ARTICLE IN PRESS

Habitat International 29 (2005) 547–558

Study of green areas and urban heat island in a tropical city

Nyuk Hien Wong*, Chen Yu
Department of Building, School of Design and Environment, National University of Singapore, Singapore 117566,
Singapore
Received 14 September 2003; received in revised form 4 January 2004; accepted 12 April 2004

Abstract

Green areas in cities have been considered as potential measure in mitigating the urban heat island (UHI)
effect. In this paper, a mobile survey was conducted to explore both the severity of UHI effect and cooling
impacts of green areas at macro-level in Singapore. Islandwide temperature distribution was mapped
relying on data derived from the mobile survey. This study has indicated a strong correlation between the
decrease of temperature and the appearance of large green areas in the city. Although there is no distinct
borderline between ‘urban’ and ‘rural’ areas in Singapore, maximum temperature difference of 4.01 C was
observed.
r 2004 Elsevier Ltd. All rights reserved.

Keywords: Green areas; Urban heat island; Mobile survey; Macro-level; Island wide temperature distribution

1. Introduction

With rapid urbanization, there has been a tremendous growth in population and buildings in
cities. The high concentration of hard surfaces actually triggered many environmental issues. The
urban heat island (UHI) effect, one of these environmental issues, is a phenomenon where air
temperatures in densely built cities are higher than the suburban rural areas. The primary root of
heat island in cities is due to the absorption of solar radiation by mass building structures, roads,
and other hard surfaces during daytime. The absorbed heat is subsequently re-radiated to the
surroundings and increases ambient temperatures at night.
The UHI phenomenon was first noticed by meteorologists more than a century ago (Howard,
1833). Since then, the UHI effect has been well explored worldwide (Oke, 1978; Landsberg, 1981;

*Corresponding author. Tel.: +65-6874-3423; fax: +65-6775-5502.

E-mail address: bdgwnh@nus.edu.sg (N.H. Wong).

0197-3975/$ - see front matter r 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.habitatint.2004.04.008
 ARTICLE IN PRESS

548 N.H. Wong, C. Yu / Habitat International 29 (2005) 547–558

Santamoouris, 2002; Akbari, Rosenfeld, & Taha, 1990; Tso, 1996). Some of the most important
factors which may influence the UHI effect include canyon geometry, thermal properties of
materials, anthropogenic heat, the urban greenhouse effect, and evaporation surfaces
(Santamoouris, 2002). According to Landsberg (1981), UHI, as the most obvious climatic
manifestation of urbanization, can be observed in every town and city.
Green areas are actually the ecological measure to combat the problems of the concrete jungle.
They are quite different from the built up environment. Any surface planted with vegetation has a
different Bowen ratio1 than a mineral surface since the incoming solar radiation is converted into
energy for transpiration and photosynthesis through plants and the sensible heat flux is
consequently lower. At night, the energy of the outgoing net radiation from a green surface is fed
from the thermal heat flux and the latent heat flux. Therefore, the temperature around the green
area is lower than that around the built environment.
The role of green areas in moderating the urban climate has been explored all over the
world during recent years. The studies can be roughly divided into three categories. Firstly,
meteorological data and satellite images were employed to study the effect of green areas in
cities at macro level. Saito (1990/1991) studied the relationship between meteorological
elements and green distribution in Kumamoto City in Japan and concluded that the air
temperature distribution was closely related to the distribution of greenery in the city.
Kawashima (1990/1991) examined the effects of vegetation density on the surface temperatures
in the urban and suburban areas of Tokyo Metropolis and observed lower surface temperature
in green areas. Secondly, some in-depth field measurements were carried out to explore the
cooling effect of green areas at micro-level. Jauregui (1990/1991) found that in a large urban
park (Chapultepec Park) in Mexico City, the ambient temperature was 2–3 C lower than its
surrounding built-up area and its influence reached a distance of 2 km, about the same as
its width. Sonne and Vieria (2000) conducted measurement over a 1 year period at three
Melbourne and Florida sites and found that temperature measured in a forested natural
park was consistently lower than that measured in a residential development with an
extensive tall trees canopy while the temperature measured in this residential development with
extensive planting was always lower than that measured in a residential development with very
few trees. Finally, numerical calculations were set up to predict the thermal benefits of green areas
in cities. Avissar (1996) studied the potential impact of vegetation on the urban thermal
environment by use of a mesoscale atmospheric model. Honjo and Takakura (1990/1991)
estimated the cooling effect of green areas on their surroundings with the use of a numerical
model.
The cooling effect of green areas in cities has been confirmed by the above studies. It is
doubtless that the UHI effect is aggravated mainly due to the lost of green areas in the urban
environment. In Singapore, rapid population influx has led to demands for converting natural
areas to pubic housing. The heat island for Singapore city has been documented (Tso, 1996).
However, less focus has been placed on the cooling effect of city’s green areas at macro level. To
address this issue, temperature measurements were conducted throughout the island by the use of

1
Bowen ratio is the proportion of sensible heat to latent heat leaving a surface. It ranges from o0.1 for a moist
surface to >10 for a dry surface [1].
 ARTICLE IN PRESS

N.H. Wong, C. Yu / Habitat International 29 (2005) 547–558 549

automobiles. The aim of the study was to detect the severity of UHI and cooling impacts of green
areas in Singapore.

2. Methodology

The land use of Singapore: Singapore is a garden city without distinct borderline between urban
and rural areas. However, the two existing predominant green areas, the primary forest of 75 ha in
the middle of the island and the open space/recreation area in the northeast of the island, are
basically located at the northern part of the country while most built-up regions like industry
areas, residential areas, the CBD area and the airport are located at the southern part. Therefore,
the northern part of the country can be considered as ‘rural’ area while the southern part is the
urban area (see Fig. 1).
In order to have a better understanding of how the land use pattern impact the ambient
environment (air temperature and humidity), mobile surveys were conducted to map out the
temperature and relative humidity profiles at the various regions so that correlation can be
established between the land use and the ambient conditions.
The mobile surveys: The mobile survey was conducted by vehicles equipped with observation
tubes (see Fig. 1) at midnight from 2 a.m. to 4 a.m. on 9th July 2002 and 13th September 2002,
respectively. Every observation tube consists of a Hobo temperature/RH mini-datalogger and a
well insulated plastic tube (see Fig. 2). The length of the round tube is 30 cm and its diameter is
7.5 cm The Hobo datalogger, which can automatically record ambient temperature and humidity,
was fixed in the centre of the tube and set at 2 min measuring interval. All the vehicles drove
along the highways at a speed of around 50 km/h. A single route (see Fig. 3) which crosses
the island from the west all the way to the east was chosen during the first survey and four routes
(see Figs. 4 and 5) were chosen which basically cover the whole island during the second survey.
They are western route, eastern route, center route and central business district (CBD) route,
respectively.

Fig. 1. The ‘urban’ and ‘rural’ partition of Singapore (source from Steed).
 ARTICLE IN PRESS

550 N.H. Wong, C. Yu / Habitat International 29 (2005) 547–558

Fig. 2. Mobile surveys conducted by vehicles equipped with observation tubes.

Fig. 3. The observation tube.

3. The survey results

3.1. The first survey

The first survey was to preliminarily investigate the correlation between temperatures
and different land uses. The survey route running from west to east actually passed through
quite a number of different land uses, like industrial areas, residential areas, the forest
and the airport. The results are illustrated in Fig. 6. The solid line shows the fluctuation of
ambient air temperatures while the dotted line represents the variation of relative humidity
during the first survey. The highest air temperature, 28.6 C, was detected in the Bedok industrial
area at around 2:43 a.m. The lowest temperature, 27.0 C, was observed when the car passed
through the edge of the primary forest at around 2:28 a.m. The Changi airport and Tuas
industrial area also had higher temperatures, around 28.3 C to 28.5 C while some residential
areas like Bukit Batok and Toa payoh had lower temperature around 27.7 C. Normally the
location with higher temperature had lower relative humidity. The highest relative humidity of
90.9% was observed near the forest while the lowest relative humidity of 82.1% was obtained near
an industrial area.
 ARTICLE IN PRESS

N.H. Wong, C. Yu / Habitat International 29 (2005) 547–558 551

Fig. 4. The route of the first survey.

The maximum difference of ambient air temperature is 1.6 C, which clearly reflected the
correlation between variances of temperatures and land uses. The industrial areas and the airport
have less green areas compared with other land uses. Therefore, the heat absorbed by massive
hard surfaces during daytime was re-emitted at night. This is the reason why higher temperatures
were observed in these areas during the night survey. Lower temperatures were detectable in
residential areas with more greenery. It indicates that plants within developments can actually
cool the surroundings and generate lower ambient temperature. For large green areas, they can
cool the surroundings at a wider range. The lowest temperature caused by the forest, therefore,
was detected during the survey when the mobile passed through the edge of the forest. The relative
humidity was inversely distributed.

3.2. The second survey

To map out the island wide temperature distribution, the second survey was conducted. Data
were collected simultaneously through four survey routes which covered the whole island. The
temperature distribution is shown in Fig. 7. It can be found that the lower temperatures were
mostly detected at the northern part while higher temperatures were observed at the southern
part, especially at the CBD area. The results of the survey accorded with the partition of ‘urban’
and ‘rural’ areas in Singapore (see Fig. 1). Basically, the temperatures measured near to large
green areas are relatively lower than those measured far away from them.
 ARTICLE IN PRESS

552 N.H. Wong, C. Yu / Habitat International 29 (2005) 547–558

Fig. 5. The routes of the second survey.

The lowest temperate, 24.3 C, was detected in the northwestern open space/recreation area
where less architecture was built but vegetation was well planted. It is estimated that even lower
temperature could be obtained within the primary forest. The highest temperature, 28.31 C, was
detected in the high rise and high density CBD area where less vegetation was planted. Maximally
around 4 C temperature difference was detected in this second mobile survey. It is estimated that
even lager temperature difference could be reported if the mobile survey could cover the center
area of the forest. According to the mobile survey, the occurrence of the UHI effect in Singapore
has been confirmed although it is not as severe as some mega cities in the world.
Based on the data derived from the mobile survey, the sketch of UHI profile in Singapore was
plotted (see Fig. 8). It seems that the temperatures measured within different land uses are quite
relevant to the density of greenery. In CBD area, absence of plants and high density of buildings
caused the highest temperature. Higher temperatures were also detected in some industry areas.
The possible reason may be due to the employing of metal roofs for industrial buildings and lack
of tall trees which can cast efficient shadow on buildings. The temperatures measured within
residential areas relied on their locations. For those residential areas near to large greens normally
 ARTICLE IN PRESS

N.H. Wong, C. Yu / Habitat International 29 (2005) 547–558 553

Mobile survey conducted on 9th July 2002

91.0%
28.5
90.0%
28.3
89.0%

Relative humidity (%)

28.1
Temperature (˚C)

88.0%
27.9
87.0%
27.7 86.0%
27.5 85.0%
27.3 84.0%
27.1 83.0%

26.9 82.0%
2:05
2:06
2:08
2:09
2:11
2:12
2:13
2:15
2:16
2:18
2:19
2:21
2:22
2:24
2:25
2:26
2:28
2:29
2:31
2:32
2:34
2:35
2:36
2:38
2:39
2:41
2:42
2:44
2:45
2:47
2:48
2:49
2:51
2:52
2:54
2:55
Time

Temperature Humidity

I
F
R
A
R R
I

Fig. 6. The first mobile survey (I—industry area; R—residential area; F—forest; A—airport).

get lower temperatures than those near to the city center. The open space/recreation area and the
forest are well planted. Therefore, they get lowest air temperatures compared with other areas.

4. The relationship between green areas and ambient air temperatures

To further uncover the UHI effect island wide, the ambient temperatures were analyzed
according to different regions. The Singapore Island has been administratively divided into five
regions: west region, central region, north region, northeast region and east region (see Fig. 9).
 ARTICLE IN PRESS

554 N.H. Wong, C. Yu / Habitat International 29 (2005) 547–558

Fig. 7. Mapping of temperature distribution based on the second mobile survey.

According to Fig. 7, central region, west region and east region are mostly ‘urban’ areas where less
vegetation is planted but high density of buildings are constructed. North region and northeast
region are mostly ‘rural’ areas which are near to large green areas. The statistical data collected
within every region during the mobile survey are presented in Table 1. The mean air temperatures
of ‘urban’ areas (central region, west region and east region) all exceeded 27 C. The highest mean
temperature of 27.64 C was obtained in the central region. However, the mean air temperatures
of ‘rural’ area (north region and northeast region) were lower than 27 C. The lowest mean value
of temperature, 26.41 C, was observed in the north region.
In order to determine whether there is a significant difference between the mean values of
different regions, the Tukey–Kramer test was used. The results were presented in Table 2. Using
the 0.05 level of significance, the mean temperatures between West and Northeast regions, central
and east regions, North and Northeast regions, respectively, are not different while the rest are
significantly different.
Besides the above five regions, data collected from the northwest open/recreation area and
CBD area were statistically compared and a t-test for the difference in two means was conducted
(see Table 3). It indicates that the mean values of two areas are significantly different. The mean
temperature of well-planted area is 3.07 C lower than that of CBD area. The cooling effect of
plants on surrounding environment is remarkable. It is also necessary to mention that the
standard deviation of data collected in the CBD area is quite small (0.22) compared with other
 ARTICLE IN PRESS

N.H. Wong, C. Yu / Habitat International 29 (2005) 547–558 555

Fig. 8. Sketch of UHI profile in Singapore.

regions. The possible answer may be that the ‘canyon’ effect within high density CBD area and the
weak wind produced stable distribution of temperatures.

5. Conclusion

The study has clearly indicated the presence of UHI effect in Singapore. The maximum
difference of 4.01 C was observed between well planted area and the CBD area. In addition, the
mapping of temperature has shown a clear variation of temperature from southern ‘urban’ area to
northern ‘rural’ area. All these indicate the severity of UHI effect in Singapore.
The study also confirmed the cooling effect of green areas at macro level. From both the two
mobile surveys, the survey routes near to large green areas (the northwest open space and the
primary forest) experienced lower temperatures compared with other land uses like industrial
areas, residential areas, the CBD area, and the airport. Both lowest temperature and mean
temperature of 24.3 C and 25.01 C, respectively, were observed in a well planted area—Lim Chu
Kang. On the other hand, places with less plants always have higher temperatures. CBD region,
 ARTICLE IN PRESS

556 N.H. Wong, C. Yu / Habitat International 29 (2005) 547–558

Fig. 9. The zoning of Singapore.

Table 1
The statistical data of five regions
Sample No. Mean Max Min Standard deviation Standard error
West region 22 27.01 27.52 25.56 0.58 0.12
Central region 56 27.64 28.31 25.95 0.59 0.08
North region 17 26.38 26.73 25.56 0.39 0.10
Northeast region 12 26.84 27.52 26.34 0.42 0.12
East region 24 27.54 28.31 27.12 0.41 0.08

the high density and high rise commercial area, has the highest temperature and mean
temperature of 28.31 C and 28.08 C, respectively. It can be concluded that large green areas
definitely have positive effect on mitigating UHI effect in the city.
This methodology serves as a useful tool to determine the severity of UHI in a city and to
indicate the cooling effect of green areas. However, in order that the findings can serve as useful
guide for urban planners and landscape architects, it is essential that more quantifiable data be
established. Currently, work is in progress to measure the cooling effect and range of city greens
 ARTICLE IN PRESS

N.H. Wong, C. Yu / Habitat International 29 (2005) 547–558 557

Table 2
The results of Tukey–Kramer test
Comparison Absolute Std. error of Critical range Results
difference difference
West and Central regions 0.632857 0.09249415 0.357 Means are different
West and North regions 0.623824 0.11870494 0.4582 Means are different
West and Northeast regions 0.168333 0.13191974 0.5092 Means are not
different
West and East regions 0.53625 0.10850113 0.4188 Means are different
Central and North regions 1.256681 0.10179238 0.3929 Means are different
Central and Northeast regions 0.80119 0.11693436 0.4514 Means are different
Central and East regions 0.096607 0.08968456 0.3462 Means are not
different
North and Northeast regions 0.45549 0.13859776 0.535 Means are not
different
North and East regions 1.160074 0.11652904 0.4498 Means are different
Northeast and East regions 0.704583 0.12996527 0.5017 Means are different
Level of significance = 0.05.

Table 3
Statistical comparison of northwest open area and CBD area
Sample no. Mean Max Min Standard deviation Standard error
Open area/recreation 10 25.01 25.95 24.30 0.51 0.16
CBD area 12 28.08 28.31 27.70 0.22 0.06

Null hypothesis is that mean values of open space/recreation area and CBD area are equal

Two-tailed t-test
Lower critical value 2.08596
Upper critical value 2.085962
p-value 3.12E 14
At level of significance = 0.05 reject the null hypothesis.

such as the neighbourhood parks and rooftop greenery. A study is also conducted to correlate the
leaf area index (LAI) of greenery with their cooling and insulating properties. It is hoped that in
the near future, sufficient database can be developed, which can serve as the impetus for urban
planners and building designers to incorporate greenery in their project development.

Acknowledgements

This research is funded by Building and Construction Authority (BCA), Singapore and the
National University of Singapore. The authors would like to acknowledge the contributions from
 ARTICLE IN PRESS

558 N.H. Wong, C. Yu / Habitat International 29 (2005) 547–558

the collaborators which include BCA, Housing and Development Board (HDB), Centre for
Remote Imaging, Sensing and Processing (CRISP), Meteorological Services of Singapore (MSS)
and National Parks Board (Nparks). The authors would like to express their sincere thanks to Mr.
Henry Feriadi, Ms. Priyadarsini Rajagopalan M.t Ang, Mr. Joko and those undergraduate
students for their valuable advice and kindly assistance in the measurement.

References

Akbari, H., Rosenfeld, A., & Taha, H. (1990). Summer heat islands, urban trees, and white surfaces. Proceedings of
American society heating, refrigerating and air-conditioning engineers, Atlanta, GA.
Avissar, B. (1996). Potential effects of vegetation on the urban thermal environment. Atmospheric Environment, 30(3),
437–448.
Honjo, T., & Takakura, T. (1990/1991). Simulation of thermal effects of urban green areas on their surrounding areas.
Energy and Buildings, 15–16, 443–446.
Howard, L. (1833). The climate of London, Vols. I–III, London.
Jauregui, E. (1990/1991). Influence of a large urban park on temperature and convective precipitation in a tropical city.
Energy and Buildings, 15–16, 45–63.
Kawashima, S. (1990/1991). Effect of vegetation on surface temperature in urban and suburban areas in winter. Energy
and Buildings, 15–16, 465–469.
Landsberg, H. E. (1981). The urban climate. New York: Academic Press.
Oke, T. R. (1978). Boundary layer climates. London: William Clowes and Sons.
Saito, I. (1990/1991). Study of the effect of green areas on the thermal environment in an urban area. Energy and
Buildings, 15–16, 493–498.
Santamoouris, M. (Ed.) (2002). Energy and climate in the urban built environment. London: James and James Science
Publishers.
Sonne, J. K., & Vieira, R. K. (2000). Cool neighborhoods: The measurement of small scale heat island. Proceedings of
2000 summer study on energy efficiency in buildings, American Council for an Energy-Efficient Economy.
Tso, C. P. (1996). A survey of urban heat island studies in two tropical cities. Atmospheric Environment, 30(3), 507–519.