that in morning because
of early morning radiative inversion on surface making pollutants diffuse difficultly and
thus the relative humidity in morning is higher than afternoon in south of Taiwan.
Results from atmospheric visibility in morning are much worse than in afternoon. The
statistical analysis of long-term visibility data indicated that the worst hourly average
visibility of 4.70 Km occurred at 6:00 AM, while the best visibility of 9.10 Km occurred
at 3:00 PM.
7. A high contribution of soil dusts is also an important factor. During the second intensive
observation period the visibility was affected by a strong sand storm blown from
Northern China to Taiwan in late winter and early spring, especially in late March. The
concentration of coarse particles increased approximately 2 to 3 times in average during
the sand storm period.
8. The relative humidity in the early morning is about 70% to 85% and is around 60% in
the afternoon during the observation period in the metropolitan Kaohsiung. According to
U.S. EPA data, it shows that once the relative humidity is over 70% the visual range will
be affected by atmospheric vapor due to its particle size distribution. If relative humidity
became higher and thus aerosol particle size became larger, it will make extinction
coefficient increase and visibility decrease. Bigger size of the aerosol particle caused by
the higher relative humidity is the major factor of the lower visibility.
9. According to data analysis, it shows that low wind speed and clear day can cause
inversion easily (dry and cold air in winter) but wind speed is increase and mixing effect
by turbulent and inversion is broken by convection. Low wind speed is occurred in the
early morning is about 1.9 m/s and wind speed increased about 3.8 m/s near noon.
Results from mixing height and wind speed increasing atmospheric Ventilation Index
also increase to change atmospheric stability from stable to unstable, therefore pollutant
can diffuse easily and thus the visibility in afternoon is better than that in early morning.
ACKNOWLEDGEMENTS
This study was performed under the auspicious of National Science Council of Republic of
China under the Contract No.NSC-92-2622-E-110 and Environmental Protection Bureau,
Kaohsiung Municipal Government under the Contract No. DEPK-A-91-01-12. The authors
would like to express their sincere appreciation for the financial supports to accomplish this
study.
REFERENCES
1. Yuan, C.S.; Lin, Y.C.; Lee, C.G.; Lo, C.C. ;The 14th IUAPPA International
Conference on Air Quality – Assessment and Policy at Local, Regional, and
Global Scales, Dubrovnik, Croatia, Oct. 6-10, 2003.
2. Won, C.H; Yuan, C.; Yuan, C.S.; Alternative Air Quality Index, Final Report,
Department of Environmental Protection, Kaohsiung Municipal Government.
3. Chung-Shin Yuan, Chang-Gai lee, San-Ho Liu, Ching Yuan, Horng-Yu Yang, and
Chen-Tung Arthur Chen. ; “Developing Strategies for Improving Urban Visual
Air Quality”; Aerosol and Air Quality Research.2002,Vol. 2, 9-22.
4. Yuan C. S., Chang J. C., Yuan J. S., Lin Z. T., Yuan C. and Yang H. Y.; A field
Study on the Relationship of Visibility with Suspended Particles and
Meteorological Factors in Kaohsiung city. The 90th A&WMA Annual Meeting, San
Diego, California.1998.
5. Xu J., Bergin M., Yu X., Zhao J., Carrico C., Beaumann K.;Measurement of
Aerosol Chemical, Physical and Radioactive Properties in the Yangtze Delta
Region of China. Atmos. Env.2002, 36: 161-173.
6. Luo, C.H., Liu, S.H., and Yuan, C.S. Aerosol and Air Quality Research, 2002, 2(1),
23-29.
KEY WORDS
atmospheric visibility, meteorological factor, weather patterns, relative humidity