ARTICLE IN PRESS

Building and Environment 42 (2007) 29492970

www.elsevier.com/locate/buildenv

Environmental study of the impact of greenery in an institutional

campus in the tropics
N.H. Wong, Steve Kardinal Jusuf, Aung Aung La Win, Htun Kyaw Thu,
To Syatia Negara, Wu Xuchao
Department of Building, School of Design and Environment, National University of Singapore, 4 Architecture Drive, Singapore 117566, Singapore
Received 13 February 2006; received in revised form 25 April 2006; accepted 7 June 2006

Abstract

Urban environment quality is worsening every year. It is a fact that the urban air temperature is gradually rising in all cities and some
effective measures are needed to mitigate it. Planting of vegetation is one of the main strategies to mitigate the urban heat island (UHI)
effect. Large urban parks can extend positive effects to the surrounding built environment. National University of Singapore (NUS)
complex can be considered as a city on a smaller scale. The greenery along Kent Ridge Road seems like a rural area, with a cooler
ambient temperature. Some methodologies were employed in this study, such as satellite image, eld measurement and computer
simulations. The satellite image was used to identify the hot and cool spots in NUS environment. Field measurement was used to get
the real temperature distribution across the campus and nally, computer simulation was used to predict some scenarios of different
conditions. The result shows that buildings near or surrounded by greenery have lower ambient temperature than the ones away from the
greenery and it is an effective way to lower the ambient temperature. The TAS simulation results also show that a rooftop garden has the
potential of cooling energy savings for NUS buildings.
r 2006 Published by Elsevier Ltd.

Keywords: Effect of greenery; Campus area; Satellite images; Thermal satellite images; Field measurement; Computer simulation

1. Introduction Urban environment quality is worsening every year. It is

1.1. Background
Environmental quality is an abstract concept resulting
from both human and natural factors operating at different
spatial scales. In urban areas the local scale is dominated
by individual buildings, streets and trees, but regional scale
inuences may include the whole city and beyond. Urban
environmental quality is a complex and spatially variable
parameter which is a function of interrelated factors
including the urban heat island (UHI), the distribution of
greenery, building density and geometry and air quality [1].
Abbreviations: NUH, National University Hospital; NUS, National
University of Singapore; OED, Ofce of Estate and Development; PGP,
Prince George Park; SDE, School of Design and Environment
Corresponding author.
E-mail address: bdgwnh@nus.edu.sg (N.H. Wong).
a fact that the urban air temperature is gradually rising in
all cities and some effective measures are needed to mitigate
it. Several factors become the cause of it, such as the
diminishing of green area, low wind velocity due to high
building density and change of street surface coating
materials [2]. Dousset [3] stated that the main contributing
factors are changes in the characteristics of the surface
(albedo, thermal capacity, heat conductivity), replacement
of vegetation by asphalt and concrete, the decrease of
surface moisture available for evapo-transpiration. Mod-
ication of land cover in urban areas can cause the local air
and surface temperatures to rise several degrees higher than
the simultaneous temperature of the surrounding rural
areas. This may lead to overheating by human energy
release and absorption of solar radiation on dark surfaces
and buildings. This problem will be further aggravated by
increasing demand on air conditioning, which will again
lead to further heating and CO2 release.

0360-1323/$ - see front matter r 2006 Published by Elsevier Ltd.

doi:10.1016/j.buildenv.2006.06.004
 ARTICLE IN PRESS
2950 N.H. Wong et al. / Building and Environment 42 (2007) 29492970
Planting of vegetation is one of the main strategies to
mitigate the UHI effect. A single tree can already moderate
the climate well. But its impacts are limited only to the
microclimate [4]. Kawashima [5], based on satellite images,
studied the effects of vegetation density on the surface
temperatures in the urban and rural areas of Tokyo.
According to his observation, lower surface temperature
was generally observed in green areas while higher
temperatures were found on the soil and buildings during
the daytime. However, the effect of vegetation on reducing
surface temperature in the urban area is relatively smaller
than that in the suburbs. In the urban area, the surface
temperature ranged from 1.4 to 2.7 1C in the green area
while it ranged from 2.0 to 3.4 1C in buildings and 2.3 to
4.9 1C on the soil. In the countryside, the surface
temperature ranged from 2.6 to 2.8 1C in forests while it
ranged from 3.3 to 4.2 1C in buildings and 5.1 to 5.9 1C on
the soil. At night, the lower surface temperature was
observed in green areas in the urban environment while
higher surface temperature was found in forests in the
suburbs.
From the previous study on the Singapore UHI [6], the
satellite image shows strong evidence that the UHI effect
does occur in Singapore. The hot spots are normally
observed on exposed hard surfaces in the urban context.
Cool spots are mostly observed on the surface of
greenery and water catchments. In the analysis of long-
term climatic data of Singapore, four meteorological
stations were chosen with data coverage of at least 10
years. Yearly mean dry bulb temperatures are analyzed and
found to be rising signicantly in Changi airport. At the
other three stations, regression results show that the
yearly mean temperatures, have either not changed or
are warming at a much slower rate. The analysis of the
weather data provide the concrete evidence that the
temperature is increasing in the highly built-up region
while it remains unchanged in the well planted area.
Ca [7] did some eld measurements to determine the
cooling inuence of a park in the surrounding area of
the Tama New Town, a city in the west of Tokyo. The
observations indicated that vegetation can alter the climate
of the town. The surface temperatures measured on the
grass eld in the park is much lower than those measured
on the asphalt and on the concrete surfaces. Simulta-
neously, air temperature measured at 1.2 m above the grass
land was more than 2 1C lower than those measured above
hard surfaces in commercial and parking areas. With the
size of 0.6 km2, a park can reduce the air temperature by up
to 1.5 1C at noon time in a leeward commercial area at a
distance of 1 km.
Yu studied [8] that large urban parks can extend the
positive effects to the surrounding built environment.
Through the eld measurement, the built environment,
which is close to the park, has a lower temperature of
average 1.3 1C. Thus, the more the parks are built in an
urban area, the lower the urban temperature will be. The
temperatures measured within parks also have strong
relationship with the density of plants, since plants with
higher LAIs may cause lower ambient temperatures.
Results derived from the TAS simulation shows energy
may be saved when buildings are built near to parks with
maximum 10% reduction of the cooling load. The ENVI-
Met simulation indicates that parks have signicant
cooling effect on surroundings during both day and night.
However, in a country like Singapore which has limited
land area especially in downtown area, provisions of large
parks may not be possible. Another strategy that may be
used is the rooftop greenery. The green plants could protect
the hard roof surface from solar radiation, thus it would
not emit long wave radiation to the surrounding environ-
ment at night and reduce the effect of UHI.
Brad [9] explored the role of green roofs in mitigating the
UHI effect in Toronto. The mesoscale community com-
pressible (MC2) model was employed in the study. In the
simulation of the UHI in Toronto, 0.58 1C temperature
reduction was observed when 5% of the total area of
the city was replaced with green roofs. The impact of
green roofs in the high density areas is even more
pronounced. Temperature across the city was reduced
between 1 and 2.8 1C.
In the previous study on the rooftop garden [10], the
results of eld measurements conducted on the two rooftop
gardens reveal that the installation of rooftop gardens
would signicantly improve the thermal environment of
building roofs in Singapore. Two different types of rooftop
garden systems, intensive and extensive, are tested. The
results derived from the intensive system reveals that it can
signicantly improve the thermal environment around the
roofs and it can benet not only buildings but also
surroundings. It can reduce the rooftop ambient tempera-
ture up to 3 1C. The extensive green roof systems tend to
experience lower surface temperature up to 18 1C as
compared to the original exposed roof surface, especially
in areas well covered by vegetation.
1.2. Object of study
National University of Singapore (NUS) complex can
be considered as a city on a smaller scale, as shown in
Fig. 1. The greenery along Kent Ridge Road seems like the
rural area, with a cooler ambient temperature. It is
believed that the dense greenery makes the ambient
temperature in NUS cooler. It is observed that some
open spaces are left with only grass-covered soil. It is also
observed that many buildings in NUS are constructed with
at roof. There is a great potential to apply the
rooftop greenery in these areas. The advantages, rstly
is that, it will result in a lower heat gain to the building,
which leads to a lower cooling load. Secondly, if it is
combined with improvement of the open spaces, it will
result in a lower ambient temperature which indirectly
leads also to a lower cooling load of buildings and
psychologically, provides outdoor thermal comfort for
the people.
 ARTICLE IN PRESS
N.H. Wong et al. / Building and Environment 42 (2007) 29492970 2951

DENSE GREENERY AREA

Fig. 1. National University of Singapore (NUS) map.

1.3. Objectives
This research has the following objectives:
(a) to identify the hot spots in NUS environment,
(b) to study the importance of dense green area for the
microclimate of NUS environment and
(c) to study the energy saving of a building as the result of
lower ambient temperature due to improvement of the
green area and application of rooftop greenery.
2. Methodology
2.1. Satellite images and thermal satellite image
The interactions of urban surfaces with the atmosphere
are governed by surface heat uxes, the distribution of
which is drastically modied by urbanization. The physical
processes of these interactions are difcult to monitor
solely with in situ instruments. Satellite images are a
very useful tool to get an overall picture of an environment.
It has higher spatial distribution but low temporal
resolution and shorter data record [11]. NUS environment,
specically the building density and the distribution of
the greenery, is studied using these images. The satellite
image and thermal satellite image of Singapore was
superimposed and zoomed into the campus level, as shown
in Fig. 2. Thermal satellite image and NUS map
are analyzed to identify hot and cool spots in the designed to simulate the surfaceplantair interactions in
campus. Satellite images use different colors to represent
different temperatures, which range from red (hot) to blue
(cool).
2.2. Field measurement
The major instruments in this study were HOBO RH
and temperature sensors (operating range 20 to 70  C,
RH accuracy 5%), which were used together with some
ancillary instruments such as solar cover in conducting
measurements (as shown in Fig. 3). The HOBO sensors
were congured at an interval of every 10 min.
Based on the study of satellite images and a walk-
through, the whole campus was divided into three groups
with respect to their different greenery and building
distribution conditions. The rst group was the dense and
virtually uninterrupted jungle along the Kent Ridge Road.
The second group was the less dense greenery areas and the
third group, the areas with sparse greenery. These HOBO
meters were deployed in locations as shown in Fig. 4 and
Table 1.
The eld measurements were conducted on 10th24th
September 2005. The HOBO sensors contained in solar
covers were all installed on light posts at about 3 m above
the ground.
2.3. Computer simulations
2.3.1. ENVI-Met simulations
ENVI-Met is a three-dimensional microclimate model
urban environment [12]. A base model was constructed
 ARTICLE IN PRESS
2952 N.H. Wong et al. / Building and Environment 42 (2007) 29492970

Fig. 2. Singapore thermal image superimposed with satellite image.

Fig. 3. HOBO sensor and solar cover.

based on the electronic map and building information
provided by Ofce of Estate and Development (OED).
Parametric variations, as shown in Fig. 5, were also made
to obtain predictions in different conditions. Three
scenarios were designed; besides the current condition:
replacing the dense greenery along Kent Ridge Road with
buildings, removing all the greenery from NUS environ-
ment and converting the grasslands into vertically denser
plantation. Then, the variations of ambient temperature in
these scenarios from the base case were recorded and
analyzed.
Basic settings employed in this simulation were as
follows:
1. Temperature: 303 K.
2. Wind speed (at 10 m above ground): 1.6 m/s.
3. Wind direction: south to north.
4. RH: 84%.
5. Roughness length in 10 m: 0.1.
6. Total simulation: 24 h.
2.3.2. TAS simulations
Engineering building (EA) was chosen as the building
model in the TAS simulations. It is seven storey high and
about 2000 m2 footprint area, as shown in Fig. 6. This
simulation was carried out to compare the cooling energy
consumption for two models. First model is the cooling
load due to the difference of ambient temperature
condition in the different locations and the second model
is the cooling load due to application of different types of
rooftop greenery. In each model, there are two scenarios,
without internal load and with internal load.
Basic setting for both scenarios:
1. Air conditioning was on 08.00 am22.00 pm (extended
ofce hour).
 ARTICLE IN PRESS
N.H. Wong et al. / Building and Environment 42 (2007) 29492970 2953

Fig. 4. Points of measurement location.

Table 1
Measurement location grouping and numbering

Group Location numbera Landmark nearby

Dense greenery (Kent Ridge Road) 1 Water Tank

2 Acoustical Lab
3 Bioinformatics Center
Less dense greenery 4 Faculty of Medicine car park
5 Temasek Hall
6 Computer center
7 Sports eld
Sparse greenery 8 PGP Road
9 Engineering Auditorium
10 PGP Canteen
a
The initial numbering was rearranged to facilitate better presentation.

2. Temperature and RH were input using the eld
measurement result on 15th September 2005.
3. Thermostat setting:
(a) Temperature upper limit: 24 1C and lower limit: 21 1C.
(b) Humidity upper limit: 70% and lower limit: 60%.
In the rst scenario, the internal heat load was omitted
to get the energy saving with only considering the ambient
temperature heat load and different roof heat loads. In the
second scenario, some general assumptions were made in
terms of internal load of the building, as follows:
1. Lighting gain 15 W/m2.
2. Occupants sensible heat and latent heat 15 W/m2.
3. Equipment sensible gain 20 W/m2.
3. Findings and discussions
3.1. Satellite images and thermal satellite image
In Fig. 7, the rst image is the satellite image of
Singapore which is zoomed into the NUS campus level.
The second one is the combined image of satellite image
and the thermal satellite image of NUS. The thermal
image overlaps the NUS image as a transparent image to
show the hints of thermal distribution according to the
building density and distribution of green area. The third
one is the thermal satellite image of NUS environment
in which the relative thermal distribution is presented by
different colors. The available relative temperature image
was derived from Landsat 7 ETM thermal band (28 April
 ARTICLE IN PRESS
2954 N.H. Wong et al. / Building and Environment 42 (2007) 29492970

Fig. 5. Four scenarios in ENVI-Met simulation: (A) current condition; (B) replacing dense trees with buildings; (C) removing all greenery; and (D) adding
more trees.

Fig. 6. Engineering building model.

2000, 11:09 am). Since there is not much signicant Master
Plan development in NUS campus from year 2000 until
this study was conducted, it is still reasonable to make use
of this image.
In the three images, it can obviously be seen that the
locations of red color distribution in thermal image are
almost the same as the locations of buildings in real image.
It means that these buildings and their environments create
the heat island effect as hot spots in thermal image. On the
contrary, it can also be seen that the locations of green
color distribution in thermal image are mainly at the large
proportion of dense greenery area in real image and create
cool spots in thermal image. Further away from the dense
area of building, the reddish color of hot spots began to
change to the greenish color of cool spots.
Fig. 8 shows the surface temperature distribution in
and around the University Cultural Centre and OED
building. Mostly, it is seen that, reddish color of thermal
 ARTICLE IN PRESS
N.H. Wong et al. / Building and Environment 42 (2007) 29492970 2955

Fig. 7. Satellite image and thermal satellite image of NUS.

distribution are on and around the buildings. Car parks Fig. 9(A) shows the School of Design and Environment
with sparse greenery area also create hot spots in thermal (SDE) and Faculty of Engineering environment, and
distribution. Fig. 9(B) shows the total environment of National
 ARTICLE IN PRESS
2956 N.H. Wong et al. / Building and Environment 42 (2007) 29492970

Fig. 8. University cultural center and OED ofce.

ENGINEERING FACULTY
OF
SCIENCE &
SDE MEDICINE
NUH

(A) (B)

Fig. 9. SDE and Faculty of Engineering (A), NUH, Faculty of Science and Faculty of Medicine (B).

University Hospital (NUH), Faculty of Science and
Faculty of Medicine which have a high density of
buildings. It can be observed that most buildings were
designed with a combination of pitched roof and at roof
and there is a tremendous thermal distribution in reddish
color. This is mainly because of the extensive use of
concrete and other heat absorbing surfaces, by decreasing
the surface moisture available for evapo-transpiration.
Furthermore, more solar radiation is absorbed and
reradiated into heat because dry surfaces have higher
absorptivity. So, the latent heat ux is very small compared
to the sensible heat in these areas.
Around Prince Georges Park (PGP) environment, the
yellowish color of surface temperature distribution is found
on the clusters of PGP residence which are besides the
dense area of plantation. It is believed that the evapo-
transpiration from plants and trees can reduce the ambient
temperature of environment nearby. However, in the other
part of PGP residence and King Edward VII Hall area,
they appear as reddish color in surface temperature
especially those areas far away from plantation area, as
shown in Fig. 10(A).
According to the observation on NUS campus, a large
green area is found along Kent Ridge Road and PGP Road
(Fig. 10(B)). Due to the shading provided by trees and the
evapo-transpiration process of the trees, the deep green
color distribution in surface temperature can be seen
obviously at the central part of dense green area. As
mentioned earlier on, it is believed that this environment
condition has impacts on the other zones. The buildings
surrounded by or at the perimeter of the green area
have better thermal distribution than other buildings away
from it.
As studied in the above images, approximately 40% of
the campus area is covered by building rooftops which are
mainly concrete at roofs. Therefore, there is a great
potential to develop rooftop gardens on NUS buildings to
achieve better ambient temperature for the whole campus
 ARTICLE IN PRESS
N.H. Wong et al. / Building and Environment 42 (2007) 29492970 2957

Fig. 10. PGP residence and King Edward VII Hall (A), Kent Ridge Road area (B).

Comparison of air temperature on a typical day (15th Sep.2005)

35
34 Water Tank
33 Acoustic Lab
Temperature (Degree C)

32 Bioinformatics Centre
31 Medicine Carpark
30 ddd Temasek Hall
29 Computer Centre
28 Sports Field
27 PGP Road
26 Engineering Auditorium
25 PGP Canteen
24
1: 0
00

3: 0
00

5: 0
6: 0
7: 0
00

9: 0
10 00
11 00
12 0
13 00
14 0
15 00
16 00
17 00
18 00
19 00
20 0
21 00
22 00
23 00
0
0

0
0
0

:0

:0

:0

:0
0:

2:

4:

8:

:
:
:
:
:

:
:
:

Fig. 11. Comparison of air temperature on a typical day (15th September 2005).

and also to save the cooling energy consumption for the
buildings.
3.2. Field measurement
A typical day on 15th September was randomly chosen
from others to give a snapshot of the temperature changes
throughout the whole day. It also shows the temperature
difference among different locations (see Fig. 11). The lines
representing sparse green areas are clustered on the top,
with maximum temperatures reaching 33 1C or even higher.
On the contrary, the cool spots lines are mostly at the
bottom, much nearer to the X-axis. As can be seen in the
graph, the peak temperature difference between location 1
(Kent Ridge Road-Water Tank) and location 10 (inside
PGP residence) can be as high as 4 1C at around 13:00.
When the time approaches midnight, as is shown near
the right edge of the graph, the temperature difference
between these two locations is about 3 1C. This is almost
the same time condition when temperature difference
induced by UHI effect can be quantied. The difference
of 3 1C within a community-scale environment is believed
to be large.
Considering the dominant role solar radiation plays
in air temperature and the possible UHI effect, daytime
and night-time data were analyzed separately. In this
study, daytime is dened as from 7 am to 7 pm, and the
balance is night-time. In Singapore, the daytime dened
here is approximately coincident with the solar radiation
availability on sunny days. The averages, minimums, and
 ARTICLE IN PRESS
2958 N.H. Wong et al. / Building and Environment 42 (2007) 29492970

Comparison of daytime temperature

35
34
33

Temperature (Degree C)
32
31
30
29
28
27
26
25
24
23
22
21
20
nk

tre

rk

tre

ld

ad

n
al
La

ee
ie

riu
pa

Ro
Ta

H
en

en

nt
tsF
ar
tic

ito
ek
sC

rC

Ca
er

P
eC

or
us

as

ud
at

te
ic

P
Sp
W

.P
co

pu
m
at

in

gA

G
8
m

Te
1.

7.
A

ic

.P
in
or
2.

ed

Co
5.

10
er
nf

6.

ne
oi

4.

gi
Bi

En
3.

9.
Locations
Minimum Average Maximum

Fig. 12. Comparison of daytime temperature.

Comparison of nighttime temperature

31
30
29
Temperature (Degree C)

28
27
26
25
24
23
22
21
20
nk b e
ark ll e eld d n
Ta La e ntr rp Ha e ntr Fi R oa riu
m
tee
r sti
c
sC Ca k rC
o an
ate u se te rts GP dit PC
co tic ne em
a
pu po 8.P
u
1.W 2.A ma ici 7.S gA .PG
fo
r ed 5.T Co
m
eri
n 10
ioin 4.M 6. ine
3.B ng
Locations 9.E
Maximum Average Minimum

Fig. 13. Comparison of night-time temperature.

maximums of air temperature during the whole measure- special importance, since it determines the sizing of air-
ment period are plotted in Figs. 12 and 13. conditioning systems. It is also worth taking note of the
As the average temperature is concerned, the pattern is night-time minimum temperature in Fig. 13. The line is
very obvious. There are uctuations within some groups, generally very even, but goes up to nearly 23 1C at location
especially in Fig. 13, but the dominant trend, from low to 10 (inside PGP residence). This veries the hypothesis that
high temperature, is unchanged. The daytime average the heat accumulated in the day is hard to dissipate during
temperature ranges from 27.4 to 29.6 1C, and it ranges night-time due to large concentration of buildings and
from 25.6 to 27.4 1C during night-time. There is about 2 1C sparse plantation in PGP.
difference. Location 3 (Kent Ridge Road-Bioinformatics) is an eye-
In Fig. 13, it can be perceived that temperature catching lump on the line for maximum daytime tempera-
difference of daytime maximum among different locations ture (see Fig. 12). It also deviates a bit from the normal
is very signicant. It has a peak of 3.3 1C, which is the trend on other lines, but is not so evident. Further
difference between location 1 (Kent Ridge Road-Water investigation was done after preliminary data analysis in
Tank) and location 10 (inside PGP residence). In the an effort to nd out whether it is an outlier. It was initially
tropics like Singapore, the maximum temperature is of chosen to represent dense green areas due to its location at
 ARTICLE IN PRESS
N.H. Wong et al. / Building and Environment 42 (2007) 29492970 2959

Fig. 14. Predicted daytime ambient temperaturesite plan: (A) current condition; (B) without greenery; (C) replacing forest with buildings; and
(D) introducing denser greenery.

Fig. 15. Predicted daytime ambient temperaturesection: (A) current condition; (B) without greenery; (C) replacing forest with buildings; and
(D) introducing denser greenery.

the east end of Kent Ridge Road. There is a large likely to have substantial inuence on its vicinity. More-
construction site in close vicinity. The anthropogenic heat over, the whole construction site is bare soil without any
from the workers, trucks, and construction machinery is green coverage. These two factors may well account for the
 ARTICLE IN PRESS
2960 N.H. Wong et al. / Building and Environment 42 (2007) 29492970

Fig. 16. Predicted night-time ambient temperaturesite plan: (A) current condition; (B) without greenery; (C) replacing forest with buildings; and
(D) introducing denser greenery.

Fig. 17. Predicted night-time ambient temperaturesection: (A) current condition; (B) without greenery; (C) replacing forest with buildings; and
(D) introducing denser greenery.

abnormal temperature of location 3 (Kent Ridge Road- throughout the NUS campus was quantied. Through
Bioinformatics). elaborate selection of locations according to their green
On a whole, the eld data measurement has achieved the conditions, the relationship between air temperature and
targeted objectives. The presumed temperature difference greenery was veried.
Table 2
Hourly average ambient temperature on 15th September 2005 (dense green area)

Time Dense green area

Water Tank Acoustical Lab Bioinformatics Average

Temp. Std. dev. RH Std. dev. Temp. Std. dev. RH Std. dev. Temp. Std. dev. RH Std. dev. Temp. RH

N.H. Wong et al. / Building and Environment 42 (2007) 29492970

09/15/05 00:00:00.0 27.12 0.00 76.10 1.09 27.32 0.22 77.00 1.49 27.59 0.16 74.75 1.10 20.51 75.95
09/15/05 01:00:00.0 26.86 0.20 74.33 4.90 27.12 0.00 74.02 3.71 27.39 0.21 73.15 4.34 20.34 73.83
09/15/05 02:00:00.0 26.93 0.21 74.67 3.99 26.99 0.20 74.77 3.82 27.32 0.22 74.45 3.31 20.31 74.63
09/15/05 03:00:00.0 26.47 0.20 80.90 0.77 26.73 0.00 80.65 0.61 26.73 0.00 80.93 1.16 19.98 80.83
09/15/05 04:00:00.0 26.34 0.00 84.47 1.89 26.73 0.00 83.00 1.37 26.73 0.00 83.53 1.08 19.95 83.67
09/15/05 05:00:00.0 26.34 0.00 88.90 1.24 26.34 0.00 88.17 1.83 26.73 0.00 86.97 0.82 19.85 88.01

ARTICLE IN PRESS
09/15/05 06:00:00.0 26.15 0.21 90.63 1.45 26.34 0.00 89.70 0.00 26.41 0.16 89.30 0.98 19.72 89.88
09/15/05 07:00:00.0 26.21 0.20 92.50 0.00 26.54 0.21 89.70 0.00 26.47 0.20 89.70 0.00 19.80 90.63
09/15/05 08:00:00.0 27.06 0.39 86.53 3.01 27.39 0.21 85.10 1.91 27.65 0.41 83.97 2.55 20.52 85.20
09/15/05 09:00:00.0 28.64 0.77 74.92 4.53 28.18 0.32 77.48 2.68 28.90 0.33 74.07 3.61 21.43 75.49
09/15/05 10:00:00.0 28.90 0.22 66.30 2.07 29.03 0.39 66.12 3.43 29.97 0.17 64.92 2.68 21.98 65.78
09/15/05 11:00:00.0 29.43 0.30 56.33 2.13 29.97 0.31 54.18 2.34 30.85 0.33 54.42 4.67 22.56 54.98
09/15/05 12:00:00.0 29.90 0.26 63.02 2.46 30.44 0.21 59.83 1.44 31.59 0.40 58.72 1.99 22.98 60.52
09/15/05 13:00:00.0 30.31 0.26 64.12 1.75 30.71 0.26 60.62 1.48 32.21 0.34 58.65 0.99 23.31 61.13
09/15/05 14:00:00.0 30.24 0.65 66.07 2.16 30.71 0.36 62.02 1.83 32.41 0.31 57.77 1.55 23.34 61.95
09/15/05 15:00:00.0 29.57 0.30 69.12 0.96 30.38 0.30 65.37 1.25 32.28 0.61 59.62 1.70 23.05 64.70
09/15/05 16:00:00.0 28.64 0.30 69.28 1.62 29.30 0.22 65.73 1.72 29.64 0.42 64.62 0.45 21.89 66.54
09/15/05 17:00:00.0 28.18 0.41 71.23 2.22 28.97 0.21 67.13 1.58 29.23 0.33 68.03 1.09 21.59 68.80
09/15/05 18:00:00.0 27.91 0.00 71.60 2.02 28.18 0.21 69.20 1.07 28.51 0.21 68.78 1.66 21.15 69.86
09/15/05 19:00:00.0 27.71 0.33 73.30 1.70 28.24 0.16 71.62 1.13 28.31 0.00 71.00 1.38 21.07 71.97
09/15/05 20:00:00.0 26.41 0.16 79.83 1.73 27.26 0.32 75.63 1.69 27.52 0.25 74.90 1.10 20.29 76.79
09/15/05 21:00:00.0 26.34 0.00 80.40 0.00 26.93 0.21 76.80 0.00 27.19 0.16 76.25 0.60 20.11 77.82
09/15/05 22:00:00.0 26.15 0.21 80.02 0.71 26.80 0.18 77.35 0.60 27.19 0.16 75.55 0.37 20.03 77.64
09/15/05 23:00:00.0 25.82 0.20 80.97 0.72 26.47 0.20 78.30 0.62 26.99 0.20 75.73 0.05 19.82 78.33

2961
 2962
Table 3
Hourly average ambient temperature on 15th September 2005 (less dense green area)

Time Less dense green area

Medical school car park Temasek Hall car park Computer centre Sports eld Average

Temp. Std. dev. RH Std. dev. Temp. Std. dev. RH Std. dev. Temp. Std. dev. RH Std. dev. Temp. Std. dev. RH Std. dev. Temp. RH

N.H. Wong et al. / Building and Environment 42 (2007) 29492970

09/15/05 00:00:00.0 27.52 0.00 71.43 0.84 27.85 0.16 74.42 0.77 27.59 0.16 73.98 1.06 27.59 0.16 74.45 1.11 27.63 73.57
09/15/05 01:00:00.0 27.39 0.21 69.85 3.58 27.91 0.00 69.90 4.50 27.52 0.00 71.30 5.26 27.65 0.20 72.90 3.39 27.62 70.99
09/15/05 02:00:00.0 26.99 0.20 70.62 3.88 27.72 0.21 70.25 4.24 27.39 0.21 71.18 4.50 27.06 0.30 73.90 4.00 27.29 71.49
09/15/05 03:00:00.0 26.15 0.21 80.52 1.71 27.25 0.21 78.95 1.41 26.93 0.21 78.10 0.49 26.15 0.21 83.95 2.52 26.62 80.38
09/15/05 04:00:00.0 26.34 0.00 83.27 1.27 27.12 0.00 82.23 1.89 26.99 0.20 80.73 1.71 26.28 0.16 87.70 0.98 26.68 83.48

ARTICLE IN PRESS
09/15/05 05:00:00.0 26.34 0.00 85.97 1.03 26.99 0.20 86.32 1.11 26.73 0.00 86.00 1.57 26.02 0.16 91.10 1.53 26.52 87.35
09/15/05 06:00:00.0 25.89 0.29 88.87 1.21 26.54 0.21 89.77 1.65 26.47 0.20 87.30 0.00 25.95 0.00 94.77 1.76 26.21 90.18
09/15/05 07:00:00.0 26.15 0.59 88.53 1.84 26.60 0.32 91.58 1.42 26.60 0.32 87.32 0.04 26.41 0.46 94.78 1.73 26.44 90.55
09/15/05 08:00:00.0 28.11 0.33 75.75 2.90 28.24 0.30 81.37 4.28 27.91 0.43 79.78 4.19 28.44 0.32 78.60 4.77 28.18 78.88
09/15/05 09:00:00.0 29.30 0.42 66.80 3.48 29.37 0.21 72.07 1.66 28.90 0.42 71.35 1.74 29.77 0.49 67.25 4.23 29.33 69.37
09/15/05 10:00:00.0 30.31 0.36 57.00 1.25 30.11 0.42 61.53 4.57 29.77 0.33 61.75 3.43 30.58 0.33 57.97 1.80 30.19 59.56
09/15/05 11:00:00.0 31.19 0.30 45.35 4.55 30.92 0.34 50.63 1.17 30.78 0.31 51.25 2.05 31.59 0.40 48.47 3.33 31.12 48.93
09/15/05 12:00:00.0 31.66 0.21 50.65 2.19 31.25 0.21 55.17 2.70 31.52 0.00 55.13 1.76 31.39 0.21 55.52 1.77 31.45 54.12
09/15/05 13:00:00.0 31.52 0.26 53.43 1.52 31.32 0.34 58.12 0.75 31.86 0.48 56.42 1.26 31.59 0.31 56.65 1.50 31.57 56.15
09/15/05 14:00:00.0 31.32 0.34 54.97 1.93 30.85 0.33 61.97 2.06 31.87 0.55 56.80 2.13 31.52 0.26 57.33 1.72 31.39 57.77
09/15/05 15:00:00.0 30.92 0.34 59.32 1.08 30.51 0.22 64.17 0.48 31.25 0.33 60.17 1.82 31.12 0.45 61.55 1.79 30.95 61.30
09/15/05 16:00:00.0 29.63 0.21 60.45 1.99 29.77 0.21 62.00 2.48 29.50 0.25 63.07 0.94 29.70 0.22 63.20 2.46 29.65 62.18
09/15/05 17:00:00.0 29.10 0.50 63.32 2.79 29.57 0.16 64.27 1.20 29.30 0.33 65.32 1.85 29.04 0.68 66.17 3.31 29.25 64.77
09/15/05 18:00:00.0 28.57 0.20 64.97 1.94 28.83 0.21 65.27 0.43 28.57 0.32 66.67 0.80 28.70 0.00 66.97 2.10 28.67 65.97
09/15/05 19:00:00.0 28.24 0.16 66.83 1.28 28.64 0.16 68.52 1.13 28.31 0.00 69.82 1.22 28.31 0.43 69.13 1.93 28.37 68.58
09/15/05 20:00:00.0 27.12 0.25 72.65 1.10 27.65 0.20 73.85 1.40 27.59 0.16 73.85 1.28 27.32 0.22 75.53 0.41 27.42 73.97
09/15/05 21:00:00.0 26.86 0.20 73.90 0.80 27.32 0.22 74.02 0.66 27.19 0.16 74.48 0.45 26.99 0.20 76.98 0.45 27.09 74.85
09/15/05 22:00:00.0 26.86 0.20 72.53 0.44 27.52 0.00 73.13 0.33 27.12 0.00 74.20 0.46 27.12 0.00 74.80 0.57 27.16 73.67
09/15/05 23:00:00.0 26.73 0.00 72.97 0.33 27.12 0.00 73.60 0.46 26.80 0.16 74.65 0.37 26.86 0.20 75.47 0.52 26.88 74.17
Table 4
Hourly average ambient temperature on 15th September 2005 (sparse green area)

Time Sparse green area

PGP Road Eng. Auditorium PGP inside Average

Temp. Std. dev. RH Std. dev. Temp. Std. dev. RH Std. dev. Temp. Std. dev. RH Std. dev. Temp. RH

N.H. Wong et al. / Building and Environment 42 (2007) 29492970

09/15/05 00:00:00.0 27.91 0.00 74.73 1.20 28.18 0.21 69.92 0.92 28.64 0.16 68.82 0.84 28.24 71.16
09/15/05 01:00:00.0 27.98 0.16 71.32 4.83 28.31 0.00 65.97 4.11 28.38 0.16 66.15 4.35 28.22 67.81
09/15/05 02:00:00.0 27.85 0.30 73.45 4.12 28.05 0.32 65.55 3.58 28.51 0.21 66.83 3.47 28.13 68.61
09/15/05 03:00:00.0 27.19 0.16 81.18 1.18 27.19 0.16 74.67 1.29 28.31 0.35 72.42 1.32 27.56 76.09
09/15/05 04:00:00.0 27.12 0.00 83.83 1.29 27.06 0.16 78.12 0.87 28.18 0.21 75.58 1.51 27.45 79.18
09/15/05 05:00:00.0 26.93 0.21 87.40 1.39 26.73 0.00 82.75 1.71 27.98 0.30 79.57 1.63 27.21 83.24

ARTICLE IN PRESS
09/15/05 06:00:00.0 26.54 0.21 90.63 1.45 26.54 0.21 86.30 1.10 27.85 0.30 81.40 0.77 26.97 86.11
09/15/05 07:00:00.0 26.93 0.33 90.20 1.13 27.19 0.63 84.18 1.99 28.05 0.32 80.93 1.16 27.39 85.11
09/15/05 08:00:00.0 28.57 0.48 79.45 3.78 29.30 0.55 71.47 3.74 29.44 0.64 72.03 3.35 29.10 74.32
09/15/05 09:00:00.0 29.50 0.44 72.22 1.93 30.92 0.56 58.67 3.91 30.58 0.55 65.08 2.30 30.33 65.32
09/15/05 10:00:00.0 30.85 0.21 62.02 2.49 32.07 0.50 48.18 2.70 31.59 0.31 56.02 1.93 31.50 55.41
09/15/05 11:00:00.0 32.07 0.33 49.57 3.61 32.90 0.34 40.75 2.70 32.90 0.21 44.12 2.69 32.62 44.81
09/15/05 12:00:00.0 32.62 0.22 53.75 2.67 33.31 0.22 45.85 1.30 33.52 0.32 47.48 2.59 33.15 49.03
09/15/05 13:00:00.0 32.69 0.17 56.72 1.01 33.17 0.37 49.72 1.90 32.97 0.22 52.65 1.41 32.94 53.03
09/15/05 14:00:00.0 32.27 0.31 58.02 1.49 32.21 0.56 54.77 2.58 32.41 0.31 55.15 1.72 32.30 55.98
09/15/05 15:00:00.0 31.80 0.56 61.03 1.91 31.66 0.49 57.62 1.99 31.32 0.61 59.88 2.04 31.59 59.51
09/15/05 16:00:00.0 30.04 0.21 63.83 1.46 30.51 0.34 56.83 1.17 30.31 0.26 60.12 3.14 30.29 60.26
09/15/05 17:00:00.0 29.77 0.42 65.93 1.78 29.77 0.61 59.67 2.93 30.04 0.33 61.93 1.88 29.86 62.51
09/15/05 18:00:00.0 28.90 0.22 68.58 1.07 29.30 0.22 61.85 2.10 29.30 0.33 63.67 0.65 29.17 64.70
09/15/05 19:00:00.0 28.83 0.21 70.00 1.05 28.90 0.55 63.90 2.50 29.17 0.47 65.43 1.91 28.97 66.44
09/15/05 20:00:00.0 27.85 0.16 75.07 1.13 27.52 0.25 71.20 1.09 28.57 0.20 68.02 0.97 27.98 71.43
09/15/05 21:00:00.0 27.72 0.21 75.72 0.66 27.59 0.16 70.70 0.80 28.84 0.48 67.25 2.00 28.05 71.22
09/15/05 22:00:00.0 27.65 0.20 74.87 0.41 27.52 0.00 70.82 0.53 29.03 0.30 65.07 0.91 28.07 70.25
09/15/05 23:00:00.0 27.45 0.16 74.87 0.41 27.12 0.00 70.90 0.69 28.37 0.30 67.18 1.08 27.65 70.98

2963
 ARTICLE IN PRESS
2964 N.H. Wong et al. / Building and Environment 42 (2007) 29492970

COMPARISON OF COOLING LOAD ON TYPICAL DAY AMBIENT TEMPERATURE WITHOUT

INTERNAL LOAD- CURRENT CONDITION
7,000.00

6,500.00
COOLING LOAD (KWh)

6,000.00

5,500.00

5,000.00

4,500.00
B

D
K

N
LA

IC

TR
N

EL

E
IN
PA

PA

TE
TA

AT

R
FI
EN
AL

PG

EE

AN
ER

AR

AR

S
IC

C
R

RT

IN

C
AT

C
FO
ST

P
O
TE
W

EN
U

PG
IN

SP
AL
O

PU
O
O
AC

M
H
BI

M
SC

IU
EK

R
C
AS
AL

O
IT
M
IC

D
TE
ED

AU
M

Increase of Energy
Usage (Compared
Greenery Condition Places CoolingLoad (KWh)
with Kent Ridge Road -
Water Tank)
PGP Inside 6,486.62 14.30%
SPARSE GREENERY Auditorium Engineering 6,298.47 10.98%
PGP Road 6 , 229. 32 9. 76%
SPORTS FIELD 5 , 981. 50 5. 40%
COMPUTER CENTRE 5,993.55 5.61%
LESS DENSE GREENERY
TEMASEK HALL CAR PARK 6,038.81 6.41%
MEDICAL SCHOOL CAR PARK 5,899.49 3.95%
BIO INFORMATICS 6 , 034. 02 6. 32%
DENSE GREENERY ACOUSTICAL LAB 5 , 808. 25 2. 34%
WATER TANK 5 , 675. 24 0%

Fig. 18. Cooling load in 10 points of different locations with current conditionwithout internal load.

3.3. Computer simulation much hotter in general as indicated by yellow and red color
representation. It is noted that ENVI-Met simulation
3.3.1. ENVI-Met simulation model assumes that the source of water in the soil is non-
3.3.1.1. Daytime. Fig. 14 shows temperature proles depleting. In reality, this is not the case, the water will
throughout NUS environment for four different conditions become dry at some time and hence the temperature in
at 13.00 hours. In condition (A), it can be observed that the condition (B) would be even higher. From Fig. 15(B) it can
presence of dense green area in the central region and be seen that there is no cooling effect in NUS environment
moderate green area around the campus clearly contribute after removal of greenery. In condition (C) most of the
to NUS low ambient temperature. This can be illustrated dense green areas in the central region have been replaced
by cool areas indicated by blue and green color in the with buildings where the rest of the greenery in NUS is kept
central region. However, areas near Faculty of Engineer- unaltered. It is observed that the central region of NUS has
ing, University Cultural Centre and University Sport experienced an increase in temperature and cooling effect
Centre have been generally high in temperature. This is of central green areas to NUS environment has consider-
due to lack of greenery, higher building density and ably decreased. Generally, in comparison with current
particularly presence of pavement at the sport eld. In condition, the areas have now become hotter as indicated
contrast, it is clearly observed that the presence of cooler by more yellow, orange and red color. Fig. 15(C) also
areas has disappeared due to removal of all green areas in shows that the cooling effect generated by dense greenery
NUS as shown in condition (B). The areas have become in central region has been reduced. The increase of
 ARTICLE IN PRESS
N.H. Wong et al. / Building and Environment 42 (2007) 29492970 2965

Fig. 19. Cooling load in 10 points of different locations with current conditionwith internal load.

temperature especially in the central region is due to high
building density and reduction of plants which eventually
contribute to the reduction of cooling effects of the
greenery to the surrounding areas.
In contrast, by adding denser greenery to NUS environ-
ment, it is clearly observed that NUS environment has now
become much cooler than current condition as represented
in Fig. 14(D) and (A). The hot spots which initially
occur near Faculty of Engineering, University Cultural
Center and Sport center have now become much cooler as
a result of cooling effects of much denser greenery in the
central region of NUS and other areas.
3.3.1.2. Night-time. Temperature proles and vertical
temperature distributions at 00.00 hours for four different
conditions in NUS environment are shown in Figs. 16
and 17, respectively. By comparing all the four conditions,
it can be observed that the presence of greenery is very
important in keeping low ambient temperature in NUS
environment. Without greenery, it is clearly seen in
condition (B) that during night-time areas with high
building density, such as: Faculty of Art and Social
Sciences, Faculty of Engineering, Faculty of Science and
PGP residence are much hotter than the surrounding areas.
This is due to the fact that the heat stored in the buildings
during daytime start to radiate back to the environment,
thus making the surrounding areas much hotter at night-
time. However, with the presence of greenery, the cooling
effects produced by plants are able to neutralize this heat
and even keep the areas cooler as shown in condition (A).
Furthermore, the comparison also shows that the
cooling effects of the greenery areas in NUS environment
are affected by building density as indicated by condition
(C). The higher the buildingplot ratio, the less pro-
nounced the cooling effects will be. This observation is
clearly seen during night-time, where the heat stored in the
concrete mass in buildings start to be released to the
environment. It is noticed that cooling effects still exist in
 ARTICLE IN PRESS
2966 N.H. Wong et al. / Building and Environment 42 (2007) 29492970

COMPARISON OF COOLING LOAD ON 7th FLOOR ZONE WITHOUT INTERNAL LOAD -

3 DIFFERENT ROOFTOP GREENERY WITH CURRENT CONDITION

2,100.00

1,900.00
COOLING LOAD (KWh)

1,700.00

1,500.00

1,300.00

1,100.00

900.00

700.00

G
K

E
S
B

EN
K

D
R

IN
D
R

TR
IC
LA
N

EL
PA

R
PA

TE
TA

AT

EN

EE
FI

P
AL

AR

AN
PG
AR
M
ER

TS
IC

IN
R

C
R
C
ST

G
FO
AT

R
TE
L

P
EN
L

O
O
U
W

AL

PG
IN

SP
PU
O
O

M
O

H
H
AC

M
BI

IU
SC

EK

R
C
AS

O
AL

IT
M
IC

D
TE
ED

AU
M

Current Condition Applying 100% Turfing Applying 100% Shrubs Applying 100% Trees

Applying 100% Applying 100% Trees

Applying 100% Turfing
Shrubs

Energy Savings of
Cooling Load Compared 25.82% 53.67% 56.78%
with Current Condition

Fig. 20. Comparison of seventh oor zone cooling load in 10 points of different locations with different rooftop applicationswithout internal heat load.

the central region of NUS due to dense greenery; however,
the effect is not so prevalent to NUS environment as
compared to the heat radiated by buildings.
In contrast to condition (C), it is observed in condition
(D) that by adding much denser greenery to NUS
environment, the cooling effects generated are much more
prevalent than the heat by buildings. Whole NUS
environment has substantially become much cooler as
indicated by blue color representation. Cooling height to
NUS environment can be seen in vertical temperature
distribution as shown in Fig. 17(D).
3.3.2. TAS simulation
The calculated result of energy consumption for current
condition is presented below. The hourly average ambient
temperature from the eld measurement on 15th Septem-
ber 2005 was inputted to the weather data, as seen in
Tables 24. In the Kent Ridge Road-Water Tank the
temperature was 27.65 1C and in the PGP complex
29.76 1C. Thus, the difference was 2.11 1C.
In the rst simulation, the internal heat load was omitted
to see the impact of ambient temperature condition to the
cooling load. The simulated result is presented in Fig. 18. It
is clearly seen that the ambient temperature has a
signicant effect to cooling load of the building. Building
with the PGP ambient temperature has 14.30% higher
cooling load than the building in Kent Ridge Road-Water
Tank area.
In the second simulation, the internal heat load was
included, as shown in Fig. 19. The lowest cooling load,
15,280.88 kWh, was the building along the Kent Ridge
Road-Water Tank area and the highest, 16,110.88 kWh, is
the building in PGP area. The difference of cooling load is
about 5.4%. The cooling loads of the other locations are
within the range, follows the ambient temperature condi-
tion of each location. Transforming the PGP condition to
become similar to the Kent Ridge Road condition may be
difcult. However, by adding more trees as in the less dense
green area may cut the cooling load difference by 50%, to
become only about 2.5%.
The potential of rooftop garden application in NUS
building was also simulated. There were three types of
rooftop garden, 100% turng, 100% shrubs and 100%
trees. The R-values are 0.84, 2.216 and 1.429 m2 k/W,
respectively [13]. The simulated result is shown in Fig. 20.
Similar to the previous simulation, for the rst simula-
tion, the internal heat load was omitted to see the
performance of different rooftop greenery. In the rst
 ARTICLE IN PRESS
N.H. Wong et al. / Building and Environment 42 (2007) 29492970 2967

COMPARISON OF OVERALL COOLING LOAD WITHOUT INTERNAL LOAD-

3 DIFFERENT ROOFTOP GREENERY WITH CURRENT CONDITION
7,000.00

6,500.00
COOLING LOAD (KWh)

6,000.00

5,500.00

5,000.00

4,500.00

4,000.00

EN
K
B

S
K

E
K

D
R
LA

IC

IN
D
N

TR
R

EL

TE
PA

R
TA

PA
AT

R
EN
AL

EE
P
FI

AN
M

AR
ER

PG
AR
IC

S
R

IN

C
RT
C
ST
AT

FO

R
C

P
TE
L
W

EN
L

O
U

PG
IN

AL
O

SP
PU
O
O
AC

M
H
H
BI

IU
SC

EK

R
C
AS

O
AL

IT
M
IC

D
TE
ED

AU
M

Current Condition Applying 100% Shrubs Applying 100% Trees

Applying 100% Turfing Applying 100% Applying 100% Trees

Shrubs

Energy Savings of
Cooling Load Compared 9.05% 18.85% 20.01%
with Current Condition

Fig. 21. Comparison of overall cooling load in 10 points of different locations with different rooftop applicationswithout internal heat load.

simulation, there are two simulated cooling load value, the
cooling load for the rooms below the roof (seventh oor)
and the overall value.
Application of rooftop garden, shows the potential of
energy saving. The rooftop garden has the potential of
energy savings of 25.82%, 53.67% and 56.78%, by
applying turng, shrubs and trees, respectively.
For the overall cooling load, energy savings may become
9.08%, 18.85% and 20.01%, by applying turng, shrubs
and trees, respectively, as shown in Fig. 21. These results
show that rooftop greenery has potential to reduce the
cooling load.
In the second simulation, the internal heat gain was
included. The simulated result is shown in Figs. 22 and 23.
The cooling load for the seventh oor zone has the
reduction of 14.64% by applying turng, of 29.96% by
applying shrubs and of 31.73% by applying trees as shown
in Fig. 22.
The overall cooling load of the building with different
rooftop application is presented in Fig. 23. Energy savings
of overall cooling load may become 3.29% by applying
turng, 6.73% by applying shrubs and 7.16% by applying
trees, see gure below. These simulation results really show
that rooftop greenery has the potential for energy savings,
where additional benets may follow, such as lowering the
ambient temperature, providing a better view to the
occupants and so on.
From these simulations, between applying shrubs and
trees, the energy savings does not seem too different. Thus,
the application of shrubs on the rooftop is more reasonable
to be applied, because application of trees may have
problem with the structure of the building in carrying the
additional load.
4. Limitations and constraints
The following are the limitations and constraints in
conducting this study:
(a) The thermal satellite image data could be taken only
during the noon time due to the rotation schedule of
the satellite. Thus, the UHI study by using the thermal
image could not be conducted which needs a mid-night
thermal image.
(b) In the eld measurement, there are limited numbers of
HOBO meter for data collection in order to make the
 ARTICLE IN PRESS
2968 N.H. Wong et al. / Building and Environment 42 (2007) 29492970

COMPARISON OF COOLING LOAD ON 7th FLOOR ZONE -

3 DIFFERENT ROOFTOP GREENERY WITH CURRENT CONDITION
3,400.00

3,200.00

3,000.00
COOLING LOAD (KWh)

2,800.00

2,600.00

2,400.00

2,200.00

2,000.00

1,800.00

1,600.00

D
K

EN
D

G
K

R
N

LA

TR
R

EL

IN
TA

TE
PA

PA

P
AT

R
FI
EN
AL

PG

EE

AN
M
R

AR

AR

S
C
IC
E

RT

IN

C
AT

FO

R
ST

P
O
TE
W

EN
U

PG
IN

SP
O

AL
O

PU
O
O
AC

M
H
BI

IU
EK
SC

R
C
AS

O
AL

IT
M
IC

D
TE
ED

AU
M

Current Condition Applying 100% Shrubs Applying 100% Trees

Applying 100% Turfing Applying 100% Applying 100% Trees

Shrubs

Energy Savings of
Cooling Load Compared 14.64% 29.96% 31.73%
with Current Condition

Fig. 22. Comparison of seventh oor zone cooling load in 10 points of different locations with different rooftop applications.

collected data more representative for the representa- surrounded by greenery has lower ambient temperature
tion of distribution greenery and building across the than the one away from the greenery.
complex. In the eld measurement, it was found, on a typical day,
(c) ENVI-Met simulation on rooftop greenery to get the that the peak temperature difference between dense green
temperature difference as a result of indirect cooling area (Kent Ridge Road-Water Tank) and PGP residence
effect is not possible due to time and hardware can be as high as 4 1C at around 13:00. When the time
constraints as the simulation generally takes 10 times approaches mid-night, the temperature difference between
longer than that of without rooftop greenery. Similar these two locations is about 3 1C. The temperature
effect condition, that is introducing more greenery to difference of daytime maximum among different locations
NUS environment, is then simulated as an alternative is very signicant. It has a peak of 3.3 1C. The night-time
method. minimum temperature is generally very even, but goes up
to nearly 23 1C at PGP residence. This veries the
hypothesis that the heat accumulated in the day is hard
5. Conclusion to dissipate during night-time due to large concentration of
buildings and sparse plantation in PGP residence. This
From thermal satellite image of NUS campus, it is seen study also conrms the previous study by Wong [10] on the
that, reddish color of thermal distribution are on and impact of large greenery.
around the buildings. The greenish color appears in high In the ENVI-Met simulation, it is conrmed that,
dense area of plantation and the yellowish in between these rstly, presence of dense greenery in NUS environment is
areas. The buildings surrounded by or at the perimeter of very important in keeping low ambient temperature.
the green area have better thermal distribution than other Cooling effect produced by greenery will affect NUS
buildings away from dense green area, shown as yellowish microclimate positively as it will make NUS environment
color. It can be concluded that a building near or cooler in general. Secondly, cooling effects of greenery on
 ARTICLE IN PRESS
N.H. Wong et al. / Building and Environment 42 (2007) 29492970 2969

COMPARISON OF OVERALL COOLING LOAD -

3 DIFFERENT ROOFTOP GREENERY WITH CURRENT CONDITION

16,000.00

15,500.00
COOLING LOAD (KWh)

15,000.00

14,500.00

14,000.00

13,500.00

13,000.00

12,500.00

D
K

EN
D

G
K

R
N

LA

TR
R

EL

IN
TA

TE
PA

PA

P
AT

R
FI
EN
AL

PG

EE

AN
M
R

AR

AR

S
C
IC
E

RT

IN

C
AT

FO

R
ST

P
O
TE
W

EN
U

PG
IN

SP
O

AL

PU
O

O
O
AC

M
H
BI

IU
EK
SC

R
C
AS

O
AL

IT
M
IC

D
TE
ED

AU
M

Current Condition Applying 100% Shrubs Applying 100% Trees

Applying 100% Turfing Applying 100% Applying 100% Trees

Shrubs

Energy Savings of
Cooling Load Compared 3. 29 % 6. 73% 7. 15 %
with Current Condition

Fig. 23. Comparison of overall cooling load in 10 points of different locations with different rooftop applications.

surrounding environment is affected by the building 6. Recommendations

density. The higher the building density, the less pro-
nounced the cooling effects will be. This is more clearly In further research, a more comprehensive study can be
observed during night-time when heat stored in the done by looking into the extent of the cooling effect of the
buildings during daytime start getting released to the dense green area along Kent Ridge Road to the surround-
environment. ing area. Some weather stations will be employed and the
NUS buildings also have the potential for rooftop data compiled into a climatic mapping using GIS system.
greenery application. It is observed that about 40% NUS This will serve as a very useful reference for master
buildings are using concrete at roofs. The TAS simulation planning of future NUS campus.
shows that the cooling load for the seventh oor zone has the
potential reduction of 14.6425.82% by applying turng, of
29.9653.67% by applying shrubs and of 31.7356.78% by Acknowledgments
applying trees. The overall energy savings of cooling load
range from 3.299.08% by applying turng, 6.7318.85% by This research was supported by the Department Building
applying shrubs and 7.1620.01% by applying trees. By and Ofce of Estate and Development (OED), National
planting trees and intensive rooftop system, the energy University of Singapore. The authors would like to express
saving is much higher. This is conrmed by the previous their sincere thanks to Ms. Lina Goh for the permission of
study on the two different rooftop systems [10]. conducting all the necessary methodology. Great apprecia-
This study has shown the importance of greenery area in tion also goes to Ms. Helen Yip for providing all the
keeping the NUS microclimate comfortable. Some im- drawings as part of the computer simulation.
provements may be considered by planting more trees in
the less dense greenery and especially in the sparse green References
area. The large green area can be considered as a large park
in the urban area, which has an impact on the surrounding [1] Nichol J, Wong MS. Modelling urban environmental quality in a
environment. tropical city. Lansdcape and Urban Planning 2005;75:4958.
 ARTICLE IN PRESS
2970 N.H. Wong et al. / Building and Environment 42 (2007) 29492970

[2] Takahashi K, Yoshida H, Tanaka Y, Aotake N, Wang F.
Measurement of thermal environment in Kyoto city and its
prediction by CFD simulation. Energy and Buildings 2004;16:
7719.
[3] Dousset B, Gourmelon F. Satellite multi-sensor data analysis of
urban surfaces temperatures and landcovers. Journal of Photogram-
metry and Remote Sensing 2004;58:4354.
[4] Jauregui E. Inuence of a large urban park on temperature and
convective precipitation in tropical city. Energy and Buildings 1990/
1991;1516:45763.
[5] Kawashima S. Effect of vegetation on surface temperature in urban
and suburban areas in winter. Energy and Buildings 1990/1991;
1516:4659.
[6] Hien WN. Study on the urban heat island in Singapore. Singapore:
National University of Singapore; 2002.
[7] Ca VT, Asaeda T, Abu EM. Reductions in air conditioning energy
caused by a nearby park. Energy and Buildings 1998;29:8392.
[8] Yu C, Hien WN. Thermal benets of city parks. Energy and
Buildings 2006;38:10520.
[9] Brad B. Mitigating the urban heat island with green roof
infrastructure. In: Urban heat island summitmitigation of and
adaptation to extreme summer heat, 2002, Toronto; May 14, 2002.
[10] Hien WN, Yu C. A comparison of two rooftop systems in the tropical
climate. Second iNTA conference, Jogjakarta; 35 April 2006.
[11] Streutker DR. Satellite-measured growth of the urban heat island of
Houston, Texas. Remote Sensing of Environment 2002;85:2829.
[12] hhttp://www.geographie.ruhr-uni-bochum.de/agklima/envimet/i.
[13] Hien WN, et al. Effects of rooftop garden on energy consumption of
a commercial building in Singapore. Energy and Buildings 2003;35:
35364.